From 92c67f0d3061730567e8e7900608dc0410bb35d9 Mon Sep 17 00:00:00 2001 From: cpuhrsch Date: Thu, 25 Apr 2024 16:07:58 -0700 Subject: [PATCH 1/6] Remove test_integration.py skip for CI --- test/integration/test_integration.py | 2 -- 1 file changed, 2 deletions(-) diff --git a/test/integration/test_integration.py b/test/integration/test_integration.py index 521f8a040..57cb82b2c 100644 --- a/test/integration/test_integration.py +++ b/test/integration/test_integration.py @@ -641,8 +641,6 @@ def test__int_mm(self): torch.testing.assert_close(y_ref, y_opt, atol=0, rtol=0) @unittest.skipIf(not torch.cuda.is_available(), "Need CUDA available") - @unittest.skipIf(TORCH_VERSION_AFTER_2_4 and torch.cuda.is_available(), "SystemError: AST constructor recursion depth mismatch (before=45, after=84)") - def test__int_mm_eager_and_torch_compile_numerics(self): def __int_mm_ref(x, w): x = x.cpu().to(torch.int32) From 1546e79b0289560a8d11951ed1dbe0ad03c48202 Mon Sep 17 00:00:00 2001 From: cpuhrsch Date: Thu, 25 Apr 2024 17:50:30 -0700 Subject: [PATCH 2/6] Update test_integration.py --- test/integration/test_integration.py | 10 ---------- 1 file changed, 10 deletions(-) diff --git a/test/integration/test_integration.py b/test/integration/test_integration.py index 57cb82b2c..585709491 100644 --- a/test/integration/test_integration.py +++ b/test/integration/test_integration.py @@ -948,7 +948,6 @@ def test_aq_int8_weight_only_quant_2_subclass(self, device, dtype): ) @parameterized.expand(COMMON_DEVICE_DTYPE) - @unittest.skipIf(TORCH_VERSION_AFTER_2_4 and torch.cuda.is_available(), "SystemError: AST constructor recursion depth mismatch (before=45, after=84)") def test_aq_int8_weight_only_quant_3_subclass(self, device, dtype): self._test_lin_weight_subclass_impl( AQWeightOnlyQuantizedLinearWeight3.from_float, device, 35, test_dtype=dtype @@ -1022,8 +1021,6 @@ def test_int8_dynamic_quant_subclass_api(self, device, dtype): ) @parameterized.expand(COMMON_DEVICE_DTYPE) - @unittest.skipIf(TORCH_VERSION_AFTER_2_4 and torch.cuda.is_available(), "SystemError: AST constructor recursion depth mismatch (before=45, after=84)") - def test_int8_weight_only_quant_subclass_api(self, device, dtype): self._test_lin_weight_subclass_api_impl( change_linear_weights_to_int8_woqtensors, device, 40, test_dtype=dtype @@ -1090,7 +1087,6 @@ def test_weight_only_quant(self): @parameterized.expand(COMMON_DEVICE_DTYPE) @torch.no_grad() @unittest.skipIf(not torch.cuda.is_available(), "Need CUDA available") - @unittest.skipIf(TORCH_VERSION_AFTER_2_4 and torch.cuda.is_available(), "SystemError: AST constructor recursion depth mismatch (before=45, after=84)") def test_weight_only_quant_force_mixed_mm(self, device, dtype): if device != "cuda": self.skipTest(f"weight_only_quant_force_mixed_mm can't be constructed on {device}") @@ -1117,8 +1113,6 @@ def test_weight_only_quant_force_mixed_mm(self, device, dtype): @parameterized.expand(COMMON_DEVICE_DTYPE) @unittest.skipIf(not torch.cuda.is_available(), "Need CUDA available") - @unittest.skipIf(TORCH_VERSION_AFTER_2_4 and torch.cuda.is_available(), "SystemError: AST constructor recursion depth mismatch (before=45, after=84)") - def test_weight_only_quant_use_mixed_mm(self, device, dtype): if device != "cuda": self.skipTest(f"weight_only_quant_force_mixed_mm can't be constructed on {device}") @@ -1355,8 +1349,6 @@ class TestAutoQuant(unittest.TestCase): # (256, 256, 128), TODO: Runs out of shared memory on T4 ])) @unittest.skipIf(not TORCH_VERSION_AFTER_2_3, "autoquant requires 2.3+.") - @unittest.skipIf(TORCH_VERSION_AFTER_2_4 and torch.cuda.is_available(), "SystemError: AST constructor recursion depth mismatch (before=45, after=84)") - def test_autoquant_one_input(self, device, dtype, m, k, n): print("(m, k, n): ", (m, k, n)) if device != "cuda" or not torch.cuda.is_available(): @@ -1390,8 +1382,6 @@ def test_autoquant_one_input(self, device, dtype, m, k, n): (32, 32, 128, 128), ])) @unittest.skipIf(not TORCH_VERSION_AFTER_2_3, "autoquant requires 2.3+.") - @unittest.skipIf(TORCH_VERSION_AFTER_2_4 and torch.cuda.is_available(), "SystemError: AST constructor recursion depth mismatch (before=45, after=84)") - def test_autoquant_multi_input(self, device, dtype, m1, m2, k, n): if device != "cuda" or not torch.cuda.is_available(): self.skipTest(f"autoquant currently does not support {device}") From 9d70517bf624e5cbc2ed8846e1e3506b370d399d Mon Sep 17 00:00:00 2001 From: Christian Puhrsch Date: Thu, 25 Apr 2024 19:33:19 -0700 Subject: [PATCH 3/6] torch-2.4.0.dev20240425 --- .github/workflows/regression_test.yml | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/.github/workflows/regression_test.yml