Stowage/go
2
0
Fork 0
mirror of https://github.com/golang/go.git synced 2025-10-20 03:23:18 +00:00
Code Issues Projects Releases Packages Wiki Activity
go/src/cmd/compile/internal/ssa/gen/PPC64Ops.go

519 lines
26 KiB
Go
Raw Normal View History

[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
// Copyright 2016 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// +build ignore
package main
import "strings"
[dev.ssa] cmd/compile: PPC64, load/store by type, shifts, divisions, bools Updates #16010. Change-Id: Ie520d64fd1c4f881f45623303ed0b7cbdf0e4764 Reviewed-on: https://go-review.googlesource.com/25493 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-04 11:34:43 -07:00
// Notes:
// - Less-than-64-bit integer types live in the low portion of registers.
// For now, the upper portion is junk; sign/zero-extension might be optimized in the future, but not yet.
// - Boolean types are zero or 1; stored in a byte, but loaded with AMOVBZ so the upper bytes of a register are zero.
all: fix misspellings Change-Id: I429637ca91f7db4144f17621de851a548dc1ce76 Reviewed-on: https://go-review.googlesource.com/34923 Reviewed-by: Ian Lance Taylor <iant@golang.org> Reviewed-by: Daniel Martí <mvdan@mvdan.cc> Run-TryBot: Ian Lance Taylor <iant@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2017-01-07 08:23:11 -08:00
// - *const instructions may use a constant larger than the instruction can encode.
[dev.ssa] cmd/compile: PPC64, load/store by type, shifts, divisions, bools Updates #16010. Change-Id: Ie520d64fd1c4f881f45623303ed0b7cbdf0e4764 Reviewed-on: https://go-review.googlesource.com/25493 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-04 11:34:43 -07:00
// In this case the assembler expands to multiple instructions and uses tmp
// register (R31).
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
var regNamesPPC64 = []string{
cmd/compile: use ISEL, cleanup use of zero & extensions Abandoned earlier efforts to expose zero register, but left it in numbering to decrease squirrelyness of register allocator. ISELrelOp used in code generation of bool := x relOp y. Some patterns added to better elide zero case and some sign extension. Updates: #17109 Change-Id: Ida7839f0023ca8f0ffddc0545f0ac269e65b05d9 Reviewed-on: https://go-review.googlesource.com/29380 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-09-16 15:02:47 -07:00
"R0", // REGZERO, not used, but simplifies counting in regalloc
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
"SP", // REGSP
"SB", // REGSB
"R3",
"R4",
"R5",
"R6",
"R7",
"R8",
"R9",
"R10",
"R11", // REGCTXT for closures
"R12",
"R13", // REGTLS
"R14",
"R15",
"R16",
"R17",
"R18",
"R19",
"R20",
"R21",
"R22",
"R23",
"R24",
"R25",
"R26",
"R27",
"R28",
"R29",
[dev.ssa] cmd/compile: PPC64, basic support for all calls and "miscellaneous" Added support for ClosureCall, DeferCall, InterCall (GoCall not yet tested). Added support for GetClosurePtr, IsNonNil, IsInBounds, IsSliceInBounds, NilCheck (Convert and GetG not yet tested) Still need to implement NilCheck optimizations. Fixed move boolean constant, order of operands to subtract. Updates #16010. Change-Id: Ibe0f6a6e688df4396cd77de0e9095997e4ca8ed2 Reviewed-on: https://go-review.googlesource.com/25241 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-26 09:24:18 -07:00
"g", // REGG. Using name "g" and setting Config.hasGReg makes it "just happen".
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
"R31", // REGTMP
[dev.ssa] cmd/compile: PPC64, basic support for all calls and "miscellaneous" Added support for ClosureCall, DeferCall, InterCall (GoCall not yet tested). Added support for GetClosurePtr, IsNonNil, IsInBounds, IsSliceInBounds, NilCheck (Convert and GetG not yet tested) Still need to implement NilCheck optimizations. Fixed move boolean constant, order of operands to subtract. Updates #16010. Change-Id: Ibe0f6a6e688df4396cd77de0e9095997e4ca8ed2 Reviewed-on: https://go-review.googlesource.com/25241 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-26 09:24:18 -07:00
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
"F0",
"F1",
"F2",
"F3",
"F4",
"F5",
"F6",
"F7",
"F8",
"F9",
"F10",
"F11",
"F12",
"F13",
"F14",
"F15",
"F16",
"F17",
"F18",
"F19",
"F20",
"F21",
"F22",
"F23",
"F24",
"F25",
"F26",
cmd/compile, runtime, etc: get rid of constant FP registers On ARM64, MIPS64, and PPC64, some floating point registers were reserved for constants 0, 1, 2, 0.5, etc. This CL removes them. On ARM64, they are never used. On MIPS64 and PPC64, the only use case is a multiplication-by-2 in the old backend of the compiler, which is replaced with an addition. Change-Id: I737cbf43283756e3408964fc88c567a938c57036 Reviewed-on: https://go-review.googlesource.com/28095 Run-TryBot: Cherry Zhang <cherryyz@google.com> Reviewed-by: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-08-30 14:46:25 -04:00
"F27",
"F28",
"F29",
"F30",
"F31",
[dev.ssa] cmd/compile: PPC64, basic support for all calls and "miscellaneous" Added support for ClosureCall, DeferCall, InterCall (GoCall not yet tested). Added support for GetClosurePtr, IsNonNil, IsInBounds, IsSliceInBounds, NilCheck (Convert and GetG not yet tested) Still need to implement NilCheck optimizations. Fixed move boolean constant, order of operands to subtract. Updates #16010. Change-Id: Ibe0f6a6e688df4396cd77de0e9095997e4ca8ed2 Reviewed-on: https://go-review.googlesource.com/25241 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-26 09:24:18 -07:00
// "CR0",
// "CR1",
// "CR2",
// "CR3",
// "CR4",
// "CR5",
// "CR6",
// "CR7",
[dev.ssa] cmd/compile: remove flags from regMask Reg allocator skips flag-typed values. Flag allocator uses the type and whether the op has "clobberFlags" set. Tested on AMD64, ARM, ARM64, 386. Passed 'toolstash -cmp' on AMD64. PPC64 is coded blindly. Change-Id: Ib1cc27efecef6a1bb27f7d7ed035a582660d244f Reviewed-on: https://go-review.googlesource.com/25480 Run-TryBot: Cherry Zhang <cherryyz@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2016-08-04 06:57:34 -04:00
// "CR",
[dev.ssa] cmd/compile: PPC64, basic support for all calls and "miscellaneous" Added support for ClosureCall, DeferCall, InterCall (GoCall not yet tested). Added support for GetClosurePtr, IsNonNil, IsInBounds, IsSliceInBounds, NilCheck (Convert and GetG not yet tested) Still need to implement NilCheck optimizations. Fixed move boolean constant, order of operands to subtract. Updates #16010. Change-Id: Ibe0f6a6e688df4396cd77de0e9095997e4ca8ed2 Reviewed-on: https://go-review.googlesource.com/25241 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-26 09:24:18 -07:00
// "XER",
// "LR",
// "CTR",
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
}
func init() {
// Make map from reg names to reg integers.
if len(regNamesPPC64) > 64 {
panic("too many registers")
}
num := map[string]int{}
for i, name := range regNamesPPC64 {
num[name] = i
}
buildReg := func(s string) regMask {
m := regMask(0)
for _, r := range strings.Split(s, " ") {
if n, ok := num[r]; ok {
m |= regMask(1) << uint(n)
continue
}
panic("register " + r + " not found")
}
return m
}
var (
[dev.ssa] cmd/compile: PPC64, basic support for all calls and "miscellaneous" Added support for ClosureCall, DeferCall, InterCall (GoCall not yet tested). Added support for GetClosurePtr, IsNonNil, IsInBounds, IsSliceInBounds, NilCheck (Convert and GetG not yet tested) Still need to implement NilCheck optimizations. Fixed move boolean constant, order of operands to subtract. Updates #16010. Change-Id: Ibe0f6a6e688df4396cd77de0e9095997e4ca8ed2 Reviewed-on: https://go-review.googlesource.com/25241 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-26 09:24:18 -07:00
gp = buildReg("R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29")
[dev.ssa] cmd/compile: PPC64, FP to/from int conversions. Passes ssa_test. Requires a few new instructions and some scratchpad memory to move data between G and F registers. Also fixed comparisons to be correct in case of NaN. Added missing instructions for run.bash. Removed some FP registers that are apparently "reserved" (but that are also apparently also unused except for a gratuitous multiplication by two when y = x+x would work just as well). Currently failing stack splits. Updates #16010. Change-Id: I73b161bfff54445d72bd7b813b1479f89fc72602 Reviewed-on: https://go-review.googlesource.com/26813 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-10 11:44:57 -07:00
fp = buildReg("F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15 F16 F17 F18 F19 F20 F21 F22 F23 F24 F25 F26")
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
sp = buildReg("SP")
[dev.ssa] cmd/compile: some improvements to PPC codegen Runs fibonacci for all integer types. Fold addressing arithmetic into stores. Updates #16010. Change-Id: I257982c82c00c80b00679757c3da345045968022 Reviewed-on: https://go-review.googlesource.com/25103 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com>
2016-07-06 13:32:52 -07:00
sb = buildReg("SB")
cmd/compile: ppc64le working, not optimized enough This time with the cherry-pick from the proper patch of the old CL. Stack size increased. Corrected NaN-comparison glitches. Marked g register as clobbered by calls. Fixed shared libraries. live_ssa.go still disabled because of differences. Presumably turning on more optimization will fix both the stack size and the live_ssa.go glitches. Enhanced debugging output for shared libs test. Rebased onto master. Updates #16010. Change-Id: I40864faf1ef32c118fb141b7ef8e854498e6b2c4 Reviewed-on: https://go-review.googlesource.com/27159 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-15 13:51:00 -07:00
gr = buildReg("g")
[dev.ssa] cmd/compile: PPC64, FP to/from int conversions. Passes ssa_test. Requires a few new instructions and some scratchpad memory to move data between G and F registers. Also fixed comparisons to be correct in case of NaN. Added missing instructions for run.bash. Removed some FP registers that are apparently "reserved" (but that are also apparently also unused except for a gratuitous multiplication by two when y = x+x would work just as well). Currently failing stack splits. Updates #16010. Change-Id: I73b161bfff54445d72bd7b813b1479f89fc72602 Reviewed-on: https://go-review.googlesource.com/26813 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-10 11:44:57 -07:00
// cr = buildReg("CR")
// ctr = buildReg("CTR")
// lr = buildReg("LR")
