A single-model (Bonsai-27B) inference engine for Intel Arc GPUs
qxmx tests
20 files

attn_forward_batch_test.cpp
chat_template_test.cpp
deltanet_naive_test.cpp
deltanet_prefill_test.cpp
deltanet_wyut_test.cpp
elementwise_batch_test.cpp
fa_fused_bench.cpp
fa_kernel_test.cpp
ffn_batch_test.cpp
ffn_timing_test.cpp
gemm_engine_test.cpp
gemm_tf32_test.cpp
gemm_tune.cpp
prefix_cache_engine_test.cpp
prefix_cache_test.cpp
scheduler_test.cpp
ssm_batch_test.cpp
ssm_timing_test.cpp
tool_call_test.cpp
tpl_crosscheck.py