CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-13 10:57:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

ggml-cpu: templateify ggml_compute_forward_rope_f32 and _f16 (#16805)

Browse Source 
* extract rotate_pairs logic from ggml_compute_forward_rope_f32

* templateify ggml_compute_forward_rope_f32 and _f16

* abort when rope type not supported, remove GLM from test-rope

* add imrope branch to switch

* add rope tests for perf

* Update ggml/src/ggml-cpu/ops.cpp

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

* Update ggml/src/ggml-cpu/ops.cpp

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
duduta
2025-11-11 13:33:24 +02:00
committed by GitHub
parent 8c583242ad
commit 73460f6278
3 changed files with 76 additions and 268 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
tests/test-backend-ops.cpp
View File

@@ -7603,6 +7603,22 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_perf() {
test_cases.emplace_back(new test_add_id(GGML_TYPE_F32, GGML_TYPE_F32, 2880, 32, 4, n_token));
}
for (bool fw : {true, false}) { // fw == forward
for (ggml_type type : {GGML_TYPE_F32, GGML_TYPE_F16}) {
for (bool ff : {false, true}) { // freq_factors
for (float v : { 0, 1 }) {
test_cases.emplace_back(new test_rope(type, {128, 32, 512, 1}, 128, GGML_ROPE_TYPE_NORMAL, 512, 1.0f, 0.0f, 1.0f, ff, v, fw)); // llama 7B
test_cases.emplace_back(new test_rope(type, {128, 64, 512, 1}, 128, GGML_ROPE_TYPE_NORMAL, 512, 1.0f, 0.0f, 1.0f, ff, v, fw)); // llama 65B
test_cases.emplace_back(new test_rope(type, { 80, 32, 512, 1}, 20, GGML_ROPE_TYPE_NEOX, 512, 1.0f, 0.0f, 1.0f, ff, v, fw)); // neox (stablelm)
test_cases.emplace_back(new test_rope(type, { 64, 8, 512, 1}, 64, GGML_ROPE_TYPE_NEOX, 512, 1.0f, 0.0f, 1.0f, ff, v, fw)); // neox (falcon 40B)
test_cases.emplace_back(new test_rope(type, {128, 12, 512, 1}, 128, GGML_ROPE_TYPE_MROPE, 512, 1.0f, 0.0f, 1.0f, ff, v, fw)); // rope_multi,m-rope (qwen2vl 2B)
test_cases.emplace_back(new test_rope(type, {128, 12, 2, 1}, 128, GGML_ROPE_TYPE_IMROPE, 512, 1.0f, 0.0f, 1.0f, ff, v, fw)); // rope_multi,imrope (qwen3vl 2B)
test_cases.emplace_back(new test_rope(type, { 80, 16, 2, 1}, 80, GGML_ROPE_TYPE_VISION, 512, 1.0f, 0.0f, 1.0f, ff, v, fw)); // rope_multi,m-rope (qwen2vl ViT)
}
}
}
}
std::vector<std::array<int64_t, 4>> reduce_rows_cases = {
{ 8192, 1, 1, 1 },
{ 8192, 8192, 1, 1 },