Commit graph

8 commits

Author SHA1 Message Date
Russ Cox
17eba6e6b7 cmd/compile, cmd/link: create from 5g, 5l, etc
Trivial merging of 5g, 6g, ... into go tool compile,
and similarlly 5l, 6l, ... into go tool link.
The files compile/main.go and link/main.go are new.
Everything else in those directories is a move followed by
change of imports and package name.

This CL breaks the build. Manual fixups are in the next CL.

See golang-dev thread titled "go tool compile, etc" for background.

Change-Id: Id35ff5a5859ad9037c61275d637b1bd51df6828b
Reviewed-on: https://go-review.googlesource.com/10287
Reviewed-by: Dave Cheney <dave@cheney.net>
Reviewed-by: Rob Pike <r@golang.org>
2015-05-21 17:31:51 +00:00
Russ Cox
653d56075d cmd/internal/gc: inline writeBarrierEnabled check before calling writebarrierptr
I believe the benchmarks that get slower are under register pressure,
and not making the call unconditionally makes the pressure worse,
and the register allocator doesn't do a great job. But part of the point
of this sequence is to get the write barriers out of the way so I can work
on the register allocator, so that's okay.

name                                       old                     new          delta
BenchmarkBinaryTree17              17.9s × (1.00,1.01)     18.0s × (0.99,1.01)  ~
BenchmarkFannkuch11                4.43s × (1.00,1.00)     4.43s × (1.00,1.00)  ~
BenchmarkFmtFprintfEmpty           110ns × (1.00,1.06)     114ns × (0.95,1.05)  ~
BenchmarkFmtFprintfString          487ns × (0.99,1.00)     468ns × (0.99,1.01)  -4.00%
BenchmarkFmtFprintfInt             450ns × (0.99,1.00)     433ns × (1.00,1.01)  -3.88%
BenchmarkFmtFprintfIntInt          762ns × (1.00,1.00)     748ns × (0.99,1.01)  -1.84%
BenchmarkFmtFprintfPrefixedInt     584ns × (0.99,1.01)     547ns × (0.99,1.01)  -6.26%
BenchmarkFmtFprintfFloat           738ns × (1.00,1.00)     756ns × (1.00,1.01)  +2.37%
BenchmarkFmtManyArgs              2.80µs × (1.00,1.01)    2.79µs × (1.00,1.01)  ~
BenchmarkGobDecode                39.0ms × (0.99,1.00)    39.6ms × (0.99,1.00)  +1.54%
BenchmarkGobEncode                37.8ms × (0.98,1.01)    37.6ms × (1.00,1.01)  ~
BenchmarkGzip                      661ms × (0.99,1.01)     663ms × (0.99,1.02)  ~
BenchmarkGunzip                    142ms × (1.00,1.00)     142ms × (1.00,1.00)  ~
BenchmarkHTTPClientServer          132µs × (0.99,1.01)     132µs × (0.99,1.01)  ~
BenchmarkJSONEncode               56.3ms × (0.99,1.01)    56.2ms × (0.99,1.01)  ~
BenchmarkJSONDecode                138ms × (0.99,1.01)     138ms × (1.00,1.00)  ~
BenchmarkMandelbrot200            6.01ms × (1.00,1.00)    6.03ms × (1.00,1.01)  +0.23%
BenchmarkGoParse                  10.2ms × (0.87,1.05)     9.8ms × (0.93,1.10)  ~
BenchmarkRegexpMatchEasy0_32       208ns × (1.00,1.00)     207ns × (1.00,1.00)  ~
BenchmarkRegexpMatchEasy0_1K       588ns × (1.00,1.00)     581ns × (1.00,1.01)  -1.27%
BenchmarkRegexpMatchEasy1_32       182ns × (0.99,1.01)     185ns × (0.99,1.01)  +1.65%
BenchmarkRegexpMatchEasy1_1K       986ns × (1.00,1.01)     975ns × (1.00,1.01)  -1.17%
BenchmarkRegexpMatchMedium_32      323ns × (1.00,1.01)     328ns × (0.99,1.00)  +1.55%
BenchmarkRegexpMatchMedium_1K     89.9µs × (1.00,1.00)    88.6µs × (1.00,1.01)  -1.38%
BenchmarkRegexpMatchHard_32       4.72µs × (0.95,1.01)    4.69µs × (0.95,1.03)  ~
BenchmarkRegexpMatchHard_1K        133µs × (1.00,1.01)     133µs × (1.00,1.01)  ~
BenchmarkRevcomp                   900ms × (1.00,1.05)     902ms × (0.99,1.05)  ~
BenchmarkTemplate                  168ms × (0.99,1.01)     174ms × (0.99,1.01)  +3.30%
BenchmarkTimeParse                 637ns × (1.00,1.00)     639ns × (1.00,1.00)  +0.31%
BenchmarkTimeFormat                738ns × (1.00,1.00)     736ns × (1.00,1.01)  ~

