ClaudeBot/llama.cpp
0
0
Fork 1
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-09-17 20:31:47 +02:00
Code Issues Packages Projects Releases Wiki Activity

cuda: support row-contiguous SUM_ROWS (#26308)

Browse Source 
* cuda: support row-contiguous SUM_ROWS

* organize the code and add GGML_OP_MEAN to support row-contiguous tensors using the same shared kernel, and add a test to MEAN permute/slice

* Keep original comments and add if/else branch
This commit is contained in:
Mohamed Elashri
2026-09-15 12:39:29 +02:00
committed by GitHub
parent 9e71716247
commit 5431581326
5 changed files with 83 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/test-backend-ops.cpp
+18 -3
View File
@@ -7126,20 +7126,32 @@ struct test_sum_rows : public test_case {
struct test_mean : public test_case {
const ggml_type type;
const std::array<int64_t, 4> ne;
const bool permute;
const bool slice;
std::string vars() override {
return VARS_TO_STR2(type, ne);
return VARS_TO_STR4(type, ne, permute, slice);
}
test_mean(ggml_type type = GGML_TYPE_F32,
std::array<int64_t, 4> ne = {10, 5, 4, 3})
: type(type), ne(ne) {}
std::array<int64_t, 4> ne = {10, 5, 4, 3},
bool permute = false, bool slice = false)
: type(type), ne(ne), permute(permute), slice(slice) {}
ggml_tensor * build_graph(ggml_context * ctx) override {
ggml_tensor * a = ggml_new_tensor(ctx, type, 4, ne.data());
ggml_set_param(a);
ggml_set_name(a, "a");
if (slice) {
a = ggml_view_4d(ctx, a,
ne[0], ne[1], ne[2] / 2, ne[3] - 1,
a->nb[1], a->nb[2] * 2, a->nb[3], /*offset=*/a->nb[3]);
}
if (permute) {
a = ggml_permute(ctx, a, 0, 2, 3, 1);
}
ggml_tensor * out = ggml_mean(ctx, a);
ggml_set_name(out, "out");
@@ -10470,6 +10482,9 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
test_cases.emplace_back(new test_mean(GGML_TYPE_F32, { 32, 1, 1, 1 }));
test_cases.emplace_back(new test_mean(GGML_TYPE_F32, { 32, 256, 1, 1 }));
test_cases.emplace_back(new test_mean(GGML_TYPE_F32, { 32768, 1, 1, 1 }));
test_cases.emplace_back(new test_mean(GGML_TYPE_F32, { 11, 5, 6, 3 }, true, false));
test_cases.emplace_back(new test_mean(GGML_TYPE_F32, { 11, 5, 6, 3 }, false, true));
test_cases.emplace_back(new test_mean(GGML_TYPE_F32, { 11, 5, 6, 3 }, true, true));
test_cases.emplace_back(new test_sum(GGML_TYPE_F32, { 33, 1, 1, 1 }));
test_cases.emplace_back(new test_sum(GGML_TYPE_F32, { 33, 1024, 1, 1 }));
test_cases.emplace_back(new test_sum(GGML_TYPE_F32, { 33, 256, 1, 1 }));