Skip to content

S
swiftshader
Switch branch/tag
  1. 11 Sep, 2020 1 commit
  3. 10 Sep, 2020 1 commit
  5. 09 Sep, 2020 4 commits
  7. 07 Sep, 2020 1 commit
  9. 04 Sep, 2020 1 commit
  11. 03 Sep, 2020 3 commits
  13. 02 Sep, 2020 2 commits
  15. 01 Sep, 2020 5 commits
  17. 28 Aug, 2020 4 commits
    • Optimize multisample resolve with SSE2 instructions · a2e6c1a1
      Nicolas Capens authored 
      Benchmark results:

Run on (48 X 2594 MHz CPU s)
CPU Caches:
  L1 Data 32 KiB (x24)
  L1 Instruction 32 KiB (x24)
  L2 Unified 256 KiB (x24)
  L3 Unified 30720 KiB (x2)
---------------------------------------------------------------
Benchmark                     Time             CPU   Iterations
---------------------------------------------------------------
(LLVM, before)
Triangle/Hello            0.845 ms        0.439 ms         1673
Triangle/Multisample       6.95 ms        0.781 ms         1000

(LLVM, after)
Triangle/Hello            0.861 ms        0.450 ms         1493
Triangle/Multisample       4.03 ms        0.753 ms          747

(Subzero, before)
Triangle/Hello             1.19 ms        0.474 ms         1120
Triangle/Multisample       11.8 ms        0.920 ms          747

(Subzero, after)
Triangle/Hello            0.907 ms        0.486 ms         1673
Triangle/Multisample       4.62 ms        0.781 ms         1000

Bug: b/147802090
Change-Id: Iea8498f2b745c86cf578db5c0f7ef2329b73c736
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/47970
Presubmit-Ready: Nicolas Capens <nicolascapens@google.com>
Tested-by: 's avatarNicolas Capens <nicolascapens@google.com>
Reviewed-by: 's avatarAlexis Hétu <sugoi@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
    • Add a fast multisample resolve code path · 4487e589
      Nicolas Capens authored 
      For whole-image 4x8-bit normalized format multisample resolves, use a
specialized code path instead of a generic blit routine.

Benchmark results:

Run on (48 X 2594 MHz CPU s)
CPU Caches:
  L1 Data 32 KiB (x24)
  L1 Instruction 32 KiB (x24)
  L2 Unified 256 KiB (x24)
  L3 Unified 30720 KiB (x2)
---------------------------------------------------------------
Benchmark                     Time             CPU   Iterations
---------------------------------------------------------------
(LLVM, before)
Triangle/Hello             1.02 ms        0.500 ms         1000
Triangle/Multisample       19.3 ms        0.984 ms         1000

(LLVM, after)
Triangle/Hello            0.845 ms        0.439 ms         1673
Triangle/Multisample       6.95 ms        0.781 ms         1000

(Subzero, before)
Triangle/Hello             1.15 ms        0.516 ms         1120
Triangle/Multisample       40.3 ms        0.469 ms          100

(Subzero, after)
Triangle/Hello             1.19 ms        0.474 ms         1120
Triangle/Multisample       11.8 ms        0.920 ms          747

Bug: b/147802090
Change-Id: I15729552f01a509a5cfce20cd7de06d0b764cf0a
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/47969
Presubmit-Ready: Nicolas Capens <nicolascapens@google.com>
Tested-by: 's avatarNicolas Capens <nicolascapens@google.com>
Reviewed-by: 's avatarAlexis Hétu <sugoi@google.com>
    • Make multisample resolve a Blitter method · 64ed1218
      Nicolas Capens authored 
      The Vulkan spec explicitly states that "vkCmdBlitImage must not be used
for multisampled source or destination images. Use vkCmdResolveImage for
this purpose." And the only other way to obtain resolve multisample
results is by using resolve attachments as part of a subpass.

This split between blit operations and resolve operations should be
reflected by the Blitter interface so we have less confusion about
its blit() method being used to perform resolves. It will also
facilitate adding a fast path for common resolve operations.

Bug: b/147802090
Change-Id: I2549a5e7acd7ef9ec3f70f8ceb88ff5fc65a0d17
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/47988
Presubmit-Ready: Nicolas Capens <nicolascapens@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: 's avatarNicolas Capens <nicolascapens@google.com>
Reviewed-by: 's avatarAlexis Hétu <sugoi@google.com>
    • Work around std::is_trivially_copyable issue with MSVC · 2014e7b9
      Nicolas Capens authored 
      Microsoft Visual Studio Community 2019 Version 16.7.1 produced the
following errors:

C2976 'llvm::SmallVectorTemplateBase': too few template argument
SwiftShader\third_party\llvm-10.0\llvm\include\llvm\ADT\SmallVector.h:315

C2338 inconsistent behavior between llvm:: and std:: implementation of is_trivially_copyable
SwiftShader\third_party\llvm-10.0\llvm\include\llvm\Support\type_traits.h:181

Not defining LLVM's HAVE_STD_IS_TRIVIALLY_COPYABLE config makes it fall
back to code which doesn't use std::is_trivially_copyable and avoids the
issue.

Note this may hide a potentially serious MSVC compiler bug since C2338
is produced by a static_assert. Without that static assert we may not
have known about the unexpected std::is_trivially_copyable behavior
until run-time issues are caused by it.

Bug: swiftshader:153
Change-Id: I04ab6d8ae725ecab45ac27fd89a37213baa86943
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/47930
Presubmit-Ready: Nicolas Capens <nicolascapens@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: 's avatarNicolas Capens <nicolascapens@google.com>
Reviewed-by: 's avatarAntonio Maiorano <amaiorano@google.com>
  19. 27 Aug, 2020 1 commit
  21. 26 Aug, 2020 3 commits
  23. 21 Aug, 2020 3 commits
  25. 20 Aug, 2020 1 commit
  27. 18 Aug, 2020 1 commit
  29. 17 Aug, 2020 2 commits
  31. 13 Aug, 2020 2 commits
  33. 12 Aug, 2020 3 commits
  35. 11 Aug, 2020 2 commits