For the tessellation op this allows it to use either the record-time
or flush-time allocator instead of making its own.
Bug: skia:10419
Change-Id: Ida13283feffda1976bd8fd221d5be0013a5de990
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/356516
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Change-Id: Ib150e6d6d3de34a85ce8051eea843ab3b2d7ab75
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/356921
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Also cleanup some of the duplicate code in SkRecords
Bug: skia:7650
Change-Id: I4d3167a892c126c19a54002beab25c9a6c96fa5d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/357000
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Rather than copying these triangles twice, we can run "pathToPolys"
during onPrePrepare and "polysToTriangles" during onPrepare.
Also adds a benchmark for normal and inner-fan triangulation.
Bug: skia:10419
Change-Id: Id301afde5de11d93ae026e75e42ac03a50867687
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/355177
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
This reduces complexity and will also allow us to add new ops that
take advantage of this same core logic.
Bug: skia:10419
Change-Id: I4ec8717a6d9510dea967d11467eeea0b5b7c7f4c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/354296
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
SkPaintPriv methods are just an internal stopgap
Change-Id: Ibe6e37c5871068d8cd67dc0948961444dfd2b62a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/347041
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Made it private and accessible internally via SkCanvasPriv.
Update SkGpuDevice methods/variables after rename of GrDrawSurfaceContext.
Cq-Include-Trybots: luci.skia.skia.primary:Canary-G3,Canary-Flutter,Canary-Android
Change-Id: I3da64cee1de03c201243ee6c7ccd4b4c44cad8c9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/346498
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
... and lots and lots of IWYU
Change-Id: Ie5157dcdd2e6d29b95c71b39153278ab48ef4eb3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/346778
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
This maps to usage better, and makes some code simpler to understand.
Note that there is still a PipelineStage *back-end*, which is specific
to the runtime-effect FP. A kRuntimeEffect_Kind program can be used to
generate a PipelineStage (for the GPU backend), or an skvm program (for
the CPU backend).
Change-Id: Id3f535db93a239726c595225aafe9467f0d19817
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/344969
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
None of these earliest testing tools are useful anymore
now that we can do useful work with SkVM and SkVMBlitter.
Change-Id: I8b25ef6ddd101c4ff8617c6742343dedb4764922
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/345456
Commit-Queue: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This is a reland of ad8b9fa2d7
Original change's description:
> Disable HW tessellation shaders by default
>
> The GLSL backdoor for tessellation shaders is only experimental at
> this point. In order to enable the tessellation path renderer, we need
> to turn off HW tessellation and let it use the indirect draw modes.
>
> Bug: skia:10419
> Change-Id: Ic979978a331c7ad016907cf42b2562b35caab8ab
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/344336
> Commit-Queue: Chris Dalton <csmartdalton@google.com>
> Reviewed-by: Brian Salomon <bsalomon@google.com>
TBR=bsalomon@google.com
Bug: skia:10419
Change-Id: I0bdf34da134a3a34a5c78b35cc0eb86d6e989bb4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/344796
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
No more
friend class ::SkArenaAlloc; // for access to ctor
Change-Id: I76fa3319498a965623e6865b75d1fb507ab845a6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/344236
Auto-Submit: Mike Klein <mtklein@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Just the class/files. variable names and additional comments to follow.
Change-Id: Ic03d07fd5009eaf3d706c2536486a117328963fc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342617
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Adds a vertex shader that maps a variable-length triangle strip to a
stroke and its preceding join. Adds a new op that generates stroke
instances from a path, bins them by log2 triangle strip length (using
SIMD for the calculations), and renders them with indirect draws.
Bug: skia:10419
Change-Id: I6d52df02cffe97d14827c6d66136957f1859f53b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/339716
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This was just wrong. It should be determine perspective
using the matrix that we are drawing with, and not the initial
matrix.
Change-Id: I8410ced714d2c766305656bdbd797f9dea59b71e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/339796
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Herb Derby <herb@google.com>
I think this is vestigial from some time in the past where RTC was
public.
Also just expose the methods that add ops rather than have so many
friends + testingOnly versions.
Change-Id: I60d9fdff23b2d67039a7b37815da7ff9e73d8999
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/339158
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Instead of using sub classing to generate the atlas API, give
the atlas portion its own interface. Then the atlas subrun
variants can subclass both interfaces.
Change-Id: I8a0ca3d19bd362877224fa64f6c49a5f50d0ceb5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336958
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
do not land
Change-Id: I5fa7b2a0d1eb7e893d9b333f850a2f515d7ce065
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336956
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
afaict the perf surprises associated with:
https://skia-review.googlesource.com/c/skia/+/334417 (Remove SkBaseDevice::flush)
were bc Ganesh resolves MSAA buffers for SkCanvas::flush but doesn't do so for GrDirectContext::flush.
