add axiswise granularity to Float8Tensor (#919)

Summary:

This is a copy-paste of meta-pytorch/float8_experimental#352
which never landed.

Test Plan:


Reviewers:

Subscribers:

Tasks:

Tags:

Author: vkuzo

test/float8/test_base.py

@@ -22,14 +22,20 @@
     pytest.skip("Unsupported PyTorch version", allow_module_level=True)
 
 
-from torchao.float8.config import CastConfig, Float8LinearConfig, ScalingType
+from torchao.float8.config import (
+    CastConfig, 
+    Float8LinearConfig, 
+    ScalingGranularity,
+    ScalingType,
+)
 from torchao.float8.float8_linear import Float8Linear
 from torchao.float8.float8_linear_utils import (
     convert_to_float8_training,
     linear_requires_sync,
     sync_float8_amax_and_scale_history,
 )
 from torchao.float8.float8_python_api import addmm_float8_unwrapped
+from torchao.float8.float8_scaling_utils import hp_tensor_to_float8_dynamic
 from torchao.float8.float8_tensor import (
     Float8Tensor,
     GemmInputRole,
@@ -51,14 +57,15 @@
 
 
 is_cuda_8_9 = torch.cuda.is_available() and torch.cuda.get_device_capability() >= (8, 9)
+is_cuda_9_0 = torch.cuda.is_available() and torch.cuda.get_device_capability() >= (9, 0)
 
 def bitwise_identical(a: Float8Tensor, b: Float8Tensor) -> bool:
     assert torch.all(a._scale == b._scale).item(), "scales are not identical"
     assert torch.all(a._data == b._data).item(), "data is not identical"
     return True
 
 
-class TestFloat8Tensor(unittest.TestCase):
+class TestFloat8Tensor:
     def test_preserves_dtype(self) -> None:
         # hp means high precision, lp means low precision
         hp_dtypes = (torch.float32, torch.float16, torch.bfloat16)
@@ -68,7 +75,7 @@ def test_preserves_dtype(self) -> None:
             x1_s = tensor_to_scale(x1_hp, lp_dtype)
             x2_lp = hp_tensor_and_scale_to_float8(x1_hp, x1_s, lp_dtype)
             x3_hp = x2_lp.to_original_precision()
-            self.assertTrue(x3_hp.dtype == hp_dtype)
+            assert x3_hp.dtype == hp_dtype
 
     def test_differentiable_casts(self) -> None:
         lp_dtypes = (e4m3_dtype, e5m2_dtype)
@@ -103,7 +110,7 @@ def test_index_put(self):
         fp8_b = hp_tensor_and_scale_to_float8(b, scale_a, torch.float8_e4m3fn)
         fp8_b_bad = hp_tensor_and_scale_to_float8(b, scale_b, torch.float8_e4m3fn)
 
-        with self.assertRaises(AssertionError):
+        with pytest.raises(AssertionError):
             b[index] = fp8_a
             fp8_b[index] = a
             fp8_b_bad[index] = fp8_a
@@ -117,7 +124,7 @@ def test_copy_(self):
         b = torch.empty(16, dtype=torch.bfloat16)
         b.copy_(fp8_a)  # Should work
         torch.testing.assert_close(b, fp8_a.to_original_precision())
-        with self.assertRaises(RuntimeError):
+        with pytest.raises(RuntimeError):
             fp8_a.copy_(b)  # Should fail
 
         fp8_b = Float8Tensor(
@@ -129,6 +136,105 @@ def test_copy_(self):
         fp8_b.copy_(fp8_a)
         torch.testing.assert_close(fp8_a._data, fp8_b._data)
 
