GitLab

**Description**

Cherry-pick bug fixes from v0.12.0 to main.

**Major Revisions**

* #725
* #727
* #728
Co-authored-by: Hongtao Zhang <hongtaozhang@microsoft.com>
Co-authored-by: Yifan Xiong <yixio@microsoft.com>
Co-authored-by: Guoshuai Zhao <guzhao@microsoft.com>

---------
Co-authored-by: Hongtao Zhang <hongtaozhang@microsoft.com>

Added MoE model using MixtralConfig. 

1. Added 8x7b and 8x22b variants 
2. Requires high VRAM as all experts are loaded in memory. Thus,
disabled training due to memory constraint on test worker.

---------
Co-authored-by: Hongtao Zhang <garyworkzht@gmail.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Hongtao Zhang <hongtaozhang@microsoft.com>

**Description**
Add deepseek megatron-lm benchmark.

---------
Co-authored-by: yukirora <yuting.jiang@microsoft.com>
Co-authored-by: Hongtao Zhang <garyworkzht@gmail.com>
Co-authored-by: Hongtao Zhang <hongtaozhang@microsoft.com>

**Description**
Add cuda 12.9 dockerfile and build in pipeline.

---------
Co-authored-by: Guoshuai Zhao <microsoft@microsoft.com>
Co-authored-by: Hongtao Zhang <hongtaozhang@microsoft.com>
Co-authored-by: Hongtao Zhang <garyworkzht@gmail.com>

**Description**
Add FP4 precision support for cublaslt_gemm benchmark.

**Major Revision**
- Add new type `fp4e2m1` and `__nv_fp4_e2m1`.
- For FP4 matmul, precision of MatrixC (add) should be FP16, precision
of MatricD (output) should be FP4, otherwise, it will not work.
- Add macro `CUDA_VERSION` to resolve the compatibility issue of
different CUDA versions.

---------
Co-authored-by: Ubuntu <aiperf@aiperf000000.hp5z1gqeinfufbj2u3jcty5fme.cdmx.internal.cloudapp.net>
Co-authored-by: AVA <39534996+avazr@users.noreply.github.com>
Co-authored-by: Guoshuai Zhao <microsoft@microsoft.com>

**Description**
Enable autotuning as an opt-in mode when benchmarking cublasLt via
`cublaslt_gemm`

The implementation is based on
https://github.com/NVIDIA/CUDALibrarySamples/blob/master/cuBLASLt/LtSgemmSimpleAutoTuning/sample_cublasLt_LtSgemmSimpleAutoTuning.cu

The behavior of original benchmark command remains unchanged, e.g.:
- `cublaslt_gemm -m 2048 -n 12288 -k 1536 -w10000 -i 1000 -t fp8e4m3`

The new opt-in options are `-a` (for autotune) and `-I` (for autotune
iterations, default is 50, same as the default for `-i`) and `-W` (for
autotune warmups, default=20, same as the default for `-w`), e.g.:
- `cublaslt_gemm -m 2048 -n 12288 -k 1536 -w 10000 -i 1000 -t fp8e4m3
-a`
- `cublaslt_gemm -m 2048 -n 12288 -k 1536 -w 10000 -i 1000 -t fp8e4m3 -a
-I 10 -W 10`

**Note:** This PR also changes the default `gemm_compute_type` for BF16
and FP16 to `CUBLAS_COMPUTE_32F`.

**Further observations:** 
1. The support matrix of the `cublaslt_gemm` could be furt...

**Description**
Added support for Grace CPU neo2 architecture in CPU Stream. Now CPU
Stream supports dual socket benchmarking.

Example config for this arch support:
```yaml
    cpu-stream:numa0:
      timeout: *default_timeout
      modes:
      - name: local
        parallel: no
      parameters:
        cpu_arch: neo2
        numa_mem_nodes: 0
        cores: 0 1 2 3 4 5 6 7 8
    cpu-stream:numa1:
      timeout: *default_timeout
      modes:
      - name: local
        parallel: no
      parameters:
        cpu_arch: neo2
        numa_mem_nodes: 1
        cores: 64 65 66 67 68 69 70 71 72
    cpu-stream:numa-spread:
      timeout: *default_timeout
      modes:
      - name: local
        parallel: no
      parameters:
        cpu_arch: neo2
        numa_mem_nodes: 0 1
        cores: 0 1 2 3 4 5 6 7 8 64 65 66 67 68 69 70 71 72
```

---------
Co-authored-by: dpower4 <dilipreddi@gmail.com>

Added GPU Stream benchmark - measures the GPU memory bandwidth and
efficiency for double datatype through various memory operations
including copy, scale, add, and triad.
- added documentation for `gpu-stream` detailing its introduction,
metrics, and descriptions.
- added unit tests for `gpu-stream`. Example output is in
`superbenchmark/tests/data/gpu_stream.log`.

