OpenCL Kernels: Improve performance preview accuracy in --benchmark, --speed-only and --progress-only mode
This commit is contained in:
+1
-1
@@ -110,7 +110,7 @@ but this is nededed for VS compiler which doesn't have inline keyword but has __
|
||||
#define PARAMCNT 64
|
||||
#define DEVICES_MAX 128
|
||||
#define EXEC_CACHE 128
|
||||
#define SPEED_CACHE 128
|
||||
#define SPEED_CACHE 4096
|
||||
#define SPEED_MAXAGE 4096
|
||||
#define EXPECTED_ITERATIONS 10000
|
||||
|
||||
|
||||
Reference in New Issue
Block a user