vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Continue implementation

Browse source
This commit is contained in:
0cc4m 2023-06-21 00:26:48 +02:00
parent 2471728a9d
commit 8ce84c2747
4 changed files with 163 additions and 69 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml-opencl.cpp
View file

@ -1489,7 +1489,7 @@ static void ggml_cl_mul_mat_f32(const ggml_tensor * src0, const ggml_tensor * sr
if (src0->backend == GGML_BACKEND_GPU) { // NOLINT
d_X = (cl_mem) src0->data;
} else {
d_X = ggml_cl_pool_malloc(sizeof(ggml_fp16_t) * x_ne, &x_size);
d_X = ggml_cl_pool_malloc(ggml_type_size(src0->type) * x_ne, &x_size);
}
cl_mem d_Y = ggml_cl_pool_malloc(sizeof(float) * y_ne, &y_size);
cl_mem d_D = ggml_cl_pool_malloc(sizeof(float) * d_ne, &d_size);

148
ggml-vulkan.cpp
View file

@ -13,7 +13,7 @@
#else
inline static void* ggml_aligned_malloc(size_t size, size_t alignment) {
void* aligned_memory = NULL;
int result = posix_memalign(&aligned_memory, alignment, size);
int result = posix_memalign(&aligned_memory, alignment >= 8 ? alignment : 8, size);
if (result != 0) {
// Handle allocation failure
return NULL;
@ -143,6 +143,7 @@ struct scoped_spin_lock {
struct vk_buffer {
vk::Buffer buffer;
VmaAllocation allocation;
VmaAllocationInfo info;
size_t size = 0;
};
@ -170,9 +171,7 @@ static void ggml_vk_pool_malloc(size_t size, vk_buffer* buf) {
if(best_i != -1) {
//found the smallest buffer that fits our needs
vk_buffer& b = g_vk_buffer_pool[best_i];
buf->buffer = b.buffer;
buf->allocation = b.allocation;
buf->size = b.size;
*buf = b;
b.size = 0;
return;
}
@ -194,14 +193,22 @@ static void ggml_vk_pool_malloc(size_t size, vk_buffer* buf) {
};
VmaAllocationCreateInfo allocation_info = {};
allocation_info.usage = VMA_MEMORY_USAGE_CPU_TO_GPU;
allocation_info.usage = VMA_MEMORY_USAGE_AUTO;
allocation_info.flags = VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT | VMA_ALLOCATION_CREATE_HOST_ACCESS_ALLOW_TRANSFER_INSTEAD_BIT | VMA_ALLOCATION_CREATE_MAPPED_BIT;
vmaCreateBuffer(vk_allocator,
(VkBufferCreateInfo*)&buffer_create_info,
&allocation_info,
(VkBuffer*)&buf->buffer,
&buf->allocation,
nullptr);
&buf->info);
VkMemoryPropertyFlags mem_prop_flags;
vmaGetAllocationMemoryProperties(vk_allocator, buf->allocation, &mem_prop_flags);
if(!(mem_prop_flags & VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT)) {
printf("Nope\n");
}
}
static void ggml_vk_pool_free(vk_buffer* buffer) {
@ -210,9 +217,7 @@ static void ggml_vk_pool_free(vk_buffer* buffer) {
for (int i = 0; i < MAX_VK_BUFFERS; ++i) {
vk_buffer& b = g_vk_buffer_pool[i];
if (b.size == 0) {
b.buffer = buffer->buffer;
b.allocation = buffer->allocation;
b.size = buffer->size;
b = *buffer;
return;
}
}
@ -221,7 +226,87 @@ static void ggml_vk_pool_free(vk_buffer* buffer) {
vmaDestroyBuffer(vk_allocator, buffer->buffer, buffer->allocation);
}
static void ggml_vk_h2d_tensor_2d(vk_buffer* dst, size_t offset, const struct ggml_tensor * src, uint64_t i3, uint64_t i2) {
static void ggml_vk_buffer_write(VkCommandBuffer cmd_buf, vk_buffer* dst, size_t offset, const void * src, size_t size) {
VkMemoryPropertyFlags mem_prop_flags;
vmaGetAllocationMemoryProperties(vk_allocator, dst->allocation, &mem_prop_flags);
if(mem_prop_flags & VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT) {