cmd/compile: request r12 for indirect calls on ppc64le On ppc64le, functions compiled with -shared expect r12 to hold the function's address for indirect calls. Previously this was enforced by generating a move instruction if the address wasn't already in r12. This change avoids that extra move by requesting r12 in the CALL ops that do indirect calls. As a result of adding support for plugins on ppc64le, it was discovered that there would be more cases where this extra move was needed, so this seemed like a better solution. Updates #20756 Change-Id: I6770885a46990f78c6d2902a715dcdaa822192a1 Reviewed-on: https://go-review.googlesource.com/62890 Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Crawshaw <crawshaw@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2017-09-11 10:39:24 -04:00
tmp = buildReg("R31")
ctxt = buildReg("R11")
callptr = buildReg("R12")
[dev.ssa] cmd/compile: PPC64, FP to/from int conversions. Passes ssa_test. Requires a few new instructions and some scratchpad memory to move data between G and F registers. Also fixed comparisons to be correct in case of NaN. Added missing instructions for run.bash. Removed some FP registers that are apparently "reserved" (but that are also apparently also unused except for a gratuitous multiplication by two when y = x+x would work just as well). Currently failing stack splits. Updates #16010. Change-Id: I73b161bfff54445d72bd7b813b1479f89fc72602 Reviewed-on: https://go-review.googlesource.com/26813 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-10 11:44:57 -07:00
// tls = buildReg("R13")
[dev.ssa] cmd/compile: remove flags from regMask Reg allocator skips flag-typed values. Flag allocator uses the type and whether the op has "clobberFlags" set. Tested on AMD64, ARM, ARM64, 386. Passed 'toolstash -cmp' on AMD64. PPC64 is coded blindly. Change-Id: Ib1cc27efecef6a1bb27f7d7ed035a582660d244f Reviewed-on: https://go-review.googlesource.com/25480 Run-TryBot: Cherry Zhang <cherryyz@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2016-08-04 06:57:34 -04:00
gp01 = regInfo{inputs: nil, outputs: []regMask{gp}}
[dev.ssa] cmd/compile: replace storeconst w/ storezero, fold addressing Because PPC lacks store-immediate, remove the instruction that implies that it exists. Replace it with storezero for the special case of storing zero, because R0 is reserved zero for Go (though the assembler knows this, do it in SSA). Also added address folding for storezero. (Now corrected to use right-sized stores in bulk-zero code.) Hello.go now compiles to genssa main 00000 (...hello.go:7) TEXT "".main(SB), $0 00001 (...hello.go:7) FUNCDATA $0, "".gcargs·0(SB) 00002 (...hello.go:7) FUNCDATA $1, "".gclocals·1(SB) v23 00003 (...hello.go:8) MOVD $go.string."Hello, World!\n"(SB), R3 v11 00004 (...hello.go:8) MOVD R3, 32(R1) v22 00005 (...hello.go:8) MOVD $14, R3 v6 00006 (...hello.go:8) MOVD R3, 40(R1) v20 00007 (...hello.go:8) MOVD R0, 48(R1) v18 00008 (...hello.go:8) MOVD R0, 56(R1) v9 00009 (...hello.go:8) MOVD R0, 64(R1) v10 00010 (...hello.go:8) CALL fmt.Printf(SB) b2 00011 (...hello.go:9) RET 00012 (<unknown line number>) END Updates #16010 Change-Id: I33cfd98c21a1617502260ac753fa8cad68c8d85a Reviewed-on: https://go-review.googlesource.com/25151 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-22 10:47:16 -07:00
gp11 = regInfo{inputs: []regMask{gp | sp | sb}, outputs: []regMask{gp}}
gp21 = regInfo{inputs: []regMask{gp | sp | sb, gp | sp | sb}, outputs: []regMask{gp}}
[dev.ssa] cmd/compile: remove flags from regMask Reg allocator skips flag-typed values. Flag allocator uses the type and whether the op has "clobberFlags" set. Tested on AMD64, ARM, ARM64, 386. Passed 'toolstash -cmp' on AMD64. PPC64 is coded blindly. Change-Id: Ib1cc27efecef6a1bb27f7d7ed035a582660d244f Reviewed-on: https://go-review.googlesource.com/25480 Run-TryBot: Cherry Zhang <cherryyz@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2016-08-04 06:57:34 -04:00
gp1cr = regInfo{inputs: []regMask{gp | sp | sb}}
gp2cr = regInfo{inputs: []regMask{gp | sp | sb, gp | sp | sb}}
crgp = regInfo{inputs: nil, outputs: []regMask{gp}}
[dev.ssa] cmd/compile: replace storeconst w/ storezero, fold addressing Because PPC lacks store-immediate, remove the instruction that implies that it exists. Replace it with storezero for the special case of storing zero, because R0 is reserved zero for Go (though the assembler knows this, do it in SSA). Also added address folding for storezero. (Now corrected to use right-sized stores in bulk-zero code.) Hello.go now compiles to genssa main 00000 (...hello.go:7) TEXT "".main(SB), $0 00001 (...hello.go:7) FUNCDATA $0, "".gcargs·0(SB) 00002 (...hello.go:7) FUNCDATA $1, "".gclocals·1(SB) v23 00003 (...hello.go:8) MOVD $go.string."Hello, World!\n"(SB), R3 v11 00004 (...hello.go:8) MOVD R3, 32(R1) v22 00005 (...hello.go:8) MOVD $14, R3 v6 00006 (...hello.go:8) MOVD R3, 40(R1) v20 00007 (...hello.go:8) MOVD R0, 48(R1) v18 00008 (...hello.go:8) MOVD R0, 56(R1) v9 00009 (...hello.go:8) MOVD R0, 64(R1) v10 00010 (...hello.go:8) CALL fmt.Printf(SB) b2 00011 (...hello.go:9) RET 00012 (<unknown line number>) END Updates #16010 Change-Id: I33cfd98c21a1617502260ac753fa8cad68c8d85a Reviewed-on: https://go-review.googlesource.com/25151 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-22 10:47:16 -07:00
gpload = regInfo{inputs: []regMask{gp | sp | sb}, outputs: []regMask{gp}}
[dev.ssa] cmd/compile: remove flags from regMask Reg allocator skips flag-typed values. Flag allocator uses the type and whether the op has "clobberFlags" set. Tested on AMD64, ARM, ARM64, 386. Passed 'toolstash -cmp' on AMD64. PPC64 is coded blindly. Change-Id: Ib1cc27efecef6a1bb27f7d7ed035a582660d244f Reviewed-on: https://go-review.googlesource.com/25480 Run-TryBot: Cherry Zhang <cherryyz@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2016-08-04 06:57:34 -04:00
gpstore = regInfo{inputs: []regMask{gp | sp | sb, gp | sp | sb}}
gpstorezero = regInfo{inputs: []regMask{gp | sp | sb}} // ppc64.REGZERO is reserved zero value
cmd/compile: intrinsify atomics on ppc64x This adds the necessary changes so that atomics are treated as intrinsics on ppc64x. The implementations of And8 and Or8 require power8 for both ppc64 and ppc64le. This is a new requirement for ppc64. Fixes #8739 Change-Id: Icb85e2755a49166ee3652668279f6ed5ebbca901 Reviewed-on: https://go-review.googlesource.com/36832 Reviewed-by: Keith Randall <khr@golang.org>
2017-02-10 17:45:12 -05:00
gpxchg = regInfo{inputs: []regMask{gp | sp | sb, gp}, outputs: []regMask{gp}}
gpcas = regInfo{inputs: []regMask{gp | sp | sb, gp, gp}, outputs: []regMask{gp}}
[dev.ssa] cmd/compile: remove flags from regMask Reg allocator skips flag-typed values. Flag allocator uses the type and whether the op has "clobberFlags" set. Tested on AMD64, ARM, ARM64, 386. Passed 'toolstash -cmp' on AMD64. PPC64 is coded blindly. Change-Id: Ib1cc27efecef6a1bb27f7d7ed035a582660d244f Reviewed-on: https://go-review.googlesource.com/25480 Run-TryBot: Cherry Zhang <cherryyz@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2016-08-04 06:57:34 -04:00
fp01 = regInfo{inputs: nil, outputs: []regMask{fp}}
[dev.ssa] cmd/compile: PPC: FP load/store/const/cmp/neg; div/mod FP<->int conversions remain. Updates #16010. Change-Id: I38d7a4923e34d0a489935fffc4c96c020cafdba2 Reviewed-on: https://go-review.googlesource.com/25589 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
2016-08-08 14:19:56 -07:00
fp11 = regInfo{inputs: []regMask{fp}, outputs: []regMask{fp}}
[dev.ssa] cmd/compile: PPC64, FP to/from int conversions. Passes ssa_test. Requires a few new instructions and some scratchpad memory to move data between G and F registers. Also fixed comparisons to be correct in case of NaN. Added missing instructions for run.bash. Removed some FP registers that are apparently "reserved" (but that are also apparently also unused except for a gratuitous multiplication by two when y = x+x would work just as well). Currently failing stack splits. Updates #16010. Change-Id: I73b161bfff54445d72bd7b813b1479f89fc72602 Reviewed-on: https://go-review.googlesource.com/26813 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-10 11:44:57 -07:00
fpgp = regInfo{inputs: []regMask{fp}, outputs: []regMask{gp}}
gpfp = regInfo{inputs: []regMask{gp}, outputs: []regMask{fp}}
[dev.ssa] cmd/compile: PPC: FP load/store/const/cmp/neg; div/mod FP<->int conversions remain. Updates #16010. Change-Id: I38d7a4923e34d0a489935fffc4c96c020cafdba2 Reviewed-on: https://go-review.googlesource.com/25589 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
2016-08-08 14:19:56 -07:00
fp21 = regInfo{inputs: []regMask{fp, fp}, outputs: []regMask{fp}}
cmd/compile: emit fused multiply-{add,subtract} on ppc64x A follow on to CL 36963 adding support for ppc64x. Performance changes (as posted on the issue): poly1305: benchmark old ns/op new ns/op delta Benchmark64-16 172 151 -12.21% Benchmark1K-16 1828 1523 -16.68% Benchmark64Unaligned-16 172 151 -12.21% Benchmark1KUnaligned-16 1827 1523 -16.64% math: BenchmarkAcos-16 43.9 39.9 -9.11% BenchmarkAcosh-16 57.0 45.8 -19.65% BenchmarkAsin-16 35.8 33.0 -7.82% BenchmarkAsinh-16 68.6 60.8 -11.37% BenchmarkAtan-16 19.8 16.2 -18.18% BenchmarkAtanh-16 65.5 57.5 -12.21% BenchmarkAtan2-16 45.4 34.2 -24.67% BenchmarkGamma-16 37.6 26.0 -30.85% BenchmarkLgamma-16 40.0 28.2 -29.50% BenchmarkLog1p-16 35.1 29.1 -17.09% BenchmarkSin-16 22.7 18.4 -18.94% BenchmarkSincos-16 31.7 23.7 -25.24% BenchmarkSinh-16 146 131 -10.27% BenchmarkY0-16 130 107 -17.69% BenchmarkY1-16 127 107 -15.75% BenchmarkYn-16 278 235 -15.47% Updates #17895. Change-Id: I1c16199715d20c9c4bd97c4a950bcfa69eb688c1 Reviewed-on: https://go-review.googlesource.com/38095 Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com> Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
2017-03-13 14:39:17 -04:00
fp31 = regInfo{inputs: []regMask{fp, fp, fp}, outputs: []regMask{fp}}
[dev.ssa] cmd/compile: PPC: FP load/store/const/cmp/neg; div/mod FP<->int conversions remain. Updates #16010. Change-Id: I38d7a4923e34d0a489935fffc4c96c020cafdba2 Reviewed-on: https://go-review.googlesource.com/25589 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
2016-08-08 14:19:56 -07:00
fp2cr = regInfo{inputs: []regMask{fp, fp}}
fpload = regInfo{inputs: []regMask{gp | sp | sb}, outputs: []regMask{fp}}
fpstore = regInfo{inputs: []regMask{gp | sp | sb, fp}}
cmd/compile: ppc64le working, not optimized enough This time with the cherry-pick from the proper patch of the old CL. Stack size increased. Corrected NaN-comparison glitches. Marked g register as clobbered by calls. Fixed shared libraries. live_ssa.go still disabled because of differences. Presumably turning on more optimization will fix both the stack size and the live_ssa.go glitches. Enhanced debugging output for shared libs test. Rebased onto master. Updates #16010. Change-Id: I40864faf1ef32c118fb141b7ef8e854498e6b2c4 Reviewed-on: https://go-review.googlesource.com/27159 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-15 13:51:00 -07:00
