sleepy/llama.cpp
1
0
Fork 0
Code Issues 12 Pull Requests Actions 217 Packages Projects Releases Wiki Activity

vulkan: Implement GGML_OP_TRI (#17503)

Browse Source 
* vulkan: Implement GGML_OP_TRI

* check types match
This commit is contained in:
Jeff Bolz
2025-11-28 03:07:29 -06:00
committed by GitHub
parent 15d2b46b4d
commit 35cf8887e1
3 changed files with 73 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-vulkan/ggml-vulkan.cpp
+27 -1
View File
@@ -649,6 +649,7 @@ struct vk_device_struct {
vk_pipeline pipeline_sin_f32;
vk_pipeline pipeline_cos_f32;
vk_pipeline pipeline_log[2];
vk_pipeline pipeline_tri[2];
vk_pipeline pipeline_clamp_f32;
vk_pipeline pipeline_pad_f32;
vk_pipeline pipeline_roll_f32;
@@ -3876,6 +3877,9 @@ static void ggml_vk_load_shaders(vk_device& device) {
ggml_vk_create_pipeline(device, device->pipeline_log[1], "log_f16", log_f16_len, log_f16_data, "main", 2, sizeof(vk_op_unary_push_constants), {512, 1, 1}, {}, 1);
}
ggml_vk_create_pipeline(device, device->pipeline_tri[0], "tri_f32", tri_f32_len, tri_f32_data, "main", 2, sizeof(vk_op_unary_push_constants), {512, 1, 1}, {}, 1);
ggml_vk_create_pipeline(device, device->pipeline_tri[1], "tri_f16", tri_f16_len, tri_f16_data, "main", 2, sizeof(vk_op_unary_push_constants), {512, 1, 1}, {}, 1);
ggml_vk_create_pipeline(device, device->pipeline_clamp_f32, "clamp_f32", clamp_f32_len, clamp_f32_data, "main", 2, sizeof(vk_op_unary_push_constants), {512, 1, 1}, {}, 1);
ggml_vk_create_pipeline(device, device->pipeline_pad_f32, "pad_f32", pad_f32_len, pad_f32_data, "main", 2, sizeof(vk_op_pad_push_constants), {512, 1, 1}, {}, 1);
@@ -8290,6 +8294,12 @@ static vk_pipeline ggml_vk_op_get_pipeline(ggml_backend_vk_context * ctx, const
return ctx->device->pipeline_log[dst->type == GGML_TYPE_F16];
}
return nullptr;
case GGML_OP_TRI:
if (src0->type == dst->type &&
(src0->type == GGML_TYPE_F32 || src0->type == GGML_TYPE_F16)) {
return ctx->device->pipeline_tri[dst->type == GGML_TYPE_F16];
}
return nullptr;
case GGML_OP_CLAMP:
if (src0->type == GGML_TYPE_F32 && dst->type == GGML_TYPE_F32) {
return ctx->device->pipeline_clamp_f32;
@@ -8991,6 +9001,7 @@ static void ggml_vk_op_f32(ggml_backend_vk_context * ctx, vk_context& subctx, co
case GGML_OP_SIN:
case GGML_OP_COS:
case GGML_OP_LOG:
case GGML_OP_TRI:
case GGML_OP_CLAMP:
case GGML_OP_PAD:
case GGML_OP_ROLL:
@@ -9671,6 +9682,13 @@ static void ggml_vk_log(ggml_backend_vk_context * ctx, vk_context& subctx, const
ggml_vk_op_f32(ctx, subctx, src0, nullptr, nullptr, nullptr, dst, GGML_OP_LOG, vk_op_unary_push_constants_init(src0, dst));
}
static void ggml_vk_tri(ggml_backend_vk_context * ctx, vk_context& subctx, const ggml_tensor * src0, ggml_tensor * dst) {
vk_op_unary_push_constants p = vk_op_unary_push_constants_init(src0, dst);
p.param1 = ggml_get_op_params_f32(dst, 0);
ggml_vk_op_f32(ctx, subctx, src0, nullptr, nullptr, nullptr, dst, GGML_OP_TRI, std::move(p));
}
static void ggml_vk_clamp(ggml_backend_vk_context * ctx, vk_context& subctx, const ggml_tensor * src0, ggml_tensor * dst) {
vk_op_unary_push_constants p = vk_op_unary_push_constants_init(src0, dst);
p.param1 = ggml_get_op_params_f32(dst, 0);
@@ -11794,6 +11812,10 @@ static bool ggml_vk_build_graph(ggml_backend_vk_context * ctx, ggml_cgraph * cgr
case GGML_OP_LOG:
ggml_vk_log(ctx, compute_ctx, src0, node);
break;
case GGML_OP_TRI:
ggml_vk_tri(ctx, compute_ctx, src0, node);
break;
case GGML_OP_CLAMP:
ggml_vk_clamp(ctx, compute_ctx, src0, node);
@@ -13919,7 +13941,9 @@ static bool ggml_backend_vk_device_supports_op(ggml_backend_dev_t dev, const ggm
case GGML_OP_OPT_STEP_SGD:
return ggml_is_contiguous(op->src[0]) && op->src[0]->type == GGML_TYPE_F32;
case GGML_OP_LOG:
return op->src[0]->type == GGML_TYPE_F32 || op->src[0]->type == GGML_TYPE_F16;
case GGML_OP_TRI:
return (op->src[0]->type == GGML_TYPE_F32 || op->src[0]->type == GGML_TYPE_F16) &&
op->type == op->src[0]->type;
case GGML_OP_ARGSORT:
{
if (!ggml_is_contiguous(op) || !ggml_is_contiguous(op->src[0])) {
@@ -14510,6 +14534,8 @@ static void ggml_vk_check_results_0(ggml_backend_vk_context * ctx, ggml_cgraph *
tensor_clone = ggml_cos(ggml_ctx, src_clone[0]);
} else if (tensor->op == GGML_OP_LOG) {
tensor_clone = ggml_log(ggml_ctx, src_clone[0]);
} else if (tensor->op == GGML_OP_TRI) {
tensor_clone = ggml_tri(ggml_ctx, src_clone[0], ggml_get_op_params_i32(tensor, 0));
} else if (tensor->op == GGML_OP_CLAMP) {
const float * params = (const float *)tensor->op_params;
tensor_clone = ggml_clamp(ggml_ctx, src_clone[0], params[0], params[1]);