This is a reland of 78cb579f33
This time, lowp stages are controlled by !defined(JUMPER_IS_SCALAR), not
by defined(__clang__). The two are usually the same, except when we opt
Clang builds into JUMPER_IS_SCALAR artificially.
Some Google3 builds use compilers old enough that they barf when
compiling our NEON code. It's conceivably also possible to define
JUMPER_IS_SCALAR yourself, but I don't think anyone does that.
Original change's description:
> Reland "make SkJumper stages normal Skia code"
>
> This is a reland of 22e536e3a1
>
> Now with fixed #include paths in SkRasterPipeline_opts.h,
> and -ffp-contract=fast for the :hsw target to minimize
> diffs on non-Windows Clang AVX2/AVX-512 bots.
>
> Original change's description:
> > make SkJumper stages normal Skia code
> >
> > Enough clients are using Clang now that we can say, use Clang to build
> > if you want these software pipeline stages to go fast.
> >
> > This lets us drop the offline build aspect of SkJumper stages, instead
> > building as part of Skia using the SkOpts framework.
> >
> > I think everything should work, except I've (temporarily) removed
> > AVX-512 support. I will put this back in a follow up.
> >
> > I have had to drop Windows down to __vectorcall and our narrower
> > stage calling convention that keeps the d-registers on the stack.
> > I tried forcing sysv_abi, but that crashed Clang. :/
> >
> > Added a TODO to up the same narrower stage calling convention
> > for lowp stages... we just *don't* today, for no good reason.
> >
> > Change-Id: Iaaa792ffe4deab3508d2dc5d0008c163c24b3383
> > Reviewed-on: https://skia-review.googlesource.com/110641
> > Commit-Queue: Mike Klein <mtklein@chromium.org>
> > Reviewed-by: Herb Derby <herb@google.com>
> > Reviewed-by: Florin Malita <fmalita@chromium.org>
>
> Change-Id: I44f2c03d33958e3807747e40904b6351957dd448
> Reviewed-on: https://skia-review.googlesource.com/112742
> Reviewed-by: Mike Klein <mtklein@chromium.org>
Change-Id: I3d71197d4bbb19ca4a94961a97fa2e54d5cbfb0d
Reviewed-on: https://skia-review.googlesource.com/112744
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 78cb579f33.
Reason for revert: lowp should be controlled by defined(JUMPER_IS_SCALAR), not defined(__clang__). So close.
Original change's description:
> Reland "make SkJumper stages normal Skia code"
>
> This is a reland of 22e536e3a1
>
> Now with fixed #include paths in SkRasterPipeline_opts.h,
> and -ffp-contract=fast for the :hsw target to minimize
> diffs on non-Windows Clang AVX2/AVX-512 bots.
>
> Original change's description:
> > make SkJumper stages normal Skia code
> >
> > Enough clients are using Clang now that we can say, use Clang to build
> > if you want these software pipeline stages to go fast.
> >
> > This lets us drop the offline build aspect of SkJumper stages, instead
> > building as part of Skia using the SkOpts framework.
> >
> > I think everything should work, except I've (temporarily) removed
> > AVX-512 support. I will put this back in a follow up.
> >
> > I have had to drop Windows down to __vectorcall and our narrower
> > stage calling convention that keeps the d-registers on the stack.
> > I tried forcing sysv_abi, but that crashed Clang. :/
> >
> > Added a TODO to up the same narrower stage calling convention
> > for lowp stages... we just *don't* today, for no good reason.
> >
> > Change-Id: Iaaa792ffe4deab3508d2dc5d0008c163c24b3383
> > Reviewed-on: https://skia-review.googlesource.com/110641
> > Commit-Queue: Mike Klein <mtklein@chromium.org>
> > Reviewed-by: Herb Derby <herb@google.com>
> > Reviewed-by: Florin Malita <fmalita@chromium.org>
>
> Change-Id: I44f2c03d33958e3807747e40904b6351957dd448
> Reviewed-on: https://skia-review.googlesource.com/112742
> Reviewed-by: Mike Klein <mtklein@chromium.org>
TBR=mtklein@chromium.org,herb@google.com,fmalita@chromium.org
Change-Id: Ie64da98f5187d44e03c0ce05d7cb189d4a6e6663
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/112743
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This is a reland of 22e536e3a1
Now with fixed #include paths in SkRasterPipeline_opts.h,
and -ffp-contract=fast for the :hsw target to minimize
diffs on non-Windows Clang AVX2/AVX-512 bots.
