bench: handle cryptonight variants + V100 fix

bench.cpp

@@ -156,6 +156,22 @@ bool bench_algo_switch_next(int thr_id)
 	if (algo == ALGO_SCRYPT) algo++;
 	if (algo == ALGO_SCRYPT_JANE) algo++;
 
+	// Set cryptonight variant
+	switch (algo) {
+		case ALGO_MONERO:
+			cryptonight_fork = 7;
+			break;
+		case ALGO_GRAFT:
+			cryptonight_fork = 8;
+			break;
+		case ALGO_STELLITE:
+			cryptonight_fork = 3;
+			break;
+		case ALGO_CRYPTONIGHT:
+			cryptonight_fork = 1;
+			break;
+	}
+
 	// free current algo memory and track mem usage
 	mused = cuda_available_memory(thr_id);
 	algo_free_all(thr_id);

ccminer.cpp

@@ -2384,6 +2384,9 @@ static void *miner_thread(void *userdata)
 		case ALGO_CRYPTOLIGHT:
 			rc = scanhash_cryptolight(thr_id, &work, max_nonce, &hashes_done, 1);
 			break;
+		case ALGO_MONERO:
+		case ALGO_STELLITE:
+		case ALGO_GRAFT:
 		case ALGO_CRYPTONIGHT:
 		{
 			int cn_variant = 0;

crypto/cryptonight.cu

@@ -50,8 +50,11 @@ extern "C" int scanhash_cryptonight(int thr_id, struct work* work, uint32_t max_
 			gpulog_init(LOG_INFO, thr_id, "%s, %d MB available, %hd SMX", device_name[dev_id],
 				mem, device_mpcount[dev_id]);
 
-		if (!device_config[thr_id] && strcmp(device_name[dev_id], "TITAN V") == 0) {
+		if (!device_config[thr_id]) {
-			device_config[thr_id] = strdup("80x24");
+			if(strcmp(device_name[dev_id], "TITAN V") == 0)
+				device_config[thr_id] = strdup("80x24");
+			if(strstr(device_name[dev_id], "V100"))
+				device_config[thr_id] = strdup("80x24");
 		}
 
 		if (device_config[thr_id]) {
@@ -83,7 +86,7 @@ extern "C" int scanhash_cryptonight(int thr_id, struct work* work, uint32_t max_
 			CUDA_LOG_ERROR();
 		}
 
-		const size_t alloc = MEMORY * throughput;
+		const size_t alloc = MEMORY * size_t(throughput);
 		cryptonight_extra_init(thr_id);
 
 		cudaMalloc(&d_long_state[thr_id], alloc);

equi/equihash.cpp

@@ -183,8 +183,7 @@ extern "C" int scanhash_equihash(int thr_id, struct work *work, uint32_t max_non
 				return -1;
 			}
 			size_t memSz = solvers[thr_id]->equi_mem_sz / (1024*1024);
-			gpus_intensity[thr_id] = (uint32_t) solvers[thr_id]->throughput;
+			api_set_throughput(thr_id, (uint32_t) solvers[thr_id]->throughput);
-			api_set_throughput(thr_id, gpus_intensity[thr_id]);
 			gpulog(LOG_DEBUG, thr_id, "Allocated %u MB of context memory", (u32) memSz);
 			cuda_get_arch(thr_id);
 			init[thr_id] = true;