warp-lang 1.7.0__py3-none-manylinux_2_34_aarch64.whl

warp/torch.py

@@ -0,0 +1,391 @@
+# SPDX-FileCopyrightText: Copyright (c) 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import ctypes
+
+import numpy
+
+import warp
+import warp.context
+
+
+# return the warp device corresponding to a torch device
+def device_from_torch(torch_device) -> warp.context.Device:
+    """Return the Warp device corresponding to a Torch device.
+
+    Args:
+        torch_device (`torch.device` or `str`): Torch device identifier
+
+    Raises:
+        RuntimeError: Torch device does not have a corresponding Warp device
+    """
+    if type(torch_device) is str:
+        warp_device = warp.context.runtime.device_map.get(torch_device)
+        if warp_device is not None:
+            return warp_device
+        elif torch_device == "cuda":
+            return warp.context.runtime.get_current_cuda_device()
+        else:
+            raise RuntimeError(f"Unsupported Torch device {torch_device}")
+    else:
+        try:
+            if torch_device.type == "cuda":
+                return warp.context.runtime.cuda_devices[torch_device.index]
+            elif torch_device.type == "cpu":
+                return warp.context.runtime.cpu_device
+            else:
+                raise RuntimeError(f"Unsupported Torch device type {torch_device.type}")
+        except Exception as e:
+            import torch
+
+            if not isinstance(torch_device, torch.device):
+                raise ValueError("Argument must be a torch.device object or a string") from e
+            raise
+
+
+def device_to_torch(warp_device: warp.context.Devicelike) -> str:
+    """Return the Torch device string corresponding to a Warp device.
+
+    Args:
+        warp_device: An identifier that can be resolved to a :class:`warp.context.Device`.
+
+    Raises:
+        RuntimeError: The Warp device is not compatible with PyTorch.
+    """
+    device = warp.get_device(warp_device)
+    if device.is_cpu or device.is_primary:
+        return str(device)
+    elif device.is_cuda and device.is_uva:
+        # it's not a primary context, but torch can access the data ptr directly thanks to UVA
+        return f"cuda:{device.ordinal}"
+    raise RuntimeError(f"Warp device {device} is not compatible with torch")
+
+
+def dtype_to_torch(warp_dtype):
+    """Return the Torch dtype corresponding to a Warp dtype.
+
+    Args:
+        warp_dtype: A Warp data type that has a corresponding ``torch.dtype``.
+            ``warp.uint16``, ``warp.uint32``, and ``warp.uint64`` are mapped
+            to the signed integer ``torch.dtype`` of the same width.
+    Raises:
+        TypeError: Unable to find a corresponding PyTorch data type.
+    """
+    # initialize lookup table on first call to defer torch import
+    if dtype_to_torch.type_map is None:
+        import torch
+
+        dtype_to_torch.type_map = {
+            warp.float16: torch.float16,
+            warp.float32: torch.float32,
+            warp.float64: torch.float64,
+            warp.int8: torch.int8,
+            warp.int16: torch.int16,
+            warp.int32: torch.int32,
+            warp.int64: torch.int64,
+            warp.uint8: torch.uint8,
+            # torch doesn't support unsigned ints bigger than 8 bits
+            warp.uint16: torch.int16,
+            warp.uint32: torch.int32,
+            warp.uint64: torch.int64,
+            warp.bool: torch.bool,
+        }
+
+    torch_dtype = dtype_to_torch.type_map.get(warp_dtype)
+    if torch_dtype is not None:
+        return torch_dtype
+    else:
+        raise TypeError(f"Cannot convert {warp_dtype} to a Torch type")
+
+
+def dtype_from_torch(torch_dtype):
+    """Return the Warp dtype corresponding to a Torch dtype.
+
+    Args:
+        torch_dtype: A ``torch.dtype`` that has a corresponding Warp data type.
+            Currently ``torch.bfloat16``, ``torch.complex64``, and
+            ``torch.complex128`` are not supported.
+
+    Raises:
+        TypeError: Unable to find a corresponding Warp data type.
+    """
+    # initialize lookup table on first call to defer torch import
+    if dtype_from_torch.type_map is None:
+        import torch
+
+        dtype_from_torch.type_map = {
+            torch.float16: warp.float16,
+            torch.float32: warp.float32,
+            torch.float64: warp.float64,
+            torch.int8: warp.int8,
+            torch.int16: warp.int16,
+            torch.int32: warp.int32,
+            torch.int64: warp.int64,
+            torch.uint8: warp.uint8,
+            torch.bool: warp.bool,
+            # currently unsupported by Warp
+            # torch.bfloat16:
+            # torch.complex64:
+            # torch.complex128:
+        }
+
+    warp_dtype = dtype_from_torch.type_map.get(torch_dtype)
+
+    if warp_dtype is not None:
+        return warp_dtype
+    else:
+        raise TypeError(f"Cannot convert {torch_dtype} to a Warp type")
+
+
+def dtype_is_compatible(torch_dtype, warp_dtype) -> bool:
+    """Evaluates whether the given torch dtype is compatible with the given Warp dtype."""
