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

Fix Vulkan check results

Browse source
This commit is contained in:
0cc4m 2024-02-11 09:59:36 +01:00
parent 5e0a9a2d37
commit c19c59812e
1 changed files with 16 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

28
ggml-vulkan.cpp
View file

@ -5344,8 +5344,9 @@ static void ggml_vk_print_tensor(ggml_backend_vk_context * ctx, const ggml_tenso
tensor_data = malloc(tensor_size);
ggml_tensor_extra_gpu * extra = (ggml_tensor_extra_gpu *) tensor->extra;
vk_buffer buf = extra->buffer_gpu.lock();
ggml_vk_buffer_read(ctx, extra->buffer_gpu, extra->offset, tensor_data, tensor_size);
ggml_vk_buffer_read(ctx, buf, extra->offset, tensor_data, tensor_size);
}
std::cerr << "TENSOR CHECK " << name << " (" << tensor->name << "): " << ggml_op_name(tensor->op) << std::endl;
@ -5452,12 +5453,13 @@ static void ggml_vk_check_results_0(ggml_backend_vk_context * ctx, ggml_compute_
memcpy(src0_clone->nb, src0->nb, sizeof(size_t) * GGML_MAX_DIMS);
} else if (src0->backend == GGML_BACKEND_GPU) {
ggml_tensor_extra_gpu * extra = (ggml_tensor_extra_gpu *) src0->extra;
vk_buffer buf = extra->buffer_gpu.lock();
uint64_t offset = extra->offset;
if (!ggml_is_contiguous(src0) && ggml_vk_dim01_contiguous(src0)) {
for (int i3 = 0; i3 < src0->ne[3]; i3++) {
for (int i2 = 0; i2 < src0->ne[2]; i2++) {
const int idx = i3*src0->ne[2] + i2;
ggml_vk_buffer_read(ctx, extra->buffer_gpu, offset + idx * src0->nb[2], ((char *)src0_clone->data + idx * src0_clone->nb[2]), src0->ne[1] * src0->nb[1]);
ggml_vk_buffer_read(ctx, buf, offset + idx * src0->nb[2], ((char *)src0_clone->data + idx * src0_clone->nb[2]), src0->ne[1] * src0->nb[1]);
}
}
@ -5467,10 +5469,10 @@ static void ggml_vk_check_results_0(ggml_backend_vk_context * ctx, ggml_compute_
src0_clone->nb[i] = src0_clone->nb[i - 1]*src0_clone->ne[i - 1];
}
} else {
if (offset + src0_size >= extra->buffer_gpu->size) {
src0_size = extra->buffer_gpu->size - offset;
if (offset + src0_size >= buf->size) {
src0_size = buf->size - offset;
}
ggml_vk_buffer_read(ctx, extra->buffer_gpu, offset, src0_clone->data, src0_size);
ggml_vk_buffer_read(ctx, buf, offset, src0_clone->data, src0_size);
memcpy(src0_clone->nb, src0->nb, sizeof(size_t) * GGML_MAX_DIMS);
}
} else {
@ -5495,12 +5497,13 @@ static void ggml_vk_check_results_0(ggml_backend_vk_context * ctx, ggml_compute_
memcpy(src1_clone->nb, src1->nb, sizeof(size_t) * GGML_MAX_DIMS);
} else if (src1->backend == GGML_BACKEND_GPU) {
ggml_tensor_extra_gpu * extra = (ggml_tensor_extra_gpu *) src1->extra;
vk_buffer buf = extra->buffer_gpu.lock();
uint64_t offset = extra->offset;
if (!ggml_is_contiguous(src1) && ggml_vk_dim01_contiguous(src1)) {
for (int i3 = 0; i3 < src1->ne[3]; i3++) {
for (int i2 = 0; i2 < src1->ne[2]; i2++) {
const int idx = i3*src1->ne[2] + i2;
ggml_vk_buffer_read(ctx, extra->buffer_gpu, offset + idx * src1->nb[2], ((char *)src1_clone->data + idx * src1_clone->nb[2]), src1->ne[1] * src1->nb[1]);
ggml_vk_buffer_read(ctx, buf, offset + idx * src1->nb[2], ((char *)src1_clone->data + idx * src1_clone->nb[2]), src1->ne[1] * src1->nb[1]);
}
}
@ -5510,10 +5513,10 @@ static void ggml_vk_check_results_0(ggml_backend_vk_context * ctx, ggml_compute_
src1_clone->nb[i] = src1_clone->nb[i - 1]*src1_clone->ne[i - 1];
}
} else {
if (offset + src1_size >= extra->buffer_gpu->size) {
src1_size = extra->buffer_gpu->size - offset;
if (offset + src1_size >= buf->size) {
src1_size = buf->size - offset;
}
ggml_vk_buffer_read(ctx, extra->buffer_gpu, offset, src1_clone->data, src1_size);
ggml_vk_buffer_read(ctx, buf, offset, src1_clone->data, src1_size);
memcpy(src1_clone->nb, src1->nb, sizeof(size_t) * GGML_MAX_DIMS);
}
} else {
@ -5669,12 +5672,13 @@ static void ggml_vk_check_results_1(ggml_backend_vk_context * ctx, ggml_compute_
tensor_data = malloc(tensor_size);
ggml_tensor_extra_gpu * extra = (ggml_tensor_extra_gpu *) tensor->extra;
vk_buffer buf = extra->buffer_gpu.lock();
if (extra->offset + tensor_size >= extra->buffer_gpu->size) {
tensor_size = extra->buffer_gpu->size - (extra->offset);
if (extra->offset + tensor_size >= buf->size) {
tensor_size = buf->size - (extra->offset);
}
ggml_vk_buffer_read(ctx, extra->buffer_gpu, extra->offset, tensor_data, tensor_size);
ggml_vk_buffer_read(ctx, buf, extra->offset, tensor_data, tensor_size);
}
float first_error_result = -1.0f;