memcpy(dst->info.pMappedData, src, size);
} else {
// Allocation ended up in a non-mappable memory - need to transfer.
VkBufferCreateInfo staging_buf_create_info = { VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO };
staging_buf_create_info.size = size;
staging_buf_create_info.usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT;
VmaAllocationCreateInfo staging_alloc_create_info = {};
staging_alloc_create_info.usage = VMA_MEMORY_USAGE_AUTO;
staging_alloc_create_info.flags = VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT |
VMA_ALLOCATION_CREATE_MAPPED_BIT;
VkBuffer staging_buf;
VmaAllocation staging_alloc;
VmaAllocationInfo staging_alloc_info;
vmaCreateBuffer(vk_allocator,
&staging_buf_create_info,
&staging_alloc_create_info,
&staging_buf,
&staging_alloc,
&staging_alloc_info);
// [Executed in runtime]:
memcpy(staging_alloc_info.pMappedData + offset, src, size);
vmaFlushAllocation(vk_allocator, staging_alloc, 0, VK_WHOLE_SIZE);
//vkCmdPipelineBarrier: VK_ACCESS_HOST_WRITE_BIT --> VK_ACCESS_TRANSFER_READ_BIT
VkBufferCopy buf_copy = {
0, // srcOffset
0, // dstOffset,
size}; // size
vkCmdCopyBuffer(cmd_buf, staging_buf, dst->buffer, 1, &buf_copy);
vmaDestroyBuffer(vk_allocator, staging_buf, staging_alloc);
}
}
static void ggml_vk_buffer_read(VkCommandBuffer cmd_buf, vk_buffer* src, size_t offset, void * dst, size_t size) {
VkMemoryPropertyFlags mem_prop_flags;
vmaGetAllocationMemoryProperties(vk_allocator, src->allocation, &mem_prop_flags);
if(mem_prop_flags & VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT) {
memcpy(dst, src->info.pMappedData, size);
} else {
// Allocation ended up in a non-mappable memory - need to transfer.
VkBufferCreateInfo staging_buf_create_info = { VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO };
staging_buf_create_info.size = size;
staging_buf_create_info.usage = VK_BUFFER_USAGE_TRANSFER_DST_BIT;
VmaAllocationCreateInfo staging_alloc_create_info = {};
staging_alloc_create_info.usage = VMA_MEMORY_USAGE_AUTO;
staging_alloc_create_info.flags = VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT |
VMA_ALLOCATION_CREATE_MAPPED_BIT;
VkBuffer staging_buf;
VmaAllocation staging_alloc;
VmaAllocationInfo staging_alloc_info;
vmaCreateBuffer(vk_allocator,
&staging_buf_create_info,
&staging_alloc_create_info,
&staging_buf,
&staging_alloc,
&staging_alloc_info);
//vkCmdPipelineBarrier: VK_ACCESS_HOST_WRITE_BIT --> VK_ACCESS_TRANSFER_READ_BIT
VkBufferCopy buf_copy = {
offset, // srcOffset
0, // dstOffset,
size}; // size
vkCmdCopyBuffer(cmd_buf, src->buffer, staging_buf, 1, &buf_copy);
vmaInvalidateAllocation(vk_allocator, staging_alloc, 0, VK_WHOLE_SIZE);
// [Executed in runtime]:
memcpy(dst, staging_alloc_info.pMappedData, size);
vmaDestroyBuffer(vk_allocator, staging_buf, staging_alloc);
}
}
static void ggml_vk_h2d_tensor_2d(VkCommandBuffer cmd_buf, vk_buffer* dst, size_t offset, const struct ggml_tensor * src, uint64_t i3, uint64_t i2) {
const uint64_t ne0 = src->ne[0];
const uint64_t ne1 = src->ne[1];
const uint64_t nb0 = src->nb[0];
@ -234,31 +319,23 @@ static void ggml_vk_h2d_tensor_2d(vk_buffer* dst, size_t offset, const struct gg
const void * x = (const void *) ((const char *) src->data + i2*nb2 + i3*nb3);
if (nb0 == ts && nb1 == ts*ne0/bs) {
void* dst_ptr = nullptr;
vmaMapMemory(vk_allocator, dst->allocation, &dst_ptr);
memcpy(dst_ptr + offset, x, ne1*nb1);
vmaUnmapMemory(vk_allocator, dst->allocation);
ggml_vk_buffer_write(cmd_buf, dst, offset, x, ne1*nb1);
return;
}
if (nb0 == ts) {
void* dst_ptr = nullptr;
// Might be better to use vkCmdCopyBuffer here
vmaMapMemory(vk_allocator, dst->allocation, &dst_ptr);
for (uint64_t i1 = 0; i1 < ne1; i1++) {
memcpy(dst_ptr + offset + ne0 * i1, x + ts*ne0/bs, ne0*nb0);
ggml_vk_buffer_write(cmd_buf, dst, offset + ne0 * i1, x + ts*ne0/bs, ne0*nb0);
}
vmaUnmapMemory(vk_allocator, dst->allocation);
return;
}
uint8_t* dst_ptr = nullptr;
uint8_t* dst_ptr = (uint8_t*) dst->info.pMappedData;
uint8_t* xc = (uint8_t*)x;
vmaMapMemory(vk_allocator, dst->allocation, (void**) &dst_ptr);
for (uint64_t i1 = 0; i1 < ne1; i1++) {
for (uint64_t i0 = 0; i0 < ne0; i0++) {
dst_ptr[offset + i1 * ts*ne0/bs + i0 * ts] = xc[i1 * nb1 + i0 * nb0];
}
}
vmaUnmapMemory(vk_allocator, dst->allocation);
return;
}
@ -286,20 +363,11 @@ static void ggml_vk_mul_mat_f32(const ggml_tensor * src0, const ggml_tensor * sr
if (src0->backend == GGML_BACKEND_GPU) {
d_X = *(vk_buffer*) src0->data;
} else {
ggml_vk_pool_malloc(sizeof(ggml_fp16_t) * x_ne, &d_X);
ggml_vk_pool_malloc(ggml_type_size(src0->type) * x_ne, &d_X);
}
ggml_vk_pool_malloc(sizeof(float) * y_ne, &d_Y);
ggml_vk_pool_malloc(sizeof(float) * d_ne, &d_D);
for (int64_t i03 = 0; i03 < ne03; i03++) {
for (int64_t i02 = 0; i02 < ne02; i02++) {
// copy data to device
if (src0->backend != GGML_BACKEND_GPU) {
ggml_vk_h2d_tensor_2d(&d_X, 0, src0, i03, i02);
}
ggml_vk_h2d_tensor_2d(&d_Y, 0, src1, i03, i02);
// compute
vk::DescriptorPoolSize descriptor_pool_size(vk::DescriptorType::eStorageBuffer, 3);
vk::DescriptorPoolCreateInfo descriptor_pool_create_info(vk::DescriptorPoolCreateFlags(), 1, descriptor_pool_size);
vk::DescriptorPool descriptor_pool = vk_device.createDescriptorPool(descriptor_pool_create_info);
@ -328,6 +396,17 @@ static void ggml_vk_mul_mat_f32(const ggml_tensor * src0, const ggml_tensor * sr
const std::vector<vk::CommandBuffer> cmd_buffers = vk_device.allocateCommandBuffers(command_buffer_alloc_info);
vk::CommandBuffer cmd_buffer = cmd_buffers.front();
for (int64_t i03 = 0; i03 < ne03; i03++) {
for (int64_t i02 = 0; i02 < ne02; i02++) {
// copy data to device
if (src0->backend != GGML_BACKEND_GPU) {
ggml_vk_h2d_tensor_2d(cmd_buffer, &d_X, 0, src0, i03, i02);
}
ggml_vk_h2d_tensor_2d(cmd_buffer, &d_Y, 0, src1, i03, i02);
printf("Beginning Vulkan kernel call\n");
// compute
vk::CommandBufferBeginInfo cmd_buffer_begin_info(vk::CommandBufferUsageFlagBits::eOneTimeSubmit);
cmd_buffer.begin(cmd_buffer_begin_info);
cmd_buffer.bindPipeline(vk::PipelineBindPoint::eCompute, vk_pipeline_matmul);
@ -352,12 +431,11 @@ static void ggml_vk_mul_mat_f32(const ggml_tensor * src0, const ggml_tensor * sr
true,
uint64_t(-1));
printf("Vulkan kernel call done\n");
// copy dst to host
void* src_ptr = nullptr;
float * d = (float *) ((char *) dst->data + i02*nb2 + i03*nb3);
vmaMapMemory(vk_allocator, d_D.allocation, &src_ptr);
memcpy(d, src_ptr, sizeof(float) * d_ne);
vmaUnmapMemory(vk_allocator, d_D.allocation);
ggml_vk_buffer_read(cmd_buffer, &d_D, 0, d, sizeof(float) * d_ne);
}
}

26
ggml-vulkan.h
View file

@ -1,23 +1,25 @@
#pragma once
#include "ggml.h"
#ifdef __cplusplus
extern "C" {
#endif
void ggml_vk_init(void);
// enum ggml_blas_order {
// GGML_BLAS_ORDER_ROW_MAJOR = 101,
// GGML_BLAS_ORDER_COLUMN_MAJOR = 102,
// };
//
// enum ggml_blas_op {
// GGML_BLAS_OP_N = 111,
// GGML_BLAS_OP_T = 112,
// GGML_BLAS_OP_C = 113,
// };
//
// void ggml_cl_sgemm_wrapper(const enum ggml_blas_order order, const enum ggml_blas_op trans_a, const enum ggml_blas_op trans_b, const int m, const int n, const int k, const float alpha, const void *host_a, const int lda, const float *host_b, const int ldb, const float beta, float *host_c, const int ldc, const int btype);
void ggml_vk_mul(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
bool ggml_vk_can_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
size_t ggml_vk_mul_mat_get_wsize(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
void ggml_vk_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst, void * wdata, size_t wsize);
void * ggml_vk_host_malloc(size_t size);
void ggml_vk_host_free(void * ptr);
void ggml_vk_free_data(const struct ggml_tensor* tensor);
void ggml_vk_transform_tensor(struct ggml_tensor * tensor);
void ggml_vk_load_data(const char * fname, struct ggml_tensor * tensor, size_t offset);
#ifdef __cplusplus
}

14
ggml.c
View file

@ -10660,6 +10660,13 @@ static void ggml_compute_forward_mul_mat_f32(
}
return;
}
#elif defined(GGML_USE_VULKAN)
if (ggml_vk_can_mul_mat(src0, src1, dst)) {
if (params->ith == 0 && params->type == GGML_TASK_COMPUTE) {
ggml_vk_mul_mat(src0, src1, dst, params->wdata, params->wsize);
}
return;
}
#endif
#if defined(GGML_USE_ACCELERATE) || defined(GGML_USE_OPENBLAS)
@ -10825,6 +10832,13 @@ static void ggml_compute_forward_mul_mat_f16_f32(
}
return;
}
#elif defined(GGML_USE_VULKAN)
if (ggml_vk_can_mul_mat(src0, src1, dst)) {
if (params->ith == 0 && params->type == GGML_TASK_COMPUTE) {
ggml_vk_mul_mat(src0, src1, dst, params->wdata, params->wsize);
}
return;
}
#endif
#if defined(GGML_USE_ACCELERATE) || defined(GGML_USE_OPENBLAS)