In the current implementation, the CPU‑stream benchmark code renames the
binary before the microbench base class can verify its existence,
causing the default‐binary check to fail.

This PR adds a “default” binary—built with the standard compile
parameters—so that the base class can always find and validate it. Once
the default binary is in place, the CPU‑stream code will rename it as
needed and re‑check its presence before running the benchmark.

The PR also enable CPU stream in the default settings.

---------
Co-authored-by: Hongtao Zhang <hongtaozhang@microsoft.com>

Update CODEOWNERS.

adding gb200 cuda arch flag for cublaslt compilation

- Upgrade OS of github runner used by lint to the latest.
- Add symbolic link for clang-format to version 14.
- Update importlib_metadata version since it is too old (inside
nvcr.io/nvidia/pytorch:20.12-py3) and failed the 11.1 build.

---------
Co-authored-by: hongtaozhang <hongtaozhang@microsoft.com>
Co-authored-by: Yifan Xiong <yifan.xiong@microsoft.com>

Merge multi-arch image in build pipeline.

**Description**
Updated docker for 12.8
Use cutlass latest relase 3.8 with ARCH 100(blackwell) support
add latest nccl-test release with ARCH 100(blackwell) 
Updated msccl to support build for sm_100
No breaking changes, so backward compatible tested with  cuda 12.4

---------
Co-authored-by: Hongtao Zhang <garyworkzht@gmail.com>

Due to the matrix strategy’s default "fail-fast" setting. In GitHub
Actions, when running a job with a matrix, the individual configurations
run in parallel. By default, if one matrix job (for example, the one
labeled "rocm6_2_rocm6_2_x_superbe") fails, the remaining parallel jobs
are canceled automatically.

In our current build image pipeline, the arm64 build job always are
canceled by the rocm build job. So, using a non-existent label in the
job config to prevent rocm build job from scheduling for a temporary
solution.

---------
Co-authored-by: hongtaozhang <hongtaozhang@microsoft.com>

This enhancement addresses an issue in mypy where it may report missing
pkg_resources even when ignore_missing_imports = True is set and the
package is installed. Adding this configuration ensures that
pkg_resources is properly skipped during type checking.
Co-authored-by: hongtaozhang <hongtaozhang@microsoft.com>

Add image build on arm64 arch.

Improves logging info for diagnosis rule op baseline errors. This allows
developers to easily detect errors in their rule files as well as
baseline files, improving end-user experience.

Co-authored-by: Yifan Xiong <yifan.xiong@microsoft.com>
Co-authored-by: Hongtao Zhang <garyworkzht@gmail.com>

Added support for Python 3.11, 3.12 and 3.13.

yapf is not compatiable with python3.12+, so we disable yapf in py3.12
for now.
https://github.com/google/yapf/issues/1258
https://github.com/google/yapf/issues/1266



---------
Co-authored-by: hongtaozhang <hongtaozhang@microsoft.com>

Root Cause:

1. '_get_all_test_cases()' was called in '_parser' while '_parser' was
defined in the base class.
2.  in '_get_all_test_cases()', cmd path was not included.

Fix:

1. Remove '_get_all_test_cases()' from '_parser'.
2. Construct path for cmd.

---------
Co-authored-by: hongtaozhang <hongtaozhang@microsoft.com>

**Description**

1. Fixed the bug that nvbandwidth benchmark need to handle 'N/A' values
in nvbandwidth cmd output.
2. Replaced the input format of test cases with a list.
3. Add nvbandwidth configuration example in default config files.

---------
Co-authored-by: hongtaozhang <hongtaozhang@microsoft.com>
Co-authored-by: Yifan Xiong <yifan.xiong@microsoft.com>

**Description**
Today I was running a benchmark on my machine. And encountered a fancy
issue with tensorrt-inference.
I got code 33, which according to the source code is:
```
MICROBENCHMARK_RESULT_PARSING_FAILURE = 33
```
I dived deep into the code and found out the following problem. The
parser stumbled upon getting to the following line:
```
[11/28/2024-17:03:11] [I] Latency: min = 7.2793 ms, max = 10.1606 ms, mean = 7.41642 ms, median = 7.39551 ms, percentile(99%) = 8 ms
```
I ran it separately on the code and found out that the regular
expression was not suitable for the cases like this, when you encounter
an INT as a result in milliseconds.
That's why this pull request is created.
I came up with the closest possible regular expression to fix this issue
and not to introduce any other bug.

**Major Revision**
- 0.11.0

Flake8 has moved away from gitlab to github.
Updating the repo path in the pre commit config.