callerSave = regMask(gp | fp | gr)
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
)
ops := []opData{
cmd/compile/internal/ssa: populate SymEffects for SSA Ops Changes to ${GOARCH}Ops.go files were mechanically produced using github.com/mdempsky/ssa-symops, a one-off tool that inserts "SymEffect: X" elements by pattern matching against the Op names. Change-Id: Ibf3e481ffd588647f2a31662d72114b740ccbfcf Reviewed-on: https://go-review.googlesource.com/38084 Run-TryBot: Matthew Dempsky <mdempsky@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
2017-03-09 14:46:43 -08:00
{name: "ADD", argLength: 2, reg: gp21, asm: "ADD", commutative: true}, // arg0 + arg1
{name: "ADDconst", argLength: 1, reg: gp11, asm: "ADD", aux: "SymOff", symEffect: "Addr"}, // arg0 + auxInt + aux.(*gc.Sym)
{name: "FADD", argLength: 2, reg: fp21, asm: "FADD", commutative: true}, // arg0+arg1
{name: "FADDS", argLength: 2, reg: fp21, asm: "FADDS", commutative: true}, // arg0+arg1
{name: "SUB", argLength: 2, reg: gp21, asm: "SUB"}, // arg0-arg1
{name: "FSUB", argLength: 2, reg: fp21, asm: "FSUB"}, // arg0-arg1
{name: "FSUBS", argLength: 2, reg: fp21, asm: "FSUBS"}, // arg0-arg1
[dev.ssa] cmd/compile: PPC64, add cmp->bool, some shifts, hmul Includes hmul (all widths) compare for boolean result and simplifications shift operations plus changes/additions for implementation (ORN, ADDME, ADDC) Also fixed a backwards-operand CMP. Change-Id: Id723c4e25125c38e0d9ab9ec9448176b75f4cdb4 Reviewed-on: https://go-review.googlesource.com/25410 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-02 13:17:09 -07:00
[dev.ssa] cmd/compile: PPC: FP load/store/const/cmp/neg; div/mod FP<->int conversions remain. Updates #16010. Change-Id: I38d7a4923e34d0a489935fffc4c96c020cafdba2 Reviewed-on: https://go-review.googlesource.com/25589 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
2016-08-08 14:19:56 -07:00
{name: "MULLD", argLength: 2, reg: gp21, asm: "MULLD", typ: "Int64", commutative: true}, // arg0*arg1 (signed 64-bit)
{name: "MULLW", argLength: 2, reg: gp21, asm: "MULLW", typ: "Int32", commutative: true}, // arg0*arg1 (signed 32-bit)
[dev.ssa] cmd/compile: PPC64, add cmp->bool, some shifts, hmul Includes hmul (all widths) compare for boolean result and simplifications shift operations plus changes/additions for implementation (ORN, ADDME, ADDC) Also fixed a backwards-operand CMP. Change-Id: Id723c4e25125c38e0d9ab9ec9448176b75f4cdb4 Reviewed-on: https://go-review.googlesource.com/25410 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-02 13:17:09 -07:00
{name: "MULHD", argLength: 2, reg: gp21, asm: "MULHD", commutative: true}, // (arg0 * arg1) >> 64, signed
{name: "MULHW", argLength: 2, reg: gp21, asm: "MULHW", commutative: true}, // (arg0 * arg1) >> 32, signed
{name: "MULHDU", argLength: 2, reg: gp21, asm: "MULHDU", commutative: true}, // (arg0 * arg1) >> 64, unsigned
{name: "MULHWU", argLength: 2, reg: gp21, asm: "MULHWU", commutative: true}, // (arg0 * arg1) >> 32, unsigned
[dev.ssa] cmd/compile: PPC64, basic support for all calls and "miscellaneous" Added support for ClosureCall, DeferCall, InterCall (GoCall not yet tested). Added support for GetClosurePtr, IsNonNil, IsInBounds, IsSliceInBounds, NilCheck (Convert and GetG not yet tested) Still need to implement NilCheck optimizations. Fixed move boolean constant, order of operands to subtract. Updates #16010. Change-Id: Ibe0f6a6e688df4396cd77de0e9095997e4ca8ed2 Reviewed-on: https://go-review.googlesource.com/25241 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-26 09:24:18 -07:00
{name: "FMUL", argLength: 2, reg: fp21, asm: "FMUL", commutative: true}, // arg0*arg1
{name: "FMULS", argLength: 2, reg: fp21, asm: "FMULS", commutative: true}, // arg0*arg1
[dev.ssa] cmd/compile: PPC64, add cmp->bool, some shifts, hmul Includes hmul (all widths) compare for boolean result and simplifications shift operations plus changes/additions for implementation (ORN, ADDME, ADDC) Also fixed a backwards-operand CMP. Change-Id: Id723c4e25125c38e0d9ab9ec9448176b75f4cdb4 Reviewed-on: https://go-review.googlesource.com/25410 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-02 13:17:09 -07:00
cmd/compile: emit fused multiply-{add,subtract} on ppc64x A follow on to CL 36963 adding support for ppc64x. Performance changes (as posted on the issue): poly1305: benchmark old ns/op new ns/op delta Benchmark64-16 172 151 -12.21% Benchmark1K-16 1828 1523 -16.68% Benchmark64Unaligned-16 172 151 -12.21% Benchmark1KUnaligned-16 1827 1523 -16.64% math: BenchmarkAcos-16 43.9 39.9 -9.11% BenchmarkAcosh-16 57.0 45.8 -19.65% BenchmarkAsin-16 35.8 33.0 -7.82% BenchmarkAsinh-16 68.6 60.8 -11.37% BenchmarkAtan-16 19.8 16.2 -18.18% BenchmarkAtanh-16 65.5 57.5 -12.21% BenchmarkAtan2-16 45.4 34.2 -24.67% BenchmarkGamma-16 37.6 26.0 -30.85% BenchmarkLgamma-16 40.0 28.2 -29.50% BenchmarkLog1p-16 35.1 29.1 -17.09% BenchmarkSin-16 22.7 18.4 -18.94% BenchmarkSincos-16 31.7 23.7 -25.24% BenchmarkSinh-16 146 131 -10.27% BenchmarkY0-16 130 107 -17.69% BenchmarkY1-16 127 107 -15.75% BenchmarkYn-16 278 235 -15.47% Updates #17895. Change-Id: I1c16199715d20c9c4bd97c4a950bcfa69eb688c1 Reviewed-on: https://go-review.googlesource.com/38095 Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com> Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
2017-03-13 14:39:17 -04:00
{name: "FMADD", argLength: 3, reg: fp31, asm: "FMADD"}, // arg0*arg1 + arg2
{name: "FMADDS", argLength: 3, reg: fp31, asm: "FMADDS"}, // arg0*arg1 + arg2
{name: "FMSUB", argLength: 3, reg: fp31, asm: "FMSUB"}, // arg0*arg1 - arg2
{name: "FMSUBS", argLength: 3, reg: fp31, asm: "FMSUBS"}, // arg0*arg1 - arg2
[dev.ssa] cmd/compile: PPC64, add cmp->bool, some shifts, hmul Includes hmul (all widths) compare for boolean result and simplifications shift operations plus changes/additions for implementation (ORN, ADDME, ADDC) Also fixed a backwards-operand CMP. Change-Id: Id723c4e25125c38e0d9ab9ec9448176b75f4cdb4 Reviewed-on: https://go-review.googlesource.com/25410 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-02 13:17:09 -07:00
{name: "SRAD", argLength: 2, reg: gp21, asm: "SRAD"}, // arg0 >>a arg1, 64 bits (all sign if arg1 & 64 != 0)
{name: "SRAW", argLength: 2, reg: gp21, asm: "SRAW"}, // arg0 >>a arg1, 32 bits (all sign if arg1 & 32 != 0)
{name: "SRD", argLength: 2, reg: gp21, asm: "SRD"}, // arg0 >> arg1, 64 bits (0 if arg1 & 64 != 0)
{name: "SRW", argLength: 2, reg: gp21, asm: "SRW"}, // arg0 >> arg1, 32 bits (0 if arg1 & 32 != 0)
{name: "SLD", argLength: 2, reg: gp21, asm: "SLD"}, // arg0 << arg1, 64 bits (0 if arg1 & 64 != 0)
{name: "SLW", argLength: 2, reg: gp21, asm: "SLW"}, // arg0 << arg1, 32 bits (0 if arg1 & 32 != 0)
cmd/compile, math/bits: add rotate rules to PPC64.rules This adds rules to match the code in math/bits RotateLeft, RotateLeft32, and RotateLef64 to allow them to be inlined. The rules are complicated because the code in these function use different types, and the non-const version of these shifts generate Mask and Carry instructions that become subexpressions during the match process. Also adds a testcase to asm_test.go. Improvement in math/bits: BenchmarkRotateLeft-16 1.57 1.32 -15.92% BenchmarkRotateLeft32-16 1.60 1.37 -14.37% BenchmarkRotateLeft64-16 1.57 1.32 -15.92% Updates #21390 Change-Id: Ib6f17669ecc9cab54f18d690be27e2225ca654a4 Reviewed-on: https://go-review.googlesource.com/59932 Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2017-08-29 11:49:08 -04:00
{name: "ROTL", argLength: 2, reg: gp21, asm: "ROTL"}, // arg0 rotate left by arg1 mod 64
{name: "ROTLW", argLength: 2, reg: gp21, asm: "ROTLW"}, // uint32(arg0) rotate left by arg1 mod 32
[dev.ssa] cmd/compile: remove flags from regMask Reg allocator skips flag-typed values. Flag allocator uses the type and whether the op has "clobberFlags" set. Tested on AMD64, ARM, ARM64, 386. Passed 'toolstash -cmp' on AMD64. PPC64 is coded blindly. Change-Id: Ib1cc27efecef6a1bb27f7d7ed035a582660d244f Reviewed-on: https://go-review.googlesource.com/25480 Run-TryBot: Cherry Zhang <cherryyz@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2016-08-04 06:57:34 -04:00
{name: "ADDconstForCarry", argLength: 1, reg: regInfo{inputs: []regMask{gp | sp | sb}, clobbers: tmp}, aux: "Int16", asm: "ADDC", typ: "Flags"}, // _, carry := arg0 + aux
{name: "MaskIfNotCarry", argLength: 1, reg: crgp, asm: "ADDME", typ: "Int64"}, // carry - 1 (if carry then 0 else -1)
[dev.ssa] cmd/compile: PPC64, add cmp->bool, some shifts, hmul Includes hmul (all widths) compare for boolean result and simplifications shift operations plus changes/additions for implementation (ORN, ADDME, ADDC) Also fixed a backwards-operand CMP. Change-Id: Id723c4e25125c38e0d9ab9ec9448176b75f4cdb4 Reviewed-on: https://go-review.googlesource.com/25410 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-02 13:17:09 -07:00
{name: "SRADconst", argLength: 1, reg: gp11, asm: "SRAD", aux: "Int64"}, // arg0 >>a aux, 64 bits
{name: "SRAWconst", argLength: 1, reg: gp11, asm: "SRAW", aux: "Int64"}, // arg0 >>a aux, 32 bits
{name: "SRDconst", argLength: 1, reg: gp11, asm: "SRD", aux: "Int64"}, // arg0 >> aux, 64 bits
{name: "SRWconst", argLength: 1, reg: gp11, asm: "SRW", aux: "Int64"}, // arg0 >> aux, 32 bits
{name: "SLDconst", argLength: 1, reg: gp11, asm: "SLD", aux: "Int64"}, // arg0 << aux, 64 bits
{name: "SLWconst", argLength: 1, reg: gp11, asm: "SLW", aux: "Int64"}, // arg0 << aux, 32 bits
cmd/compile: add rotates to PPC64.rules This updates PPC64.rules to include rules to generate rotates for ADD, OR, XOR operators that combine two opposite shifts that sum to 32 or 64. To support this change opcodes for ROTL and ROTLW were added to be used like the rotldi and rotlwi extended mnemonics. This provides the following improvement in sha3: BenchmarkPermutationFunction-8 302.83 376.40 1.24x BenchmarkSha3_512_MTU-8 98.64 121.92 1.24x BenchmarkSha3_384_MTU-8 136.80 168.30 1.23x BenchmarkSha3_256_MTU-8 169.21 211.29 1.25x BenchmarkSha3_224_MTU-8 179.76 221.19 1.23x BenchmarkShake128_MTU-8 212.87 263.23 1.24x BenchmarkShake256_MTU-8 196.62 245.60 1.25x BenchmarkShake256_16x-8 163.57 194.37 1.19x BenchmarkShake256_1MiB-8 199.02 248.74 1.25x BenchmarkSha3_512_1MiB-8 106.55 133.13 1.25x Fixes #20030 Change-Id: I484c56f48395d32f53ff3ecb3ac6cb8191cfee44 Reviewed-on: https://go-review.googlesource.com/40992 Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com> Reviewed-by: Michael Munday <munday@ca.ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2017-04-18 17:05:31 -04:00
