issue/205 - 添加Sub算子

issue/205 - 添加Sub算子的头文件、CPU实现、cuda实现、及Python测试

include/infiniop.h

@@ -19,6 +19,7 @@
 #include "infiniop/ops/relu.h"
 #include "infiniop/ops/rms_norm.h"
 #include "infiniop/ops/rope.h"
+#include "infiniop/ops/sub.h"
 #include "infiniop/ops/swiglu.h"
 #include "infiniop/tensor_descriptor.h"
 

include/infiniop/ops/sub.h

+#ifndef __INFINIOP_SUB_API_H__
+#define __INFINIOP_SUB_API_H__
+
+#include "../operator_descriptor.h"
+
+typedef struct InfiniopDescriptor *infiniopSubDescriptor_t;
+
+__C __export infiniStatus_t infiniopCreateSubDescriptor(infiniopHandle_t handle,
+                                                        infiniopSubDescriptor_t *desc_ptr,
+                                                        infiniopTensorDescriptor_t c,
+                                                        infiniopTensorDescriptor_t a,
+                                                        infiniopTensorDescriptor_t b);
+
+__C __export infiniStatus_t infiniopGetSubWorkspaceSize(infiniopSubDescriptor_t desc, size_t *size);
+
+__C __export infiniStatus_t infiniopSub(infiniopSubDescriptor_t desc,
+                                        void *workspace,
+                                        size_t workspace_size,
+                                        void *c,
+                                        void const *a,
+                                        void const *b,
+                                        void *stream);
+
+__C __export infiniStatus_t infiniopDestroySubDescriptor(infiniopSubDescriptor_t desc);
+
+#endif

scripts/python_test.py

@@ -17,6 +17,7 @@ def run_tests(args):
         "random_sample.py",
         "rms_norm.py",
         "rope.py",
+        "sub.py",
         "swiglu.py",
         "attention.py",
     ]:

src/infiniop/ops/sub/cpu/sub_cpu.cc

+#include "sub_cpu.h"
+
+namespace op::sub::cpu {
+
+Descriptor::~Descriptor() = default;
+
+infiniStatus_t Descriptor::create(
+    infiniopHandle_t handle_,
+    Descriptor **desc_ptr,
+    infiniopTensorDescriptor_t out_desc,
+    std::vector<infiniopTensorDescriptor_t> input_desc_vec) {
+
+    auto handle = reinterpret_cast<device::cpu::Handle *>(handle_);
+    auto dtype = out_desc->dtype();
+
+    const auto &a_desc = input_desc_vec.at(0);
+    const auto &b_desc = input_desc_vec.at(1);
+    const auto &c_shape = out_desc->shape();
+    const auto &a_shape = a_desc->shape();
+    const auto &b_shape = b_desc->shape();
+
+    CHECK_DTYPE(dtype, INFINI_DTYPE_F16, INFINI_DTYPE_F32, INFINI_DTYPE_F64);
+
+    CHECK_SAME_SHAPE(c_shape, a_shape, b_shape);
+
+    // create CPU elementwise descriptor
+    CREATE_ELEMENTWISE_CPU_DESCRIPTOR(handle, dtype, out_desc, input_desc_vec);
+
+    return INFINI_STATUS_SUCCESS;
+}
+
+infiniStatus_t Descriptor::calculate(
+    void *workspace,
+    size_t workspace_size,
+    void *output,
+    std::vector<const void *> inputs,
+    void *stream) const {
+
+    switch (_dtype) {
+    case INFINI_DTYPE_F16:
+        return _device_info->calculate<SubOp, fp16_t>(_info, output, inputs, stream);
+    case INFINI_DTYPE_F32:
+        return _device_info->calculate<SubOp, float>(_info, output, inputs, stream);
+    case INFINI_DTYPE_F64:
+        return _device_info->calculate<SubOp, double>(_info, output, inputs, stream);
+    default:
+        return INFINI_STATUS_BAD_TENSOR_DTYPE;
+    }
+
+    return INFINI_STATUS_SUCCESS;
+}
+} // namespace op::sub::cpu

