Skip to content

Add inline to avoid duplicated-symbols-error #7

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jun 28, 2024
Merged

Add inline to avoid duplicated-symbols-error #7

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3affc31
Add inline to avoid duplicated-symbols-error
junjihashimoto Jun 27, 2024
3958b23
Merge branch 'main' into fix/duplicated-symbols
austinvhuang Jun 28, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
70 changes: 35 additions & 35 deletions gpu.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,23 +43,23 @@ struct Shape {
// dynamic allocation for shape data
std::array<size_t, kMaxRank> data = {0};
size_t rank = 0;
Shape() = default;
Shape(std::initializer_list<size_t> dims) {
inline Shape() = default;
inline Shape(std::initializer_list<size_t> dims) {
assert(dims.size() <= kMaxRank);
std::copy(dims.begin(), dims.end(), data.begin());
rank = dims.size();
}
size_t &operator[](size_t index) {
inline size_t &operator[](size_t index) {
assert(index < rank);
return data[index];
}
const size_t &operator[](size_t index) const {
inline const size_t &operator[](size_t index) const {
assert(index < rank);
return data[index];
}
};

size_t size(const Shape &shape) {
inline size_t size(const Shape &shape) {
size_t numels = 1;
for (size_t i = 0; i < shape.rank; i++) {
numels *= shape.data[i];
Expand Down Expand Up @@ -102,7 +102,7 @@ struct Context; // Forward declaration so that TensorPool can have a pointer to
// Context

struct TensorPool {
TensorPool(Context *ctx) : ctx(ctx), data() {};
inline TensorPool(Context *ctx) : ctx(ctx), data() {};
Context *ctx;
std::unordered_map<WGPUBuffer, Tensor> data;
~TensorPool();
Expand All @@ -113,7 +113,7 @@ enum NumType { kf32 };
/**
* @brief Converts NumType to string.
*/
std::string ToString(NumType type) {
inline std::string ToString(NumType type) {
switch (type) {
case kf32:
return "f32";
Expand All @@ -128,7 +128,7 @@ std::string ToString(NumType type) {
* is meant to be slotted into shader code (hence no additional parentheses or
* brackets).
*/
std::string ToString(const Shape &shape) {
inline std::string ToString(const Shape &shape) {
std::string str;
for (size_t i = 0; i < shape.rank; i++) {
str += std::to_string(shape.data[i]);
Expand All @@ -146,11 +146,11 @@ std::string ToString(const Shape &shape) {
* consistent.
*/
struct ShaderCode {
ShaderCode(const std::string &data = "", size_t workgroupSize = 256,
inline ShaderCode(const std::string &data = "", size_t workgroupSize = 256,
NumType precision = kf32)
: data(data), workgroupSize({workgroupSize, 1, 1}), precision(precision) {
}
ShaderCode(const std::string &data, const Shape &workgroupSize = {256, 1, 1},
inline ShaderCode(const std::string &data, const Shape &workgroupSize = {256, 1, 1},
NumType precision = kf32)
: data(data), workgroupSize(workgroupSize), precision(precision) {}
std::string data;
Expand Down Expand Up @@ -192,7 +192,7 @@ struct Kernel {
* @param[in] rhs Second Kernel instance to compare
* @return True if lhs < rhs, false otherwise
*/
bool operator<(const Kernel &lhs, const Kernel &rhs) {
inline bool operator<(const Kernel &lhs, const Kernel &rhs) {
return lhs.commandBuffer < rhs.commandBuffer;
}

Expand All @@ -202,10 +202,10 @@ bool operator<(const Kernel &lhs, const Kernel &rhs) {
* have multiple resource pools of kernels in more complex scenarios.
*/
struct KernelPool {
KernelPool(Context *ctx) : ctx(ctx), data() {}
inline KernelPool(Context *ctx) : ctx(ctx), data() {}
Context *ctx;
std::set<Kernel *> data;
~KernelPool() {
inline ~KernelPool() {
// Note : Some kernel resources such as commandBuffer are harvested by
// queue submission, explicitly destroying readback and callback buffers
// produces runtime errors.
Expand Down Expand Up @@ -274,7 +274,7 @@ struct Context {
* @return Tensor instance representing the created tensor
* @example Tensor tensor = CreateTensor(pool, device, {256, 256}, kf32);
*/
Tensor CreateTensor(TensorPool &pool, WGPUDevice &device, const Shape &shape,
inline Tensor CreateTensor(TensorPool &pool, WGPUDevice &device, const Shape &shape,
NumType dtype,
WGPUBufferUsageFlags usage = WGPUBufferUsage_Storage |
WGPUBufferUsage_CopyDst |
Expand Down Expand Up @@ -314,7 +314,7 @@ Tensor CreateTensor(TensorPool &pool, WGPUDevice &device, const Shape &shape,
* @return Tensor instance representing the created tensor
* @example Tensor tensor = CreateTensor(ctx, {256, 256}, kf32);
*/
Tensor CreateTensor(Context &ctx, const Shape &shape, NumType dtype) {
inline Tensor CreateTensor(Context &ctx, const Shape &shape, NumType dtype) {
return CreateTensor(ctx.pool, ctx.device, shape, dtype);
}