{name: "ROTLconst", argLength: 1, reg: gp11, asm: "ROTL", aux: "Int64"}, // arg0 rotate left by auxInt bits
{name: "ROTLWconst", argLength: 1, reg: gp11, asm: "ROTLW", aux: "Int64"}, // uint32(arg0) rotate left by auxInt bits
cmd/compile: ppc64x intrinsics for math/bits This adds math/bits intrinsics for OnesCount, Len, TrailingZeros on ppc64x. benchmark old ns/op new ns/op delta BenchmarkLeadingZeros-16 4.26 1.71 -59.86% BenchmarkLeadingZeros16-16 3.04 1.83 -39.80% BenchmarkLeadingZeros32-16 3.31 1.82 -45.02% BenchmarkLeadingZeros64-16 3.69 1.71 -53.66% BenchmarkTrailingZeros-16 2.55 1.62 -36.47% BenchmarkTrailingZeros32-16 2.55 1.77 -30.59% BenchmarkTrailingZeros64-16 2.78 1.62 -41.73% BenchmarkOnesCount-16 3.19 0.93 -70.85% BenchmarkOnesCount32-16 2.55 1.18 -53.73% BenchmarkOnesCount64-16 3.22 0.93 -71.12% Update #18616 I also made a change to bits_test.go because when debugging some failures the output was not quite providing the right argument information. Change-Id: Ia58d31d1777cf4582a4505f85b11a1202ca07d3e Reviewed-on: https://go-review.googlesource.com/41630 Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com> Reviewed-by: Keith Randall <khr@golang.org>
2017-04-24 15:11:39 -04:00
{name: "CNTLZD", argLength: 1, reg: gp11, asm: "CNTLZD", clobberFlags: true}, // count leading zeros
{name: "CNTLZW", argLength: 1, reg: gp11, asm: "CNTLZW", clobberFlags: true}, // count leading zeros (32 bit)
{name: "POPCNTD", argLength: 1, reg: gp11, asm: "POPCNTD"}, // number of set bits in arg0
{name: "POPCNTW", argLength: 1, reg: gp11, asm: "POPCNTW"}, // number of set bits in each word of arg0 placed in corresponding word
{name: "POPCNTB", argLength: 1, reg: gp11, asm: "POPCNTB"}, // number of set bits in each byte of arg0 placed in corresonding byte
[dev.ssa] cmd/compile: PPC64, add cmp->bool, some shifts, hmul Includes hmul (all widths) compare for boolean result and simplifications shift operations plus changes/additions for implementation (ORN, ADDME, ADDC) Also fixed a backwards-operand CMP. Change-Id: Id723c4e25125c38e0d9ab9ec9448176b75f4cdb4 Reviewed-on: https://go-review.googlesource.com/25410 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-02 13:17:09 -07:00
{name: "FDIV", argLength: 2, reg: fp21, asm: "FDIV"}, // arg0/arg1
{name: "FDIVS", argLength: 2, reg: fp21, asm: "FDIVS"}, // arg0/arg1
[dev.ssa] cmd/compile: PPC: FP load/store/const/cmp/neg; div/mod FP<->int conversions remain. Updates #16010. Change-Id: I38d7a4923e34d0a489935fffc4c96c020cafdba2 Reviewed-on: https://go-review.googlesource.com/25589 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
2016-08-08 14:19:56 -07:00
{name: "DIVD", argLength: 2, reg: gp21, asm: "DIVD", typ: "Int64"}, // arg0/arg1 (signed 64-bit)
{name: "DIVW", argLength: 2, reg: gp21, asm: "DIVW", typ: "Int32"}, // arg0/arg1 (signed 32-bit)
{name: "DIVDU", argLength: 2, reg: gp21, asm: "DIVDU", typ: "Int64"}, // arg0/arg1 (unsigned 64-bit)
{name: "DIVWU", argLength: 2, reg: gp21, asm: "DIVWU", typ: "Int32"}, // arg0/arg1 (unsigned 32-bit)
// MOD is implemented as rem := arg0 - (arg0/arg1) * arg1
[dev.ssa] cmd/compile: PPC64, load/store by type, shifts, divisions, bools Updates #16010. Change-Id: Ie520d64fd1c4f881f45623303ed0b7cbdf0e4764 Reviewed-on: https://go-review.googlesource.com/25493 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-04 11:34:43 -07:00
[dev.ssa] cmd/compile: PPC64, FP to/from int conversions. Passes ssa_test. Requires a few new instructions and some scratchpad memory to move data between G and F registers. Also fixed comparisons to be correct in case of NaN. Added missing instructions for run.bash. Removed some FP registers that are apparently "reserved" (but that are also apparently also unused except for a gratuitous multiplication by two when y = x+x would work just as well). Currently failing stack splits. Updates #16010. Change-Id: I73b161bfff54445d72bd7b813b1479f89fc72602 Reviewed-on: https://go-review.googlesource.com/26813 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-10 11:44:57 -07:00
// Conversions are all float-to-float register operations. "Integer" refers to encoding in the FP register.
{name: "FCTIDZ", argLength: 1, reg: fp11, asm: "FCTIDZ", typ: "Float64"}, // convert float to 64-bit int round towards zero
{name: "FCTIWZ", argLength: 1, reg: fp11, asm: "FCTIWZ", typ: "Float64"}, // convert float to 32-bit int round towards zero
{name: "FCFID", argLength: 1, reg: fp11, asm: "FCFID", typ: "Float64"}, // convert 64-bit integer to float
cmd/compile,math: improve int<->float conversions on ppc64x The functions Float64bits and Float64frombits perform poorly on ppc64x because the int<->float conversions often result in load and store sequences to handle the type change. This patch adds more rules to recognize those sequences and use register to register moves and avoid unnecessary loads and stores where possible. There were some existing rules to improve these conversions, but this provides additional improvements. Included here: - New instruction FCFIDS to improve on conversion to 32 bit - Rename Xf2i64 and Xi2f64 as MTVSRD, MFVSRD, to match the asm - Add rules to lower some of the load/store sequences for - Added new go asm to ppc64.s testcase. conversions Improvements: BenchmarkAbs-16 2.16 0.93 -56.94% BenchmarkCopysign-16 2.66 1.18 -55.64% BenchmarkRound-16 4.82 2.69 -44.19% BenchmarkSignbit-16 1.71 1.14 -33.33% BenchmarkFrexp-16 11.4 7.94 -30.35% BenchmarkLogb-16 10.4 7.34 -29.42% BenchmarkLdexp-16 15.7 11.2 -28.66% BenchmarkIlogb-16 10.2 7.32 -28.24% BenchmarkPowInt-16 69.6 55.9 -19.68% BenchmarkModf-16 10.1 8.19 -18.91% BenchmarkLog2-16 17.4 14.3 -17.82% BenchmarkCbrt-16 45.0 37.3 -17.11% BenchmarkAtanh-16 57.6 48.3 -16.15% BenchmarkRemainder-16 76.6 65.4 -14.62% BenchmarkGamma-16 26.0 22.5 -13.46% BenchmarkPowFrac-16 197 174 -11.68% BenchmarkMod-16 112 99.8 -10.89% BenchmarkAsinh-16 59.9 53.7 -10.35% BenchmarkAcosh-16 44.8 40.3 -10.04% Updates #21390 Change-Id: I56cc991fc2e55249d69518d4e1ba76cc23904e35 Reviewed-on: https://go-review.googlesource.com/63290 Reviewed-by: Michael Munday <mike.munday@ibm.com>
2017-09-12 12:33:21 -04:00
{name: "FCFIDS", argLength: 1, reg: fp11, asm: "FCFIDS", typ: "Float32"}, // convert 32-bit integer to float
[dev.ssa] cmd/compile: PPC64, FP to/from int conversions. Passes ssa_test. Requires a few new instructions and some scratchpad memory to move data between G and F registers. Also fixed comparisons to be correct in case of NaN. Added missing instructions for run.bash. Removed some FP registers that are apparently "reserved" (but that are also apparently also unused except for a gratuitous multiplication by two when y = x+x would work just as well). Currently failing stack splits. Updates #16010. Change-Id: I73b161bfff54445d72bd7b813b1479f89fc72602 Reviewed-on: https://go-review.googlesource.com/26813 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-10 11:44:57 -07:00
{name: "FRSP", argLength: 1, reg: fp11, asm: "FRSP", typ: "Float64"}, // round float to 32-bit value
// Movement between float and integer registers with no change in bits; accomplished with stores+loads on PPC.
// Because the 32-bit load-literal-bits instructions have impoverished addressability, always widen the
// data instead and use FMOVDload and FMOVDstore instead (this will also dodge endianess issues).
// There are optimizations that should apply -- (Xi2f64 (MOVWload (not-ADD-ptr+offset) ) ) could use
// the word-load instructions. (Xi2f64 (MOVDload ptr )) can be (FMOVDload ptr)
cmd/compile,math: improve int<->float conversions on ppc64x The functions Float64bits and Float64frombits perform poorly on ppc64x because the int<->float conversions often result in load and store sequences to handle the type change. This patch adds more rules to recognize those sequences and use register to register moves and avoid unnecessary loads and stores where possible. There were some existing rules to improve these conversions, but this provides additional improvements. Included here: - New instruction FCFIDS to improve on conversion to 32 bit - Rename Xf2i64 and Xi2f64 as MTVSRD, MFVSRD, to match the asm - Add rules to lower some of the load/store sequences for - Added new go asm to ppc64.s testcase. conversions Improvements: BenchmarkAbs-16 2.16 0.93 -56.94% BenchmarkCopysign-16 2.66 1.18 -55.64% BenchmarkRound-16 4.82 2.69 -44.19% BenchmarkSignbit-16 1.71 1.14 -33.33% BenchmarkFrexp-16 11.4 7.94 -30.35% BenchmarkLogb-16 10.4 7.34 -29.42% BenchmarkLdexp-16 15.7 11.2 -28.66% BenchmarkIlogb-16 10.2 7.32 -28.24% BenchmarkPowInt-16 69.6 55.9 -19.68% BenchmarkModf-16 10.1 8.19 -18.91% BenchmarkLog2-16 17.4 14.3 -17.82% BenchmarkCbrt-16 45.0 37.3 -17.11% BenchmarkAtanh-16 57.6 48.3 -16.15% BenchmarkRemainder-16 76.6 65.4 -14.62% BenchmarkGamma-16 26.0 22.5 -13.46% BenchmarkPowFrac-16 197 174 -11.68% BenchmarkMod-16 112 99.8 -10.89% BenchmarkAsinh-16 59.9 53.7 -10.35% BenchmarkAcosh-16 44.8 40.3 -10.04% Updates #21390 Change-Id: I56cc991fc2e55249d69518d4e1ba76cc23904e35 Reviewed-on: https://go-review.googlesource.com/63290 Reviewed-by: Michael Munday <mike.munday@ibm.com>
2017-09-12 12:33:21 -04:00
{name: "MFVSRD", argLength: 1, reg: fpgp, asm: "MFVSRD", typ: "Int64"}, // move 64 bits of F register into G register
{name: "MTVSRD", argLength: 1, reg: gpfp, asm: "MTVSRD", typ: "Float64"}, // move 64 bits of G register into F register
[dev.ssa] cmd/compile: PPC64, FP to/from int conversions. Passes ssa_test. Requires a few new instructions and some scratchpad memory to move data between G and F registers. Also fixed comparisons to be correct in case of NaN. Added missing instructions for run.bash. Removed some FP registers that are apparently "reserved" (but that are also apparently also unused except for a gratuitous multiplication by two when y = x+x would work just as well). Currently failing stack splits. Updates #16010. Change-Id: I73b161bfff54445d72bd7b813b1479f89fc72602 Reviewed-on: https://go-review.googlesource.com/26813 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-10 11:44:57 -07:00
[dev.ssa] cmd/compile: PPC64, load/store by type, shifts, divisions, bools Updates #16010. Change-Id: Ie520d64fd1c4f881f45623303ed0b7cbdf0e4764 Reviewed-on: https://go-review.googlesource.com/25493 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-04 11:34:43 -07:00
{name: "AND", argLength: 2, reg: gp21, asm: "AND", commutative: true}, // arg0&arg1
{name: "ANDN", argLength: 2, reg: gp21, asm: "ANDN"}, // arg0&^arg1
{name: "OR", argLength: 2, reg: gp21, asm: "OR", commutative: true}, // arg0|arg1
{name: "ORN", argLength: 2, reg: gp21, asm: "ORN"}, // arg0|^arg1
