Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦
C
build: let the compiler supply the cap test's header prerequisites (#1758)
Jaapjan Tinbergen45d7a102cdfa
diff --git a/c/Makefile b/c/Makefileindex 25bdb85f..7251b080 100644--- a/c/Makefile+++ b/c/Makefile@@ -1591,8 +1591,8 @@ tests/test_qwen36_slot_int8$(EXE): tests/test_qwen36_slot_int8.c qwen36.c $(QWEN$(CC) $(QWEN36_CFLAGS) $< $(QWEN36_TIER_OBJ) $(CUDA_OBJ) $(VK_OBJ) -o $@ $(QWEN36_LDFLAGS)# qwen36_cap_for_ram(): the cap<=0 "auto" sentinel's RAM-derived slot count.-tests/test_qwen36_cap_precedence$(EXE): tests/test_qwen36_cap_precedence.c qwen36.c qwen36_tier.h st.h json.h compat.h $(QWEN36_TIER_SRC) $(CUDA_OBJ)- $(CC) $(QWEN36_CFLAGS) $< $(QWEN36_TIER_SRC) $(CUDA_OBJ) -o $@ $(QWEN36_LDFLAGS)+tests/test_qwen36_cap_precedence$(EXE): tests/test_qwen36_cap_precedence.c qwen36.c $(QWEN36_TIER_OBJ) $(CUDA_OBJ) $(VK_OBJ)+ $(CC) $(QWEN36_CFLAGS) $< $(QWEN36_TIER_OBJ) $(CUDA_OBJ) $(VK_OBJ) -o $@ $(QWEN36_LDFLAGS)# the dense trunk's integer path: quantizer contract, int4 planar packing, dispatch.tests/test_qwen36_dense_idot$(EXE): tests/test_qwen36_dense_idot.c qwen36.c $(QWEN36_TIER_OBJ) $(CUDA_OBJ) $(VK_OBJ)