diff --git a/RCKangaroo.cpp b/RCKangaroo.cpp
index 0ae7c12..6aa69e7 100644
--- a/RCKangaroo.cpp
+++ b/RCKangaroo.cpp
@@ -96,9 +96,9 @@ void InitGpus()
 		cudaGetDeviceProperties(&deviceProp, i);
 		printf("GPU %d: %s, %.2f GB, %d CUs, cap %d.%d, PCI %d, L2 size: %d KB\r\n", i, deviceProp.name, ((float)(deviceProp.totalGlobalMem / (1024 * 1024))) / 1024.0f, deviceProp.multiProcessorCount, deviceProp.major, deviceProp.minor, deviceProp.pciBusID, deviceProp.l2CacheSize / 1024);
 		
-		if (deviceProp.l2CacheSize < 16 * 1024 * 1024) //we need large L2 cache which is available for ADA and later. 16MB means RTX4050, the slowest card with minimal L2
+		if (deviceProp.major < 8)
 		{
-			printf("GPU %d - L2 cache size is too small for this application, skip\r\n", i);
+			printf("GPU %d - not supported, skip\r\n", i);
 			continue;
 		}
 
@@ -536,12 +536,12 @@ int main(int argc, char* argv[])
 #endif
 
 	printf("********************************************************************************\r\n");
-	printf("*                    RCKangaroo v1.0  (c) 2024 RetiredCoder                    *\r\n");
+	printf("*                    RCKangaroo v1.1  (c) 2024 RetiredCoder                    *\r\n");
 	printf("********************************************************************************\r\n\r\n");
 
 	printf("This software is free and open-source: https://github.com/RetiredC\r\n");
 	printf("It demonstrates fast GPU implementation of SOTA Kangaroo method for solving ECDLP\r\n");
-	printf("NOTE: Only RTX 40xx and newer cards are supported.\r\n\r\n");
+	printf("NOTE: Only RTX 40xx and 30xx cards are supported.\r\n\r\n");
 #ifdef DEBUG_MODE
 	printf("DEBUG MODE\r\n\r\n");
 #endif
diff --git a/RCKangaroo.vcxproj b/RCKangaroo.vcxproj
index c5607f7..c89b5a5 100644
--- a/RCKangaroo.vcxproj
+++ b/RCKangaroo.vcxproj
@@ -115,7 +115,7 @@
       <GenerateDebugInformation>true</GenerateDebugInformation>
     </Link>
     <CudaCompile>
-      <CodeGeneration>compute_89,sm_89</CodeGeneration>
+      <CodeGeneration>compute_89,sm_89;compute_86,sm_86</CodeGeneration>
       <MaxRegCount>0</MaxRegCount>
       <FastMath>false</FastMath>
     </CudaCompile>
@@ -138,7 +138,7 @@
       <GenerateDebugInformation>true</GenerateDebugInformation>
     </Link>
     <CudaCompile>
-      <CodeGeneration>compute_89,sm_89</CodeGeneration>
+      <CodeGeneration>compute_89,sm_89;compute_86,sm_86</CodeGeneration>
     </CudaCompile>
   </ItemDefinitionGroup>
   <ItemGroup>
diff --git a/README.md b/README.md
index 5deaaa0..fe84075 100644
--- a/README.md
+++ b/README.md
@@ -16,7 +16,7 @@ Discussion thread: https://bitcointalk.org/index.php?topic=5517607
 
 <b>Limitations:</b>
 
-- Only RTX 40xx and newer GPUs are supported.
+- Only RTX 40xx and 30xx GPUs are supported, but the software is optimized for 40xx only, so 30xx cards have bad speed.
 - No advanced features like networking, saving/loading DPs, etc.
 
 <b>Command line parameters:</b>