The following performance improvements have been made to the
low-level atomic functions for ppc64le & ppc64:
- For those cases containing a lwarx and stwcx (or other sizes):
sync, lwarx, maybe something, stwcx, loop to sync, sync, isync
The sync is moved before (outside) the lwarx/stwcx loop, and the
sync after is removed, so it becomes:
sync, lwarx, maybe something, stwcx, loop to lwarx, isync
- For the Or8 and And8, the shifting and manipulation of the
address to the word aligned version were removed and the
instructions were changed to use lbarx, stbcx instead of
register shifting, xor, then lwarx, stwcx.
- New instructions LWSYNC, LBAR, STBCC were tested and added.
runtime/atomic_ppc64x.s was changed to use the LWSYNC opcode
instead of the WORD encoding.
Fixes#15469
Ran some of the benchmarks in the runtime and sync directories.
Some results varied from run to run but the trend was improvement
based on best times for base and new:
runtime.test:
BenchmarkChanNonblocking-128 0.88 0.89 +1.14%
BenchmarkChanUncontended-128 569 511 -10.19%
BenchmarkChanContended-128 63110 53231 -15.65%
BenchmarkChanSync-128 691 598 -13.46%
BenchmarkChanSyncWork-128 11355 11649 +2.59%
BenchmarkChanProdCons0-128 2402 2090 -12.99%
BenchmarkChanProdCons10-128 1348 1363 +1.11%
BenchmarkChanProdCons100-128 1002 746 -25.55%
BenchmarkChanProdConsWork0-128 2554 2720 +6.50%
BenchmarkChanProdConsWork10-128 1909 1804 -5.50%
BenchmarkChanProdConsWork100-128 1624 1580 -2.71%
BenchmarkChanCreation-128 237 212 -10.55%
BenchmarkChanSem-128 705 667 -5.39%
BenchmarkChanPopular-128 5081190 4497566 -11.49%
BenchmarkCreateGoroutines-128 532 473 -11.09%
BenchmarkCreateGoroutinesParallel-128 35.0 34.7 -0.86%
BenchmarkCreateGoroutinesCapture-128 4923 4200 -14.69%
sync.test:
BenchmarkUncontendedSemaphore-128 112 94.2 -15.89%
BenchmarkContendedSemaphore-128 133 128 -3.76%
BenchmarkMutexUncontended-128 1.90 1.67 -12.11%
BenchmarkMutex-128 353 310 -12.18%
BenchmarkMutexSlack-128 304 283 -6.91%
BenchmarkMutexWork-128 554 541 -2.35%
BenchmarkMutexWorkSlack-128 567 556 -1.94%
BenchmarkMutexNoSpin-128 275 242 -12.00%
BenchmarkMutexSpin-128 1129 1030 -8.77%
BenchmarkOnce-128 1.08 0.96 -11.11%
BenchmarkPool-128 29.8 27.4 -8.05%
BenchmarkPoolOverflow-128 40564 36583 -9.81%
BenchmarkSemaUncontended-128 3.14 2.63 -16.24%
BenchmarkSemaSyntNonblock-128 1087 1069 -1.66%
BenchmarkSemaSyntBlock-128 897 893 -0.45%
BenchmarkSemaWorkNonblock-128 1034 1028 -0.58%
BenchmarkSemaWorkBlock-128 949 886 -6.64%
Change-Id: I4403fb29d3cd5254b7b1ce87a216bd11b391079e
Reviewed-on: https://go-review.googlesource.com/22549
Reviewed-by: Michael Munday <munday@ca.ibm.com>
Reviewed-by: Minux Ma <minux@golang.org>
This change breaks out most of the atomics functions in the runtime
into package runtime/internal/atomic. It adds some basic support
in the toolchain for runtime packages, and also modifies linux/arm
atomics to remove the dependency on the runtime's mutex. The mutexes
have been replaced with spinlocks.
all trybots are happy!
In addition to the trybots, I've tested on the darwin/arm64 builder,
on the darwin/arm builder, and on a ppc64le machine.
Change-Id: I6698c8e3cf3834f55ce5824059f44d00dc8e3c2f
Reviewed-on: https://go-review.googlesource.com/14204
Run-TryBot: Michael Matloob <matloob@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
Replace the confusing game where a frame size of $-8 would suppress the
implicit setting up of a stack frame with a nice explicit flag.
The code to set up the function prologue is still a little confusing but better
than it was.
Change-Id: I1d49278ff42c6bc734ebfb079998b32bc53f8d9a
Reviewed-on: https://go-review.googlesource.com/15670
Reviewed-by: Minux Ma <minux@golang.org>
Currently its possible for the garbage collector to observe
uninitialized memory or stale heap bitmap bits on weakly ordered
architectures such as ARM and PPC. On such architectures, the stores
that zero newly allocated memory and initialize its heap bitmap may
move after a store in user code that makes the allocated object
observable by the garbage collector.
To fix this, add a "publication barrier" (also known as an "export
barrier") before returning from mallocgc. This is a store/store
barrier that ensures any write done by user code that makes the
returned object observable to the garbage collector will be ordered
after the initialization performed by mallocgc. No barrier is
necessary on the reading side because of the data dependency between
loading the pointer and loading the contents of the object.
Fixes one of the issues raised in #9984.
Change-Id: Ia3d96ad9c5fc7f4d342f5e05ec0ceae700cd17c8
Reviewed-on: https://go-review.googlesource.com/11083
Reviewed-by: Rick Hudson <rlh@golang.org>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Minux Ma <minux@golang.org>
Reviewed-by: Martin Capitanio <capnm9@gmail.com>
Reviewed-by: Russ Cox <rsc@golang.org>
All of the architectures except ppc64 have only "RET" for the return
mnemonic. ppc64 used to have only "RETURN", but commit cf06ea6
introduced RET as a synonym for RETURN to make ppc64 consistent with
the other architectures. However, that commit was never followed up to
make the code itself consistent by eliminating uses of RETURN.
This commit replaces all uses of RETURN in the ppc64 assembly with
RET.
This was done with
sed -i 's/\<RETURN\>/RET/' **/*_ppc64x.s
plus one manual change to syscall/asm.s.
Change-Id: I3f6c8d2be157df8841d48de988ee43f3e3087995
Reviewed-on: https://go-review.googlesource.com/10672
Reviewed-by: Rob Pike <r@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
Reviewed-by: Minux Ma <minux@golang.org>
References to FP must now have a symbol.
Change-Id: I3f06b99cc48cbd4ccd6f23f2e4b0830af40f7f3d
Reviewed-on: https://go-review.googlesource.com/5281
Reviewed-by: Russ Cox <rsc@golang.org>