Where possible this CL switches SkCanvas::flush to SkSurface::flush (which will also resolve MSAA buffers) so that when https://skia-review.googlesource.com/c/skia/+/334417 relands there should not be any performance surprises.
Change-Id: I705ad6219f0f625a88cf3f9e8b2418a3182d298c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335866
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Adlai Holler <adlai@google.com>
We were always pre-loading the fragment and vertex modules, but deferred
loading all others. Those two take up about 300 KB of heap. Now, all
modules are deferred, so compiler instances that don't need them (like
the one used for runtime effects) are much smaller.
Now that we can get better fine-grained numbers, added two more
benchmarks, to track actual baseline usage, plus the usage in the two
most likely configurations.
Change-Id: Idfbcd52c8afee566ac42ab827c80c940f91c4ad7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/337176
Commit-Queue: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
In follow on CLs we need to know what the load op is when we try to use
discardable msaa attachments. For vulkan the load op affects how we
copy the resolve attachment into the msaa attachment, which changes the
render pass we use (adds extra subpass). We need to be able to make a
compatible render pass to compile programs.
Bug: skia:10979
Change-Id: I40c23a18b251af6a2ad3b78a1f6382bdba0b90c4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336598
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Previously, any builtin functions would be optimized as a side-effect of
optimizing programs that used them. Now that shared elements aren't
being optimized in that way, we explicitly optimize any shared modules
when they are first created. We don't remove dead elements, but we
we do substitute settings, simplify, and inline.
Bug: skia:10905
Change-Id: I701b5e9f52fb880ef3e6f4c67694d08602f47e95
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336440
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
The important part is in GrOpFlushState.h were previously
we were taking a mutable pointer to the view, which should
at least be a const pointer and was making us do funky things
in some of the calling code. But I decided to go all the way
and do a const ref instead which is The Way It Should Be (tm).
Change-Id: I399d102e8b5e0a5059168cc450ae66f12ad47e13
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336451
Reviewed-by: Greg Daniel <egdaniel@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Adlai Holler <adlai@google.com>
This change will facilitate experimenting with direct-to-op
text draws. There should be no performance change because
almost all blobs are a single run.
Change-Id: I07fb3487e0601e00507403d94bc611c5022c1408
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336447
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
This is a reland of 26ad8ccdec
... now with MSAN support.
Original change's description:
> add ERMS (enhanced rep mov/sto) SkOpts slice
>
> Intel's got two CPUID bits indicating the speed of rep mov/sto
> (memcpy/memset),
>
> - ERMS, Enhanced Rep Mov/Sto, older, large copies are fast?
> - FSRM, Fast Short Rep Mov, newer, small copies are fast?
>
> ERMS has been around a long time on Intel, but is relatively recent on
> Ryzen, and FSRM is new across the board. The startup cost for
> ERMS-but-not-FSRM copies really is noticeable, so we cut over to the
> previous SSE/AVX routines when N is small.
>
> I've left the memset benchmarks as I found them most useful when
> tuning the small/large cutoff in this CL.
>
> Change-Id: I3ac4e3f34796aba0ea86aabbe9dda7526919456a
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332580
> Reviewed-by: Herb Derby <herb@google.com>
> Commit-Queue: Mike Klein <mtklein@google.com>
Cq-Include-Trybots: luci.skia.skia.primary:Test-Debian10-Clang-GCE-CPU-AVX2-x86_64-Release-All-MSAN
Change-Id: Ia293bba90022c48c884599331ef35aa67644729b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334343
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
So we don't incidentally measure the cost of the copy
Change-Id: I6f7e2119792423b0857764b3b557a9f838bb43fb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334423
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
This reverts commit 26ad8ccdec.
Reason for revert: gonna need to teach MSAN about this to reland.
Original change's description:
> add ERMS (enhanced rep mov/sto) SkOpts slice
>
> Intel's got two CPUID bits indicating the speed of rep mov/sto
> (memcpy/memset),
>
> - ERMS, Enhanced Rep Mov/Sto, older, large copies are fast?
> - FSRM, Fast Short Rep Mov, newer, small copies are fast?
>
> ERMS has been around a long time on Intel, but is relatively recent on
> Ryzen, and FSRM is new across the board. The startup cost for
> ERMS-but-not-FSRM copies really is noticeable, so we cut over to the
> previous SSE/AVX routines when N is small.
>
> I've left the memset benchmarks as I found them most useful when
> tuning the small/large cutoff in this CL.
