remove path

Author: LeiWang1999

benchmark/matmul_fp8/README.md

@@ -1,6 +1,6 @@
 # FP8 Matmul Benchmark (8192×8192)
 
-This document records the throughput achieved by `benchmark_matmul.py` when multiplying FP8 matrices sized `M = N = 8192` across different `K` dimensions. Each measurement relies on the default autotuning search space bundled with the benchmark. The file lives in `/weka-hg/prod/deepseek/permanent/wanglei/tilelang/benchmark/matmul_fp8`.
+This document records the throughput achieved by `benchmark_matmul.py` when multiplying FP8 matrices sized `M = N = 8192` across different `K` dimensions. Each measurement relies on the default autotuning search space bundled with the benchmark.
 
 ## Environment
 

benchmark/matmul_fp8/benchmark_matmul.py

@@ -219,9 +219,9 @@ def main(
 if __name__ == "__main__":
     # Parse command-line arguments for matrix dimensions
     parser = argparse.ArgumentParser(description="Autotuned MatMul Benchmark")
-    parser.add_argument("--m", type=int, default=8192, help="Matrix dimension M")
-    parser.add_argument("--n", type=int, default=8192, help="Matrix dimension N")
-    parser.add_argument("--k", type=int, default=8192, help="Matrix dimension K")
+    parser.add_argument("--m", type=int, default=16384, help="Matrix dimension M")
+    parser.add_argument("--n", type=int, default=16384, help="Matrix dimension N")
+    parser.add_argument("--k", type=int, default=16384, help="Matrix dimension K")
     parser.add_argument(
         "--with_roller",
         action="store_true",
@@ -237,7 +237,6 @@ def main(
 
     # matmul(...) returns (best_latency, best_config, ref_latency)
     best_result = matmul(M, N, K, with_roller)
-    print(best_result.get_kernel_source())
     best_latency = best_result.latency
     best_config = best_result.config