Change-Id: I03ce152852edec404538f6c20eb650fac82e2aa2
Reviewed-on: https://go-review.googlesource.com/9224
Reviewed-by: Austin Clements <austin@google.com>
2015-04-28 01:38:06 +00:00
Russ Cox
2050f57141 runtime: change unused argument in fat write barriers from pointer to scalar
The argument is unused, only present for alignment of the
following argument. The compiler today always passes a zero
but I'd rather not write anything there during the call sequence,
so mark it as a scalar so the garbage collector won't look at it.

As expected, no significant performance change.

name                                       old                     new          delta
BenchmarkBinaryTree17              17.9s × (0.99,1.00)     17.9s × (0.99,1.01)  ~
BenchmarkFannkuch11                4.35s × (1.00,1.00)     4.35s × (1.00,1.00)  ~
BenchmarkFmtFprintfEmpty           120ns × (0.94,1.05)     120ns × (0.93,1.06)  ~
BenchmarkFmtFprintfString          477ns × (1.00,1.00)     479ns × (0.99,1.00)  ~
BenchmarkFmtFprintfInt             450ns × (0.99,1.01)     452ns × (0.99,1.02)  ~
BenchmarkFmtFprintfIntInt          765ns × (0.99,1.01)     766ns × (0.99,1.01)  ~
BenchmarkFmtFprintfPrefixedInt     569ns × (0.99,1.01)     576ns × (0.98,1.01)  ~
BenchmarkFmtFprintfFloat           728ns × (1.00,1.00)     730ns × (1.00,1.01)  ~
BenchmarkFmtManyArgs              2.82µs × (0.99,1.01)    2.84µs × (0.99,1.00)  ~
BenchmarkGobDecode                39.1ms × (0.99,1.01)    39.3ms × (0.98,1.01)  ~
BenchmarkGobEncode                39.4ms × (0.99,1.01)    39.5ms × (0.99,1.01)  ~
BenchmarkGzip                      661ms × (0.99,1.01)     663ms × (1.00,1.01)  ~
BenchmarkGunzip                    143ms × (1.00,1.00)     143ms × (1.00,1.00)  ~
BenchmarkHTTPClientServer          133µs × (0.99,1.01)     132µs × (0.99,1.01)  ~
BenchmarkJSONEncode               57.3ms × (0.99,1.04)    57.4ms × (0.99,1.01)  ~
BenchmarkJSONDecode                139ms × (0.99,1.00)     139ms × (0.99,1.00)  ~
BenchmarkMandelbrot200            6.02ms × (1.00,1.00)    6.03ms × (1.00,1.00)  ~
BenchmarkGoParse                  9.72ms × (0.92,1.11)   10.31ms × (0.89,1.14)  ~
BenchmarkRegexpMatchEasy0_32       209ns × (1.00,1.01)     209ns × (1.00,1.00)  ~
BenchmarkRegexpMatchEasy0_1K       592ns × (0.99,1.00)     591ns × (0.99,1.00)  ~
BenchmarkRegexpMatchEasy1_32       183ns × (0.98,1.01)     184ns × (0.99,1.02)  ~
BenchmarkRegexpMatchEasy1_1K      1.01µs × (1.00,1.01)    1.01µs × (1.00,1.00)  ~
BenchmarkRegexpMatchMedium_32      330ns × (1.00,1.00)     330ns × (1.00,1.00)  ~
BenchmarkRegexpMatchMedium_1K     92.4µs × (1.00,1.00)    92.6µs × (1.00,1.00)  ~
BenchmarkRegexpMatchHard_32       4.77µs × (0.95,1.01)    4.80µs × (0.95,1.00)  ~
BenchmarkRegexpMatchHard_1K        136µs × (1.00,1.00)     136µs × (1.00,1.00)  ~
BenchmarkRevcomp                   906ms × (0.99,1.05)     900ms × (0.99,1.04)  ~
BenchmarkTemplate                  171ms × (0.99,1.01)     172ms × (1.00,1.00)  ~
BenchmarkTimeParse                 638ns × (1.00,1.00)     637ns × (1.00,1.00)  ~
BenchmarkTimeFormat                745ns × (0.99,1.02)     744ns × (1.00,1.01)  ~

