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

Replace uint64_t(-1) with UINT64_MAX, rename function for clarity

Browse source
This commit is contained in:
0cc4m 2024-01-18 19:34:21 +01:00
parent 2d14b22a99
commit 1811c4ec9b
1 changed files with 18 additions and 15 deletions
Download patch file Download diff file Expand all files Collapse all files

33
ggml-vulkan.cpp
View file

@ -1576,7 +1576,7 @@ static void ggml_vk_buffer_write_2d(vk_buffer* dst, size_t offset, const void *
ggml_vk_buffer_write_2d_async(*ctx, dst, offset, src, spitch, width, height, q, nullptr, nullptr, true); ggml_vk_buffer_write_2d_async(*ctx, dst, offset, src, spitch, width, height, q, nullptr, nullptr, true);
ggml_vk_ctx_end(*ctx); ggml_vk_ctx_end(*ctx);
ggml_vk_submit(q, ctx->seqs, vk_fence); ggml_vk_submit(q, ctx->seqs, vk_fence);
VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, uint64_t(-1)), "vk_buffer_write_2d waitForFences"); VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, UINT64_MAX), "vk_buffer_write_2d waitForFences");
vk_device.device.resetFences({ vk_fence }); vk_device.device.resetFences({ vk_fence });
} }
} }
@ -1676,7 +1676,7 @@ static void ggml_vk_buffer_read(vk_buffer* src, size_t offset, void * dst, size_
ggml_vk_buffer_read_async(*ctx, src, offset, dst, size, q, &staging); ggml_vk_buffer_read_async(*ctx, src, offset, dst, size, q, &staging);
ggml_vk_ctx_end(*ctx); ggml_vk_ctx_end(*ctx);
ggml_vk_submit(q, ctx->seqs, vk_fence); ggml_vk_submit(q, ctx->seqs, vk_fence);
VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, uint64_t(-1)), "vk_buffer_read waitForFences"); VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, UINT64_MAX), "vk_buffer_read waitForFences");
vk_device.device.resetFences({ vk_fence }); vk_device.device.resetFences({ vk_fence });
for (auto& cpy : staging) { for (auto& cpy : staging) {
@ -1696,7 +1696,7 @@ static void ggml_vk_buffer_copy(vk_buffer * dst, size_t dst_offset, vk_buffer *
vkCmdCopyBuffer(ctx->s->buffer, src->buffer, dst->buffer, 1, &bc); vkCmdCopyBuffer(ctx->s->buffer, src->buffer, dst->buffer, 1, &bc);
ggml_vk_ctx_end(*ctx); ggml_vk_ctx_end(*ctx);
ggml_vk_submit(q, ctx->seqs, vk_fence); ggml_vk_submit(q, ctx->seqs, vk_fence);
VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, uint64_t(-1)), "vk_buffer_copy waitForFences"); VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, UINT64_MAX), "vk_buffer_copy waitForFences");
vk_device.device.resetFences({ vk_fence }); vk_device.device.resetFences({ vk_fence });
} }
@ -1713,7 +1713,7 @@ static void ggml_vk_buffer_memset(vk_buffer* dst, size_t offset, uint32_t c, siz
std::vector<vk_sequence> s = { { submission } }; std::vector<vk_sequence> s = { { submission } };
ggml_vk_submit(q, s, vk_fence); ggml_vk_submit(q, s, vk_fence);
VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, uint64_t(-1)), "vk_memset waitForFences"); VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, UINT64_MAX), "vk_memset waitForFences");
vk_device.device.resetFences({ vk_fence }); vk_device.device.resetFences({ vk_fence });
} }
@ -3425,7 +3425,7 @@ void ggml_vk_test_transfer(size_t ne, bool pinned) {
ggml_vk_submit(vk_device.transfer_queue, seqs, vk_fence); ggml_vk_submit(vk_device.transfer_queue, seqs, vk_fence);
VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, uint64_t(-1)), "ggml_vk_compute_forward waitForFences"); VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, UINT64_MAX), "ggml_vk_compute_forward waitForFences");
vk_device.device.resetFences({ vk_fence }); vk_device.device.resetFences({ vk_fence });
auto end = std::chrono::high_resolution_clock::now(); auto end = std::chrono::high_resolution_clock::now();
@ -3438,7 +3438,7 @@ void ggml_vk_test_transfer(size_t ne, bool pinned) {
ggml_vk_submit(vk_device.transfer_queue, seqs, vk_fence); ggml_vk_submit(vk_device.transfer_queue, seqs, vk_fence);
VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, uint64_t(-1)), "ggml_vk_compute_forward waitForFences"); VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, UINT64_MAX), "ggml_vk_compute_forward waitForFences");
vk_device.device.resetFences({ vk_fence }); vk_device.device.resetFences({ vk_fence });
for (auto& cpy : memcpys) { for (auto& cpy : memcpys) {
@ -3578,9 +3578,9 @@ static size_t ggml_vk_tensor_size(const ggml_tensor * tensor) {
return tensor->ne[order[3]]*tensor->nb[order[3]]; return tensor->ne[order[3]]*tensor->nb[order[3]];
} }
static void ggml_vk_preallocate_buffers(ggml_tensor * tensor, size_t d_size, size_t qx_size, size_t qy_size, size_t x_size, size_t y_size, size_t split_k_size, int inplace_d, size_t staging_size) { static void ggml_vk_preallocate_buffers_reserve(ggml_tensor * tensor, size_t d_size, size_t qx_size, size_t qy_size, size_t x_size, size_t y_size, size_t split_k_size, int inplace_d, size_t staging_size) {
#ifdef VK_DEBUG #ifdef VK_DEBUG
std::cerr << "ggml_vk_preallocate_buffers(tensor=" << tensor << ", d_size=" << d_size << ", qx_size=" << qx_size << ", qy_size=" << qy_size << ", x_size=" << x_size << ", y_size=" << y_size << ", split_k_size=" << split_k_size << ", inplace_d=" << inplace_d << ", staging_size=" << staging_size << ")" << std::endl; std::cerr << "ggml_vk_preallocate_buffers_reserve(tensor=" << tensor << ", d_size=" << d_size << ", qx_size=" << qx_size << ", qy_size=" << qy_size << ", x_size=" << x_size << ", y_size=" << y_size << ", split_k_size=" << split_k_size << ", inplace_d=" << inplace_d << ", staging_size=" << staging_size << ")" << std::endl;
#endif #endif
ggml_tensor_extra_gpu * extra = (ggml_tensor_extra_gpu *) tensor->extra; ggml_tensor_extra_gpu * extra = (ggml_tensor_extra_gpu *) tensor->extra;
GGML_ASSERT(extra != nullptr); GGML_ASSERT(extra != nullptr);
@ -3740,17 +3740,17 @@ void ggml_vk_preallocate_buffers_graph(ggml_tensor * node, ggml_cgraph * graph){
switch (node->op) { switch (node->op) {
case GGML_OP_REPEAT: case GGML_OP_REPEAT:
ggml_vk_preallocate_buffers(node, d_sz, qx_sz, 0, 0, 0, 0, inplace, qx_sz); ggml_vk_preallocate_buffers_reserve(node, d_sz, qx_sz, 0, 0, 0, 0, inplace, qx_sz);
break; break;
case GGML_OP_GET_ROWS: case GGML_OP_GET_ROWS:
ggml_vk_preallocate_buffers(node, d_sz, 0, 0, 0, 0, 0, inplace, 0); ggml_vk_preallocate_buffers_reserve(node, d_sz, 0, 0, 0, 0, 0, inplace, 0);
break; break;
case GGML_OP_RESHAPE: case GGML_OP_RESHAPE:
case GGML_OP_VIEW: case GGML_OP_VIEW:
case GGML_OP_PERMUTE: case GGML_OP_PERMUTE:
case GGML_OP_TRANSPOSE: case GGML_OP_TRANSPOSE:
