Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(46)

Issue 14122002: Use rep_movsb for copying one-byte strings in substring intrinsic. (Closed)

Created:
7 years, 8 months ago by srdjan
Modified:
7 years, 8 months ago
Reviewers:
Florian Schneider
CC:
reviews_dartlang.org, Vyacheslav Egorov (Google)
Visibility:
Public.

Description

Use rep_movsb for copying one-byte strings in substring intrinsic. Committed: https://code.google.com/p/dart/source/detail?r=21293

Patch Set 1 #

Patch Set 2 : #

Patch Set 3 : #

Unified diffs Side-by-side diffs Delta from patch set Stats (+105 lines, -43 lines) Patch
M runtime/vm/assembler_ia32.h View 1 2 1 chunk +2 lines, -0 lines 0 comments Download
M runtime/vm/assembler_ia32.cc View 1 2 1 chunk +7 lines, -0 lines 0 comments Download
M runtime/vm/assembler_ia32_test.cc View 1 2 2 chunks +29 lines, -1 line 0 comments Download
M runtime/vm/assembler_x64.h View 1 2 1 chunk +2 lines, -0 lines 0 comments Download
M runtime/vm/assembler_x64.cc View 1 2 1 chunk +7 lines, -0 lines 0 comments Download
M runtime/vm/assembler_x64_test.cc View 1 2 1 chunk +30 lines, -0 lines 0 comments Download
M runtime/vm/disassembler_ia32.cc View 1 2 1 chunk +5 lines, -0 lines 0 comments Download
M runtime/vm/intrinsifier_ia32.cc View 1 2 2 chunks +12 lines, -20 lines 0 comments Download
M runtime/vm/intrinsifier_x64.cc View 1 2 1 chunk +11 lines, -22 lines 0 comments Download

Messages

Total messages: 5 (0 generated)
srdjan
7 years, 8 months ago (2013-04-11 00:43:45 UTC) #1
Florian Schneider
LGTM.
7 years, 8 months ago (2013-04-11 09:04:02 UTC) #2
srdjan
Committed patchset #3 manually as r21293 (presubmit successful).
7 years, 8 months ago (2013-04-11 16:39:43 UTC) #3
srdjan
Unfortunately this CL has a >10% regression for Json benchmarks (it did not have it ...
7 years, 8 months ago (2013-04-11 18:35:03 UTC) #4
Florian Schneider
7 years, 8 months ago (2013-04-12 08:14:45 UTC) #5
Message was sent while issue was closed.
On 2013/04/11 18:35:03, srdjan wrote:
> Unfortunately  this CL has a >10% regression for Json benchmarks (it did not
> have it on my Mac).  I will either revert to loop or will specialize for
> lengths/architectures. Any suggestions?

Ok, that is surprising. It would be interesting to see what role the length of
strings plays. The ramp-up cost of rep movsb may be too high for small trip
counts. Maybe let's leave the normal loop for now.

Powered by Google App Engine
This is Rietveld 408576698