ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 2 ROCm devices: Device 0: AMD Radeon Graphics, gfx1201 (0x1201), VMM: no, Wave Size: 32 Device 1: AMD Radeon Graphics, gfx1201 (0x1201), VMM: no, Wave Size: 32 | model | size | params | backend | ngl | n_ubatch | fa | test | t/s | | ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: | /opt/llama.cpp/ggml/src/ggml-cuda/ggml-cuda.cu:92: ROCm error /usr/local/lib64/libggml-base.so.0(+0x3565) [0x7f28a4d6b565] /usr/local/lib64/libggml-base.so.0(ggml_print_backtrace+0x1eb) [0x7f28a4d6b92b] /usr/local/lib64/libggml-base.so.0(ggml_abort+0x11f) [0x7f28a4d6baaf] /usr/local/lib64/libggml-hip.so.0(+0x2efaf92) [0x7f28a7d22f92] /usr/local/lib64/libggml-hip.so.0(+0x2f0f7f7) [0x7f28a7d377f7] /usr/local/lib64/libggml-hip.so.0(+0x2f0ddcc) [0x7f28a7d35dcc] /usr/local/lib64/libggml-hip.so.0(+0x2f0cf97) [0x7f28a7d34f97] /usr/local/lib64/libggml-hip.so.0(+0x2f07a7b) [0x7f28a7d2fa7b] /usr/local/lib64/libggml-hip.so.0(+0x2f03d4b) [0x7f28a7d2bd4b] /usr/local/lib64/libggml-hip.so.0(+0x2f0020f) [0x7f28a7d2820f] /usr/local/lib64/libggml-base.so.0(ggml_backend_sched_graph_compute_async+0x7f3) [0x7f28a4d85f63] /usr/local/lib64/libllama.so.0(_ZN13llama_context13graph_computeEP11ggml_cgraphb+0xa0) [0x7f28a83f0a90] /usr/local/lib64/libllama.so.0(_ZN13llama_context14process_ubatchERK12llama_ubatch14llm_graph_typeP22llama_memory_context_iR11ggml_status+0xe2) [0x7f28a83f2722] /usr/local/lib64/libllama.so.0(_ZN13llama_context6decodeERK11llama_batch+0x3bf) [0x7f28a83f75df] /usr/local/lib64/libllama.so.0(llama_decode+0xe) [0x7f28a83f83fe] /usr/local/bin/llama-bench() [0x40acdb] /usr/local/bin/llama-bench() [0x4087ec] /lib64/libc.so.6(+0x35b5) [0x7f28a47015b5] /lib64/libc.so.6(__libc_start_main+0x88) [0x7f28a4701668] /usr/local/bin/llama-bench() [0x409b65] ✖ ! [rocm-7.9-rocwmma] gemma-3-27b-it-BF16-00001-of-00002__hblt0__fa1 __longctx16384 failed (exit 0)