+    @pytest.mark.parametrize("shape", [(8, 16), (4, 8, 16), (2, 4, 8, 16)])
+    @pytest.mark.parametrize("axiswise_dim", [0, -1])
+    def test_axiswise_dynamic_cast(self, shape, axiswise_dim):
+        a = torch.randn(*shape, dtype=torch.bfloat16)
+        linear_mm_config = LinearMMConfig()
+        a_fp8 = hp_tensor_to_float8_dynamic(
+            a,
+            e4m3_dtype,
+            linear_mm_config,
+            scaling_granularity=ScalingGranularity.AXISWISE,
+            axiswise_dim=axiswise_dim,
+        )
+        a_dq = a_fp8.to_original_precision()
+        sqnr = compute_error(a, a_dq)
+        assert sqnr >= 25.0
+
+    def test_axiswise_reshape(self):
+        a = torch.randn(3, 5, 7, dtype=torch.bfloat16)
+        linear_mm_config = LinearMMConfig()
+
+        # if we scale across dim0, we can only reshape to [3, -1]
+        a_fp8_d0 = hp_tensor_to_float8_dynamic(
+            a,
+            e4m3_dtype,
+            linear_mm_config,
+            scaling_granularity=ScalingGranularity.AXISWISE,
+            axiswise_dim=0,
+        )
+        assert list(a_fp8_d0._data.shape) == [3, 5, 7]
+        assert list(a_fp8_d0._scale.shape) == [1, 5, 7]
+
+        a_fp8_d0_r = a_fp8_d0.reshape(3, -1)
+        assert list(a_fp8_d0_r.shape) == [3, 5 * 7]
+        assert list(a_fp8_d0_r._scale.shape) == [1, 5 * 7]
+        # verify numerics did not change
+        assert torch.allclose(
+            a_fp8_d0.to_original_precision(),
+            a_fp8_d0_r.to_original_precision().reshape(3, 5, 7),
+            atol=0,
+            rtol=0,
+        )
+        with pytest.raises(RuntimeError):
+            a_fp8_d0_r2 = a_fp8_d0.reshape(-1, 7)
+
+        # if we scale across dim2, we can only reshape to [-1, 7]
+        a_fp8_d2 = hp_tensor_to_float8_dynamic(
+            a,
+            e4m3_dtype,
+            linear_mm_config,
+            scaling_granularity=ScalingGranularity.AXISWISE,
+            axiswise_dim=-1,
+        )
+        assert list(a_fp8_d2._data.shape) == [3, 5, 7]
+        assert list(a_fp8_d2._scale.shape) == [3, 5, 1]
+
+        a_fp8_d2_r = a_fp8_d2.reshape(-1, 7)
+        assert list(a_fp8_d2_r.shape) == [3 * 5, 7]
+        assert list(a_fp8_d2_r._scale.shape) == [3 * 5, 1]
+        # verify numerics did not change
+        assert torch.allclose(
+            a_fp8_d2.to_original_precision(),
+            a_fp8_d2_r.to_original_precision().reshape(3, 5, 7),
+            atol=0,
+            rtol=0,
+        )
+        with pytest.raises(RuntimeError):
+            a_fp8_d2_r2 = a_fp8_d2.reshape(3, -1)
+
+    @pytest.mark.parametrize("a_shape", [(16, 32), (2, 16, 32), (1, 2, 16, 32)])
+    @unittest.skipIf(not torch.cuda.is_available(), "CUDA not available")
+    @unittest.skipIf(not is_cuda_9_0, "Requires CUDA capability >= 9.0")
+    def test_axiswise_gemm(self, a_shape):
+        a = torch.randn(*a_shape, dtype=torch.bfloat16, device="cuda")
+        b = torch.randn(64, 32, dtype=torch.bfloat16, device="cuda")
+
+        linear_mm_config = LinearMMConfig()
+
+        a_fp8 = hp_tensor_to_float8_dynamic(
+            a,
+            e4m3_dtype,
+            linear_mm_config,
+            gemm_input_role=GemmInputRole.INPUT,
+            scaling_granularity=ScalingGranularity.AXISWISE,
+            axiswise_dim=-1,
+        )
+        a_fp8 = a_fp8.reshape(-1, a_shape[-1])
+        b_fp8 = hp_tensor_to_float8_dynamic(
+            b,
+            e4m3_dtype,
+            linear_mm_config,
+            gemm_input_role=GemmInputRole.WEIGHT,
+            scaling_granularity=ScalingGranularity.AXISWISE,
+            axiswise_dim=-1,  # will be transposed
+        )
+        c_fp8_compute = torch.mm(a_fp8, b_fp8.t())
+        a = a.reshape(-1, a_shape[-1])
+        c_ref = torch.mm(a, b.t())
+        sqnr = compute_error(c_ref, c_fp8_compute)
+        assert sqnr >= 25.0
 
 
 class TestFloat8Linear:

torchao/float8/config.py

@@ -26,6 +26,18 @@ def short_str(self):
             return "sta"
 
 
+class ScalingGranularity(enum.Enum):
+    """
+    Defines the granularity of scaling strategies for casting to float8
+    """
+
+    # A single scaling factor for the entire tensor
+    TENSORWISE = "tensorwise"
+    # Scaling factors computed along one axis of the tensor, reducing it to
+    # size 1.
+    AXISWISE = "axiswise"
+
+
 @dataclass(frozen=True)
 class CastConfig:
     """
@@ -146,6 +158,8 @@ class Float8LinearConfig:
     # save the fp8_weight_transpose for backward, which is an un-sahrded weight and costs a high memory utilization.
     # The longer-term solution is to let compile decide how to partition the graph with optimal computation and memory savings.
     # For now, we use the checkpointing api to force the recomputation of fp8 weight in backward.
+    # TODO(future PR): either enable by default or have a warning and set up the
+    # tests so that the warning does not spam the CI stdout.
 
     force_recompute_fp8_weight_in_bwd: bool = False
 

torchao/float8/float8_linear.py

@@ -9,7 +9,6 @@
 
 import dataclasses
 import enum
-import logging
 
 from typing import Optional
 
@@ -50,8 +49,6 @@
     WeightWithStaticFloat8CastTensor,
 )
 
-logger = logging.getLogger(__name__)
-
 
 # this code was resurrected from https://github.com/pytorch-labs/torchao.float8/pull/128/files
 @torch._dynamo.allow_in_graph
@@ -191,15 +188,6 @@ def __init__(self, *args, **kwargs):
         # would be initialized in every iteration.
         self.enable_pre_and_post_forward = self.config.enable_pre_and_post_forward
 
-        # See the comments in config.py for more details of this option.
-        if (
-            self.config.enable_pre_and_post_forward
-            and not self.config.force_recompute_fp8_weight_in_bwd
-        ):
-            logger.warning(
-                "When using FSDP, it's recommended to enable config.force_recompute_fp8_weight_in_bwd."
-            )
-
     def create_buffers(self):
         # Default values for history buffers, see above TODO
         history_len = self.config.delayed_scaling_config.history_len