cmd/compile: automatically handle commuting ops in rewrite rules Note that this is a redo of an undo of the original buggy CL 38666. We have lots of rewrite rules that vary only in the fact that we have 2 versions for the 2 different orderings of various commuting ops. For example: (ADDL x (MOVLconst [c])) -> (ADDLconst [c] x) (ADDL (MOVLconst [c]) x) -> (ADDLconst [c] x) It can get unwieldly quickly, especially when there is more than one commuting op in a rule. Our existing "fix" for this problem is to have rules that canonicalize the operations first. For example: (Eq64 x (Const64 <t> [c])) && x.Op != OpConst64 -> (Eq64 (Const64 <t> [c]) x) Subsequent rules can then assume if there is a constant arg to Eq64, it will be the first one. This fix kinda works, but it is fragile and only works when we remember to include the required extra rules. The fundamental problem is that the rule matcher doesn't know anything about commuting ops. This CL fixes that fact. We already have information about which ops commute. (The register allocator takes advantage of commutivity.) The rule generator now automatically generates multiple rules for a single source rule when there are commutative ops in the rule. We can now drop all of our almost-duplicate source-level rules and the canonicalization rules. I have some CLs in progress that will be a lot less verbose when the rule generator handles commutivity for me. I had to reorganize the load-combining rules a bit. The 8-way OR rules generated 128 different reorderings, which was causing the generator to put too much code in the rewrite*.go files (the big ones were going from 25K lines to 132K lines). Instead I reorganized the rules to combine pairs of loads at a time. The generated rule files are now actually a bit (5%) smaller. Make.bash times are ~unchanged. Compiler benchmarks are not observably different. Probably because we don't spend much compiler time in rule matching anyway. I've also done a pass over all of our ops adding commutative markings for ops which hadn't had them previously. Fixes #18292 Change-Id: Ic1c0e43fbf579539f459971625f69690c9ab8805 Reviewed-on: https://go-review.googlesource.com/38801 Run-TryBot: Keith Randall <khr@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2017-03-30 03:30:22 +00:00
{name: "NOR", argLength: 2, reg: gp21, asm: "NOR", commutative: true}, // ^(arg0|arg1)
[dev.ssa] cmd/compile: PPC64, load/store by type, shifts, divisions, bools Updates #16010. Change-Id: Ie520d64fd1c4f881f45623303ed0b7cbdf0e4764 Reviewed-on: https://go-review.googlesource.com/25493 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-04 11:34:43 -07:00
{name: "XOR", argLength: 2, reg: gp21, asm: "XOR", typ: "Int64", commutative: true}, // arg0^arg1
{name: "EQV", argLength: 2, reg: gp21, asm: "EQV", typ: "Int64", commutative: true}, // arg0^^arg1
[dev.ssa] cmd/compile: PPC: FP load/store/const/cmp/neg; div/mod FP<->int conversions remain. Updates #16010. Change-Id: I38d7a4923e34d0a489935fffc4c96c020cafdba2 Reviewed-on: https://go-review.googlesource.com/25589 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
2016-08-08 14:19:56 -07:00
{name: "NEG", argLength: 1, reg: gp11, asm: "NEG"}, // -arg0 (integer)
{name: "FNEG", argLength: 1, reg: fp11, asm: "FNEG"}, // -arg0 (floating point)
[dev.ssa] cmd/compile: PPC64, FP to/from int conversions. Passes ssa_test. Requires a few new instructions and some scratchpad memory to move data between G and F registers. Also fixed comparisons to be correct in case of NaN. Added missing instructions for run.bash. Removed some FP registers that are apparently "reserved" (but that are also apparently also unused except for a gratuitous multiplication by two when y = x+x would work just as well). Currently failing stack splits. Updates #16010. Change-Id: I73b161bfff54445d72bd7b813b1479f89fc72602 Reviewed-on: https://go-review.googlesource.com/26813 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-10 11:44:57 -07:00
{name: "FSQRT", argLength: 1, reg: fp11, asm: "FSQRT"}, // sqrt(arg0) (floating point)
{name: "FSQRTS", argLength: 1, reg: fp11, asm: "FSQRTS"}, // sqrt(arg0) (floating point, single precision)
cmd/compile: intrinsics for trunc, floor, ceil on ppc64x This implements trunc, floor, and ceil in the math package as intrinsics on ppc64x. Significant improvement mainly due to avoiding call overhead of args and return value. BenchmarkCeil-16 5.95 0.69 -88.40% BenchmarkFloor-16 5.95 0.69 -88.40% BenchmarkTrunc-16 5.82 0.69 -88.14% Updates #21390 Change-Id: I951e182694f6e0c431da79c577272b81fb0ebad0 Reviewed-on: https://go-review.googlesource.com/54654 Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com> Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com> Reviewed-by: David Chase <drchase@google.com>
2017-08-09 13:53:34 -04:00
{name: "FFLOOR", argLength: 1, reg: fp11, asm: "FRIM"}, // floor(arg0), float64
{name: "FCEIL", argLength: 1, reg: fp11, asm: "FRIP"}, // ceil(arg0), float64
{name: "FTRUNC", argLength: 1, reg: fp11, asm: "FRIZ"}, // trunc(arg0), float64
[dev.ssa] cmd/compile: PPC64, load/store by type, shifts, divisions, bools Updates #16010. Change-Id: Ie520d64fd1c4f881f45623303ed0b7cbdf0e4764 Reviewed-on: https://go-review.googlesource.com/25493 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-04 11:34:43 -07:00
[dev.ssa] cmd/compile: remove flags from regMask Reg allocator skips flag-typed values. Flag allocator uses the type and whether the op has "clobberFlags" set. Tested on AMD64, ARM, ARM64, 386. Passed 'toolstash -cmp' on AMD64. PPC64 is coded blindly. Change-Id: Ib1cc27efecef6a1bb27f7d7ed035a582660d244f Reviewed-on: https://go-review.googlesource.com/25480 Run-TryBot: Cherry Zhang <cherryyz@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2016-08-04 06:57:34 -04:00
{name: "ORconst", argLength: 1, reg: gp11, asm: "OR", aux: "Int64"}, // arg0|aux
{name: "XORconst", argLength: 1, reg: gp11, asm: "XOR", aux: "Int64"}, // arg0^aux
{name: "ANDconst", argLength: 1, reg: regInfo{inputs: []regMask{gp | sp | sb}, outputs: []regMask{gp}}, asm: "ANDCC", aux: "Int64", clobberFlags: true}, // arg0&aux // and-immediate sets CC on PPC, always.
cmd/compile: PPC64, elide unnecessary sign extension Inputs to store[BHW] and cmpW(U) need not be correct in more bits than are used by the instruction. Added a pattern tailored to what appears to be cgo boilerplate. Added a pattern (also seen in cgo boilerplate and hashing) to replace {EQ,NE}-CMP-ANDconst with {EQ-NE}-ANDCCconst. Added a pattern to clean up ANDconst shift distance inputs (this was seen in hashing). Simplify repeated and,or,xor. Fixes #17109. Change-Id: I68eac83e3e614d69ffe473a08953048c8b066d88 Reviewed-on: https://go-review.googlesource.com/30455 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
2016-10-05 13:21:09 -07:00
{name: "ANDCCconst", argLength: 1, reg: regInfo{inputs: []regMask{gp | sp | sb}}, asm: "ANDCC", aux: "Int64", typ: "Flags"}, // arg0&aux == 0 // and-immediate sets CC on PPC, always.
[dev.ssa] cmd/compile: PPC64, load/store by type, shifts, divisions, bools Updates #16010. Change-Id: Ie520d64fd1c4f881f45623303ed0b7cbdf0e4764 Reviewed-on: https://go-review.googlesource.com/25493 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-04 11:34:43 -07:00
cmd/compile/internal/ssa: populate SymEffects for SSA Ops Changes to ${GOARCH}Ops.go files were mechanically produced using github.com/mdempsky/ssa-symops, a one-off tool that inserts "SymEffect: X" elements by pattern matching against the Op names. Change-Id: Ibf3e481ffd588647f2a31662d72114b740ccbfcf Reviewed-on: https://go-review.googlesource.com/38084 Run-TryBot: Matthew Dempsky <mdempsky@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
2017-03-09 14:46:43 -08:00
{name: "MOVBreg", argLength: 1, reg: gp11, asm: "MOVB", typ: "Int64"}, // sign extend int8 to int64
{name: "MOVBZreg", argLength: 1, reg: gp11, asm: "MOVBZ", typ: "Int64"}, // zero extend uint8 to uint64
{name: "MOVHreg", argLength: 1, reg: gp11, asm: "MOVH", typ: "Int64"}, // sign extend int16 to int64
{name: "MOVHZreg", argLength: 1, reg: gp11, asm: "MOVHZ", typ: "Int64"}, // zero extend uint16 to uint64
{name: "MOVWreg", argLength: 1, reg: gp11, asm: "MOVW", typ: "Int64"}, // sign extend int32 to int64
{name: "MOVWZreg", argLength: 1, reg: gp11, asm: "MOVWZ", typ: "Int64"}, // zero extend uint32 to uint64
{name: "MOVBZload", argLength: 2, reg: gpload, asm: "MOVBZ", aux: "SymOff", typ: "UInt8", faultOnNilArg0: true, symEffect: "Read"}, // zero extend uint8 to uint64
{name: "MOVHload", argLength: 2, reg: gpload, asm: "MOVH", aux: "SymOff", typ: "Int16", faultOnNilArg0: true, symEffect: "Read"}, // sign extend int16 to int64
{name: "MOVHZload", argLength: 2, reg: gpload, asm: "MOVHZ", aux: "SymOff", typ: "UInt16", faultOnNilArg0: true, symEffect: "Read"}, // zero extend uint16 to uint64
{name: "MOVWload", argLength: 2, reg: gpload, asm: "MOVW", aux: "SymOff", typ: "Int32", faultOnNilArg0: true, symEffect: "Read"}, // sign extend int32 to int64
{name: "MOVWZload", argLength: 2, reg: gpload, asm: "MOVWZ", aux: "SymOff", typ: "UInt32", faultOnNilArg0: true, symEffect: "Read"}, // zero extend uint32 to uint64
{name: "MOVDload", argLength: 2, reg: gpload, asm: "MOVD", aux: "SymOff", typ: "Int64", faultOnNilArg0: true, symEffect: "Read"},
{name: "FMOVDload", argLength: 2, reg: fpload, asm: "FMOVD", aux: "SymOff", typ: "Float64", faultOnNilArg0: true, symEffect: "Read"},
{name: "FMOVSload", argLength: 2, reg: fpload, asm: "FMOVS", aux: "SymOff", typ: "Float32", faultOnNilArg0: true, symEffect: "Read"},
{name: "MOVBstore", argLength: 3, reg: gpstore, asm: "MOVB", aux: "SymOff", typ: "Mem", faultOnNilArg0: true, symEffect: "Write"},
{name: "MOVHstore", argLength: 3, reg: gpstore, asm: "MOVH", aux: "SymOff", typ: "Mem", faultOnNilArg0: true, symEffect: "Write"},
{name: "MOVWstore", argLength: 3, reg: gpstore, asm: "MOVW", aux: "SymOff", typ: "Mem", faultOnNilArg0: true, symEffect: "Write"},
{name: "MOVDstore", argLength: 3, reg: gpstore, asm: "MOVD", aux: "SymOff", typ: "Mem", faultOnNilArg0: true, symEffect: "Write"},
{name: "FMOVDstore", argLength: 3, reg: fpstore, asm: "FMOVD", aux: "SymOff", typ: "Mem", faultOnNilArg0: true, symEffect: "Write"},
{name: "FMOVSstore", argLength: 3, reg: fpstore, asm: "FMOVS", aux: "SymOff", typ: "Mem", faultOnNilArg0: true, symEffect: "Write"},
{name: "MOVBstorezero", argLength: 2, reg: gpstorezero, asm: "MOVB", aux: "SymOff", typ: "Mem", faultOnNilArg0: true, symEffect: "Write"}, // store zero byte to arg0+aux. arg1=mem
{name: "MOVHstorezero", argLength: 2, reg: gpstorezero, asm: "MOVH", aux: "SymOff", typ: "Mem", faultOnNilArg0: true, symEffect: "Write"}, // store zero 2 bytes to ...