src/infiniop/ops/sub/cpu/sub_cpu.h

+#ifndef __SUB_CPU_H__
+#define __SUB_CPU_H__
+
+#include "../../../elementwise/cpu/elementwise_cpu.h"
+
+ELEMENTWISE_DESCRIPTOR(sub, cpu)
+
+namespace op::sub::cpu {
+typedef struct SubOp {
+public:
+    static constexpr size_t num_inputs = 2;
+    template <typename T>
+    T operator()(const T &a, const T &b) const {
+        return a - b;
+    }
+} SubOp;
+} // namespace op::sub::cpu
+
+#endif // __SUB_CPU_H__

src/infiniop/ops/sub/cuda/sub_cuda.cu

+#include "sub_cuda.cuh"
+#include "sub_cuda_internal.cuh"
+
+namespace op::sub::cuda {
+
+Descriptor::~Descriptor() = default;
+
+infiniStatus_t Descriptor::create(
+    infiniopHandle_t handle_,
+    Descriptor **desc_ptr,
+    infiniopTensorDescriptor_t out_desc,
+    std::vector<infiniopTensorDescriptor_t> input_desc_vec) {
+
+    auto handle = reinterpret_cast<device::cuda::Handle *>(handle_);
+    auto dtype = out_desc->dtype();
+
+    const auto &a_desc = input_desc_vec.at(0);
+    const auto &b_desc = input_desc_vec.at(1);
+    const auto &c_shape = out_desc->shape();
+    const auto &a_shape = a_desc->shape();
+    const auto &b_shape = b_desc->shape();
+
+    CHECK_DTYPE(dtype, INFINI_DTYPE_F16, INFINI_DTYPE_F32, INFINI_DTYPE_F64);
+
+    CHECK_SAME_SHAPE(c_shape, a_shape, b_shape);
+
+    // create CUDA elementwise descriptor
+    CREATE_ELEMENTWISE_CUDA_DESCRIPTOR(handle, dtype, out_desc, input_desc_vec)
+
+    return INFINI_STATUS_SUCCESS;
+}
+
+infiniStatus_t Descriptor::calculate(
+    void *workspace,
+    size_t workspace_size,
+    void *output,
+    std::vector<const void *> inputs,
+    void *stream) const {
+
+    if (workspace_size < _workspace_size) {
+        return INFINI_STATUS_INSUFFICIENT_WORKSPACE;
+    }
+
+    switch (_dtype) {
+    case INFINI_DTYPE_F16:
+        return _device_info->calculate<256, SubOp, half>(_info, workspace, output, inputs, stream);
+    case INFINI_DTYPE_F32:
+        return _device_info->calculate<256, SubOp, float>(_info, workspace, output, inputs, stream);
+    case INFINI_DTYPE_F64:
+        return _device_info->calculate<256, SubOp, double>(_info, workspace, output, inputs, stream);
+    default:
+        return INFINI_STATUS_BAD_TENSOR_DTYPE;
+    }
+
+    return INFINI_STATUS_SUCCESS;
+}
+} // namespace op::sub::cuda

src/infiniop/ops/sub/cuda/sub_cuda.cuh

+#ifndef __SUB_CUDA_API_H__
+#define __SUB_CUDA_API_H__
+
+#include "../../../elementwise/cuda/elementwise_cuda_api.cuh"
+
+ELEMENTWISE_DESCRIPTOR(sub, cuda)
+
+#endif // __SUB_CUDA_API_H__

src/infiniop/ops/sub/cuda/sub_cuda_internal.cuh

+#ifndef __SUB_CUDA_H__
+#define __SUB_CUDA_H__
+
+#include "../../../elementwise/cuda/elementwise_cuda.cuh"
+#include <cuda_fp16.h>
+
+namespace op::sub::cuda {
+typedef struct SubOp {
+public:
+    static constexpr size_t num_inputs = 2;
+    template <typename T>
+    __device__ __forceinline__ T operator()(const T &a, const T &b) const {
+        if constexpr (std::is_same_v<T, half2>) {
+            return __hsub2(a, b);
+        } else if constexpr (std::is_same_v<T, half>) {
+            return __hsub(a, b);
+        } else if constexpr (std::is_same_v<T, float>) {
+            return __fsub_rd(a, b);
+        } else {
+            return a - b;
+        }
+    }
+} SubOp;
+} // namespace op::sub::cuda
+
+#endif // __SUB_CUDA_H__