Original change's description:
> make SkJumper stages normal Skia code
>
> Enough clients are using Clang now that we can say, use Clang to build
> if you want these software pipeline stages to go fast.
>
> This lets us drop the offline build aspect of SkJumper stages, instead
> building as part of Skia using the SkOpts framework.
>
> I think everything should work, except I've (temporarily) removed
> AVX-512 support. I will put this back in a follow up.
>
> I have had to drop Windows down to __vectorcall and our narrower
> stage calling convention that keeps the d-registers on the stack.
> I tried forcing sysv_abi, but that crashed Clang. :/
>
> Added a TODO to up the same narrower stage calling convention
> for lowp stages... we just *don't* today, for no good reason.
>
> Change-Id: Iaaa792ffe4deab3508d2dc5d0008c163c24b3383
> Reviewed-on: https://skia-review.googlesource.com/110641
> Commit-Queue: Mike Klein <mtklein@chromium.org>
> Reviewed-by: Herb Derby <herb@google.com>
> Reviewed-by: Florin Malita <fmalita@chromium.org>
Change-Id: I44f2c03d33958e3807747e40904b6351957dd448
Reviewed-on: https://skia-review.googlesource.com/112742
Reviewed-by: Mike Klein <mtklein@chromium.org>
This reverts commit 22e536e3a1.
Reason for revert: wrong include path :/
Original change's description:
> make SkJumper stages normal Skia code
>
> Enough clients are using Clang now that we can say, use Clang to build
> if you want these software pipeline stages to go fast.
>
> This lets us drop the offline build aspect of SkJumper stages, instead
> building as part of Skia using the SkOpts framework.
>
> I think everything should work, except I've (temporarily) removed
> AVX-512 support. I will put this back in a follow up.
>
> I have had to drop Windows down to __vectorcall and our narrower
> stage calling convention that keeps the d-registers on the stack.
> I tried forcing sysv_abi, but that crashed Clang. :/
>
> Added a TODO to up the same narrower stage calling convention
> for lowp stages... we just *don't* today, for no good reason.
>
> Change-Id: Iaaa792ffe4deab3508d2dc5d0008c163c24b3383
> Reviewed-on: https://skia-review.googlesource.com/110641
> Commit-Queue: Mike Klein <mtklein@chromium.org>
> Reviewed-by: Herb Derby <herb@google.com>
> Reviewed-by: Florin Malita <fmalita@chromium.org>
TBR=mtklein@chromium.org,herb@google.com,fmalita@chromium.org
Change-Id: I2bdc709c80cdfa6b13ff24e024b3721bef887f46
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/112741
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Enough clients are using Clang now that we can say, use Clang to build
if you want these software pipeline stages to go fast.
This lets us drop the offline build aspect of SkJumper stages, instead
building as part of Skia using the SkOpts framework.
I think everything should work, except I've (temporarily) removed
AVX-512 support. I will put this back in a follow up.
I have had to drop Windows down to __vectorcall and our narrower
stage calling convention that keeps the d-registers on the stack.
I tried forcing sysv_abi, but that crashed Clang. :/
Added a TODO to up the same narrower stage calling convention
for lowp stages... we just *don't* today, for no good reason.
Change-Id: Iaaa792ffe4deab3508d2dc5d0008c163c24b3383
Reviewed-on: https://skia-review.googlesource.com/110641
Commit-Queue: Mike Klein <mtklein@chromium.org>
Reviewed-by: Herb Derby <herb@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
CQ_INCLUDE_TRYBOTS=skia.primary:Test-Debian9-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD
Bug: skia:
Change-Id: If6f0d0a57463bf99a66d674e65a62ce3931d0116
Reviewed-on: https://skia-review.googlesource.com/24644
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
There are only a couple of these, primarily focused on index8 srcs and 565 dsts. The burden's starting to outweigh the benefit. No one on the team knows MIPS assembly.
If we're going to try this again, I'd rather we try some sort of SkNx / compiler-intrinsic based approach, probably targeting MIPS SIMD (MSA), not this older instruction set.
We already ignore these files for 64-bit MIPS. This just closes the loop on 32-bit MIPS.