{name: "MOVWstorezero", argLength: 2, reg: gpstorezero, asm: "MOVW", aux: "SymOff", typ: "Mem", faultOnNilArg0: true, symEffect: "Write"}, // store zero 4 bytes to ...
{name: "MOVDstorezero", argLength: 2, reg: gpstorezero, asm: "MOVD", aux: "SymOff", typ: "Mem", faultOnNilArg0: true, symEffect: "Write"}, // store zero 8 bytes to ...
{name: "MOVDaddr", argLength: 1, reg: regInfo{inputs: []regMask{sp | sb}, outputs: []regMask{gp}}, aux: "SymOff", asm: "MOVD", rematerializeable: true, symEffect: "Addr"}, // arg0 + auxInt + aux.(*gc.Sym), arg0=SP/SB
[dev.ssa] cmd/compile: some improvements to PPC codegen Runs fibonacci for all integer types. Fold addressing arithmetic into stores. Updates #16010. Change-Id: I257982c82c00c80b00679757c3da345045968022 Reviewed-on: https://go-review.googlesource.com/25103 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com>
2016-07-06 13:32:52 -07:00
cmd/compile: PPC64, elide unnecessary sign extension Inputs to store[BHW] and cmpW(U) need not be correct in more bits than are used by the instruction. Added a pattern tailored to what appears to be cgo boilerplate. Added a pattern (also seen in cgo boilerplate and hashing) to replace {EQ,NE}-CMP-ANDconst with {EQ-NE}-ANDCCconst. Added a pattern to clean up ANDconst shift distance inputs (this was seen in hashing). Simplify repeated and,or,xor. Fixes #17109. Change-Id: I68eac83e3e614d69ffe473a08953048c8b066d88 Reviewed-on: https://go-review.googlesource.com/30455 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
2016-10-05 13:21:09 -07:00
{name: "MOVDconst", argLength: 0, reg: gp01, aux: "Int64", asm: "MOVD", typ: "Int64", rematerializeable: true}, //
{name: "FMOVDconst", argLength: 0, reg: fp01, aux: "Float64", asm: "FMOVD", rematerializeable: true}, //
{name: "FMOVSconst", argLength: 0, reg: fp01, aux: "Float32", asm: "FMOVS", rematerializeable: true}, //
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
{name: "FCMPU", argLength: 2, reg: fp2cr, asm: "FCMPU", typ: "Flags"},
{name: "CMP", argLength: 2, reg: gp2cr, asm: "CMP", typ: "Flags"}, // arg0 compare to arg1
{name: "CMPU", argLength: 2, reg: gp2cr, asm: "CMPU", typ: "Flags"}, // arg0 compare to arg1
{name: "CMPW", argLength: 2, reg: gp2cr, asm: "CMPW", typ: "Flags"}, // arg0 compare to arg1
{name: "CMPWU", argLength: 2, reg: gp2cr, asm: "CMPWU", typ: "Flags"}, // arg0 compare to arg1
[dev.ssa] cmd/compile: PPC64, add cmp->bool, some shifts, hmul Includes hmul (all widths) compare for boolean result and simplifications shift operations plus changes/additions for implementation (ORN, ADDME, ADDC) Also fixed a backwards-operand CMP. Change-Id: Id723c4e25125c38e0d9ab9ec9448176b75f4cdb4 Reviewed-on: https://go-review.googlesource.com/25410 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-02 13:17:09 -07:00
{name: "CMPconst", argLength: 1, reg: gp1cr, asm: "CMP", aux: "Int64", typ: "Flags"},
{name: "CMPUconst", argLength: 1, reg: gp1cr, asm: "CMPU", aux: "Int64", typ: "Flags"},
{name: "CMPWconst", argLength: 1, reg: gp1cr, asm: "CMPW", aux: "Int32", typ: "Flags"},
{name: "CMPWUconst", argLength: 1, reg: gp1cr, asm: "CMPWU", aux: "Int32", typ: "Flags"},
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
// pseudo-ops
[dev.ssa] cmd/compile: PPC64, FP to/from int conversions. Passes ssa_test. Requires a few new instructions and some scratchpad memory to move data between G and F registers. Also fixed comparisons to be correct in case of NaN. Added missing instructions for run.bash. Removed some FP registers that are apparently "reserved" (but that are also apparently also unused except for a gratuitous multiplication by two when y = x+x would work just as well). Currently failing stack splits. Updates #16010. Change-Id: I73b161bfff54445d72bd7b813b1479f89fc72602 Reviewed-on: https://go-review.googlesource.com/26813 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-10 11:44:57 -07:00
{name: "Equal", argLength: 1, reg: crgp}, // bool, true flags encode x==y false otherwise.
{name: "NotEqual", argLength: 1, reg: crgp}, // bool, true flags encode x!=y false otherwise.
{name: "LessThan", argLength: 1, reg: crgp}, // bool, true flags encode x<y false otherwise.
{name: "FLessThan", argLength: 1, reg: crgp}, // bool, true flags encode x<y false otherwise.
{name: "LessEqual", argLength: 1, reg: crgp}, // bool, true flags encode x<=y false otherwise.
{name: "FLessEqual", argLength: 1, reg: crgp}, // bool, true flags encode x<=y false otherwise; PPC <= === !> which is wrong for NaN
{name: "GreaterThan", argLength: 1, reg: crgp}, // bool, true flags encode x>y false otherwise.
{name: "FGreaterThan", argLength: 1, reg: crgp}, // bool, true flags encode x>y false otherwise.
{name: "GreaterEqual", argLength: 1, reg: crgp}, // bool, true flags encode x>=y false otherwise.
{name: "FGreaterEqual", argLength: 1, reg: crgp}, // bool, true flags encode x>=y false otherwise.; PPC >= === !< which is wrong for NaN
[dev.ssa] cmd/compile: PPC64, basic support for all calls and "miscellaneous" Added support for ClosureCall, DeferCall, InterCall (GoCall not yet tested). Added support for GetClosurePtr, IsNonNil, IsInBounds, IsSliceInBounds, NilCheck (Convert and GetG not yet tested) Still need to implement NilCheck optimizations. Fixed move boolean constant, order of operands to subtract. Updates #16010. Change-Id: Ibe0f6a6e688df4396cd77de0e9095997e4ca8ed2 Reviewed-on: https://go-review.googlesource.com/25241 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-26 09:24:18 -07:00
// Scheduler ensures LoweredGetClosurePtr occurs only in entry block,
// and sorts it to the very beginning of the block to prevent other
// use of the closure pointer.
{name: "LoweredGetClosurePtr", reg: regInfo{outputs: []regMask{ctxt}}},
//arg0=ptr,arg1=mem, returns void. Faults if ptr is nil.
cmd/compile: remove duplicate nilchecks Mark nil check operations as faulting if their arg is zero. This lets the late nilcheck pass remove duplicates. Fixes #17242. Change-Id: I4c9938d8a5a1e43edd85b4a66f0b34004860bcd9 Reviewed-on: https://go-review.googlesource.com/29952 Run-TryBot: Keith Randall <khr@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-09-27 14:39:27 -07:00
{name: "LoweredNilCheck", argLength: 2, reg: regInfo{inputs: []regMask{gp | sp | sb}, clobbers: tmp}, clobberFlags: true, nilCheck: true, faultOnNilArg0: true},
cmd/compile: emit fused multiply-{add,subtract} on ppc64x A follow on to CL 36963 adding support for ppc64x. Performance changes (as posted on the issue): poly1305: benchmark old ns/op new ns/op delta Benchmark64-16 172 151 -12.21% Benchmark1K-16 1828 1523 -16.68% Benchmark64Unaligned-16 172 151 -12.21% Benchmark1KUnaligned-16 1827 1523 -16.64% math: BenchmarkAcos-16 43.9 39.9 -9.11% BenchmarkAcosh-16 57.0 45.8 -19.65% BenchmarkAsin-16 35.8 33.0 -7.82% BenchmarkAsinh-16 68.6 60.8 -11.37% BenchmarkAtan-16 19.8 16.2 -18.18% BenchmarkAtanh-16 65.5 57.5 -12.21% BenchmarkAtan2-16 45.4 34.2 -24.67% BenchmarkGamma-16 37.6 26.0 -30.85% BenchmarkLgamma-16 40.0 28.2 -29.50% BenchmarkLog1p-16 35.1 29.1 -17.09% BenchmarkSin-16 22.7 18.4 -18.94% BenchmarkSincos-16 31.7 23.7 -25.24% BenchmarkSinh-16 146 131 -10.27% BenchmarkY0-16 130 107 -17.69% BenchmarkY1-16 127 107 -15.75% BenchmarkYn-16 278 235 -15.47% Updates #17895. Change-Id: I1c16199715d20c9c4bd97c4a950bcfa69eb688c1 Reviewed-on: https://go-review.googlesource.com/38095 Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com> Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
2017-03-13 14:39:17 -04:00
// Round ops to block fused-multiply-add extraction.
{name: "LoweredRound32F", argLength: 1, reg: fp11, resultInArg0: true},
{name: "LoweredRound64F", argLength: 1, reg: fp11, resultInArg0: true},
[dev.ssa] cmd/compile: PPC64, basic support for all calls and "miscellaneous" Added support for ClosureCall, DeferCall, InterCall (GoCall not yet tested). Added support for GetClosurePtr, IsNonNil, IsInBounds, IsSliceInBounds, NilCheck (Convert and GetG not yet tested) Still need to implement NilCheck optimizations. Fixed move boolean constant, order of operands to subtract. Updates #16010. Change-Id: Ibe0f6a6e688df4396cd77de0e9095997e4ca8ed2 Reviewed-on: https://go-review.googlesource.com/25241 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-26 09:24:18 -07:00
// Convert pointer to integer, takes a memory operand for ordering.