src/infiniop/ops/sub/operator.cc

+#include "../../operator.h"
+#include "../../handle.h"
+#include "infiniop/ops/sub.h"
+
+#ifdef ENABLE_CPU_API
+#include "cpu/sub_cpu.h"
+#endif
+#ifdef ENABLE_CUDA_API
+#include "cuda/sub_cuda.cuh"
+#endif
+
+__C infiniStatus_t infiniopCreateSubDescriptor(
+    infiniopHandle_t handle,
+    infiniopSubDescriptor_t *desc_ptr,
+    infiniopTensorDescriptor_t c_desc,
+    infiniopTensorDescriptor_t a_desc,
+    infiniopTensorDescriptor_t b_desc) {
+
+#define CREATE(CASE, NAMESPACE)                                            \
+    case CASE:                                                             \
+        return op::sub::NAMESPACE::Descriptor::create(                     \
+            handle,                                                        \
+            reinterpret_cast<op::sub::NAMESPACE::Descriptor **>(desc_ptr), \
+            c_desc,                                                        \
+            {a_desc,                                                       \
+             b_desc})
+
+    switch (handle->device) {
+
+#ifdef ENABLE_CPU_API
+        CREATE(INFINI_DEVICE_CPU, cpu);
+#endif
+#ifdef ENABLE_CUDA_API
+        CREATE(INFINI_DEVICE_NVIDIA, cuda);
+#endif
+
+    default:
+        return INFINI_STATUS_DEVICE_TYPE_NOT_SUPPORTED;
+    }
+
+#undef CREATE
+}
+
+__C infiniStatus_t infiniopGetSubWorkspaceSize(infiniopSubDescriptor_t desc, size_t *size) {
+
+#define GET(CASE, NAMESPACE)                                                               \
+    case CASE:                                                                             \
+        *size = reinterpret_cast<op::sub::NAMESPACE::Descriptor *>(desc)->workspaceSize(); \
+        return INFINI_STATUS_SUCCESS;
+
+    switch (desc->device_type) {
+#ifdef ENABLE_CPU_API
+        GET(INFINI_DEVICE_CPU, cpu)
+#endif
+#ifdef ENABLE_CUDA_API
+        GET(INFINI_DEVICE_NVIDIA, cuda)
+#endif
+    default:
+        return INFINI_STATUS_DEVICE_TYPE_NOT_SUPPORTED;
+    }
+#undef GET
+
+    return INFINI_STATUS_DEVICE_TYPE_NOT_SUPPORTED;
+}
+
+__C infiniStatus_t infiniopSub(
+    infiniopSubDescriptor_t desc,
+    void *workspace,
+    size_t workspace_size,
+    void *c,
+    const void *a,
+    const void *b,
+    void *stream) {
+
+#define CALCULATE(CASE, NAMESPACE)                                            \
+    case CASE:                                                                \
+        return reinterpret_cast<const op::sub::NAMESPACE::Descriptor *>(desc) \
+            ->calculate(workspace, workspace_size, c, {a, b}, stream)
+
+    switch (desc->device_type) {
+
+#ifdef ENABLE_CPU_API
+        CALCULATE(INFINI_DEVICE_CPU, cpu);
+#endif
+#ifdef ENABLE_CUDA_API
+        CALCULATE(INFINI_DEVICE_NVIDIA, cuda);
+#endif
+
+    default:
+        return INFINI_STATUS_DEVICE_TYPE_NOT_SUPPORTED;
+    }
+
+#undef CALCULATE
+}
+
+__C infiniStatus_t
+infiniopDestroySubDescriptor(infiniopSubDescriptor_t desc) {
+
+#define DELETE(CASE, NAMESPACE)                                                \
+    case CASE:                                                                 \
+        delete reinterpret_cast<const op::sub::NAMESPACE::Descriptor *>(desc); \
+        return INFINI_STATUS_SUCCESS;
+
+    switch (desc->device_type) {
+
+#ifdef ENABLE_CPU_API
+        DELETE(INFINI_DEVICE_CPU, cpu);
+#endif
+#ifdef ENABLE_CUDA_API
+        DELETE(INFINI_DEVICE_NVIDIA, cuda);
+#endif
+
+    default:
+        return INFINI_STATUS_DEVICE_TYPE_NOT_SUPPORTED;
+    }
+
+#undef DELETE
+}

