diff --git a/cmake/Modules/Packages/GPU.cmake b/cmake/Modules/Packages/GPU.cmake
index 141b086592..4c52eee68b 100644
--- a/cmake/Modules/Packages/GPU.cmake
+++ b/cmake/Modules/Packages/GPU.cmake
@@ -97,9 +97,9 @@ if(GPU_API STREQUAL "CUDA")
   if(CUDA_VERSION VERSION_GREATER_EQUAL "10.0")
     string(APPEND GPU_CUDA_GENCODE " -gencode arch=compute_75,code=[sm_75,compute_75]")
   endif()
-  # Ampere (GPU Arch 8.0) is supported by CUDA 11 and later
+  # Ampere (GPU Arch 8.0 and 8.6) is supported by CUDA 11 and later
   if(CUDA_VERSION VERSION_GREATER_EQUAL "11.0")
-    string(APPEND GPU_CUDA_GENCODE " -gencode arch=compute_80,code=[sm_80,compute_80]")
+    string(APPEND GPU_CUDA_GENCODE " -gencode arch=compute_80,code=[sm_80,compute_80] -gencode arch=compute_86,code=[sm_86,compute_86]")
   endif()
   if(CUDA_VERSION VERSION_GREATER_EQUAL "12.0")
     message(WARNING "Unsupported CUDA version. Use at your own risk.")
diff --git a/lib/gpu/Makefile.cuda b/lib/gpu/Makefile.cuda
index b18e4620eb..e03b59979a 100644
--- a/lib/gpu/Makefile.cuda
+++ b/lib/gpu/Makefile.cuda
@@ -38,7 +38,8 @@ NVCC = nvcc
 
 CUDA_ARCH = -gencode arch=compute_50,code=[sm_50,compute_50] -gencode arch=compute_52,code=[sm_52,compute_52] \
 	    -gencode arch=compute_60,code=[sm_60,compute_60] -gencode arch=compute_61,code=[sm_61,compute_61] \
-	    -gencode arch=compute_70,code=[sm_70,compute_70] -gencode arch=compute_75,code=[sm_75,compute_75]
+	    -gencode arch=compute_70,code=[sm_70,compute_70] -gencode arch=compute_75,code=[sm_75,compute_75] \
+	    -gencode arch=compute_80,code=[sm_80,compute_80] -gencode arch=compute_86,code=[sm_86,compute_86]
 CUDA_INCLUDE = -I$(CUDA_HOME)/include
 CUDA_LIB = -L$(CUDA_HOME)/lib64 -L$(CUDA_HOME)/lib64/stubs
 CUDA_OPTS = -DUNIX -O3 --use_fast_math $(LMP_INC) -Xcompiler -fPIC
diff --git a/lib/gpu/Makefile.cuda_mps b/lib/gpu/Makefile.cuda_mps
index c6e5202adc..172640ce6a 100644
--- a/lib/gpu/Makefile.cuda_mps
+++ b/lib/gpu/Makefile.cuda_mps
@@ -38,7 +38,8 @@ NVCC = nvcc
 
 CUDA_ARCH = -gencode arch=compute_50,code=[sm_50,compute_50] -gencode arch=compute_52,code=[sm_52,compute_52] \
 	    -gencode arch=compute_60,code=[sm_60,compute_60] -gencode arch=compute_61,code=[sm_61,compute_61] \
-	    -gencode arch=compute_70,code=[sm_70,compute_70] -gencode arch=compute_75,code=[sm_75,compute_75]
+	    -gencode arch=compute_70,code=[sm_70,compute_70] -gencode arch=compute_75,code=[sm_75,compute_75] \
+	    -gencode arch=compute_80,code=[sm_80,compute_80] -gencode arch=compute_86,code=[sm_86,compute_86]
 CUDA_INCLUDE = -I$(CUDA_HOME)/include
 CUDA_LIB = -L$(CUDA_HOME)/lib64 -L$(CUDA_HOME)/lib64/stubs
 CUDA_OPTS = -DUNIX -O3 --use_fast_math $(LMP_INC) -Xcompiler -fPIC
diff --git a/lib/gpu/Makefile.linux b/lib/gpu/Makefile.linux
index da18ae41ca..5fff81072b 100644
--- a/lib/gpu/Makefile.linux
+++ b/lib/gpu/Makefile.linux
@@ -41,6 +41,10 @@ CUDA_ARCH = -arch=sm_50
 # Turing hardware
 #CUDA_ARCH = -arch=sm_75
 
+# Ampere hardware
+#CUDA_ARCH = -arch=sm_80
+#CUDA_ARCH = -arch=sm_86
+
 # this setting should match LAMMPS Makefile
 # one of LAMMPS_SMALLBIG (default), LAMMPS_BIGBIG and LAMMPS_SMALLSMALL
 
diff --git a/lib/gpu/Makefile.linux_multi b/lib/gpu/Makefile.linux_multi
index e0c250a951..aa7d92572d 100644
--- a/lib/gpu/Makefile.linux_multi
+++ b/lib/gpu/Makefile.linux_multi
@@ -31,9 +31,24 @@ NVCC = nvcc
 CUDA_ARCH = -arch=sm_50
 #CUDA_ARCH = -arch=sm_52
 
+# Pascal hardware
+#CUDA_ARCH = -arch=sm_60
+#CUDA_ARCH = -arch=sm_61
+
+# Volta hardware
+#CUDA_ARCH = -arch=sm_70
+
+# Turing hardware
+#CUDA_ARCH = -arch=sm_75
+
+# Ampere hardware
+#CUDA_ARCH = -arch=sm_80
+#CUDA_ARCH = -arch=sm_86
+
 CUDA_CODE = -gencode arch=compute_50,code=[sm_50,compute_50] -gencode arch=compute_52,code=[sm_52,compute_52] \
 	    -gencode arch=compute_60,code=[sm_60,compute_60] -gencode arch=compute_61,code=[sm_61,compute_61] \
-	    -gencode arch=compute_70,code=[sm_70,compute_70] -gencode arch=compute_75,code=[sm_75,compute_75]
+	    -gencode arch=compute_70,code=[sm_70,compute_70] -gencode arch=compute_75,code=[sm_75,compute_75] \
+	    -gencode arch=compute_80,code=[sm_80,compute_80] -gencode arch=compute_86,code=[sm_86,compute_86]
 
 CUDA_ARCH += $(CUDA_CODE)