Merge pull request #2 from SChernykh/master

Added auto-config for RandomX

src/cuda_extra.cu

@@ -571,6 +571,20 @@ int cuda_get_deviceinfo(nvid_ctx *ctx)
     ctx->device_pciDeviceID     = props.pciDeviceID;
     ctx->device_pciDomainID     = props.pciDomainID;
 
+    if ((ctx->algorithm.family() == Algorithm::RANDOM_X) && ((ctx->device_blocks < 0) || (ctx->device_threads < 0))) {
+        ctx->device_threads = 32;
+        ctx->device_blocks = props.multiProcessorCount * 2;
+        ctx->device_bfactor = 6;
+        ctx->device_bsleep = 0;
+
+        // Leave memory for 2080 MB dataset + 64 MB free
+        // Each thread uses 1 scratchpad plus a few small buffers on GPU
+        const size_t max_blocks = (freeMemory - (2080u << 20) - (64u << 20)) / (ctx->algorithm.l3() + 32768) / 32;
+        if (ctx->device_blocks > max_blocks) {
+            ctx->device_blocks = max_blocks;
+        }
+    }
+
     // set all device option those marked as auto (-1) to a valid value
     if (ctx->device_blocks < 0) {
         /* good values based of my experience