test/infiniop/sub.py

+import torch
+import ctypes
+from ctypes import POINTER, Structure, c_int32, c_void_p, c_uint64
+from libinfiniop import (
+    infiniopHandle_t,
+    infiniopTensorDescriptor_t,
+    open_lib,
+    to_tensor,
+    get_test_devices,
+    check_error,
+    rearrange_if_needed,
+    test_operator,
+    get_args,
+    debug,
+    get_tolerance,
+    profile_operation,
+    create_workspace,
+)
+from enum import Enum, auto
+
+# ==============================================================================
+#  Configuration (Internal Use Only)
+# ==============================================================================
+# These are not meant to be imported from other modules
+_TEST_CASES_ = [
+    # shape, a_stride, b_stride, c_stride
+    ((13, 4), None, None, None),
+    ((13, 4), (10, 1), (10, 1), (10, 1)),
+    ((13, 4), (0, 1), None, None),
+    ((13, 4, 4), None, None, None),
+    ((13, 4, 4), (20, 4, 1), (20, 4, 1), (20, 4, 1)),
+    ((13, 4, 4), (4, 0, 1), (0, 4, 1), None),
+    ((16, 5632), None, None, None),
+    ((16, 5632), (13312, 1), (13312, 1), (13312, 1)),
+    ((4, 4, 5632), None, None, None),
+    ((4, 4, 5632), (45056, 5632, 1), (45056, 5632, 1), (45056, 5632, 1)),
+]
+
+
+class Inplace(Enum):
+    OUT_OF_PLACE = auto()
+    INPLACE_A = auto()
+    INPLACE_B = auto()
+
+
+# Inplace options applied for each test case in _TEST_CASES_
+_INPLACE = [
+    Inplace.OUT_OF_PLACE,
+    Inplace.INPLACE_A,
+    Inplace.INPLACE_B,
+]
+
+# Form the test cases by appending each element of _INPLACE to each tuple in _TEST_CASES_
+_TEST_CASES = [
+    test_case + (inplace_item,)
+    for test_case in _TEST_CASES_
+    for inplace_item in _INPLACE
+]
+
+# Data types used for testing
+_TENSOR_DTYPES = [torch.float16, torch.float32]
+
+# Tolerance map for different data types
+_TOLERANCE_MAP = {
+    torch.float16: {"atol": 1e-3, "rtol": 1e-3},
+    torch.float32: {"atol": 1e-7, "rtol": 1e-7},
+}
+
+DEBUG = False
+PROFILE = False
+NUM_PRERUN = 10
+NUM_ITERATIONS = 1000
+
+
+class SubDescriptor(Structure):
+    _fields_ = [("device", c_int32)]
+
+
+infiniopSubDescriptor_t = POINTER(SubDescriptor)
+
+
+def sub(x, y):
+    return torch.sub(x, y)
+
+
+def process_tensors(c, c_strides, a, a_stride, b, b_stride, inplace):
+    """
+    rearrange the tensors if needed and apply the inplace config.
+    if inplace is true and the output (i.e., c) is placed to the broadcasted input,
+    the inplace config is ignored and out-of-place is used
+    """
+    original_c_strides = c_strides if c_strides else c.stride()
+
+    def _rearrange(tensor, strides):
+        if strides and 0 in strides:
+            tensor.set_(tensor.untyped_storage(), 0, tensor.shape, strides)
+            return tensor
+        else:
+            return rearrange_if_needed(tensor, strides)
+
+    a, b, c = [
+        _rearrange(tensor, stride)
+        for tensor, stride in zip([a, b, c], [a_stride, b_stride, c_strides])
+    ]
+    c = (
+        c
+        if inplace == Inplace.OUT_OF_PLACE
+        else (a if inplace == Inplace.INPLACE_A else b)
+    )
+    # if inplace is true and c has broadcasted config, reset it to the original unbroadcasted strides
+    if 0 in c.stride():
+        c.set_(c.untyped_storage(), 0, c.shape, original_c_strides)
+
+    return a, b, c
+
+
+def test(
+    lib,
+    handle,
+    torch_device,
+    shape,
+    a_stride=None,
+    b_stride=None,
+    c_stride=None,
+    inplace=Inplace.OUT_OF_PLACE,
+    dtype=torch.float16,
+    sync=None,
+):
+    print(
