CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-27 08:21:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

ggml webgpu: actually add softmax, fix rms_norm offset (#16400)

Browse Source 
* implement soft_max

* Fix soft_max data race

* Temporary fix, wait on each submit
This commit is contained in:
Reese Levine
2025-10-04 20:59:31 -07:00
committed by GitHub
parent 86df2c9ae4
commit 35266573b9
3 changed files with 10 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
ggml/src/ggml-webgpu/ggml-webgpu.cpp
View File

@@ -424,6 +424,7 @@ static void ggml_backend_webgpu_build_and_enqueue(webgpu_context &
ctx->staged_param_bufs.push_back(params_bufs);
if (ctx->staged_command_bufs.size() == WEBGPU_COMMAND_SUBMIT_BATCH_SIZE) {
ggml_backend_webgpu_submit_queue(ctx);
ggml_backend_webgpu_wait_on_submission(ctx);
}
}
}
@@ -1060,6 +1061,9 @@ static bool ggml_webgpu_encode_node(webgpu_context ctx, ggml_tensor * node) {
case GGML_OP_SCALE:
ggml_webgpu_scale(ctx, src0, node);
break;
case GGML_OP_SOFT_MAX:
ggml_webgpu_soft_max(ctx, src0, src1, src2, node);
break;
default:
return false;
}
@@ -1806,6 +1810,9 @@ static bool ggml_backend_webgpu_device_supports_op(ggml_backend_dev_t dev, const
case GGML_OP_SCALE:
supports_op = op->type == GGML_TYPE_F32;
break;
case GGML_OP_SOFT_MAX:
supports_op = op->type == GGML_TYPE_F32;
break;
default:
break;
}
@@ -1949,6 +1956,7 @@ static ggml_backend_dev_t ggml_backend_webgpu_reg_get_device(ggml_backend_reg_t
ggml_webgpu_init_rope_pipeline(ctx);
ggml_webgpu_init_glu_pipeline(ctx);
ggml_webgpu_init_scale_pipeline(ctx);
ggml_webgpu_init_soft_max_pipeline(ctx);
#ifdef GGML_WEBGPU_DEBUG
// Initialize debug buffers

2
ggml/src/ggml-webgpu/wgsl-shaders/rms_norm.wgsl
View File

@@ -84,7 +84,7 @@ fn main(@builtin(workgroup_id) wid: vec3<u32>,
let i2 = i / params.ne1;
let i1 = i % params.ne1;
let i_src_row = params.offset_src + i3 * params.stride_src3 + i2 * params.stride_src2 + i1 * params.stride_src1;
let i_dst_row = params.offset_src + i3 * params.stride_dst3 + i2 * params.stride_dst2 + i1 * params.stride_dst1;
let i_dst_row = params.offset_dst + i3 * params.stride_dst3 + i2 * params.stride_dst2 + i1 * params.stride_dst1;
let elems = (params.ne0 + wg_size - 1) / wg_size;

1
ggml/src/ggml-webgpu/wgsl-shaders/soft_max.tmpl.wgsl
View File

@@ -300,6 +300,7 @@ fn main(@builtin(workgroup_id) wid: vec3<u32>,
workgroupBarrier();
}
let row_max = scratch[0];
workgroupBarrier();
var sum = 0.0f;
col = lid.x;