ggml_vk_preallocate_buffers(node, 0, 0, 0, 0, 0, 0, inplace, 0); ggml_vk_preallocate_buffers_reserve(node, 0, 0, 0, 0, 0, 0, inplace, 0);
break; break;
case GGML_OP_ADD: case GGML_OP_ADD:
case GGML_OP_SCALE: case GGML_OP_SCALE:
@ -3765,21 +3765,21 @@ void ggml_vk_preallocate_buffers_graph(ggml_tensor * node, ggml_cgraph * graph){
case GGML_OP_DIAG_MASK_INF: case GGML_OP_DIAG_MASK_INF:
case GGML_OP_SOFT_MAX: case GGML_OP_SOFT_MAX:
case GGML_OP_ROPE: case GGML_OP_ROPE:
ggml_vk_preallocate_buffers(node, d_sz, transfer_src0 ? qx_sz : 0, transfer_src1 ? qy_sz : 0, 0, 0, 0, inplace, qx_sz + qy_sz); ggml_vk_preallocate_buffers_reserve(node, d_sz, transfer_src0 ? qx_sz : 0, transfer_src1 ? qy_sz : 0, 0, 0, 0, inplace, qx_sz + qy_sz);
break; break;
case GGML_OP_UNARY: case GGML_OP_UNARY:
switch (ggml_get_unary_op(node)) { switch (ggml_get_unary_op(node)) {
case GGML_UNARY_OP_SILU: case GGML_UNARY_OP_SILU:
case GGML_UNARY_OP_GELU: case GGML_UNARY_OP_GELU:
case GGML_UNARY_OP_RELU: case GGML_UNARY_OP_RELU:
ggml_vk_preallocate_buffers(node, d_sz, transfer_src0 ? qx_sz : 0, 0, 0, 0, 0, inplace, qx_sz); ggml_vk_preallocate_buffers_reserve(node, d_sz, transfer_src0 ? qx_sz : 0, 0, 0, 0, 0, inplace, qx_sz);
break; break;
default: default:
return; return;
} }
break; break;
case GGML_OP_MUL_MAT: case GGML_OP_MUL_MAT:
ggml_vk_preallocate_buffers(node, d_sz, transfer_src0 ? qx_sz : 0, transfer_src1 ? qy_sz : 0, qx_needs_dequant ? x_sz : 0, qy_needs_dequant ? y_sz : 0, split_k > 1 ? d_sz * 4 : 0, inplace, qx_sz + qy_sz); ggml_vk_preallocate_buffers_reserve(node, d_sz, transfer_src0 ? qx_sz : 0, transfer_src1 ? qy_sz : 0, qx_needs_dequant ? x_sz : 0, qy_needs_dequant ? y_sz : 0, split_k > 1 ? d_sz * 4 : 0, inplace, qx_sz + qy_sz);
break; break;
default: default:
return; return;
@ -4167,7 +4167,7 @@ bool ggml_vk_compute_forward(ggml_compute_params * params, ggml_tensor * tensor)
ggml_vk_submit(vk_device.compute_queue, ctx.seqs, vk_fence); ggml_vk_submit(vk_device.compute_queue, ctx.seqs, vk_fence);
if (tensor == ctx.exit_tensor) { if (tensor == ctx.exit_tensor) {
VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, uint64_t(-1)), "ggml_vk_compute_forward waitForFences"); VK_CHECK(vk_device.device.waitForFences({ vk_fence }, true, UINT64_MAX), "ggml_vk_compute_forward waitForFences");
vk_device.device.resetFences({ vk_fence }); vk_device.device.resetFences({ vk_fence });
// Do staging buffer copies // Do staging buffer copies
@ -4399,6 +4399,9 @@ GGML_CALL static const char * ggml_backend_vk_buffer_type_name(ggml_backend_buff
} }
GGML_CALL static ggml_backend_buffer_t ggml_backend_vk_buffer_type_alloc_buffer(ggml_backend_buffer_type_t buft, size_t size) { GGML_CALL static ggml_backend_buffer_t ggml_backend_vk_buffer_type_alloc_buffer(ggml_backend_buffer_type_t buft, size_t size) {
#ifdef VK_DEBUG
std::cerr << "ggml_backend_vk_buffer_type_alloc_buffer(" << size << ")" << std::endl;
#endif
// vk_buffer dev_buffer = ggml_vk_create_buffer(size, vk::MemoryPropertyFlagBits::eDeviceLocal); // vk_buffer dev_buffer = ggml_vk_create_buffer(size, vk::MemoryPropertyFlagBits::eDeviceLocal);
ggml_backend_vk_buffer_context * ctx = new ggml_backend_vk_buffer_context(); ggml_backend_vk_buffer_context * ctx = new ggml_backend_vk_buffer_context();