diff --git a/CMakeLists.txt b/CMakeLists.txt
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -375,7 +375,7 @@
 option(WITH_CYCLES_CUBIN_COMPILER   "Build cubins with nvrtc based compiler instead of nvcc" OFF)
 option(WITH_CYCLES_CUDA_BUILD_SERIAL "Build cubins one after another (useful on machines with limited RAM)" OFF)
 mark_as_advanced(WITH_CYCLES_CUDA_BUILD_SERIAL)
-set(CYCLES_CUDA_BINARIES_ARCH sm_30 sm_35 sm_37 sm_50 sm_52 sm_60 sm_61 sm_70 sm_75 CACHE STRING "CUDA architectures to build binaries for")
+set(CYCLES_CUDA_BINARIES_ARCH sm_30 sm_35 sm_37 sm_50 sm_52 sm_60 sm_61 sm_70 sm_75 sm_80 CACHE STRING "CUDA architectures to build binaries for")
 mark_as_advanced(CYCLES_CUDA_BINARIES_ARCH)
 unset(PLATFORM_DEFAULT)
 option(WITH_CYCLES_LOGGING  "Build Cycles with logging support" ON)
diff --git a/build_files/cmake/config/blender_release.cmake b/build_files/cmake/config/blender_release.cmake
--- a/build_files/cmake/config/blender_release.cmake
+++ b/build_files/cmake/config/blender_release.cmake
@@ -52,7 +52,7 @@
 set(WITH_MEM_JEMALLOC          ON  CACHE BOOL "" FORCE)
 set(WITH_CYCLES_CUDA_BINARIES  ON  CACHE BOOL "" FORCE)
 set(WITH_CYCLES_CUBIN_COMPILER OFF CACHE BOOL "" FORCE)
-set(CYCLES_CUDA_BINARIES_ARCH sm_30;sm_35;sm_37;sm_50;sm_52;sm_60;sm_61;sm_70;sm_75 CACHE STRING "" FORCE)
+set(CYCLES_CUDA_BINARIES_ARCH sm_30;sm_35;sm_37;sm_50;sm_52;sm_60;sm_61;sm_70;sm_75;sm_80 CACHE STRING "" FORCE)
 set(WITH_CYCLES_DEVICE_OPTIX   ON CACHE BOOL "" FORCE)
 
 # platform dependent options
diff --git a/intern/cycles/CMakeLists.txt b/intern/cycles/CMakeLists.txt
--- a/intern/cycles/CMakeLists.txt
+++ b/intern/cycles/CMakeLists.txt
@@ -313,7 +313,7 @@
       set(MAX_MSVC 1910)
     elseif(${CUDA_VERSION} EQUAL "9.1")
       set(MAX_MSVC 1911)
-    elseif(${CUDA_VERSION} LESS "11.0")
+    elseif(${CUDA_VERSION} LESS "12.0")
       set(MAX_MSVC 1999)
     endif()
     if(NOT MSVC_VERSION LESS ${MAX_MSVC} OR CMAKE_C_COMPILER_ID MATCHES "Clang")
diff --git a/intern/cycles/kernel/CMakeLists.txt b/intern/cycles/kernel/CMakeLists.txt
--- a/intern/cycles/kernel/CMakeLists.txt
+++ b/intern/cycles/kernel/CMakeLists.txt
@@ -485,8 +485,12 @@
   foreach(arch ${CYCLES_CUDA_BINARIES_ARCH})
     if(${arch} MATCHES "sm_2.")
       message(STATUS "CUDA binaries for ${arch} are no longer supported, skipped.")
+    elseif(${arch} MATCHES "sm_30" AND ${CUDA_VERSION} GREATER 109)
+      message(STATUS "CUDA binaries for ${arch} are no longer supported, skipped.")
     elseif(${arch} MATCHES "sm_7." AND ${CUDA_VERSION} LESS 100)
       message(STATUS "CUDA binaries for ${arch} require CUDA 10.0+, skipped.")
+    elseif(${arch} MATCHES "sm_8." AND ${CUDA_VERSION} LESS 110)
+      message(STATUS "CUDA binaries for ${arch} require CUDA 11.0+, skipped.")
     else()
       # Compile regular kernel
       CYCLES_CUDA_KERNEL_ADD(${arch} ${prev_arch} filter "" "${cuda_filter_sources}" FALSE)
@@ -525,6 +529,11 @@
       set(cuda_flags ${cuda_flags}
         -D __KERNEL_DEBUG__)
     endif()
+    set(OPTIX_TARGET 30)
+    if(${CUDA_VERSION} GREATER 109) #cuda 11
+      set(OPTIX_TARGET 52)
+    endif()
+
     if(WITH_CYCLES_CUBIN_COMPILER)
 
       # Needed to find libnvrtc-builtins.so. Can't do it from inside
@@ -536,7 +545,6 @@
         set(CUBIN_CC_ENV ${CMAKE_COMMAND}
           -E env LD_LIBRARY_PATH="${CUDA_TOOLKIT_ROOT_DIR}/lib64")
       endif()
-
       add_custom_command(
         OUTPUT ${output}
         DEPENDS
@@ -551,7 +559,7 @@
           ${SRC_UTIL_HEADERS}
         COMMAND ${CUBIN_CC_ENV}
             "$<TARGET_FILE:cycles_cubin_cc>"
-            -target 30
+            -target ${OPTIX_TARGET}
             -ptx
             -i ${CMAKE_CURRENT_SOURCE_DIR}/${input}
             ${cuda_flags}
@@ -575,7 +583,7 @@
         COMMAND
           ${CUDA_NVCC_EXECUTABLE}
           --ptx
-          -arch=sm_30
+          -arch=sm_${OPTIX_TARGET}
           ${cuda_flags}
           ${input}
         WORKING_DIRECTORY
diff --git a/intern/cycles/kernel/kernels/cuda/kernel_config.h b/intern/cycles/kernel/kernels/cuda/kernel_config.h
--- a/intern/cycles/kernel/kernels/cuda/kernel_config.h
+++ b/intern/cycles/kernel/kernels/cuda/kernel_config.h
@@ -70,8 +70,8 @@
 #  endif
 #  define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 63
 
-/* 7.x */
-#elif __CUDA_ARCH__ <= 799
+/* 7.x / 8.x */
+#elif __CUDA_ARCH__ <= 899
 #  define CUDA_MULTIPRESSOR_MAX_REGISTERS 65536
 #  define CUDA_MULTIPROCESSOR_MAX_BLOCKS 32
 #  define CUDA_BLOCK_MAX_THREADS 1024