api: report throughput when default

Conflicts: JHA/jackpotcoin.cu api/index.php cuda_nist5.cu fuguecoin.cpp groestlcoin.cpp heavy/heavy.cu lyra2/lyra2RE.cu myriadgroestl.cpp pentablake.cu quark/animecoin.cu quark/quarkcoin.cu qubit/deep.cu qubit/qubit.cu x11/fresh.cu x11/s3.cu x11/x11.cu x13/x13.cu x15/whirlpool.cu x15/x14.cu x15/x15.cu x17/x17.cu
KlausT · Jan 22, 2015 · c92ef71 · c92ef71
1 parent ef022e7
commit c92ef71
Show file tree

Hide file tree

Showing 26 changed files with 75 additions and 39 deletions.
diff --git a/Algo256/blake256.cu b/Algo256/blake256.cu
@@ -390,6 +390,7 @@ extern "C" int scanhash_blake256(int thr_id, uint32_t *pdata, const uint32_t *pt
 #endif
 	int intensity = (device_sm[device_map[thr_id]] > 500) ? 22 : 20;
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << intensity);
+	apiReportThroughput(thr_id, (uint32_t) throughput);
 	throughput = min(throughput, max_nonce - first_nonce);
 
 	int rc = 0;

diff --git a/Algo256/keccak256.cu b/Algo256/keccak256.cu
@@ -43,6 +43,7 @@ extern "C" int scanhash_keccak256(int thr_id, uint32_t *pdata,
 {
 	const uint32_t first_nonce = pdata[19];
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << 21); // 256*256*8*4
+	apiReportThroughput(thr_id, throughput);
 	throughput = min(throughput, (max_nonce - first_nonce));
 
 	if (opt_benchmark)

diff --git a/JHA/jackpotcoin.cu b/JHA/jackpotcoin.cu
@@ -95,7 +95,8 @@ extern "C" int scanhash_jackpot(int thr_id, uint32_t *pdata,
 		((uint32_t*)ptarget)[7] = 0x000f;
 
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << 20); // 256*4096
-	throughput = min(throughput, max_nonce - first_nonce);
+	apiReportThroughput(thr_id, throughput);
+	throughput = min(throughput, (int)(max_nonce - first_nonce));
 
 	if (!init[thr_id])
 	{

diff --git a/api.cpp b/api.cpp
@@ -126,17 +126,6 @@ static void gpustatus(int thr_id)
 #endif
 		cuda_gpu_clocks(cgpu);
 
-		// todo: can be 0 if set by algo (auto)
-		if (opt_intensity == 0 && opt_work_size) {
-			int i = 0;
-			uint32_t ws = opt_work_size;
-			while (ws > 1 && i++ < 32)
-				ws = ws >> 1;
-			cgpu->intensity = i;
-		} else {
-			cgpu->intensity = opt_intensity;
-		}
-
 		// todo: per gpu
 		cgpu->accepted = accepted_count;
 		cgpu->rejected = rejected_count;
@@ -146,10 +135,10 @@ static void gpustatus(int thr_id)
 		card = device_name[gpuid];
 
 		snprintf(buf, sizeof(buf), "GPU=%d;BUS=%hd;CARD=%s;"
-			"TEMP=%.1f;FAN=%hu;RPM=%hu;FREQ=%d;KHS=%.2f;HWF=%d;I=%d|THR=%u",
+			"TEMP=%.1f;FAN=%hu;RPM=%hu;FREQ=%d;KHS=%.2f;HWF=%d;I=%.2f;THR=%u|",
 			gpuid, cgpu->gpu_bus, card, cgpu->gpu_temp, cgpu->gpu_fan,
 			cgpu->gpu_fan_rpm, cgpu->gpu_clock, cgpu->khashes,
-			cgpu->hw_errors, cgpu->intensity, opt_work_size);
+			cgpu->hw_errors, cgpu->intensity, cgpu->throughput);
 
