aboutsummaryrefslogtreecommitdiffhomepage
path: root/src/core/SkVarAlloc.h
Commit message (Collapse)AuthorAge
* Have SkVarAlloc::alloc() use sk_malloc_throw.Gravatar mtklein2015-09-14
| | | | | | | | Very right, it's not prepared to handle return-NULL mallocs at all. BUG=530759 Review URL: https://codereview.chromium.org/1339093002
* Use static_assert instead of SK_COMPILE_ASSERT.Gravatar bungeman2015-08-20
| | | | | | | Now that static_assert is allowed, there is no need to use a non- standard compile time assertion Review URL: https://codereview.chromium.org/1306443004
* O(1) SkPictureUtils::ApproxBytesUsed()Gravatar mtklein2015-04-21
| | | | | | | | | | Chrome wants to call this more often, and it's quite slow today. Seems like this could be clearer if SkPictureUtils::ApproxBytesUsed() were SkPicture::approxBytesUsed(). BUG=chromium:471873 Review URL: https://codereview.chromium.org/1090943004
* Rearrange SkRecord with small N in mindGravatar mtklein2015-04-09
| | | | | | | | | | | | | | This rearranges the record pointers and types so they can go in a single array, then preallocates some space for them and for the SkVarAlloc. picture_overhead_draw bench drops from ~1000ns to 500-600ns, with no effect on picture_overhead_nodraw. I don't see any significant effect on large picture recording times from our .skps. BUG=chromium:470553 Committed: https://skia.googlesource.com/skia/+/e2dd9408cd711777afaa9410427fb0d761ab004a Review URL: https://codereview.chromium.org/1061783002
* Revert of Rearrange SkRecord with small N in mind (patchset #8 id:120001 of ↵Gravatar mtklein2015-04-08
| | | | | | | | | | | | | | | | | | | | | | | | | | | | https://codereview.chromium.org/1061783002/) Reason for revert: https://uberchromegw.corp.google.com/i/client.skia/builders/Test-Ubuntu-GCC-GCE-CPU-AVX2-x86-Debug/builds/149/steps/dm/logs/stdio Original issue's description: > Rearrange SkRecord with small N in mind > > This rearranges the record pointers and types so they can go in a single array, then preallocates some space for them and for the SkVarAlloc. > > picture_overhead_draw bench drops from ~1000ns to 500-600ns, with no effect on picture_overhead_nodraw. > > I don't see any significant effect on large picture recording times from our .skps. > > BUG=chromium:470553 > > Committed: https://skia.googlesource.com/skia/+/e2dd9408cd711777afaa9410427fb0d761ab004a TBR=reed@google.com,mtklein@chromium.org NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG=chromium:470553 Review URL: https://codereview.chromium.org/1068383003
* Rearrange SkRecord with small N in mindGravatar mtklein2015-04-08
| | | | | | | | | | | | This rearranges the record pointers and types so they can go in a single array, then preallocates some space for them and for the SkVarAlloc. picture_overhead_draw bench drops from ~1000ns to 500-600ns, with no effect on picture_overhead_nodraw. I don't see any significant effect on large picture recording times from our .skps. BUG=chromium:470553 Review URL: https://codereview.chromium.org/1061783002
* SkRecord: increase min block to 512B, remove max.Gravatar mtklein2014-12-12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When we added the 64K allocation cap, the bots showed we took a perf hit on some large .skps like desk_pokemonwiki.skp, despite not seeing a local effect. I'm still not seeing that locally, but I'd like to try removing the cap on the bots to see what happens. For big monolithic pictures, really packing into memory tightly is probably not as important as it is for tiny ones. Similarly, we're probably being too cautious about making tiny allocations. Today we start at 16 bytes, which isn't really enough to record anything. Even the smallest picture, say, save clipRect drawRect restore requires ~200 bytes, so we might as