>
> Change-Id: I3ac4e3f34796aba0ea86aabbe9dda7526919456a
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332580
> Reviewed-by: Herb Derby <herb@google.com>
> Commit-Queue: Mike Klein <mtklein@google.com>
TBR=mtklein@google.com,herb@google.com
Change-Id: I3264af132272dbbaac8fc8b62e139a6a112bbadb
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334342
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Intel's got two CPUID bits indicating the speed of rep mov/sto
(memcpy/memset),
- ERMS, Enhanced Rep Mov/Sto, older, large copies are fast?
- FSRM, Fast Short Rep Mov, newer, small copies are fast?
ERMS has been around a long time on Intel, but is relatively recent on
Ryzen, and FSRM is new across the board. The startup cost for
ERMS-but-not-FSRM copies really is noticeable, so we cut over to the
previous SSE/AVX routines when N is small.
I've left the memset benchmarks as I found them most useful when
tuning the small/large cutoff in this CL.
Change-Id: I3ac4e3f34796aba0ea86aabbe9dda7526919456a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332580
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Moves the "BenchmarkTarget" class from tessellation benchmarks into a
mock header where it can be reused by other tests and benchmarks.
Change-Id: I344d9ba3d391ff99e10c4ab238684b0b6ada87d0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333220
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
There's no reason to use std::aligned_storage when it's simpler to use
an array and alignas(). This way you don't have to remember whether the
template arguments are size-then-align or align-then-size, you don't
have to remember to use the _t variant or typename ... ::type, and
there's no risk to forgetting the alignment parameter entirely.
It doesn't look like this was deprecated, but I still think this paper
makes good arguments for why we shouldn't use it:
http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2019/p1413r1.pdf
Bug: skia:10921
Change-Id: Ia64a2e43c4cba9b4d64138a7474e353a8eaf01a6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333258
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 1277971939.
Change-Id: I7985ef22ddd19adcab468acc684b330ce6978c8d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332738
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
This reverts commit a2d6b31f66.
Reason for revert: breaking bots
Original change's description:
> Additional SkSL benches
>
> This adds additional benchmarks to give us better insight into how long
> the various compilation phases take. The four benches per size now
> cover just parsing, parsing + converting to IR, parsing + converting to
> IR + optimizing, and finally parsing + converting to IR + optimizing +
> generating GLSL.
>
> Change-Id: I7099a5a9f40ae5031e330dc4e1bb08c2a20ada63
> Bug: skia:10805
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332262
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com
Change-Id: Idb3e3082d11f72978131068b2229ce2578d645c3
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:10805
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332601
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
This adds additional benchmarks to give us better insight into how long
the various compilation phases take. The four benches per size now
cover just parsing, parsing + converting to IR, parsing + converting to
IR + optimizing, and finally parsing + converting to IR + optimizing +
generating GLSL.
Change-Id: I7099a5a9f40ae5031e330dc4e1bb08c2a20ada63
Bug: skia:10805
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332262
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
The previous behavior leaked Skia-internal concepts into public SkSL.
Users coming from GLSL will expect that bindable/sampleable objects are
uniform (just like texture2D). This keeps the old support around (and
tested), but updates all of our examples to use 'uniform'.
Bug: skia:10679
Change-Id: I0c98162f5e21dad7014d9778ceb26143d2f6030e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332376
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
This ties the caps to the compiler instance, paving the way for
pre-optimizing the shared code. Most of the time, the compiler is
created and owned the GPU instance, so this is fine. For runtime
effects, we now use the shared (device-agnostic) compiler instance
for the first compile, even on GPU. It's configured with caps that
apply no workarounds. We pass the user's SkSL to the backend as
cleanly as possible, and then apply any workarounds once it's part
of the full program.
Bug: skia:10905
Bug: skia:10868
Change-Id: Ifcf8d7ebda5d43ad8e180f06700a261811da83de
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/331493
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
There's really no reason for these classes to be separate at this
point. Also extracts a "GrStrokeOp" base class that has the
functionality that will be shared with indirect stroking.
Bug: skia:10419
Change-Id: I960d5e6d64f0814ccb4a3852bc627af2b8082a1f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/331860
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
This method finds the locations a cubic needs to be chopped at before
it can be passed to the stroke tessellation shader. It's an integral
part of CPU stroke preparation and therefore extremely perf sensitive.
Bug: skia:10419
Change-Id: Ib23c2583b8cfc78814ce52425f7af2c8b2f8b420
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/330314
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
With the addition of vertex data to the thread safe cache we also have
to handle the case where a given SkPath becomes inaccessible and
proactively invalidate the matching entry.
Bug: 1108408
Change-Id: Id11ce2aa10517f7c0772a253634d3c0d13e13460
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/330261
Reviewed-by: Adlai Holler <adlai@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
