Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Inductor][CPP] Fix int8 cvt half #136353

Conversation

leslie-fang-intel
Copy link
Collaborator

@leslie-fang-intel leslie-fang-intel commented Sep 20, 2024

Stack from ghstack (oldest at bottom):

Fix the correctness issue of pytorch/ao#884. The current implementation for converting between Half/BFloat16 and int8/uint8 incorrectly assumes that 1/4 of the int8/uint8 vector lane maps to 1/2 of the Half/BFloat16 vector lane. This assumption leads to accuracy issues after the full bit-width vectorization of the Half data type was introduced. When converting between int8 weights and the half data type, the generated code is as the following:

#include "/tmp/torchinductor_leslie/xw/cxww3s7wxrujoyxna7mlcjktid2uu6nntixqwm542xfkd756gl3x.h"
extern "C"  void kernel(const int8_t* in_ptr0,
                       half* out_ptr0)
{
    {
        for(int64_t x0=static_cast<int64_t>(0L); x0<static_cast<int64_t>(2048L); x0+=static_cast<int64_t>(32L))
        {
            auto tmp0 = at::vec::Vectorized<int8_t>::loadu(in_ptr0 + static_cast<int64_t>(x0), static_cast<int64_t>(32));
            auto tmp1 = at::vec::convert<half>(tmp0);
            tmp1.store(out_ptr0 + static_cast<int64_t>(x0), static_cast<int64_t>(32));
        }
    }
}

In this PR, we address the issue by changing the implementation to convert 1/2 of the int8/uint8 vector lane into a full vector lane of Half/BFloat16.

TestPlan

  • AO: python test/integration/test_integration.py -k test_int8_weight_only_quant_subclass_api
  • python -u -m pytest -s -v test/inductor/test_cpu_repro.py -k test_convert_int8_to_half_vec
  • Due to the CPP backend legalization pass, we are unable to create a unit test to simulate the conversion from Half to int8. Instead, we rely on a C++ test case.
    • ./build/bin/vec_test_all_types_AVX512 --gtest_filter="VecConvertTestsReducedFloat/*.ConvertReduced"
    • ./build/bin/vec_test_all_types_AVX2 --gtest_filter="VecConvertTestsReducedFloat/*.ConvertReduced"

cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10 @voznesenskym @penguinwu @EikanWang @Guobing-Chen @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @ColinPeppler @amjames @desertfire @chauhang

[ghstack-poisoned]
Copy link

pytorch-bot bot commented Sep 20, 2024

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/136353

Note: Links to docs will display an error until the docs builds have been completed.

✅ You can merge normally! (2 Unrelated Failures)

As of commit 3a43c94 with merge base failed to retrieve merge base, please contact dev infra:

FLAKY - The following jobs failed but were likely due to flakiness present on trunk:

This comment was automatically generated by Dr. CI and updates every 15 minutes.

leslie-fang-intel added a commit that referenced this pull request Sep 20, 2024
ghstack-source-id: 3a3a2525167d209cb087a0e3294b7362893bf2bd
Pull Request resolved: #136353
@pytorch-bot pytorch-bot bot added module: cpu CPU specific problem (e.g., perf, algorithm) module: inductor labels Sep 20, 2024
@leslie-fang-intel leslie-fang-intel added ciflow/trunk Trigger trunk jobs on your pull request topic: not user facing topic category labels Sep 20, 2024
[ghstack-poisoned]
leslie-fang-intel added a commit that referenced this pull request Sep 20, 2024
ghstack-source-id: 0f08e4c6f85916fe0b80247b3faa4965b1542ee8
Pull Request resolved: #136353
[ghstack-poisoned]
leslie-fang-intel added a commit that referenced this pull request Sep 23, 2024
ghstack-source-id: 6bc8a27d35cab306b15f0b1150a817ad951eaee7
Pull Request resolved: #136353
@leslie-fang-intel
Copy link
Collaborator Author

@pytorchbot merge

@pytorchmergebot
Copy link
Collaborator

Merge failed

Reason: Approvers from one of the following sets are needed:

  • superuser (pytorch/metamates)
  • Core Reviewers (mruberry, lezcano, Skylion007, ngimel, peterbell10, ...)
  • Core Maintainers (soumith, gchanan, ezyang, dzhulgakov, malfet, ...)
Details for Dev Infra team Raised by workflow job

