diff --git a/src/s2_generate.cpp b/src/s2_generate.cpp index 6ee4c83..5ec059d 100755 --- a/src/s2_generate.cpp +++ b/src/s2_generate.cpp @@ -48,7 +48,8 @@ GenerateResult generate( std::cout << "[Generate] Prefilling " << prompt.cols << " tokens..." << std::endl; } const auto prefill_t0 = std::chrono::steady_clock::now(); - if (!model.prefill_fast(prompt_tm, prompt.cols, params.n_threads, state)) { + // using prefill_fast causes segfault on vulkan with specific architectures (RDNA2 radv gfx 1031) + if (!model.prefill(prompt_tm, prompt.cols, params.n_threads, state)) { std::cerr << "[Generate] Prefill failed." << std::endl; return out; }