public_git/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/whisper.cpp.git synced 2026-10-12 17:25:39 +02:00
Code Issues Packages Projects Releases Wiki Activity

tests : refactor test-recurrent-state-rollback (llama/29426)

Browse Source 
* tests : use llama_context_ptr in test-recurrent-state-rollback

Replace raw llama_context pointers with llama_context_ptr and drop the
manual llama_free calls and cleanup lambda.

Assisted-by: pi:llama.cpp/MiMo-V2.6-Flash-RL

* tests : run test-recurrent-state-rollback over all dummy models

Add a --models DIR mode that mirrors test-save-load-state: iterate every
dummy model, report PASS/FAIL/SKIP in a table and fail only when a model
fails. Register a single ctest entry with ARGS --models instead of the four
per-model registrations.

Assisted-by: pi:llama.cpp/MiMo-V2.6-Flash-RL

* cont : fix typo

* metal : allow fusing 0-element nodes to keep graph packing shape-independent

The fusion packing in ggml_metal_fusion_max excluded 0-element tensors and
the topk_moe/moe_reduce checks rejected n_tokens == 0, so graphs decoding
batches with no outputs packed differently from the worst-case reserved
graph. The Metal optimizer then reordered the nodes differently and
ggml_gallocr_needs_realloc failed on the layout mismatch, forcing an
unexpected graph re-reserve (caught by GGML_SCHED_DEBUG_REALLOC).

Treat empty tensors like their non-empty counterparts: match them in the
pattern sequence and only reject genuinely malformed shapes. Fused kernels
dispatch zero threadgroups for empty graphs, which is a legal no-op.

Assisted-by: pi:llama.cpp/MiMo-V2.6-Flash-RL

* tests : run test_multi_seq_split_replay as a separate test

test_multi_seq_split_replay was invoked at the end of test_rollback,
so its result was folded into the rollback status and it only ran when
the rollback part passed.

Give it its own test_status return, run both tests independently over
both cache fills via a shared run_tests helper, and report them as
separate rollback / split replay columns in the --models table with
per-test summaries. The exit code fails when either test fails.

Assisted-by: pi:llama.cpp/MiMo-V2.6-Flash-RL

* tests : loosen the split replay nmse bound to 1e-4

test-generate-models seeds its weights from std::random_device, and some
generated lfm2 models drift up to ~1.7e-5 nmse on the split replay due to
rounding noise, tripping the previous 1e-5 bound. Raise the bound to 1e-4
so the random generations stop flaking.

Assisted-by: pi:llama.cpp/MiMo-V2.6-Flash-RL

* tests : reuse run_tests_for_model in single-model mode

The single-model path duplicated the model init and the non-recurrent
check from run_tests_for_model; route it through the shared helper
instead. Model load failures now return FAIL rather than SKIP so that
--model with a broken file still exits non-zero, and the helper loads
with model_only like the --models loop does since the tests create
their own contexts.

Assisted-by: pi:llama.cpp/MiMo-V2.6-Flash-RL
This commit is contained in:
Georgi Gerganov
2026-10-06 10:38:07 +03:00
parent 83fd1554ca
commit ee86a150bd
1 changed files with 9 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-metal/ggml-metal-fusion.cpp
+9 -6
View File
@@ -418,7 +418,8 @@ static bool ggml_metal_fusion_check_topk_moe(
const int64_t n_tokens = logits->ne[1];
const int64_t n_expert_used = ids->ne[0];
if (n_expert <= 0 || n_tokens <= 0 || n_expert_used <= 0 || n_expert_used > n_expert ||
// note: n_tokens == 0 (no-output batch) must match so that the packing stays shape-independent
if (n_expert <= 0 || n_expert_used <= 0 || n_expert_used > n_expert ||
n_expert > GGML_METAL_TOPK_MOE_MAX_EXPERTS || n_expert_used > GGML_METAL_TOPK_MOE_MAX_EXPERTS) {
return false;
}
@@ -545,7 +546,8 @@ static bool ggml_metal_fusion_match_moe_reduce(
const int64_t n_embd = experts->ne[0];
const int64_t n_tokens = experts->ne[2];
if (n_embd <= 0 || n_tokens <= 0 || experts->ne[1] != n_expert_used || experts->ne[3] != 1 ||
// note: n_tokens == 0 (no-output batch) must match so that the packing stays shape-independent
if (n_embd <= 0 || experts->ne[1] != n_expert_used || experts->ne[3] != 1 ||
weights->ne[0] != 1 || weights->ne[1] != n_expert_used || weights->ne[2] != n_tokens || weights->ne[3] != 1 ||
dst->ne[0] != n_embd || dst->ne[1] != n_tokens || dst->ne[2] != 1 || dst->ne[3] != 1) {
return false;
@@ -1079,16 +1081,17 @@ const ggml_metal_fusion * ggml_metal_fusion_next(
// transparent) node sequence that the compute phase uses, so the returned count is the raw index
// span from idx to the last matched node (intermediate views are packed along).
int ggml_metal_fusion_max(const ggml_cgraph * gf, int idx) {
// an empty/view node cannot start a pattern - pack it alone
if (ggml_op_is_empty(gf->nodes[idx]->op) || ggml_is_empty(gf->nodes[idx])) {
// a view node cannot start a pattern - pack it alone
if (ggml_op_is_empty(gf->nodes[idx]->op)) {
return 1;
}
// collect the non-empty node indices starting at idx
// collect the non-view node indices starting at idx; 0-element tensors are included so
// that empty graphs pack like their non-empty counterparts (see ggml_metal_fusion_filter_ops)
int idxs[GGML_METAL_FUSION_MAX];
int n_idxs = 0;
for (int i = idx; i < gf->n_nodes && n_idxs < GGML_METAL_FUSION_MAX; i++) {
if (!ggml_op_is_empty(gf->nodes[i]->op) && !ggml_is_empty(gf->nodes[i])) {
if (!ggml_op_is_empty(gf->nodes[i]->op)) {
idxs[n_idxs++] = i;
}
}