patch mm segfault & patch cubin avail. (#1628)

<!-- .github/pull_request_template.md -->

## 📌 Description

<!-- What does this PR do? Briefly describe the changes and why they’re
needed. -->

## 🔍 Related Issues

<!-- Link any related issues here -->

## 🚀 Pull Request Checklist

Thank you for contributing to FlashInfer! Before we review your pull
request, please make sure the following items are complete.

### ✅ Pre-commit Checks

- [x] I have installed `pre-commit` by running `pip install pre-commit`
(or used your preferred method).
- [x] I have installed the hooks with `pre-commit install`.
- [x] I have run the hooks manually with `pre-commit run --all-files`
and fixed any reported issues.

> If you are unsure about how to set up `pre-commit`, see [the
pre-commit documentation](https://pre-commit.com/).

## 🧪 Tests

- [x] Tests have been added or updated as needed.
- [x] All tests are passing (`unittest`, etc.).

## Reviewer Notes

<!-- Optional: anything you'd like reviewers to focus on, concerns, etc.
-->

Author: aleozlx

csrc/trtllm_gemm_runner.cu

@@ -207,7 +207,7 @@ class TrtllmGenGemmRunner {
         return optionsA.mUseUnrollLoop2xForMma;
       }
 
-      return true;
+      return false;
     });
 
     bool findLoop2xMma = false;

flashinfer/artifacts.py

@@ -48,27 +48,27 @@ def get_available_cubin_files(source, retries=3, delay=5, timeout=10):
 
 
 class ArtifactPath:
-    TRTLLM_GEN_FMHA: str = "6c74964c96684c3e674340f7e35fc20ad909a9a0/fmha/trtllm-gen/"
+    TRTLLM_GEN_FMHA: str = "037e528e719ec3456a7d7d654f26b805e44c63b1/fmha/trtllm-gen/"
     TRTLLM_GEN_BMM: str = (
-        "6c74964c96684c3e674340f7e35fc20ad909a9a0/batched_gemm-8704aa4-ba3b00d/"
+        "037e528e719ec3456a7d7d654f26b805e44c63b1/batched_gemm-8704aa4-ba3b00d/"
     )
     TRTLLM_GEN_GEMM: str = (
-        "6c74964c96684c3e674340f7e35fc20ad909a9a0/gemm-8704aa4-f91dc9e/"
+        "037e528e719ec3456a7d7d654f26b805e44c63b1/gemm-8704aa4-f91dc9e/"
     )
     CUDNN_SDPA: str = "4c623163877c8fef5751c9c7a59940cd2baae02e/fmha/cudnn/"
     DEEPGEMM: str = "d25901733420c7cddc1adf799b0d4639ed1e162f/deep-gemm/"
 
 
 class MetaInfoHash:
     TRTLLM_GEN_FMHA: str = (
-        "5a41a165d4d5e956d4cccd0a7d1627dbdcaccf4d07a9cfcc8055ef0cb52e0c87"
+        "0ff77215b86997665cf75973e13cd2932f551d46b4e008f851d32d47e1d9560f"
     )
     TRTLLM_GEN_BMM: str = (
-        "3edf4847059d465182779436397ece3d5fb45c3360a1d1abda3b71e35f957caa"
+        "34bdfe7acfd49f5fb8b48e06d56e6a5ad88b951c730552f228fc5f614f7632a8"
     )
     DEEPGEMM: str = "69aa277b7f3663ed929e73f9c57301792b8c594dac15a465b44a5d151b6a1d50"
     TRTLLM_GEN_GEMM: str = (
-        "c6265cf047fc5d2208a37c54b5d720f4755b1215de5f7434ad24ffbc81c31c27"
+        "0345358c916d990709f9670e113e93f35c76aa22715e2d5128ec2ca8740be5ba"
     )
 
 

tests/test_mm_fp4.py

@@ -25,10 +25,6 @@ def test_mm_fp4(m, n, k, res_dtype, backend, use_128x4_sf_layout, auto_tuning):
     if auto_tuning and backend == "cudnn":
         pytest.skip("Skipping test for cudnn fp4 with auto_tuning=True")
 
-    if not use_128x4_sf_layout and backend == "trtllm":
-        # FIXME (bringup) quantization failure from main
-        pytest.xfail("Skipping test for non-trtllm fp4 with use_128x4_sf_layout=False")
-
     input = torch.randn([m, k], device="cuda", dtype=torch.bfloat16)
     mat2 = torch.randn([n, k], device="cuda", dtype=torch.bfloat16)
     a_sf_layout = SfLayout.layout_128x4 if use_128x4_sf_layout else SfLayout.layout_8x4