{name: "MOVDconvert", argLength: 2, reg: gp11, asm: "MOVD"},
cmd/compile: experiment which clobbers all dead pointer fields The experiment "clobberdead" clobbers all pointer fields that the compiler thinks are dead, just before and after every safepoint. Useful for debugging the generation of live pointer bitmaps. Helped find the following issues: Update #15936 Update #16026 Update #16095 Update #18860 Change-Id: Id1d12f86845e3d93bae903d968b1eac61fc461f9 Reviewed-on: https://go-review.googlesource.com/23924 Run-TryBot: Keith Randall <khr@golang.org> Reviewed-by: Matthew Dempsky <mdempsky@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-06-08 22:02:08 -07:00
{name: "CALLstatic", argLength: 1, reg: regInfo{clobbers: callerSave}, aux: "SymOff", clobberFlags: true, call: true, symEffect: "None"}, // call static function aux.(*obj.LSym). arg0=mem, auxint=argsize, returns mem
cmd/compile: request r12 for indirect calls on ppc64le On ppc64le, functions compiled with -shared expect r12 to hold the function's address for indirect calls. Previously this was enforced by generating a move instruction if the address wasn't already in r12. This change avoids that extra move by requesting r12 in the CALL ops that do indirect calls. As a result of adding support for plugins on ppc64le, it was discovered that there would be more cases where this extra move was needed, so this seemed like a better solution. Updates #20756 Change-Id: I6770885a46990f78c6d2902a715dcdaa822192a1 Reviewed-on: https://go-review.googlesource.com/62890 Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Crawshaw <crawshaw@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2017-09-11 10:39:24 -04:00
{name: "CALLclosure", argLength: 3, reg: regInfo{inputs: []regMask{callptr, ctxt, 0}, clobbers: callerSave}, aux: "Int64", clobberFlags: true, call: true}, // call function via closure. arg0=codeptr, arg1=closure, arg2=mem, auxint=argsize, returns mem
{name: "CALLinter", argLength: 2, reg: regInfo{inputs: []regMask{callptr}, clobbers: callerSave}, aux: "Int64", clobberFlags: true, call: true}, // call fn by pointer. arg0=codeptr, arg1=mem, auxint=argsize, returns mem
[dev.ssa] cmd/compile: PPC64, basic support for all calls and "miscellaneous" Added support for ClosureCall, DeferCall, InterCall (GoCall not yet tested). Added support for GetClosurePtr, IsNonNil, IsInBounds, IsSliceInBounds, NilCheck (Convert and GetG not yet tested) Still need to implement NilCheck optimizations. Fixed move boolean constant, order of operands to subtract. Updates #16010. Change-Id: Ibe0f6a6e688df4396cd77de0e9095997e4ca8ed2 Reviewed-on: https://go-review.googlesource.com/25241 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-26 09:24:18 -07:00
[dev.ssa] cmd/compile: PPC64, add more zeroing and moves Passes light testing. Modified to avoid possible exposure of "exterior" pointers to GC. Updates #16010. Change-Id: I41fced4fa83cefb9542dff8c8dee1a0c48056b3c Reviewed-on: https://go-review.googlesource.com/25310 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-27 13:54:07 -07:00
// large or unaligned zeroing
// arg0 = address of memory to zero (in R3, changed as side effect)
// returns mem
cmd/compile: improve LoweredZero performance for ppc64x This change improves the performance of the LoweredZero rule on ppc64x. The improvement can be seen in the runtime ClearFat benchmarks: BenchmarkClearFat12-16 2.40 0.69 -71.25% BenchmarkClearFat16-16 9.98 0.93 -90.68% BenchmarkClearFat24-16 4.75 0.93 -80.42% BenchmarkClearFat32-16 6.02 0.93 -84.55% BenchmarkClearFat40-16 7.19 1.16 -83.87% BenchmarkClearFat48-16 15.0 1.39 -90.73% BenchmarkClearFat56-16 9.95 1.62 -83.72% BenchmarkClearFat64-16 18.0 1.86 -89.67% BenchmarkClearFat128-16 30.0 8.08 -73.07% BenchmarkClearFat256-16 52.5 11.3 -78.48% BenchmarkClearFat512-16 97.0 19.0 -80.41% BenchmarkClearFat1024-16 244 34.2 -85.98% Fixes: #19532 Change-Id: If493e28bc1d8e61bc79978498be9f5336a36cd3f Reviewed-on: https://go-review.googlesource.com/38096 Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Michael Munday <munday@ca.ibm.com>
2017-03-13 10:16:30 -04:00
//
// a loop is generated when there is more than one iteration
// needed to clear 4 doublewords
//
// MOVD $len/32,R31
// MOVD R31,CTR
// loop:
// MOVD R0,(R3)
// MOVD R0,8(R3)
// MOVD R0,16(R3)
// MOVD R0,24(R3)
// ADD R3,32
// BC loop
// remaining doubleword clears generated as needed
// MOVD R0,(R3)
// MOVD R0,8(R3)
// MOVD R0,16(R3)
// MOVD R0,24(R3)
// one or more of these to clear remainder < 8 bytes
// MOVW R0,n1(R3)
// MOVH R0,n2(R3)
// MOVB R0,n3(R3)
[dev.ssa] cmd/compile: PPC64, add more zeroing and moves Passes light testing. Modified to avoid possible exposure of "exterior" pointers to GC. Updates #16010. Change-Id: I41fced4fa83cefb9542dff8c8dee1a0c48056b3c Reviewed-on: https://go-review.googlesource.com/25310 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-27 13:54:07 -07:00
{
name: "LoweredZero",
aux: "Int64",
cmd/compile: improve LoweredZero performance for ppc64x This change improves the performance of the LoweredZero rule on ppc64x. The improvement can be seen in the runtime ClearFat benchmarks: BenchmarkClearFat12-16 2.40 0.69 -71.25% BenchmarkClearFat16-16 9.98 0.93 -90.68% BenchmarkClearFat24-16 4.75 0.93 -80.42% BenchmarkClearFat32-16 6.02 0.93 -84.55% BenchmarkClearFat40-16 7.19 1.16 -83.87% BenchmarkClearFat48-16 15.0 1.39 -90.73% BenchmarkClearFat56-16 9.95 1.62 -83.72% BenchmarkClearFat64-16 18.0 1.86 -89.67% BenchmarkClearFat128-16 30.0 8.08 -73.07% BenchmarkClearFat256-16 52.5 11.3 -78.48% BenchmarkClearFat512-16 97.0 19.0 -80.41% BenchmarkClearFat1024-16 244 34.2 -85.98% Fixes: #19532 Change-Id: If493e28bc1d8e61bc79978498be9f5336a36cd3f Reviewed-on: https://go-review.googlesource.com/38096 Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Michael Munday <munday@ca.ibm.com>
2017-03-13 10:16:30 -04:00
argLength: 2,
[dev.ssa] cmd/compile: PPC64, add more zeroing and moves Passes light testing. Modified to avoid possible exposure of "exterior" pointers to GC. Updates #16010. Change-Id: I41fced4fa83cefb9542dff8c8dee1a0c48056b3c Reviewed-on: https://go-review.googlesource.com/25310 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-27 13:54:07 -07:00
reg: regInfo{
cmd/compile: improve LoweredZero performance for ppc64x This change improves the performance of the LoweredZero rule on ppc64x. The improvement can be seen in the runtime ClearFat benchmarks: BenchmarkClearFat12-16 2.40 0.69 -71.25% BenchmarkClearFat16-16 9.98 0.93 -90.68% BenchmarkClearFat24-16 4.75 0.93 -80.42% BenchmarkClearFat32-16 6.02 0.93 -84.55% BenchmarkClearFat40-16 7.19 1.16 -83.87% BenchmarkClearFat48-16 15.0 1.39 -90.73% BenchmarkClearFat56-16 9.95 1.62 -83.72% BenchmarkClearFat64-16 18.0 1.86 -89.67% BenchmarkClearFat128-16 30.0 8.08 -73.07% BenchmarkClearFat256-16 52.5 11.3 -78.48% BenchmarkClearFat512-16 97.0 19.0 -80.41% BenchmarkClearFat1024-16 244 34.2 -85.98% Fixes: #19532 Change-Id: If493e28bc1d8e61bc79978498be9f5336a36cd3f Reviewed-on: https://go-review.googlesource.com/38096 Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Michael Munday <munday@ca.ibm.com>
2017-03-13 10:16:30 -04:00
inputs: []regMask{buildReg("R3")},
[dev.ssa] cmd/compile: remove flags from regMask Reg allocator skips flag-typed values. Flag allocator uses the type and whether the op has "clobberFlags" set. Tested on AMD64, ARM, ARM64, 386. Passed 'toolstash -cmp' on AMD64. PPC64 is coded blindly. Change-Id: Ib1cc27efecef6a1bb27f7d7ed035a582660d244f Reviewed-on: https://go-review.googlesource.com/25480 Run-TryBot: Cherry Zhang <cherryyz@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2016-08-04 06:57:34 -04:00
clobbers: buildReg("R3"),
[dev.ssa] cmd/compile: PPC64, add more zeroing and moves Passes light testing. Modified to avoid possible exposure of "exterior" pointers to GC. Updates #16010. Change-Id: I41fced4fa83cefb9542dff8c8dee1a0c48056b3c Reviewed-on: https://go-review.googlesource.com/25310 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-27 13:54:07 -07:00
},
cmd/compile: redo nil checks Get rid of BlockCheck. Josh goaded me into it, and I went down a rabbithole making it happen. NilCheck now panics if the pointer is nil and returns void, as before. BlockCheck is gone, and NilCheck is no longer a Control value for any block. It just exists (and deadcode knows not to throw it away). I rewrote the nilcheckelim pass to handle this case. In particular, there can now be multiple NilCheck ops per block. I moved all of the arch-dependent nil check elimination done as part of ssaGenValue into its own proper pass, so we don't have to duplicate that code for every architecture. Making the arch-dependent nil check its own pass means I needed to add a bunch of flags to the opcode table so I could write the code without arch-dependent ops everywhere. Change-Id: I419f891ac9b0de313033ff09115c374163416a9f Reviewed-on: https://go-review.googlesource.com/29120 Run-TryBot: Keith Randall <khr@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2016-09-13 17:01:01 -07:00
clobberFlags: true,
typ: "Mem",
faultOnNilArg0: true,
[dev.ssa] cmd/compile: PPC64, add more zeroing and moves Passes light testing. Modified to avoid possible exposure of "exterior" pointers to GC. Updates #16010. Change-Id: I41fced4fa83cefb9542dff8c8dee1a0c48056b3c Reviewed-on: https://go-review.googlesource.com/25310 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-27 13:54:07 -07:00
},
cmd/compile: improve LoweredMove performance on ppc64x This change improves the performance for LoweredMove on ppc64le and ppc64. benchmark old ns/op new ns/op delta BenchmarkCopyFat8-16 0.93 0.69 -25.81% BenchmarkCopyFat12-16 2.61 1.85 -29.12% BenchmarkCopyFat16-16 9.68 1.89 -80.48% BenchmarkCopyFat24-16 4.48 1.85 -58.71% BenchmarkCopyFat32-16 6.12 1.82 -70.26% BenchmarkCopyFat64-16 21.2 2.70 -87.26% BenchmarkCopyFat128-16 29.6 3.97 -86.59% BenchmarkCopyFat256-16 52.6 13.4 -74.52% BenchmarkCopyFat512-16 97.1 18.7 -80.74% BenchmarkCopyFat1024-16 186 35.3 -81.02% BenchmarkAssertE2TLarge-16 14.2 5.06 -64.37% Fixes #19785 Change-Id: I7d5e0052712b75811c02c7d86c5112e5649ad782 Reviewed-on: https://go-review.googlesource.com/38950 Reviewed-by: Keith Randall <khr@golang.org>
2017-03-30 11:07:36 -04:00
// Loop code:
// MOVD len/32,REG_TMP only for loop
// MOVD REG_TMP,CTR only for loop
// loop:
// MOVD (R4),R7
// MOVD 8(R4),R8
// MOVD 16(R4),R9
// MOVD 24(R4),R10
// ADD R4,$32 only with loop
// MOVD R7,(R3)
// MOVD R8,8(R3)
// MOVD R9,16(R3)
// MOVD R10,24(R3)
// ADD R3,$32 only with loop
// BC 16,0,loop only with loop
// Bytes not moved by this loop are moved
// with a combination of the following instructions,
// starting with the largest sizes and generating as
// many as needed, using the appropriate offset value.