**Description**
Introduce architecture support for version 10.0 in gemm-flops.

Fix installation and lint issues:

* Fix transformer installation in Python3.7 due to upgrade of safetensors.
* Fix lint issues in mypy 1.14.1.

Bumps [nanoid](https://github.com/ai/nanoid) from 3.3.6 to 3.3.8.
- [Release notes](https://github.com/ai/nanoid/releases)
- [Changelog](https://github.com/ai/nanoid/blob/main/CHANGELOG.md

)
- [Commits](ai/nanoid@3.3.6...3.3.8)

---
updated-dependencies:
- dependency-name: nanoid
  dependency-type: indirect
...
Signed-off-by: dependabot[bot] <support@github.com>

Added llama benchmark - training and inference in accordance with the
existing pytorch models implementation like gpt2, lstm etc.

- added llama fp8 unit test for better code coverage, to reduce memory
required
- updated transformers version >= 4.28.0 for LLamaConfig
- set tokenizers version <= 0.20.3 to avoid 0.20.4 version
[issues](https://github.com/huggingface/tokenizers/issues/1691

) with
py3.8
- added llama2 to tensorrt
- llama2 tests not added to test_tensorrt_inference_performance.py due
to large memory requirement for worker gpu. tests validated separately
on gh200

---------
Co-authored-by: dpatlolla <dpatlolla@microsoft.com>

Fix ordering of args in err messages.

Upgrade dependency versions in Azure pipeline:

* Remove Python 3.6 and add Python 3.10 for cpu-unit-test
* Upgrade CUDA from 11.1 to 12.4 for cuda-unit-test
* Update labels accordingly

---------
Co-authored-by: Dilip Patlolla <dilipreddi@gmail.com>

**Description**

Add nvbandwidth benchmark.

---------
Co-authored-by: hongtaozhang <hongtaozhang@microsoft.com>

**Description**
Add nvbandwidth build to repo

---------
Co-authored-by: hongtaozhang <hongtaozhang@microsoft.com>

Update CODEOWNERS for docs.

**Description**
Add micro benchmark to measure general CPU bandwidth and latency without 'mlc'.

Test output:
```
{
"cpu-memory-bw-latency/return_code": 0,
"cpu-memory-bw-latency/mem_bandwidth_matrix_numa_0_1_bw": 5388.75021,
"cpu-memory-bw-latency/mem_bandwidth_matrix_numa_0_1_lat": 0.185571786,
"cpu-memory-bw-latency/mem_bandwidth_matrix_numa_1_0_bw": 4634.82028,
"cpu-memory-bw-latency/mem_bandwidth_matrix_numa_1_0_lat": 0.215758096,
}
```

---------
Co-authored-by: hongtaozhang <hongtaozhang@microsoft.com>

**Description**

Bump onnxruntime-gpu from 1.10.0 to 1.12.0.

---------
Co-authored-by: hongtaozhang <hongtaozhang@microsoft.com>

Bumps [webpack](https://github.com/webpack/webpack) from 5.76.1 to 5.96.1.
- [Release notes](https://github.com/webpack/webpack/releases

)
- [Commits](webpack/webpack@v5.76.1...v5.96.1)

---
updated-dependencies:
- dependency-name: webpack
  dependency-type: indirect
...
Signed-off-by: dependabot[bot] <support@github.com>

Bumps [cookie](https://github.com/jshttp/cookie) and [express](https://github.com/expressjs/express). These dependencies needed to be updated together.

Updates `cookie` from 0.6.0 to 0.7.1
- [Release notes](https://github.com/jshttp/cookie/releases)
- [Commits](jshttp/cookie@v0.6.0...v0.7.1)

Updates `express` from 4.21.0 to 4.21.1
- [Release notes](https://github.com/expressjs/express/releases)
- [Changelog](https://github.com/expressjs/express/blob/4.21.1/History.md

)
- [Commits](expressjs/express@4.21.0...4.21.1)

---
updated-dependencies:
- dependency-name: cookie
  dependency-type: indirect
- dependency-name: express
  dependency-type: indirect
...
Signed-off-by: dependabot[bot] <support@github.com>

Add support for arm64 build:

- Updated dockerfile for arm64 build
- extend cpu stream compilation for neoverse 
- handle onnxruntime-gpu installation
- third party builds filtering based on arch
- disable cuda decode perf build for non x86

The current GPU Copy BW Performance fails on Nvidia Grace systems. This
is due to the memory only numa node and thus the numa_run_on_node fails
for such nodes and halts completely.

This fix checks for the presence of assigned CPU cores for the numa
node, on checking if it has no cpu cores assigned, it skips that
specific node during the args creation and continues.

**Description**

Update image build.

**Major Revision**

* Remove ROCm 6.0 image due to outdated packages
* Remove build tag for ROCm
* Preserve build cache for 30 days