lil less is okay

This commit is contained in:
George Hotz
2025-10-21 17:09:44 +08:00
parent 32af1ff84b
commit d59d4cdbe4

View File

@@ -91,11 +91,11 @@ class TestKernelSpeed(unittest.TestCase):
# theoretical is nv_tflops=165, amd_tflops=123
def test_gemm_4096(self): self._test_matmul(4096, nv_tflops=115, amd_tflops=65)
def test_gemm_8192(self): self._test_matmul(8192, nv_tflops=125, amd_tflops=60)
def test_gemm_8192(self): self._test_matmul(8192, nv_tflops=115, amd_tflops=60)
# theoretical is nv_gbs=1008, amd_gbs=960
def test_gemv_16384_4096(self): self._test_matmul(16384, 4096, 1, nv_gbs=840, amd_gbs=750)
def test_gemv_4096_16384(self): self._test_matmul(4096, 16384, 1, nv_gbs=830, amd_gbs=750)
def test_gemv_4096_16384(self): self._test_matmul(4096, 16384, 1, nv_gbs=820, amd_gbs=750)
if __name__ == '__main__':
unittest.main()