Expand All @@ -333,7 +333,7 @@ Tensor CreateTensor(Context &ctx, const Shape &shape, NumType dtype) {
* @return Tensor instance representing the created tensor
* @example Tensor tensor = CreateTensor(ctx, {256, 256}, kf32, data);
*/
Tensor CreateTensor(Context &ctx, const Shape &shape, NumType dtype,
inline Tensor CreateTensor(Context &ctx, const Shape &shape, NumType dtype,
float *data) {
Tensor tensor =
CreateTensor(ctx.pool, ctx.device, shape, dtype,
Expand All @@ -355,7 +355,7 @@ Tensor CreateTensor(Context &ctx, const Shape &shape, NumType dtype,
* @param[in] tensor Tensor instance to free
* @example FreeTensor(pool, tensor);
*/
void FreeTensor(TensorPool &pool, Tensor tensor) {
inline void FreeTensor(TensorPool &pool, Tensor tensor) {
if (tensor.data.buffer) {
wgpuBufferRelease(tensor.data.buffer);
} else {
Expand All @@ -371,7 +371,7 @@ void FreeTensor(TensorPool &pool, Tensor tensor) {
/**
* @brief Destructor for TensorPool which frees all tensors in the pool.
*/
TensorPool::~TensorPool() {
inline TensorPool::~TensorPool() {
// Need to get keys in a separate iteration, otherwise iterator is getting
// invalidated during erase.
std::vector<WGPUBuffer> keys;
Expand All @@ -397,7 +397,7 @@ TensorPool::~TensorPool() {
* @param[in] to Substring to replace with
* @example ReplaceAll(str, "{{workgroupSize}}", "256");
*/
void ReplaceAll(std::string &str, const std::string &from,
inline void ReplaceAll(std::string &str, const std::string &from,
const std::string &to) {
size_t start_pos = 0;
while ((start_pos = str.find(from, start_pos)) != std::string::npos) {
Expand All @@ -423,7 +423,7 @@ void ReplaceAll(std::string &str, const std::string &from,
* that is returned by CreateShader().
* @example ShaderCode code = CreateShader(kPuzzle1, {256, 1, 1}, kf32);
*/
ShaderCode CreateShader(const char *shaderTemplate,
inline ShaderCode CreateShader(const char *shaderTemplate,
const Shape &workgroupSize = {256, 1, 1},
NumType precision = kf32) {
std::string codeString(shaderTemplate);
Expand All @@ -445,8 +445,8 @@ ShaderCode CreateShader(const char *shaderTemplate,
* @param[in] precision Data type precision for the shader
* @example ShaderCode code = CreateShader(kPuzzle1, 256, kf32);
*/
ShaderCode CreateShader(const char *shaderTemplate, size_t workgroupSize,
NumType precision = kf32) {
inline ShaderCode CreateShader(const char *shaderTemplate, size_t workgroupSize,
NumType precision = kf32) {
return CreateShader(shaderTemplate, Shape{workgroupSize, 1, 1}, precision);
}

Expand Down Expand Up @@ -482,9 +482,9 @@ inline void check(bool condition, const char *message,
* @return Context instance representing the created GPU context
* @example Context ctx = CreateContext();
*/
Context CreateContext(const WGPUInstanceDescriptor &desc = {},
const WGPURequestAdapterOptions &adapterOpts = {},
WGPUDeviceDescriptor devDescriptor = {}) {
inline Context CreateContext(const WGPUInstanceDescriptor &desc = {},
const WGPURequestAdapterOptions &adapterOpts = {},
WGPUDeviceDescriptor devDescriptor = {}) {
Context context;
{
context.instance = wgpuCreateInstance(&desc);
Expand Down Expand Up @@ -562,7 +562,7 @@ Context CreateContext(const WGPUInstanceDescriptor &desc = {},
return context;
}