well move our minimum block size up near there. I don't know if 16 bytes is too small to start for GrTextStrikes, so I've left the behavior the same (though the max is still gone). Local recording performance is neutral-to-positive: tabl_deviantart.skp 126us -> 129us 1.02x tabl_nytimes.skp 110us -> 112us 1.02x tabl_cuteoverload.skp 521us -> 530us 1.02x desk_mobilenews.skp 673us -> 682us 1.01x desk_chalkboard.skp 843us -> 854us 1.01x desk_sfgate.skp 528us -> 535us 1.01x desk_silkfinance.skp 68.2us -> 69us 1.01x desk_youtube.skp 623us -> 629us 1.01x desk_blogger.skp 472us -> 475us 1.01x desk_jsfiddlehumperclip.skp 42.2us -> 42.5us 1.01x desk_espn.skp 255us -> 256us 1.01x desk_ebay.skp 174us -> 174us 1x desk_twitter.skp 454us -> 455us 1x tabl_pravda.skp 200us -> 201us 1x desk_wordpress.skp 782us -> 784us 1x desk_samoasvg.skp 762us -> 761us 1x tabl_mozilla.skp 1.58ms -> 1.58ms 1x tabl_slashdot.skp 107us -> 107us 1x tabl_techmeme.skp 102us -> 102us 0.99x tabl_gamedeksiam.skp 729us -> 724us 0.99x tabl_nofolo.skp 65.3us -> 64.7us 0.99x desk_gmailthread.skp 339us -> 336us 0.99x tabl_sahadan.skp 91us -> 90us 0.99x desk_yahooanswers.skp 144us -> 142us 0.99x tabl_cnet.skp 143us -> 141us 0.99x tabl_googleblog.skp 206us -> 203us 0.99x tabl_cnn.skp 160us -> 158us 0.99x tabl_frantzen.skp 50.5us -> 49.6us 0.98x desk_linkedin.skp 328us -> 323us 0.98x tabl_digg.skp 790us -> 769us 0.97x desk_jsfiddlebigcar.skp 40.6us -> 39.5us 0.97x desk_mapsvg.skp 1.57ms -> 1.52ms 0.97x tabl_gmail.skp 19.4us -> 18.6us 0.96x tabl_hsfi.skp 9.81us -> 9.11us 0.93x BUG=skia: Review URL: https://codereview.chromium.org/793033002
* Fix some win64 warnings.Gravatar bsalomon2014-12-11
| | | | Review URL: https://codereview.chromium.org/795223002
* Add SkNVRefCnt, prune down SkPicture's sizeGravatar mtklein2014-11-20
| | | | | | | | | | | | | | SkNVRefCnt is a variant of SkRefCnt that's Not Virtual, so weighs 4 bytes instead of 8 or 16. There's only benefit to doing this if the deriving class does not otherwise need a vtable, e.g. SkPicture. I've stripped out some cruft from SkPicture, rearranged fields to pack tightly, and added compile asserts for the sizes of SkPicture, SkRecord, and SkVarAlloc. BUG=skia:3144 Review URL: https://codereview.chromium.org/741793002
* SkVarAlloc::approxBytesAllocated()Gravatar mtklein2014-11-18
| | | | | | | | | | | | This is what I was getting at on the other CL. BUG=skia: Committed: https://skia.googlesource.com/skia/+/f27f1bcce50c8f95aea8469684a70b70c9baee09 CQ_EXTRA_TRYBOTS=client.skia.android:Test-Android-Nexus5-Adreno330-Arm7-Release-Trybot Review URL: https://codereview.chromium.org/730193003
* Revert of SkVarAlloc::approxBytesAllocated (patchset #5 id:80001 of ↵Gravatar mtklein2014-11-18
| | | | | | | | | | | | | | | | | | | | | | | | | https://codereview.chromium.org/730193003/) Reason for revert: Android needs dlmalloc_usable_size(). Original issue's description: > SkVarAlloc::approxBytesAllocated() > > This is what I was getting at on the other CL. > > BUG=skia: > > Committed: https://skia.googlesource.com/skia/+/f27f1bcce50c8f95aea8469684a70b70c9baee09 > > CQ_EXTRA_TRYBOTS=Test-Android-Nexus5-Adreno330-Arm7-Release-Trybot TBR=reed@google.com,mtklein@chromium.org NOTREECHECKS=true NOTRY=true BUG=skia: Review URL: https://codereview.chromium.org/741443002
* SkVarAlloc::approxBytesAllocated()Gravatar mtklein2014-11-18
| | | | | | | | This is what I was getting at on the other CL. BUG=skia: Review URL: https://codereview.chromium.org/730193003
* Cap SkVarAlloc's desired block at 64K.Gravatar mtklein2014-11-13
| | | | | | | | | | | This means we can store fLgMinSize in 4 bits (TBD). Local perf comparison calls this harmless-to-slightly-helpful. Nothing to get excited about, but seems to certainly not harm perf. BUG=skia: Review URL: https://codereview.chromium.org/722293003