Failing merge rule: Core Maintainers

@leslie-fang-intel
Copy link
Collaborator Author

Hi @jerryzh168, could you kindly help to take a look of this PR? it will fix failure of pytorch/ao#884 with FP16 input

Copy link
Contributor

@jerryzh168 jerryzh168 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

stamping

@jerryzh168
Copy link
Contributor

please make sure test passes before landing

@leslie-fang-intel
Copy link
Collaborator Author

@pytorchbot merge

@pytorchmergebot
Copy link
Collaborator

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging
Check the merge workflow status
here

BoyuanFeng pushed a commit to BoyuanFeng/pytorch that referenced this pull request Sep 25, 2024
Fix the correctness issue of pytorch/ao#884. The current implementation for converting between `Half/BFloat16` and `int8/uint8` incorrectly assumes that 1/4 of the int8/uint8 vector lane maps to 1/2 of the Half/BFloat16 vector lane. This assumption leads to accuracy issues after the full bit-width vectorization of the Half data type was introduced. When converting between int8 weights and the half data type, the generated code is as the following:
```
#include "/tmp/torchinductor_leslie/xw/cxww3s7wxrujoyxna7mlcjktid2uu6nntixqwm542xfkd756gl3x.h"
extern "C"  void kernel(const int8_t* in_ptr0,
                       half* out_ptr0)
{
    {
        for(int64_t x0=static_cast<int64_t>(0L); x0<static_cast<int64_t>(2048L); x0+=static_cast<int64_t>(32L))
        {
            auto tmp0 = at::vec::Vectorized<int8_t>::loadu(in_ptr0 + static_cast<int64_t>(x0), static_cast<int64_t>(32));
            auto tmp1 = at::vec::convert<half>(tmp0);
            tmp1.store(out_ptr0 + static_cast<int64_t>(x0), static_cast<int64_t>(32));
        }
    }
}
```

In this PR, we address the issue by changing the implementation to convert 1/2 of the int8/uint8 vector lane into a full vector lane of Half/BFloat16.

**TestPlan**
* AO: `python test/integration/test_integration.py -k test_int8_weight_only_quant_subclass_api`
* `python -u -m pytest -s -v test/inductor/test_cpu_repro.py -k test_convert_int8_to_half_vec`
* Due to the CPP backend legalization pass, we are unable to create a unit test to simulate the conversion from `Half` to `int8`. Instead, we rely on a C++ test case.
  * `./build/bin/vec_test_all_types_AVX512 --gtest_filter="VecConvertTestsReducedFloat/*.ConvertReduced"`
  * `./build/bin/vec_test_all_types_AVX2 --gtest_filter="VecConvertTestsReducedFloat/*.ConvertReduced"`

Pull Request resolved: pytorch#136353
Approved by: https://github.com/jgong5, https://github.com/jerryzh168
pytorchmergebot pushed a commit that referenced this pull request Sep 26, 2024
…136630)

**Summary**
Optimize the WOQ int8 AMX performance by changing the int8 -> bf16 conversion.
Earlier, 16 int8 elements were being loaded at a time & converted to 16 BF16 elements.
With this change, 32 int8 elements will be loaded at a time, and converted to a cache-line of 32 BF16 elements more efficiently.

Performance before
```
AUTOTUNE _weight_int8pack_mm(4096x4096, 4096x4096, 4096)
  cpp_packed_gemm_0 38.0439 ms 100.0%
  _weight_int8pack_mm 50.2524 ms 75.7%
SingleProcess AUTOTUNE benchmarking takes 1.1087 seconds and 1.9791 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x4096, 11008x4096, 11008)
  cpp_packed_gemm_4 78.2038 ms 100.0%
  _weight_int8pack_mm 119.1962 ms 65.6%
SingleProcess AUTOTUNE benchmarking takes 1.9274 seconds and 1.9949 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x11008, 4096x11008, 4096)
  cpp_packed_gemm_6 79.2368 ms 100.0%
  _weight_int8pack_mm 118.3212 ms 67.0%
SingleProcess AUTOTUNE benchmarking takes 1.9200 seconds and 2.0015 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x4096, 32000x4096, 32000)
  cpp_packed_gemm_224 225.7201 ms 100.0%
  _weight_int8pack_mm 388.5588 ms 58.1%
```