+    # initialize lookup table on first call to defer torch import
+    if dtype_is_compatible.compatible_sets is None:
+        import torch
+
+        dtype_is_compatible.compatible_sets = {
+            torch.float64: {warp.float64},
+            torch.float32: {warp.float32},
+            torch.float16: {warp.float16},
+            # allow aliasing integer tensors as signed or unsigned integer arrays
+            torch.int64: {warp.int64, warp.uint64},
+            torch.int32: {warp.int32, warp.uint32},
+            torch.int16: {warp.int16, warp.uint16},
+            torch.int8: {warp.int8, warp.uint8},
+            torch.uint8: {warp.uint8, warp.int8},
+            torch.bool: {warp.bool, warp.uint8, warp.int8},
+            # currently unsupported by Warp
+            # torch.bfloat16:
+            # torch.complex64:
+            # torch.complex128:
+        }
+
+    compatible_set = dtype_is_compatible.compatible_sets.get(torch_dtype)
+
+    if compatible_set is not None:
+        if warp_dtype in compatible_set:
+            return True
+        # check if it's a vector or matrix type
+        if hasattr(warp_dtype, "_wp_scalar_type_"):
+            return warp_dtype._wp_scalar_type_ in compatible_set
+
+    return False
+
+
+# lookup tables initialized when needed
+dtype_from_torch.type_map = None
+dtype_to_torch.type_map = None
+dtype_is_compatible.compatible_sets = None
+
+
+# wrap a torch tensor to a wp array, data is not copied
+def from_torch(t, dtype=None, requires_grad=None, grad=None, return_ctype=False):
+    """Convert a Torch tensor to a Warp array without copying the data.
+
+    Args:
+        t (torch.Tensor): The torch tensor to wrap.
+        dtype (warp.dtype, optional): The target data type of the resulting Warp array. Defaults to the tensor value type mapped to a Warp array value type.
+        requires_grad (bool, optional): Whether the resulting array should wrap the tensor's gradient, if it exists (the grad tensor will be allocated otherwise). Defaults to the tensor's `requires_grad` value.
+        return_ctype (bool, optional): Whether to return a low-level array descriptor instead of a ``wp.array`` object (faster).  The descriptor can be passed to Warp kernels.
+
+    Returns:
+        warp.array: The wrapped array or array descriptor.
+    """
+    if dtype is None:
+        dtype = dtype_from_torch(t.dtype)
+    elif not dtype_is_compatible(t.dtype, dtype):
+        raise RuntimeError(f"Cannot convert Torch type {t.dtype} to Warp type {dtype}")
+
+    # get size of underlying data type to compute strides
+    ctype_size = ctypes.sizeof(dtype._type_)
+
+    shape = tuple(t.shape)
+    strides = tuple(s * ctype_size for s in t.stride())
+
+    # if target is a vector or matrix type
+    # then check if trailing dimensions match
+    # the target type and update the shape
+    if hasattr(dtype, "_shape_"):
+        dtype_shape = dtype._shape_
+        dtype_dims = len(dtype._shape_)
+        # ensure inner shape matches
+        if dtype_dims > len(shape) or dtype_shape != shape[-dtype_dims:]:
+            raise RuntimeError(
+                f"Could not convert Torch tensor with shape {shape} to Warp array with dtype={dtype}, ensure that source inner shape is {dtype_shape}"
+            )
+        # ensure inner strides are contiguous
+        if strides[-1] != ctype_size or (dtype_dims > 1 and strides[-2] != ctype_size * dtype_shape[-1]):
+            raise RuntimeError(
+                f"Could not convert Torch tensor with shape {shape} to Warp array with dtype={dtype}, because the source inner strides are not contiguous"
+            )
+        # trim shape and strides
+        shape = tuple(shape[:-dtype_dims]) or (1,)
+        strides = tuple(strides[:-dtype_dims]) or (ctype_size,)
+
+    # gradient
+    # - if return_ctype is False, we set `grad` to a wp.array or None
+    # - if return_ctype is True, we set `grad_ptr` and set `grad` as the owner (wp.array or torch.Tensor)
+    requires_grad = t.requires_grad if requires_grad is None else requires_grad
+    grad_ptr = 0
+    if grad is not None:
+        if isinstance(grad, warp.array):
+            if return_ctype:
+                if grad.strides != strides:
+                    raise RuntimeError(
+                        f"Gradient strides must match array strides, expected {strides} but got {grad.strides}"
+                    )
+                grad_ptr = grad.ptr
+        else:
+            # assume grad is a torch.Tensor
+            if return_ctype:
+                if t.stride() != grad.stride():
+                    raise RuntimeError(
+                        f"Gradient strides must match array strides, expected {t.stride()} but got {grad.stride()}"
+                    )
