diff --git a/tests/hrx-loom-jit-test.cpp b/tests/hrx-loom-jit-test.cpp index 8ea9d23b7093..90b9a257bb89 100644 --- a/tests/hrx-loom-jit-test.cpp +++ b/tests/hrx-loom-jit-test.cpp @@ -239,10 +239,13 @@ static std::map flash_attention_config(const char * qu { "ggml.flash_attention.key_value_head_count", key_value_head_count }, { "ggml.flash_attention.qk_head_size", head_size }, { "ggml.flash_attention.value_head_size", head_size }, - // engine#2138: this line's prefill kernel takes the V row stride as a config (#95, MLA shapes); - // dispatch-flash-attention.cpp emits it, and test-hrx-loom-jit.cpp passes the same value. - { "ggml.flash_attention.value_stride", - std::to_string(std::stoll(key_value_head_count) * std::stoll(head_size)) }, + // engine#2138: this line's prefill kernel takes the V row stride as a config (#95, MLA shapes). + // It is ONE ROW: flash_attention_f32_f16_wmma.loom:331 names it `value_row_stride`, and + // dispatch-flash-attention.cpp:577 emits `match.value->nb[1] / sizeof(fp16)`. V is + // [head_size, kv_capacity, kv_head_count, 1] (the matcher requires ne[2] == key_value_head_count), so + // contiguous rows give nb[1]/2 = head_size -- multiplying by the head count would be a stride no + // dispatch produces. engine#2404. + { "ggml.flash_attention.value_stride", head_size }, { "ggml.flash_attention.attention_scale", attention_scale }, { "ggml.flash_attention.apply_gate", "0" }, { "ggml.flash_attention.gate_stride_head", "1" },