CQ_INCLUDE_TRYBOTS=skia.primary:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD,Build-Ubuntu-Clang-mips64el-Debug-Android,Build-Ubuntu-Clang-mips64el-Release-Android,Build-Ubuntu-Clang-mipsel-Debug-Android,Build-Ubuntu-Clang-mipsel-Release-Android
BUG=skia:6065
Change-Id: Iecac15b56f59625b2e743ea36e7791b90bb0b422
Reviewed-on: https://skia-review.googlesource.com/6353
Reviewed-by: Leon Scroggins <scroggo@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Motivation: Since I am activly refactoring SkXPS, I will be needing to
add and subtract sources. This is the only way to keep chrome and skia
in sync.
TODO(halcanry): follow up in Chrome build files.
BUG=skia:3495
Change-Id: I2bfcbfce7628c34f9a623c6fa806374a10cdbb75
Reviewed-on: https://skia-review.googlesource.com/6350
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Hal Canary <halcanary@google.com>
Also extend GN format presubmit check to .gni files.
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=3976
Change-Id: I292f3136437101e3e2c3271e260536ab7a24b192
Reviewed-on: https://skia-review.googlesource.com/3976
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
This canaries the idea by converting gyp/pdf.gypi to gn/pdf.gni.
If this lands and rolls quietly, I'll flesh out the rest.
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=3972
Change-Id: Iecaf1ccce7891679904c585946272bfb25eea299
Reviewed-on: https://skia-review.googlesource.com/3972
Reviewed-by: Hal Canary <halcanary@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Android's the easiest place to start supporting Vulkan, so it's up first.
ndk_api becomes user-specifiable so that the Vulkan bots can kick it up
to 24, the first release supporting Vulkan. The defaults remain the same:
18 for 32-bit and 21 for 64-bit, the first release supporting 64-bit.
To test this, I set ndk_api=24 and skia_use_vulkan=true in GN, then
$ ninja -C out dm; and droid out/dm --config vk --src gm
Seems to work! Bot scripts to follow.
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2336343002
Review-Url: https://codereview.chromium.org/2336343002
This is for the coalescing of sources in one GYP file to improve GN runtime. If
this is successful, the other sources lists will also be merged into this one
file.
Inlines "../src" and "../include" for skia_source_dir and skia_include_dir. Evaluating these with GYP's variable expansion is annoying since I think another layer of nesting is required. Coding these explicitly is also more clear. This used to be required because Chromium would reference these .gypi files directly with a different path, but this no longer happens.
Removes the chromium defines gypi which is no longer referenced.
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2302723005
Committed: https://skia.googlesource.com/skia/+/39f7a10a04a914384944d8bf62621144ac4eeaa3
Review-Url: https://codereview.chromium.org/2302723005
Reason for revert:
The broke the PDFium build. See
https://build.chromium.org/p/client.skia.compile/builders/Build-Ubuntu-GCC-x86_64-Release-PDFium/builds/2094
Original issue's description:
> Move uils and sksl to a common sources GYP file.
>
> This is for the coalescing of sources in one GYP file to improve GN runtime. If
> this is successful, the other sources lists will also be merged into this one
> file.
>
> Inlines "../src" and "../include" for skia_source_dir and skia_include_dir. Evaluating these with GYP's variable expansion is annoying since I think another layer of nesting is required. Coding these explicitly is also more clear. This used to be required because Chromium would reference these .gypi files directly with a different path, but this no longer happens.
>
> Removes the chromium defines gypi which is no longer referenced.
>
> GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2302723005
>
> Committed: https://skia.googlesource.com/skia/+/39f7a10a04a914384944d8bf62621144ac4eeaa3TBR=mtklein@chromium.org,brettw@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
Review-Url: https://codereview.chromium.org/2317683002
This is for the coalescing of sources in one GYP file to improve GN runtime. If
this is successful, the other sources lists will also be merged into this one
file.
Inlines "../src" and "../include" for skia_source_dir and skia_include_dir. Evaluating these with GYP's variable expansion is annoying since I think another layer of nesting is required. Coding these explicitly is also more clear. This used to be required because Chromium would reference these .gypi files directly with a different path, but this no longer happens.
Removes the chromium defines gypi which is no longer referenced.
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2302723005
Review-Url: https://codereview.chromium.org/2302723005
This file will be imported by Chrome to access the sources lists.
Once Chrome is updated to use this file, changes to the skia .gypi layout can
be done entirely within the skia repository as long as the resulting lists
produced by the new .gni file have the same name.
Marks skia_for_chromium_defines as obsolete and moves the definition into the new .gni file. We can remove the .gypi file when Chrome is updated.
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2302803005
Review-Url: https://codereview.chromium.org/2302803005