+                grad_ptr = grad.data_ptr()
+            else:
+                grad = from_torch(grad, dtype=dtype, requires_grad=False)
+    elif requires_grad:
+        # wrap the tensor gradient, allocate if necessary
+        if t.grad is not None:
+            if return_ctype:
+                grad = t.grad
+                if t.stride() != grad.stride():
+                    raise RuntimeError(
+                        f"Gradient strides must match array strides, expected {t.stride()} but got {grad.stride()}"
+                    )
+                grad_ptr = grad.data_ptr()
+            else:
+                grad = from_torch(t.grad, dtype=dtype, requires_grad=False)
+        else:
+            # allocate a zero-filled gradient if it doesn't exist
+            # Note: we use Warp to allocate the shared gradient with compatible strides
+            grad = warp.zeros(dtype=dtype, shape=shape, strides=strides, device=device_from_torch(t.device))
+            t.grad = to_torch(grad, requires_grad=False)
+            grad_ptr = grad.ptr
+
+    if return_ctype:
+        ptr = t.data_ptr()
+
+        # create array descriptor
+        array_ctype = warp.types.array_t(ptr, grad_ptr, len(shape), shape, strides)
+
+        # keep data and gradient alive
+        array_ctype._ref = t
+        array_ctype._gradref = grad
+
+        return array_ctype
+
+    else:
+        a = warp.array(
+            ptr=t.data_ptr(),
+            dtype=dtype,
+            shape=shape,
+            strides=strides,
+            device=device_from_torch(t.device),
+            copy=False,
+            grad=grad,
+            requires_grad=requires_grad,
+        )
+
+        # save a reference to the source tensor, otherwise it may get deallocated
+        a._tensor = t
+
+        return a
+
+
+def to_torch(a, requires_grad=None):
+    """
+    Convert a Warp array to a Torch tensor without copying the data.
+
+    Args:
+        a (warp.array): The Warp array to convert.
+        requires_grad (bool, optional): Whether the resulting tensor should convert the array's gradient, if it exists, to a grad tensor. Defaults to the array's `requires_grad` value.
+
+    Returns:
+        torch.Tensor: The converted tensor.
+    """
+    import torch
+
+    if requires_grad is None:
+        requires_grad = a.requires_grad
+
+    # Torch does not support structured arrays
+    if isinstance(a.dtype, warp.codegen.Struct):
+        raise RuntimeError("Cannot convert structured Warp arrays to Torch.")
+
+    if a.device.is_cpu:
+        # Torch has an issue wrapping CPU objects
+        # that support the __array_interface__ protocol
+        # in this case we need to workaround by going
+        # to an ndarray first, see https://pearu.github.io/array_interface_pytorch.html
+        t = torch.as_tensor(numpy.asarray(a))
+        t.requires_grad = requires_grad
+        if requires_grad and a.requires_grad:
+            t.grad = torch.as_tensor(numpy.asarray(a.grad))
+        return t
+
+    elif a.device.is_cuda:
+        # Torch does support the __cuda_array_interface__
+        # correctly, but we must be sure to maintain a reference
+        # to the owning object to prevent memory allocs going out of scope
+        t = torch.as_tensor(a, device=device_to_torch(a.device))
+        t.requires_grad = requires_grad
+        if requires_grad and a.requires_grad:
+            t.grad = torch.as_tensor(a.grad, device=device_to_torch(a.device))
+        return t
+
+    else:
+        raise RuntimeError("Unsupported device")
+
+
+def stream_from_torch(stream_or_device=None):
+    """Convert from a Torch CUDA stream to a Warp CUDA stream."""
+    import torch
+
+    if isinstance(stream_or_device, torch.cuda.Stream):
+        stream = stream_or_device
+    else:
+        # assume arg is a torch device
+        stream = torch.cuda.current_stream(stream_or_device)
+
+    device = device_from_torch(stream.device)
+
+    warp_stream = warp.Stream(device, cuda_stream=stream.cuda_stream)
+
+    # save a reference to the source stream, otherwise it may be destroyed
+    warp_stream._torch_stream = stream
+
+    return warp_stream
+
+
+def stream_to_torch(stream_or_device=None):
+    """Convert from a Warp CUDA stream to a Torch CUDA stream."""
+    import torch
+
+    if isinstance(stream_or_device, warp.Stream):
+        stream = stream_or_device
+    else:
+        # assume arg is a warp device
+        stream = warp.get_device(stream_or_device).stream
+
+    device = device_to_torch(stream.device)
+
+    torch_stream = torch.cuda.ExternalStream(stream.cuda_stream, device=device)
+
+    # save a reference to the source stream, otherwise it may be destroyed
+    torch_stream._warp_stream = stream
+
+    return torch_stream