root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-25 02:44:36 +00:00
Code Issues Actions 9 Packages Projects Releases Wiki Activity

metal : fix indent (ggml/0)

Browse Source
This commit is contained in:
Georgi Gerganov 2024-05-11 16:57:53 +03:00
parent 325756d28d
commit 6aeff24f8b
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
1 changed files with 14 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
ggml-metal.m
View File

@ -1195,24 +1195,24 @@ static enum ggml_status ggml_metal_graph_compute(
[encoder dispatchThreadgroups:MTLSizeMake(n, 1, 1) threadsPerThreadgroup:MTLSizeMake(1, 1, 1)]; [encoder dispatchThreadgroups:MTLSizeMake(n, 1, 1) threadsPerThreadgroup:MTLSizeMake(1, 1, 1)];
} break; } break;
case GGML_OP_CLAMP: case GGML_OP_CLAMP:
{ {
id<MTLComputePipelineState> pipeline = ctx->kernels[GGML_METAL_KERNEL_TYPE_CLAMP].pipeline; id<MTLComputePipelineState> pipeline = ctx->kernels[GGML_METAL_KERNEL_TYPE_CLAMP].pipeline;
float min; float min;
float max; float max;
memcpy(&min, ((int32_t *) dst->op_params) + 0, sizeof(float)); memcpy(&min, ((int32_t *) dst->op_params) + 0, sizeof(float));
memcpy(&max, ((int32_t *) dst->op_params) + 1, sizeof(float)); memcpy(&max, ((int32_t *) dst->op_params) + 1, sizeof(float));
[encoder setComputePipelineState:pipeline]; [encoder setComputePipelineState:pipeline];
[encoder setBuffer:id_src0 offset:offs_src0 atIndex:0]; [encoder setBuffer:id_src0 offset:offs_src0 atIndex:0];
[encoder setBuffer:id_dst offset:offs_dst atIndex:1]; [encoder setBuffer:id_dst offset:offs_dst atIndex:1];
[encoder setBytes:&min length:sizeof(min) atIndex:2]; [encoder setBytes:&min length:sizeof(min) atIndex:2];
[encoder setBytes:&max length:sizeof(max) atIndex:3]; [encoder setBytes:&max length:sizeof(max) atIndex:3];
const int64_t n = ggml_nelements(dst); const int64_t n = ggml_nelements(dst);
[encoder dispatchThreadgroups:MTLSizeMake(n, 1, 1) threadsPerThreadgroup:MTLSizeMake(1, 1, 1)]; [encoder dispatchThreadgroups:MTLSizeMake(n, 1, 1) threadsPerThreadgroup:MTLSizeMake(1, 1, 1)];
} break; } break;
case GGML_OP_UNARY: case GGML_OP_UNARY:
switch (ggml_get_unary_op(gf->nodes[i])) { switch (ggml_get_unary_op(gf->nodes[i])) {
// we are not taking into account the strides, so for now require contiguous tensors // we are not taking into account the strides, so for now require contiguous tensors