From 3d0112261042b356621e93db3fa4c6798a5d098f Mon Sep 17 00:00:00 2001
From: Kawrakow <48489457+ikawrakow@users.noreply.github.com>
Date: Fri, 16 Jun 2023 20:08:44 +0300
Subject: CUDA : faster k-quant dot kernels (#1862)

* cuda : faster k-quant dot kernels

* Imrove Q2_K dot kernel on older GPUs

We now have a K_QUANTS_PER_ITERATION macro, which should be
set to 1 on older and to 2 on newer GPUs.
With this, we preserve the performance of the original
PR on RTX-4080, and are faster compared to master on
GTX-1660.

* Imrove Q6_K dot kernel on older GPUs

Using the same K_QUANTS_PER_ITERATION macro as last commit,
we preserve performance on RTX-4080 and speed up
Q6_K on a GTX-1660.

* Add LLAMA_CUDA_KQUANTS_ITER to CMakeLists.txt and Makefile

Allowed values are 1 or 2. 2 gives the best performance on
modern GPUs and is set as default. On older GPUs 1 may work
better.

* PR comments

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
---
 Makefile | 5 +++++
 1 file changed, 5 insertions(+)

(limited to 'Makefile')

diff --git a/Makefile b/Makefile
index 09c8834..b24caf8 100644
--- a/Makefile
+++ b/Makefile
@@ -171,6 +171,11 @@ ifdef LLAMA_CUDA_DMMV_Y
 else
 	NVCCFLAGS += -DGGML_CUDA_DMMV_Y=1
 endif # LLAMA_CUDA_DMMV_Y
+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: ggml-cuda.cu ggml-cuda.h
 	$(NVCC) $(NVCCFLAGS) $(CXXFLAGS) -Wno-pedantic -c $< -o $@
 endif # LLAMA_CUBLAS
-- 
cgit v1.2.3