kquants_iter for hipblas and add gfx803

Author: YellowRoseCx

Makefile

@@ -167,13 +167,21 @@ ifdef LLAMA_HIPBLAS
 	ROCM_PATH  ?= /opt/rocm
 	CC         := $(ROCM_PATH)/llvm/bin/clang
 	CXX        := $(ROCM_PATH)/llvm/bin/clang++
-	GPU_TARGETS = gfx900 gfx906 gfx908 gfx90a gfx1030
+	GPU_TARGETS = gfx803 gfx900 gfx906 gfx908 gfx90a gfx1030
 	LLAMA_CUDA_DMMV_X ?= 64
 	LLAMA_CUDA_DMMV_Y ?= 2
 	CFLAGS     += -DGGML_USE_HIPBLAS -DGGML_USE_CUBLAS $(shell $(ROCM_PATH)/bin/hipconfig -C)
 	CXXFLAGS   += -DGGML_USE_HIPBLAS -DGGML_USE_CUBLAS $(shell $(ROCM_PATH)/bin/hipconfig -C)
 	LDFLAGS    += -L/opt/rocm/lib -Wl,-rpath=$(ROCM_PATH)/lib -lhipblas -lamdhip64
 	OBJS       += ggml-cuda.o ggml_v2-cuda.o
+ifdef LLAMA_CUDA_DMMV_F16
+	NVCCFLAGS += -DGGML_CUDA_DMMV_F16
+endif # LLAMA_CUDA_DMMV_F16
+ifdef LLAMA_CUDA_KQUANTS_ITER
+	NVCCFLAGS += -DK_QUANTS_PER_ITERATION=$(LLAMA_CUDA_KQUANTS_ITER)
+else
+	NVCCFLAGS += -DK_QUANTS_PER_ITERATION=2
+endif
 ggml-cuda.o: CXXFLAGS += $(addprefix --offload-arch=,$(GPU_TARGETS))
 ggml-cuda.o: CXXFLAGS += -DGGML_CUDA_DMMV_X=$(LLAMA_CUDA_DMMV_X)
 ggml-cuda.o: CXXFLAGS += -DGGML_CUDA_DMMV_Y=$(LLAMA_CUDA_DMMV_Y)