From d27949f22a0cc2241da39b059fa09288bcfb3e30 Mon Sep 17 00:00:00 2001
From: Concedo <39025047+LostRuins@users.noreply.github.com>
Date: Sat, 6 Dec 2025 21:31:44 +0800
Subject: [PATCH] Revert "try remove volta as a dedicated target b (+1 squashed
 commits)"

This reverts commit ddba580f005115ead00951959b0b17c7cc547604.
---
 CMakeLists.txt | 4 ++--
 Makefile       | 2 ++
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/CMakeLists.txt b/CMakeLists.txt
index db51133ed..1a9847f79 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -139,10 +139,10 @@ if (LLAMA_CUBLAS)
         elseif(CUDAToolkit_VERSION VERSION_GREATER 12)
                 add_compile_definitions(GGML_CUDA_USE_GRAPHS) #try enable cuda graphs on cu12 build
                 add_compile_definitions(KCPP_LIMIT_CUDA_MAX_ARCH=800)
-                set(CMAKE_CUDA_ARCHITECTURES "50-virtual;61-virtual;75-virtual;80-virtual") # lowest CUDA 12 standard + lowest for integer intrinsics
+                set(CMAKE_CUDA_ARCHITECTURES "50-virtual;61-virtual;70-virtual;75-virtual;80-virtual") # lowest CUDA 12 standard + lowest for integer intrinsics
         else()
                 add_compile_definitions(KCPP_LIMIT_CUDA_MAX_ARCH=750) #will cause issues with ggml_cuda_highest_compiled_arch if removed
-                set(CMAKE_CUDA_ARCHITECTURES "35-virtual;50-virtual;61-virtual;75-virtual") # lowest CUDA 12 standard + lowest for integer intrinsics
+                set(CMAKE_CUDA_ARCHITECTURES "35-virtual;50-virtual;61-virtual;70-virtual;75-virtual") # lowest CUDA 12 standard + lowest for integer intrinsics
         endif()
     endif()
     message(STATUS "Using CUDA architectures: ${CMAKE_CUDA_ARCHITECTURES}")
diff --git a/Makefile b/Makefile
index a8a6095bf..f083bea29 100644
--- a/Makefile
+++ b/Makefile
@@ -226,6 +226,7 @@ NVCCFLAGS += -Wno-deprecated-gpu-targets \
              -gencode arch=compute_35,code=compute_35 \
              -gencode arch=compute_50,code=compute_50 \
              -gencode arch=compute_61,code=compute_61 \
+             -gencode arch=compute_70,code=compute_70 \
              -gencode arch=compute_75,code=compute_75 \
              -DKCPP_LIMIT_CUDA_MAX_ARCH=750
 
@@ -233,6 +234,7 @@ else ifdef LLAMA_ARCHES_CU12
 NVCCFLAGS += -Wno-deprecated-gpu-targets \
              -gencode arch=compute_50,code=compute_50 \
              -gencode arch=compute_61,code=compute_61 \
+             -gencode arch=compute_70,code=compute_70 \
              -gencode arch=compute_75,code=compute_75 \
              -gencode arch=compute_80,code=compute_80 \
 			 -DKCPP_LIMIT_CUDA_MAX_ARCH=800