* Deparameterize SkVarAlloc.Gravatar mtklein2014-11-13
| | | | | | | | | | | | | | | | | | | | | | | | SkRecord performance is not sensitive to these values, so we can cut some storage. This rearranges the space-remaining logic to use a count of bytes left rather than a pointer to the end, cutting memory usage a little more. An SkVarAlloc used to weigh 20 or 32 bytes which now becomes 16 or 24. I think if I think about it a bit more I can trim off that Block* too, getting us to 12 or 16 bytes. Because we now just always grow by doubling, this CL switches from storing fSmallest to its log base 2. This has the nice effect of never having to worry about it overflowing, and means we can probably squeeze it down into a byte if we want, even 6 bits. BUG=skia: Committed: https://skia.googlesource.com/skia/+/bc415389855888af5a1282ca4b6bee30afa3d69d CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu12-ShuttleA-GTX660-x86-Debug-Trybot Review URL: https://codereview.chromium.org/721313002
* Revert of Deparameterize SkVarAlloc. (patchset #6 id:100001 of ↵Gravatar mtklein2014-11-13
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | https://codereview.chromium.org/721313002/) Reason for revert: Unit test failures on 32-bit machines. test Record_Alignment: ../../tests/RecordTest.cpp:100 is_aligned(record.alloc<uint64_t>()) Original issue's description: > Deparameterize SkVarAlloc. > > SkRecord performance is not sensitive to these values, so we can cut some > storage. This rearranges the space-remaining logic to use a count of bytes > left rather than a pointer to the end, cutting memory usage a little more. > > An SkVarAlloc used to weigh 20 or 32 bytes which now becomes 16 or 24. > > I think if I think about it a bit more I can trim off that Block* too, > getting us to 12 or 16 bytes. > > Because we now just always grow by doubling, this CL switches from storing > fSmallest to its log base 2. This has the nice effect of never having to worry > about it overflowing, and means we can probably squeeze it down into a byte > if we want, even 6 bits. > > BUG=skia: > > Committed: https://skia.googlesource.com/skia/+/bc415389855888af5a1282ca4b6bee30afa3d69d TBR=reed@google.com,mtklein@chromium.org NOTREECHECKS=true NOTRY=true BUG=skia: Review URL: https://codereview.chromium.org/718203006
* Deparameterize SkVarAlloc.Gravatar mtklein2014-11-13
| | | | | | | | | | | | | | | | | | | | SkRecord performance is not sensitive to these values, so we can cut some storage. This rearranges the space-remaining logic to use a count of bytes left rather than a pointer to the end, cutting memory usage a little more. An SkVarAlloc used to weigh 20 or 32 bytes which now becomes 16 or 24. I think if I think about it a bit more I can trim off that Block* too, getting us to 12 or 16 bytes. Because we now just always grow by doubling, this CL switches from storing fSmallest to its log base 2. This has the nice effect of never having to worry about it overflowing, and means we can probably squeeze it down into a byte if we want, even 6 bits. BUG=skia: Review URL: https://codereview.chromium.org/721313002
* SkVarAllocGravatar mtklein2014-11-12
Like SkChunkAlloc, but - does its allocation with better sympathy for malloc granularity; - the fast path inlines entirely; - smaller per-block overhead; - smaller per-SkVarAlloc overhead; - growth parameters are a little more tunable. Its main downside is less flexibility; it supports fewer methods than SkChunkAlloc. These current parameters bring the first allocation down from 4K to 1K, without affecting recording time on my desktop. skiaperf.com will tell the whole story. BUG=skia: Review URL: https://codereview.chromium.org/674263002
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-04 21:13:49 +0000