+        f"Testing Sub on {torch_device} with shape:{shape} a_stride:{a_stride} b_stride:{b_stride} c_stride:{c_stride} "
+        f"dtype:{dtype} inplace:{inplace}"
+    )
+
+    a = torch.rand(shape, dtype=dtype).to(torch_device)
+    b = torch.rand(shape, dtype=dtype).to(torch_device)
+    c = torch.rand(shape, dtype=dtype).to(torch_device)
+    a, b, c = process_tensors(c, c_stride, a, a_stride, b, b_stride, inplace)
+
+    ans = sub(a, b)
+
+    a_tensor, b_tensor = [to_tensor(tensor, lib) for tensor in [a, b]]
+    c_tensor = (
+        to_tensor(c, lib)
+        if inplace == Inplace.OUT_OF_PLACE
+        else (a_tensor if inplace == Inplace.INPLACE_A else b_tensor)
+    )
+    if sync is not None:
+        sync()
+
+    descriptor = infiniopSubDescriptor_t()
+    check_error(
+        lib.infiniopCreateSubDescriptor(
+            handle,
+            ctypes.byref(descriptor),
+            c_tensor.descriptor,
+            a_tensor.descriptor,
+            b_tensor.descriptor,
+        )
+    )
+
+    # Invalidate the shape and strides in the descriptor to prevent them from being directly used by the kernel
+    for tensor in [a_tensor, b_tensor, c_tensor]:
+        tensor.destroyDesc(lib)
+
+    workspace_size = c_uint64(0)
+    check_error(
+        lib.infiniopGetSubWorkspaceSize(descriptor, ctypes.byref(workspace_size))
+    )
+    workspace = create_workspace(workspace_size.value, c.device)
+
+    def lib_sub():
+        check_error(
+            lib.infiniopSub(
+                descriptor,
+                workspace.data_ptr() if workspace is not None else None,
+                workspace_size.value,
+                c_tensor.data,
+                a_tensor.data,
+                b_tensor.data,
+                None,
+            )
+        )
+
+    lib_sub()
+
+    atol, rtol = get_tolerance(_TOLERANCE_MAP, dtype)
+    if DEBUG:
+        debug(c, ans, atol=atol, rtol=rtol)
+    assert torch.allclose(c, ans, atol=atol, rtol=rtol)
+
+    # Profiling workflow
+    if PROFILE:
+        # fmt: off
+        profile_operation("PyTorch", lambda: sub(a, b), torch_device, NUM_PRERUN, NUM_ITERATIONS)
+        profile_operation("    lib", lambda: lib_sub(), torch_device, NUM_PRERUN, NUM_ITERATIONS)
+        # fmt: on
+    check_error(lib.infiniopDestroySubDescriptor(descriptor))
+
+
+if __name__ == "__main__":
+    args = get_args()
+    lib = open_lib()
+
+    lib.infiniopCreateSubDescriptor.restype = c_int32
+    lib.infiniopCreateSubDescriptor.argtypes = [
+        infiniopHandle_t,
+        POINTER(infiniopSubDescriptor_t),
+        infiniopTensorDescriptor_t,
+        infiniopTensorDescriptor_t,
+        infiniopTensorDescriptor_t,
+    ]
+
+    lib.infiniopGetSubWorkspaceSize.restype = c_int32
+    lib.infiniopGetSubWorkspaceSize.argtypes = [
+        infiniopSubDescriptor_t,
+        POINTER(c_uint64),
+    ]
+
+    lib.infiniopSub.restype = c_int32
+    lib.infiniopSub.argtypes = [
+        infiniopSubDescriptor_t,
+        c_void_p,
+        c_uint64,
+        c_void_p,
+        c_void_p,
+        c_void_p,
+        c_void_p,
+    ]
+
+    lib.infiniopDestroySubDescriptor.restype = c_int32
+    lib.infiniopDestroySubDescriptor.argtypes = [
+        infiniopSubDescriptor_t,
+    ]
+
+    # Configure testing options
+    DEBUG = args.debug
+    PROFILE = args.profile
+    NUM_PRERUN = args.num_prerun
+    NUM_ITERATIONS = args.num_iterations
+
+    for device in get_test_devices(args):
+        test_operator(lib, device, test, _TEST_CASES, _TENSOR_DTYPES)
+
+    print("\033[92mTest passed!\033[0m")