Change-Id: I0aeac5dc7adfd75e2223e3aabfedc7818d339f9b
Reviewed-on: https://go-review.googlesource.com/9320
Reviewed-by: Austin Clements <austin@google.com>
2015-04-28 01:37:45 +00:00
David Chase
2270133981 cmd/gc: allocate backing storage for non-escaping interfaces on stack
Extend escape analysis to convT2E and conT2I. If the interface value
does not escape supply runtime with a stack buffer for the object copy.

This is a straight port from .c to .go of Dmitry's patch

Change-Id: Ic315dd50d144d94dd3324227099c116be5ca70b6
Reviewed-on: https://go-review.googlesource.com/8201
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
2015-03-30 16:11:22 +00:00
Russ Cox
4224d81fae cmd/internal/gc: inline x := y.(*T) and x, ok := y.(*T)
These can be implemented with just a compare and a move instruction.
Do so, avoiding the overhead of a call into the runtime.

These assertions are a significant cost in Go code that uses interface{}
as a safe alternative to C's void* (or unsafe.Pointer), such as the
current version of the Go compiler.

*T here includes pointer to T but also any Go type represented as
a single pointer (chan, func, map). It does not include [1]*T or struct{*int}.
That requires more work in other parts of the compiler; there is a TODO.

Change-Id: I7ff681c20d2c3eb6ad11dd7b3a37b1f3dda23965
Reviewed-on: https://go-review.googlesource.com/7862
Reviewed-by: Rob Pike <r@golang.org>
2015-03-20 20:05:37 +00:00
Matthew Dempsky
81d4072eb0 cmd/internal/gc, runtime: change growslice to use int instead of int64
Gc already calculates n as an int, so converting to int64 to call
growslice doesn't serve any purpose except to emit slightly larger
code on 32-bit platforms.  Passing n as an int shrinks godoc's text
segment by 8kB (9472633 => 9464133) when building for ARM.

Change-Id: Ief9492c21d01afcb624d3f2a484df741450b788d
Reviewed-on: https://go-review.googlesource.com/6231
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
2015-03-04 17:17:17 +00:00
Matthew Dempsky
d68f52aa43 cmd/internal/gc: restore mkbuiltin
Change-Id: I6cc4061fef9d3c3530406ab3d4811192b7175e89
Reviewed-on: https://go-review.googlesource.com/6233
Reviewed-by: Minux Ma <minux@golang.org>
2015-02-27 09:19:47 +00:00
Russ Cox
8c195bdf12 [dev.cc] cmd/internal/gc, cmd/new6g etc: convert from cmd/gc, cmd/6g etc
First draft of converted Go compiler, using rsc.io/c2go rev 83d795a.

Change-Id: I29f4c7010de07d2ff1947bbca9865879d83c32c3
Reviewed-on: https://go-review.googlesource.com/4851
Reviewed-by: Rob Pike <r@golang.org>
2015-02-17 23:28:51 +00:00