b/.github/workflows/regression_test.yml index fa2f58bd9..97ece7c4a 100644 --- a/.github/workflows/regression_test.yml +++ b/.github/workflows/regression_test.yml @@ -31,6 +31,11 @@ jobs: torch-spec: 'torch==2.3.0' gpu-arch-type: "cuda" gpu-arch-version: "12.1" + - name: CUDA 2.4.0.dev20240424 + runs-on: linux.g5.12xlarge.nvidia.gpu + torch-spec: '--pre torch==torch-2.4.0.dev20240424 --index-url https://download.pytorch.org/whl/nightly/cu121' + gpu-arch-type: "cuda" + gpu-arch-version: "12.1" - name: CUDA Nightly runs-on: linux.g5.12xlarge.nvidia.gpu torch-spec: '--pre torch --index-url https://download.pytorch.org/whl/nightly/cu121' From b4ea259e28173509632cfe6bde8f56f19c9f9376 Mon Sep 17 00:00:00 2001 From: Christian Puhrsch Date: Thu, 25 Apr 2024 19:41:25 -0700 Subject: [PATCH 4/6] More nightlies --- .github/workflows/regression_test.yml | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/.github/workflows/regression_test.yml b/.github/workflows/regression_test.yml index 97ece7c4a..d4080508e 100644 --- a/.github/workflows/regression_test.yml +++ b/.github/workflows/regression_test.yml @@ -31,9 +31,14 @@ jobs: torch-spec: 'torch==2.3.0' gpu-arch-type: "cuda" gpu-arch-version: "12.1" + - name: CUDA 2.4.0.dev20240423 + runs-on: linux.g5.12xlarge.nvidia.gpu + torch-spec: '--pre torch==2.4.0.dev20240423+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121' + gpu-arch-type: "cuda" + gpu-arch-version: "12.1" - name: CUDA 2.4.0.dev20240424 runs-on: linux.g5.12xlarge.nvidia.gpu - torch-spec: '--pre torch==torch-2.4.0.dev20240424 --index-url https://download.pytorch.org/whl/nightly/cu121' + torch-spec: '--pre torch==2.4.0.dev20240424+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121' gpu-arch-type: "cuda" gpu-arch-version: "12.1" - name: CUDA Nightly From c3caf770b465f4d7ff9e8ebe1fc29150cc52b665 Mon Sep 17 00:00:00 2001 From: Christian Puhrsch Date: Thu, 25 Apr 2024 20:03:31 -0700 Subject: [PATCH 5/6] More nightlies --- .github/workflows/regression_test.yml | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/.github/workflows/regression_test.yml b/.github/workflows/regression_test.yml index e4e7571e1..d64916688 100644 --- a/.github/workflows/regression_test.yml +++ b/.github/workflows/regression_test.yml @@ -31,6 +31,16 @@ jobs: torch-spec: 'torch==2.3.0' gpu-arch-type: "cuda" gpu-arch-version: "12.1" + - name: CUDA 2.4.0.dev20240421 + runs-on: linux.g5.12xlarge.nvidia.gpu + torch-spec: '--pre torch==2.4.0.dev20240421+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121' + gpu-arch-type: "cuda" + gpu-arch-version: "12.1" + - name: CUDA 2.4.0.dev20240422 + runs-on: linux.g5.12xlarge.nvidia.gpu + torch-spec: '--pre torch==2.4.0.dev20240422+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121' + gpu-arch-type: "cuda" + gpu-arch-version: "12.1" - name: CUDA 2.4.0.dev20240423 runs-on: linux.g5.12xlarge.nvidia.gpu torch-spec: '--pre torch==2.4.0.dev20240423+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121' From afffad1e036b5dd872a1b2ac9f8ed1375df5c034 Mon Sep 17 00:00:00 2001 From: Christian Puhrsch Date: Thu, 25 Apr 2024 21:03:09 -0700 Subject: [PATCH 6/6] Pin nightly --- .github/workflows/regression_test.yml | 20 -------------------- 1 file changed, 20 deletions(-) diff --git a/.github/workflows/regression_test.yml b/.github/workflows/regression_test.yml index d64916688..85a79cd5c 100644 --- a/.github/workflows/regression_test.yml +++ b/.github/workflows/regression_test.yml @@ -36,26 +36,6 @@ jobs: torch-spec: '--pre torch==2.4.0.dev20240421+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121' gpu-arch-type: "cuda" gpu-arch-version: "12.1" - - name: CUDA 2.4.0.dev20240422 - runs-on: linux.g5.12xlarge.nvidia.gpu - torch-spec: '--pre torch==2.4.0.dev20240422+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121' - gpu-arch-type: "cuda" - gpu-arch-version: "12.1" - - name: CUDA 2.4.0.dev20240423 - runs-on: linux.g5.12xlarge.nvidia.gpu - torch-spec: '--pre torch==2.4.0.dev20240423+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121' - gpu-arch-type: "cuda" - gpu-arch-version: "12.1" - - name: CUDA 2.4.0.dev20240424 - runs-on: linux.g5.12xlarge.nvidia.gpu - torch-spec: '--pre torch==2.4.0.dev20240424+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121' - gpu-arch-type: "cuda" - gpu-arch-version: "12.1" - - name: CUDA Nightly - runs-on: linux.g5.12xlarge.nvidia.gpu - torch-spec: '--pre torch --index-url https://download.pytorch.org/whl/nightly/cu121' - gpu-arch-type: "cuda" - gpu-arch-version: "12.1" - name: CPU 2.2.2 runs-on: linux.4xlarge torch-spec: 'torch==2.2.2 --index-url https://download.pytorch.org/whl/cpu'