// MOVD n(R4),R7
// MOVD R7,n(R3)
// MOVW n1(R4),R7
// MOVW R7,n1(R3)
// MOVH n2(R4),R7
// MOVH R7,n2(R3)
// MOVB n3(R4),R7
// MOVB R7,n3(R3)
[dev.ssa] cmd/compile: PPC64, add more zeroing and moves Passes light testing. Modified to avoid possible exposure of "exterior" pointers to GC. Updates #16010. Change-Id: I41fced4fa83cefb9542dff8c8dee1a0c48056b3c Reviewed-on: https://go-review.googlesource.com/25310 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-27 13:54:07 -07:00
{
name: "LoweredMove",
aux: "Int64",
cmd/compile: improve LoweredMove performance on ppc64x This change improves the performance for LoweredMove on ppc64le and ppc64. benchmark old ns/op new ns/op delta BenchmarkCopyFat8-16 0.93 0.69 -25.81% BenchmarkCopyFat12-16 2.61 1.85 -29.12% BenchmarkCopyFat16-16 9.68 1.89 -80.48% BenchmarkCopyFat24-16 4.48 1.85 -58.71% BenchmarkCopyFat32-16 6.12 1.82 -70.26% BenchmarkCopyFat64-16 21.2 2.70 -87.26% BenchmarkCopyFat128-16 29.6 3.97 -86.59% BenchmarkCopyFat256-16 52.6 13.4 -74.52% BenchmarkCopyFat512-16 97.1 18.7 -80.74% BenchmarkCopyFat1024-16 186 35.3 -81.02% BenchmarkAssertE2TLarge-16 14.2 5.06 -64.37% Fixes #19785 Change-Id: I7d5e0052712b75811c02c7d86c5112e5649ad782 Reviewed-on: https://go-review.googlesource.com/38950 Reviewed-by: Keith Randall <khr@golang.org>
2017-03-30 11:07:36 -04:00
argLength: 3,
[dev.ssa] cmd/compile: PPC64, add more zeroing and moves Passes light testing. Modified to avoid possible exposure of "exterior" pointers to GC. Updates #16010. Change-Id: I41fced4fa83cefb9542dff8c8dee1a0c48056b3c Reviewed-on: https://go-review.googlesource.com/25310 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-27 13:54:07 -07:00
reg: regInfo{
cmd/compile: improve LoweredMove performance on ppc64x This change improves the performance for LoweredMove on ppc64le and ppc64. benchmark old ns/op new ns/op delta BenchmarkCopyFat8-16 0.93 0.69 -25.81% BenchmarkCopyFat12-16 2.61 1.85 -29.12% BenchmarkCopyFat16-16 9.68 1.89 -80.48% BenchmarkCopyFat24-16 4.48 1.85 -58.71% BenchmarkCopyFat32-16 6.12 1.82 -70.26% BenchmarkCopyFat64-16 21.2 2.70 -87.26% BenchmarkCopyFat128-16 29.6 3.97 -86.59% BenchmarkCopyFat256-16 52.6 13.4 -74.52% BenchmarkCopyFat512-16 97.1 18.7 -80.74% BenchmarkCopyFat1024-16 186 35.3 -81.02% BenchmarkAssertE2TLarge-16 14.2 5.06 -64.37% Fixes #19785 Change-Id: I7d5e0052712b75811c02c7d86c5112e5649ad782 Reviewed-on: https://go-review.googlesource.com/38950 Reviewed-by: Keith Randall <khr@golang.org>
2017-03-30 11:07:36 -04:00
inputs: []regMask{buildReg("R3"), buildReg("R4")},
clobbers: buildReg("R3 R4 R7 R8 R9 R10"),
[dev.ssa] cmd/compile: PPC64, add more zeroing and moves Passes light testing. Modified to avoid possible exposure of "exterior" pointers to GC. Updates #16010. Change-Id: I41fced4fa83cefb9542dff8c8dee1a0c48056b3c Reviewed-on: https://go-review.googlesource.com/25310 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-27 13:54:07 -07:00
},
cmd/compile: redo nil checks Get rid of BlockCheck. Josh goaded me into it, and I went down a rabbithole making it happen. NilCheck now panics if the pointer is nil and returns void, as before. BlockCheck is gone, and NilCheck is no longer a Control value for any block. It just exists (and deadcode knows not to throw it away). I rewrote the nilcheckelim pass to handle this case. In particular, there can now be multiple NilCheck ops per block. I moved all of the arch-dependent nil check elimination done as part of ssaGenValue into its own proper pass, so we don't have to duplicate that code for every architecture. Making the arch-dependent nil check its own pass means I needed to add a bunch of flags to the opcode table so I could write the code without arch-dependent ops everywhere. Change-Id: I419f891ac9b0de313033ff09115c374163416a9f Reviewed-on: https://go-review.googlesource.com/29120 Run-TryBot: Keith Randall <khr@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
2016-09-13 17:01:01 -07:00
clobberFlags: true,
typ: "Mem",
faultOnNilArg0: true,
faultOnNilArg1: true,
[dev.ssa] cmd/compile: PPC64, add more zeroing and moves Passes light testing. Modified to avoid possible exposure of "exterior" pointers to GC. Updates #16010. Change-Id: I41fced4fa83cefb9542dff8c8dee1a0c48056b3c Reviewed-on: https://go-review.googlesource.com/25310 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-27 13:54:07 -07:00
},
[dev.ssa] cmd/compile: PPC64, add cmp->bool, some shifts, hmul Includes hmul (all widths) compare for boolean result and simplifications shift operations plus changes/additions for implementation (ORN, ADDME, ADDC) Also fixed a backwards-operand CMP. Change-Id: Id723c4e25125c38e0d9ab9ec9448176b75f4cdb4 Reviewed-on: https://go-review.googlesource.com/25410 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-02 13:17:09 -07:00
cmd/compile: intrinsify atomics on ppc64x This adds the necessary changes so that atomics are treated as intrinsics on ppc64x. The implementations of And8 and Or8 require power8 for both ppc64 and ppc64le. This is a new requirement for ppc64. Fixes #8739 Change-Id: Icb85e2755a49166ee3652668279f6ed5ebbca901 Reviewed-on: https://go-review.googlesource.com/36832 Reviewed-by: Keith Randall <khr@golang.org>
2017-02-10 17:45:12 -05:00
{name: "LoweredAtomicStore32", argLength: 3, reg: gpstore, typ: "Mem", faultOnNilArg0: true, hasSideEffects: true},
{name: "LoweredAtomicStore64", argLength: 3, reg: gpstore, typ: "Mem", faultOnNilArg0: true, hasSideEffects: true},
{name: "LoweredAtomicLoad32", argLength: 2, reg: gpload, typ: "UInt32", clobberFlags: true, faultOnNilArg0: true},
{name: "LoweredAtomicLoad64", argLength: 2, reg: gpload, typ: "Int64", clobberFlags: true, faultOnNilArg0: true},
{name: "LoweredAtomicLoadPtr", argLength: 2, reg: gpload, typ: "Int64", clobberFlags: true, faultOnNilArg0: true},
// atomic add32, 64
// SYNC
// LDAR (Rarg0), Rout
// ADD Rarg1, Rout
// STDCCC Rout, (Rarg0)
// BNE -3(PC)
// ISYNC
// return new sum
{name: "LoweredAtomicAdd32", argLength: 3, reg: gpxchg, resultNotInArgs: true, clobberFlags: true, faultOnNilArg0: true, hasSideEffects: true},
{name: "LoweredAtomicAdd64", argLength: 3, reg: gpxchg, resultNotInArgs: true, clobberFlags: true, faultOnNilArg0: true, hasSideEffects: true},
// atomic exchange32, 64
// SYNC
// LDAR (Rarg0), Rout
// STDCCC Rarg1, (Rarg0)
// BNE -2(PC)
// ISYNC
// return old val
{name: "LoweredAtomicExchange32", argLength: 3, reg: gpxchg, resultNotInArgs: true, clobberFlags: true, faultOnNilArg0: true, hasSideEffects: true},
{name: "LoweredAtomicExchange64", argLength: 3, reg: gpxchg, resultNotInArgs: true, clobberFlags: true, faultOnNilArg0: true, hasSideEffects: true},
// atomic compare and swap.
// arg0 = pointer, arg1 = old value, arg2 = new value, arg3 = memory. auxint must be zero.
// if *arg0 == arg1 {
// *arg0 = arg2
// return (true, memory)
// } else {
// return (false, memory)
// }
// SYNC
// LDAR (Rarg0), Rtmp
// CMP Rarg1, Rtmp
// BNE 3(PC)
// STDCCC Rarg2, (Rarg0)
// BNE -4(PC)
// CBNZ Rtmp, -4(PC)
// CSET EQ, Rout
{name: "LoweredAtomicCas64", argLength: 4, reg: gpcas, resultNotInArgs: true, clobberFlags: true, faultOnNilArg0: true, hasSideEffects: true},
{name: "LoweredAtomicCas32", argLength: 4, reg: gpcas, resultNotInArgs: true, clobberFlags: true, faultOnNilArg0: true, hasSideEffects: true},
// atomic 8 and/or.
// *arg0 &= (|=) arg1. arg2=mem. returns memory. auxint must be zero.
// LBAR (Rarg0), Rtmp
// AND/OR Rarg1, Rtmp
// STBCCC Rtmp, (Rarg0), Rtmp
// BNE Rtmp, -3(PC)
{name: "LoweredAtomicAnd8", argLength: 3, reg: gpstore, asm: "AND", faultOnNilArg0: true, hasSideEffects: true},
{name: "LoweredAtomicOr8", argLength: 3, reg: gpstore, asm: "OR", faultOnNilArg0: true, hasSideEffects: true},
[dev.ssa] cmd/compile: PPC64, add cmp->bool, some shifts, hmul Includes hmul (all widths) compare for boolean result and simplifications shift operations plus changes/additions for implementation (ORN, ADDME, ADDC) Also fixed a backwards-operand CMP. Change-Id: Id723c4e25125c38e0d9ab9ec9448176b75f4cdb4 Reviewed-on: https://go-review.googlesource.com/25410 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-02 13:17:09 -07:00
// (InvertFlags (CMP a b)) == (CMP b a)
// So if we want (LessThan (CMP a b)) but we can't do that because a is a constant,
// then we do (LessThan (InvertFlags (CMP b a))) instead.
// Rewrites will convert this to (GreaterThan (CMP b a)).
// InvertFlags is a pseudo-op which can't appear in assembly output.
{name: "InvertFlags", argLength: 1}, // reverse direction of arg0
// Constant flag values. For any comparison, there are 3 possible
// outcomes: either the three from the signed total order (<,==,>)
// or the three from the unsigned total order, depending on which
// comparison operation was used (CMP or CMPU -- PPC is different from
// the other architectures, which have a single comparison producing
// both signed and unsigned comparison results.)
// These ops are for temporary use by rewrite rules. They
// cannot appear in the generated assembly.
{name: "FlagEQ"}, // equal
{name: "FlagLT"}, // signed < or unsigned <
{name: "FlagGT"}, // signed > or unsigned >
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
}
blocks := []blockData{
{name: "EQ"},
{name: "NE"},
{name: "LT"},
{name: "LE"},
{name: "GT"},
{name: "GE"},
[dev.ssa] cmd/compile: PPC64, FP to/from int conversions. Passes ssa_test. Requires a few new instructions and some scratchpad memory to move data between G and F registers. Also fixed comparisons to be correct in case of NaN. Added missing instructions for run.bash. Removed some FP registers that are apparently "reserved" (but that are also apparently also unused except for a gratuitous multiplication by two when y = x+x would work just as well). Currently failing stack splits. Updates #16010. Change-Id: I73b161bfff54445d72bd7b813b1479f89fc72602 Reviewed-on: https://go-review.googlesource.com/26813 Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
2016-08-10 11:44:57 -07:00
{name: "FLT"},
{name: "FLE"},
{name: "FGT"},
{name: "FGE"},
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
}
archs = append(archs, arch{
name: "PPC64",
pkg: "cmd/internal/obj/ppc64",
genfile: "../../ppc64/ssa.go",
ops: ops,
blocks: blocks,
regnames: regNamesPPC64,
gpregmask: gp,
fpregmask: fp,
framepointerreg: int8(num["SP"]),
cmd/compile: make link register allocatable in non-leaf functions We save and restore the link register in non-leaf functions because it is clobbered by CALLs. It is therefore available for general purpose use. Only enabled on s390x currently. The RC4 benchmarks in particular benefit from the extra register: name old speed new speed delta RC4_128 243MB/s ± 2% 341MB/s ± 2% +40.46% (p=0.008 n=5+5) RC4_1K 267MB/s ± 0% 359MB/s ± 1% +34.32% (p=0.008 n=5+5) RC4_8K 271MB/s ± 0% 362MB/s ± 0% +33.61% (p=0.008 n=5+5) Change-Id: Id23bff95e771da9425353da2f32668b8e34ba09f Reviewed-on: https://go-review.googlesource.com/30597 Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: Michael Munday <munday@ca.ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-10-06 15:06:45 -04:00
linkreg: -1, // not used
[dev.ssa] cmd/compile: Add initial SSA configuration for PPC64 This adds the initial SSA implementation for PPC64. Builds golang and all.bash runs correctly. Simple hello.go builds but does not run. Change-Id: I7cec211b934cd7a2dd75a6cdfaf9f71867063466 Reviewed-on: https://go-review.googlesource.com/24453 Reviewed-by: David Chase <drchase@google.com> Reviewed-by: Cherry Zhang <cherryyz@google.com> Run-TryBot: David Chase <drchase@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-06-24 14:37:17 -05:00
})
}
Reference in a new issue Copy permalink