void Wait(Context &ctx, std::future<void> &future) {
inline void Wait(Context &ctx, std::future<void> &future) {
while (future.wait_for(std::chrono::seconds(0)) !=
std::future_status::ready) {
wgpuInstanceProcessEvents(ctx.instance);
Expand All @@ -577,7 +577,7 @@ void Wait(Context &ctx, std::future<void> &future) {
* @param[in] bufferSize Size of the data buffer in bytes
* @example ToCPU(ctx, tensor, data, bufferSize);
*/
void ToCPU(Context &ctx, Tensor &tensor, float *data, size_t bufferSize) {
inline void ToCPU(Context &ctx, Tensor &tensor, float *data, size_t bufferSize) {
WGPUDevice device = ctx.device;
struct CopyOp {
WGPUCommandBuffer commandBuffer;
Expand Down Expand Up @@ -657,7 +657,7 @@ void ToCPU(Context &ctx, Tensor &tensor, std::array<float, N> data) {
* @param[in] size Size of the data buffer in bytes
* @example ToGPU(ctx, data, buffer, size);
*/
void ToGPU(Context &ctx, const void *data, WGPUBuffer buffer, size_t size) {
inline void ToGPU(Context &ctx, const void *data, WGPUBuffer buffer, size_t size) {
wgpuQueueWriteBuffer(ctx.queue, buffer, 0, data, size);
}

Expand All @@ -669,7 +669,7 @@ void ToGPU(Context &ctx, const void *data, WGPUBuffer buffer, size_t size) {
* @param[in] tensor Tensor instance representing the GPU buffer to copy to
* @example ToGPU(ctx, data, tensor);
*/
void ToGPU(Context &ctx, const float *data, Tensor &tensor) {
inline void ToGPU(Context &ctx, const float *data, Tensor &tensor) {
wgpuQueueWriteBuffer(ctx.queue, tensor.data.buffer, 0, data,
tensor.data.size);
}
Expand All @@ -686,7 +686,7 @@ void ToGPU(Context &ctx, const float *data, Tensor &tensor) {
* @param[in] op Kernel instance representing the kernel to reset
* @example ResetCommandBuffer(device, {256, 1, 1}, op);
*/
void ResetCommandBuffer(WGPUDevice &device, const Shape &nThreads, Kernel &op) {
inline void ResetCommandBuffer(WGPUDevice &device, const Shape &nThreads, Kernel &op) {
LOG(kDefLog, kTrace, "Create command buffer 0x%x", op.commandBuffer);
{
WGPUCommandEncoder commandEncoder =
Expand Down Expand Up @@ -738,10 +738,10 @@ template <typename T> constexpr bool IsNoParam = std::is_same_v<T, NoParam>;
* @example Kernel kernel = CreateKernel(ctx, shader, dataBindings, numInputs, output,
* nThreads, params, paramsSize);
*/
Kernel CreateKernel(Context &ctx, const ShaderCode &shader,
const Tensor *dataBindings, size_t numTensors,
const Shape &nThreads, const void *params,
size_t paramsSize = 0) {
inline Kernel CreateKernel(Context &ctx, const ShaderCode &shader,
const Tensor *dataBindings, size_t numTensors,
const Shape &nThreads, const void *params,
size_t paramsSize = 0) {
assert(nThreads.rank == 3);
WGPUDevice device = ctx.device;
WGPUQueue queue = ctx.queue;
Expand Down Expand Up @@ -922,7 +922,7 @@ Kernel CreateKernel(Context &ctx, const ShaderCode &shader,
* @param[in] kernel Kernel instance to dispatch
* @example DispatchKernel(ctx, kernel);
*/
void DispatchKernel(Context &ctx, Kernel &kernel, std::promise<void> &promise) {
inline void DispatchKernel(Context &ctx, Kernel &kernel, std::promise<void> &promise) {
// Submit the command buffer
wgpuQueueSubmit(ctx.queue, 1, &kernel.commandBuffer);
wgpuQueueOnSubmittedWorkDone(
Expand Down
6 changes: 3 additions & 3 deletions utils/array_utils.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -74,7 +74,7 @@ std::string show(const std::array<numtype, rows * cols>& a, const std::string& n


// For testing only, not optimized
void transpose(float* input, float* output, size_t M, size_t N) {
inline void transpose(float* input, float* output, size_t M, size_t N) {
for (size_t i = 0; i < M; i++) {
for (size_t j = 0; j < N; j++) {
output[j * M + i] = input[i * N + j];
Expand Down Expand Up @@ -109,15 +109,15 @@ void randn(std::array<float, size> &a, std::mt19937 &gen, float mean = 0.0,
}


void eye(float* a, size_t N) {
inline void eye(float* a, size_t N) {
for (size_t i = 0; i < N; i++) {
for (size_t j = 0; j < N; j++) {
a[i * N + j] = (i == j) ? 1.0 : 0.0;
}
}
}

void flip(float* a, size_t R, size_t C, bool horizontal = true) {
inline void flip(float* a, size_t R, size_t C, bool horizontal = true) {
if (horizontal) {
for (size_t i = 0; i < R; i++) {
for (size_t j = 0; j < C / 2; j++) {
Expand Down