]> git.djapps.eu Git - pkg/ggml/sources/whisper.cpp/commit
vulkan: Allow up to 4096 elements for mul_mat_id row_ids (llama/13326)
authorJeff Bolz <redacted>
Fri, 9 May 2025 07:23:41 +0000 (02:23 -0500)
committerGeorgi Gerganov <redacted>
Tue, 13 May 2025 10:59:21 +0000 (13:59 +0300)
commite46df4850f7ce25468cae602f404f0f1477c0a39
treee228ebe3b01ab66f2fd4d91ff2b11fe62e5823ee
parente8a7f1b7bbd8dc73dc53887d4007c0a078eaae77
vulkan: Allow up to 4096 elements for mul_mat_id row_ids (llama/13326)

This assert fired running Qwen_Qwen3-30B-A3B-Q2_K.gguf:

GGML_ASSERT(nei0 * nei1 <= 3072);

The tensor is 8 x 512. Increase this array size to accommodate.
ggml/src/ggml-vulkan/ggml-vulkan.cpp
ggml/src/ggml-vulkan/vulkan-shaders/mul_mm.comp
ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_cm2.comp
ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq.comp