Performance after this PR
```
AUTOTUNE _weight_int8pack_mm(4096x4096, 4096x4096, 4096)
  cpp_packed_gemm_0 11.0086 ms 100.0%
  _weight_int8pack_mm 50.2918 ms 21.9%
SingleProcess AUTOTUNE benchmarking takes 1.0837 seconds and 2.0301 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x4096, 11008x4096, 11008)
  cpp_packed_gemm_4 24.3528 ms 100.0%
  _weight_int8pack_mm 119.8492 ms 20.3%
SingleProcess AUTOTUNE benchmarking takes 1.8303 seconds and 1.8195 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x11008, 4096x11008, 4096)
  cpp_packed_gemm_6 24.6148 ms 100.0%
  _weight_int8pack_mm 119.1908 ms 20.7%
SingleProcess AUTOTUNE benchmarking takes 1.8315 seconds and 1.8352 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x4096, 32000x4096, 32000)
  cpp_packed_gemm_224 78.1369 ms 100.0%
  _weight_int8pack_mm 387.6289 ms 20.2%
SingleProcess AUTOTUNE benchmarking takes 4.5059 seconds and 1.8010 seconds precompiling
```

Pull Request resolved: #136630
Approved by: https://github.com/jgong5
ghstack dependencies: #136353
pytorchmergebot pushed a commit to sanchitintel/pytorch that referenced this pull request Sep 26, 2024
…ytorch#136630)

**Summary**
Optimize the WOQ int8 AMX performance by changing the int8 -> bf16 conversion.
Earlier, 16 int8 elements were being loaded at a time & converted to 16 BF16 elements.
With this change, 32 int8 elements will be loaded at a time, and converted to a cache-line of 32 BF16 elements more efficiently.

Performance before
```
AUTOTUNE _weight_int8pack_mm(4096x4096, 4096x4096, 4096)
  cpp_packed_gemm_0 38.0439 ms 100.0%
  _weight_int8pack_mm 50.2524 ms 75.7%
SingleProcess AUTOTUNE benchmarking takes 1.1087 seconds and 1.9791 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x4096, 11008x4096, 11008)
  cpp_packed_gemm_4 78.2038 ms 100.0%
  _weight_int8pack_mm 119.1962 ms 65.6%
SingleProcess AUTOTUNE benchmarking takes 1.9274 seconds and 1.9949 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x11008, 4096x11008, 4096)
  cpp_packed_gemm_6 79.2368 ms 100.0%
  _weight_int8pack_mm 118.3212 ms 67.0%
SingleProcess AUTOTUNE benchmarking takes 1.9200 seconds and 2.0015 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x4096, 32000x4096, 32000)
  cpp_packed_gemm_224 225.7201 ms 100.0%
  _weight_int8pack_mm 388.5588 ms 58.1%
```

Performance after this PR
```
AUTOTUNE _weight_int8pack_mm(4096x4096, 4096x4096, 4096)
  cpp_packed_gemm_0 11.0086 ms 100.0%
  _weight_int8pack_mm 50.2918 ms 21.9%
SingleProcess AUTOTUNE benchmarking takes 1.0837 seconds and 2.0301 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x4096, 11008x4096, 11008)
  cpp_packed_gemm_4 24.3528 ms 100.0%
  _weight_int8pack_mm 119.8492 ms 20.3%
SingleProcess AUTOTUNE benchmarking takes 1.8303 seconds and 1.8195 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x11008, 4096x11008, 4096)
  cpp_packed_gemm_6 24.6148 ms 100.0%
  _weight_int8pack_mm 119.1908 ms 20.7%
SingleProcess AUTOTUNE benchmarking takes 1.8315 seconds and 1.8352 seconds precompiling
AUTOTUNE _weight_int8pack_mm(4096x4096, 32000x4096, 32000)
  cpp_packed_gemm_224 78.1369 ms 100.0%
  _weight_int8pack_mm 387.6289 ms 20.2%
SingleProcess AUTOTUNE benchmarking takes 4.5059 seconds and 1.8010 seconds precompiling
```

Pull Request resolved: pytorch#136630
Approved by: https://github.com/jgong5
ghstack dependencies: pytorch#136353
@github-actions github-actions bot deleted the gh/leslie-fang-intel/151/head branch October 27, 2024 02:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
ciflow/trunk Trigger trunk jobs on your pull request Merged module: cpu CPU specific problem (e.g., perf, algorithm) module: inductor open source topic: not user facing topic category
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants