Skip to content

Commit

Permalink
cuda : use CUDA memory pool with async memory allocation/deallocation…
Browse files Browse the repository at this point in the history 
… when available (#3903)

* Using cuda memory pools for async alloc/dealloc.

* If cuda device doesnt support memory pool than use old implementation.

* Removed redundant cublasSetStream

---------

Co-authored-by: Oleksii Maryshchenko <omaryshchenko@dtis.com>
  • Loading branch information
@young-developer
young-developer and Oleksii Maryshchenko authored Nov 2, 2023
1 parent 4ff1046 commit d606905
Showing 1 changed file with 77 additions and 51 deletions.
128 changes: 77 additions & 51 deletions ggml-cuda.cu
View file
Original file line number Diff line number Diff line change
Expand Up @@ -181,11 +181,11 @@ static_assert(sizeof(half) == sizeof(ggml_fp16_t), "wrong fp16 size");
do { \
cudaError_t err_ = (err); \
if (err_ != cudaSuccess) { \
int id; \
cudaGetDevice(&id); \
int dev_id; \
cudaGetDevice(&dev_id); \
fprintf(stderr, "\nCUDA error %d at %s:%d: %s\n", err_, __FILE__, __LINE__, \
cudaGetErrorString(err_)); \
fprintf(stderr, "current device: %d\n", id); \
fprintf(stderr, "current device: %d\n", dev_id); \
exit(1); \
} \
} while (0)
Expand All @@ -195,11 +195,11 @@ static_assert(sizeof(half) == sizeof(ggml_fp16_t), "wrong fp16 size");
do { \
cublasStatus_t err_ = (err); \
if (err_ != CUBLAS_STATUS_SUCCESS) { \
int id; \
cudaGetDevice(&id); \
int dev_id; \
cudaGetDevice(&dev_id); \
fprintf(stderr, "\ncuBLAS error %d at %s:%d: %s\n", \
err_, __FILE__, __LINE__, cublasGetStatusString(err_)); \
fprintf(stderr, "current device: %d\n", id); \
fprintf(stderr, "current device: %d\n", dev_id); \
exit(1); \
} \
} while (0)
Expand Down Expand Up @@ -465,6 +465,7 @@ static_assert(K_QUANTS_PER_ITERATION == 1 || K_QUANTS_PER_ITERATION == 2, "K_QUA

#define MAX_STREAMS 8
static cudaStream_t g_cudaStreams[GGML_CUDA_MAX_DEVICES][MAX_STREAMS] = { nullptr };
static cudaMemPool_t g_cudaMemPools[GGML_CUDA_MAX_DEVICES] = { nullptr };

struct ggml_tensor_extra_gpu {
void * data_device[GGML_CUDA_MAX_DEVICES]; // 1 pointer for each device for split tensors
Expand Down Expand Up @@ -5772,6 +5773,16 @@ static void * ggml_cuda_pool_malloc(size_t size, size_t * actual_size) {
return ptr;
}

static void * ggml_cuda_pool_malloc_async(size_t size, size_t * actual_size, int id, cudaStream_t stream) {
if (g_cudaMemPools[id] == nullptr) {
return ggml_cuda_pool_malloc(size, actual_size);
}
void *ptr;
CUDA_CHECK(cudaMallocFromPoolAsync(&ptr, size, g_cudaMemPools[id], stream));
*actual_size = size;
return ptr;
}

static void ggml_cuda_pool_free(void * ptr, size_t size) {
scoped_spin_lock lock(g_cuda_pool_lock);
int id;
Expand All @@ -5790,6 +5801,13 @@ static void ggml_cuda_pool_free(void * ptr, size_t size) {
}


static void ggml_cuda_pool_free_async(void * ptr, size_t actual_size, int id, cudaStream_t stream) {
if (g_cudaMemPools[id] == nullptr) {
return ggml_cuda_pool_free(ptr, actual_size);
}
CUDA_CHECK(cudaFreeAsync(ptr, stream));
}

void ggml_init_cublas() {
static bool initialized = false;

Expand Down Expand Up @@ -5844,6 +5862,13 @@ void ggml_init_cublas() {
// create cublas handle
CUBLAS_CHECK(cublasCreate(&g_cublas_handles[id]));
CUBLAS_CHECK(cublasSetMathMode(g_cublas_handles[id], CUBLAS_TF32_TENSOR_OP_MATH));

// configure memory pool
cudaError_t err = cudaDeviceGetMemPool(&g_cudaMemPools[id], id);
if (err == cudaSuccess) {
size_t treshold = UINT64_MAX;
CUDA_CHECK(cudaMemPoolSetAttribute(g_cudaMemPools[id], cudaMemPoolAttrReleaseThreshold, &treshold));
}
}

// configure logging to stdout
Expand Down Expand Up @@ -6437,7 +6462,7 @@ inline void ggml_cuda_op_mul_mat_cublas(
const to_fp16_cuda_t to_fp16_cuda = ggml_get_to_fp16_cuda(src0->type);
GGML_ASSERT(to_fp16_cuda != nullptr);
size_t ne = row_diff*ne00;
src0_as_f16 = (half *) ggml_cuda_pool_malloc(ne * sizeof(half), &src0_as);
src0_as_f16 = (half *) ggml_cuda_pool_malloc_async(ne * sizeof(half), &src0_as, id, stream);
to_fp16_cuda(src0_dd_i, src0_as_f16, ne, stream);
}
const half * src0_ptr = src0->type == GGML_TYPE_F16 ? (const half *) src0_dd_i : src0_as_f16;
Expand All @@ -6448,13 +6473,12 @@ inline void ggml_cuda_op_mul_mat_cublas(
const to_fp16_cuda_t to_fp16_cuda = ggml_get_to_fp16_cuda(src1->type);
GGML_ASSERT(to_fp16_cuda != nullptr);
size_t ne = src1_ncols*ne10;
src1_as_f16 = (half *) ggml_cuda_pool_malloc(ne * sizeof(half), &src1_as);
src1_as_f16 = (half *) ggml_cuda_pool_malloc_async(ne * sizeof(half), &src1_as, id, stream);
to_fp16_cuda(src1_ddf_i, src1_as_f16, ne, stream);
}
const half * src1_ptr = src1->type == GGML_TYPE_F16 ? (const half *) src1_ddq_i : src1_as_f16;

size_t dst_as = 0;
half * dst_f16 = (half *) ggml_cuda_pool_malloc(row_diff*src1_ncols * sizeof(half), &dst_as);
size_t dst_f16_as = 0;
half * dst_f16 = (half *) ggml_cuda_pool_malloc_async(row_diff*src1_ncols * sizeof(half), &dst_f16_as, id, stream);

const half alpha_f16 = 1.0f;
const half beta_f16 = 0.0f;
Expand All @@ -6472,14 +6496,15 @@ inline void ggml_cuda_op_mul_mat_cublas(
const to_fp32_cuda_t to_fp32_cuda = ggml_get_to_fp32_cuda(GGML_TYPE_F16);
to_fp32_cuda(dst_f16, dst_dd_i, row_diff*src1_ncols, stream);

ggml_cuda_pool_free(dst_f16, dst_as);
if (dst_f16_as != 0) {
ggml_cuda_pool_free_async(dst_f16, dst_f16_as, id, stream);
}

if (src0_as != 0) {
ggml_cuda_pool_free(src0_as_f16, src0_as);
ggml_cuda_pool_free_async(src0_as_f16, src0_as, id, stream);
}

if (src1_as != 0) {
ggml_cuda_pool_free(src1_as_f16, src1_as);
ggml_cuda_pool_free_async(src1_as_f16, src1_as, id, stream);
}
}
else {
Expand All @@ -6489,7 +6514,7 @@ inline void ggml_cuda_op_mul_mat_cublas(
if (src0->type != GGML_TYPE_F32) {
const to_fp32_cuda_t to_fp32_cuda = ggml_get_to_fp32_cuda(src0->type);
GGML_ASSERT(to_fp32_cuda != nullptr);
src0_ddq_as_f32 = (float *) ggml_cuda_pool_malloc(row_diff*ne00 * sizeof(float), &src0_as); // NOLINT
src0_ddq_as_f32 = (float *) ggml_cuda_pool_malloc_async(row_diff*ne00 * sizeof(float), &src0_as, id, stream); // NOLINT
to_fp32_cuda(src0_dd_i, src0_ddq_as_f32, row_diff*ne00, stream);
}
const float * src0_ddf_i = src0->type == GGML_TYPE_F32 ? (const float *) src0_dd_i : src0_ddq_as_f32;
Expand All @@ -6506,7 +6531,7 @@ inline void ggml_cuda_op_mul_mat_cublas(
&beta, dst_dd_i, ldc));

if (src0_as != 0) {
ggml_cuda_pool_free(src0_ddq_as_f32, src0_as);
ggml_cuda_pool_free_async(src0_ddq_as_f32, src0_as, id, stream);
}
}

Expand Down Expand Up @@ -6929,29 +6954,30 @@ static void ggml_cuda_op_mul_mat(
src0_dd[id] = (char *) src0_extra->data_device[id];
} else {
const size_t size_src0_ddq = split ? (row_high[id]-row_low[id])*ne00 * src0_ts/src0_bs : ggml_nbytes(src0);
src0_dd[id] = (char *) ggml_cuda_pool_malloc(ggml_nbytes(src0), &src0_as[id]);
src0_dd[id] = (char *) ggml_cuda_pool_malloc_async(ggml_nbytes(src0), &src0_as[id], id, stream);
}

if (src1_on_device && src1_is_contiguous) {
src1_ddf[id] = (float *) src1_extra->data_device[id];
} else {
src1_ddf[id] = (float *) ggml_cuda_pool_malloc(ggml_nbytes(src1), &src1_asf[id]);
src1_ddf[id] = (float *) ggml_cuda_pool_malloc_async(ggml_nbytes(src1), &src1_asf[id], id, stream);
}

if (convert_src1_to_q8_1) {
src1_ddq[id] = (char *) ggml_cuda_pool_malloc(nrows1*src1_padded_col_size*q8_1_ts/q8_1_bs, &src1_asq[id]);
const size_t size_dst_ddq = nrows1*src1_padded_col_size*q8_1_ts/q8_1_bs;
src1_ddq[id] = (char *) ggml_cuda_pool_malloc_async(size_dst_ddq, &src1_asq[id], id, stream);

if (src1_on_device && src1_is_contiguous) {
quantize_row_q8_1_cuda(src1_ddf[id], src1_ddq[id], ne10, nrows1, src1_padded_col_size, stream);
CUDA_CHECK(cudaGetLastError());
// CUDA_CHECK(cudaGetLastError());
}
}

if (dst_on_device) {
dst_dd[id] = (float *) dst_extra->data_device[id];
} else {
const size_t size_dst_ddf = split ? (row_high[id]-row_low[id])*ne1*sizeof(float) : ggml_nbytes(dst);
dst_dd[id] = (float *) ggml_cuda_pool_malloc(size_dst_ddf, &dst_as[id]);
dst_dd[id] = (float *) ggml_cuda_pool_malloc_async(size_dst_ddf, &dst_as[id], id, stream);
}
}

Expand Down Expand Up @@ -7077,24 +7103,6 @@ static void ggml_cuda_op_mul_mat(
}
}

for (int64_t id = 0; id < g_device_count; ++id) {
CUDA_CHECK(ggml_cuda_set_device(id));

// free buffers again when done
if (src0_as[id] > 0) {
ggml_cuda_pool_free(src0_dd[id], src0_as[id]);
}
if (src1_asf[id] > 0) {
ggml_cuda_pool_free(src1_ddf[id], src1_asf[id]);
}
if (src1_asq[id] > 0) {
ggml_cuda_pool_free(src1_ddq[id], src1_asq[id]);
}
if (dst_as[id] > 0) {
ggml_cuda_pool_free(dst_dd[id], dst_as[id]);
}
}

// main device waits for all other devices to be finished
if (split && g_device_count > 1) {
int64_t is_max = (ne11 + MUL_MAT_SRC1_COL_STRIDE - 1) / MUL_MAT_SRC1_COL_STRIDE;
Expand All @@ -7112,6 +7120,21 @@ static void ggml_cuda_op_mul_mat(
CUDA_CHECK(ggml_cuda_set_device(g_main_device));
CUDA_CHECK(cudaDeviceSynchronize());
}

for (int64_t id = 0; id < g_device_count; ++id) {
if (src0_as[id] > 0) {
ggml_cuda_pool_free_async(src0_dd[id], src0_as[id], id, g_cudaStreams[id][0]);
}
if (src1_asf[id] > 0) {
ggml_cuda_pool_free_async(src1_ddf[id], src1_asf[id], id, g_cudaStreams[id][0]);
}
if (src1_asq[id] > 0) {
ggml_cuda_pool_free_async(src1_ddq[id], src1_asq[id], id, g_cudaStreams[id][0]);
}
if (dst_as[id] > 0) {
ggml_cuda_pool_free_async(dst_dd[id], dst_as[id], id, g_cudaStreams[id][0]);
}
}
}

static void ggml_cuda_repeat(const ggml_tensor * src0, const ggml_tensor * src1, ggml_tensor * dst) {
Expand Down Expand Up @@ -7298,11 +7321,11 @@ static void ggml_cuda_mul_mat_mat_batched_cublas(const ggml_tensor * src0, const
GGML_ASSERT(to_fp16_cuda != nullptr);

size_t src1_as = 0;
half * src1_as_f16 = (half *) ggml_cuda_pool_malloc(ne1 * sizeof(half), &src1_as);
half * src1_as_f16 = (half *) ggml_cuda_pool_malloc_async(ne1 * sizeof(half), &src1_as, id, main_stream);
to_fp16_cuda(src1_ddf, src1_as_f16, ne1, main_stream);

size_t dst_as = 0;
half * dst_f16 = (half *) ggml_cuda_pool_malloc(ne * sizeof(half), &dst_as);
half * dst_f16 = (half *) ggml_cuda_pool_malloc_async(ne * sizeof(half), &dst_as, id, main_stream);

GGML_ASSERT(ne12 % ne02 == 0);
GGML_ASSERT(ne13 % ne03 == 0);
Expand Down Expand Up @@ -7349,10 +7372,9 @@ static void ggml_cuda_mul_mat_mat_batched_cublas(const ggml_tensor * src0, const
} else {
// use cublasGemmBatchedEx
const int ne23 = ne12*ne13;

void ** ptrs_as = nullptr;
// allocate device memory for pointers
size_t ptrs_s = 0;
ptrs_as = (void **) ggml_cuda_pool_malloc(3*ne23*sizeof(void *), &ptrs_s);
void ** ptrs_as = (void **)ggml_cuda_pool_malloc_async(3*ne23*sizeof(void *), &ptrs_s, id, main_stream);

dim3 block_dims(ne13, ne12);
k_compute_batched_ptrs<<<1, block_dims, 0, main_stream>>>(
Expand All @@ -7365,7 +7387,6 @@ static void ggml_cuda_mul_mat_mat_batched_cublas(const ggml_tensor * src0, const
dst->nb[2], dst->nb[3],
r2, r3);
CUDA_CHECK(cudaGetLastError());

CUBLAS_CHECK(
cublasGemmBatchedEx(g_cublas_handles[id], CUBLAS_OP_T, CUBLAS_OP_N,
ne01, ne11, ne10,
Expand All @@ -7375,16 +7396,21 @@ static void ggml_cuda_mul_mat_mat_batched_cublas(const ggml_tensor * src0, const
ne23,
CUBLAS_COMPUTE_16F,
CUBLAS_GEMM_DEFAULT_TENSOR_OP));

ggml_cuda_pool_free(ptrs_as, ptrs_s);
// free device memory for pointers
if (ptrs_s != 0) {
ggml_cuda_pool_free_async(ptrs_as, ptrs_s, id, main_stream);
}
}
#endif

const to_fp32_cuda_t to_fp32_cuda = ggml_get_to_fp32_cuda(GGML_TYPE_F16);
to_fp32_cuda(dst_f16, dst_ddf, ne, main_stream);

ggml_cuda_pool_free(src1_as_f16, src1_as);
ggml_cuda_pool_free(dst_f16, dst_as);
if (src1_as != 0) {
ggml_cuda_pool_free_async(src1_as_f16, src1_as, id, main_stream);
}
if (dst_as != 0) {
ggml_cuda_pool_free_async(dst_f16, dst_as, id, main_stream);
}
}

static void ggml_cuda_mul_mat(const ggml_tensor * src0, const ggml_tensor * src1, ggml_tensor * dst) {
Expand Down

0 comments on commit d606905

Please sign in to comment.