 		// append to buffer for multi gpus
 		strcat(buffer, buf);
@@ -885,3 +874,23 @@ void *api_thread(void *userdata)
 
 	return NULL;
 }
+
+/* to be able to report the default value set in each algo */
+void apiReportThroughput(int thr_id, uint32_t throughput)
+{
+	struct cgpu_info *cgpu = &thr_info[thr_id].gpu;
+	if (cgpu) {
+		cgpu->throughput = throughput;
+		if (opt_intensity == 0) {
+			uint8_t i = 0;
+			uint32_t ws = throughput;
+			while (ws > 1 && i++ < 32)
+				ws = ws >> 1;
+			cgpu->intensity_int = i;
+		} else {
+			cgpu->intensity_int = (uint8_t) opt_intensity;
+		}
+		// dec. part to finish...
+		cgpu->intensity = (float) cgpu->intensity_int;
+	}
+}
diff --git a/api/index.php b/api/index.php
@@ -49,6 +49,7 @@ function translateField($key)
 	$intl['DIFF'] = 'Difficulty';
 	$intl['UPTIME'] = 'Miner up time';
 	$intl['TS'] = 'Last update';
+	$intl['THR'] = 'Throughput';
 
 	$intl['H'] = 'Bloc height';
 	$intl['I'] = 'Intensity';

diff --git a/cuda_nist5.cu b/cuda_nist5.cu
@@ -74,6 +74,7 @@ extern "C" int scanhash_nist5(int thr_id, uint32_t *pdata,
 		((uint32_t*)ptarget)[7] = 0x00FF;
 
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << 20); // 256*4096
+	apiReportThroughput(thr_id, (uint32_t) throughput);
 	throughput = min(throughput, (int) (max_nonce - first_nonce));
 
 	if (!init[thr_id])

diff --git a/fuguecoin.cpp b/fuguecoin.cpp
@@ -27,16 +27,17 @@ extern "C" int scanhash_fugue256(int thr_id, uint32_t *pdata, const uint32_t *pt
 {
 	uint32_t start_nonce = pdata[19]++;
 	int intensity = (device_sm[device_map[thr_id]] > 500) ? 22 : 19;
-	uint32_t throughPut = opt_work_size ? opt_work_size : (1 << intensity);
-	throughPut = min(throughPut, max_nonce - start_nonce);
+	uint32_t throughput = opt_work_size ? opt_work_size : (1 << intensity);
+	apiReportThroughput(thr_id, throughput);
+	throughput = min(throughput, max_nonce - start_nonce);
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0xf;
 
 	// init
 	if(!init[thr_id])
 	{
-		fugue256_cpu_init(thr_id, throughPut);
+		fugue256_cpu_init(thr_id, throughput);
 		init[thr_id] = true;
 	}
 
@@ -51,7 +52,7 @@ extern "C" int scanhash_fugue256(int thr_id, uint32_t *pdata, const uint32_t *pt
 	do {
 		// GPU
 		uint32_t foundNounce = 0xFFFFFFFF;
-		fugue256_cpu_hash(thr_id, throughPut, pdata[19], NULL, &foundNounce);
+		fugue256_cpu_hash(thr_id, throughput, pdata[19], NULL, &foundNounce);
 
 		if(foundNounce < 0xffffffff)
 		{

diff --git a/groestlcoin.cpp b/groestlcoin.cpp
@@ -64,18 +64,19 @@ extern "C" int scanhash_groestlcoin(int thr_id, uint32_t *pdata, const uint32_t
     uint32_t max_nonce, unsigned long *hashes_done)
 {    
     uint32_t start_nonce = pdata[19]++;
-    uint32_t throughPut = opt_work_size ? opt_work_size : (1 << 19); // 256*2048
-    throughPut = min(throughPut, max_nonce - start_nonce);
+    uint32_t throughput = opt_work_size ? opt_work_size : (1 << 19); // 256*2048
+    apiReportThroughput(thr_id, throughput);
+    throughput = min(throughput, max_nonce - start_nonce);
 
-    uint32_t *outputHash = (uint32_t*)malloc(throughPut * 16 * sizeof(uint32_t));
+    uint32_t *outputHash = (uint32_t*)malloc(throughput * 16 * sizeof(uint32_t));
 
     if (opt_benchmark)
         ((uint32_t*)ptarget)[7] = 0x000000ff;
 
     // init
     if(!init[thr_id])
     {
-        groestlcoin_cpu_init(thr_id, throughPut);
+        groestlcoin_cpu_init(thr_id, throughput);
         init[thr_id] = true;
     }
 
@@ -92,7 +93,7 @@ extern "C" int scanhash_groestlcoin(int thr_id, uint32_t *pdata, const uint32_t
         uint32_t foundNounce = 0xFFFFFFFF;
         const uint32_t Htarg = ptarget[7];
 
-        groestlcoin_cpu_hash(thr_id, throughPut, pdata[19], outputHash, &foundNounce);
+        groestlcoin_cpu_hash(thr_id, throughput, pdata[19], outputHash, &foundNounce);
 
         if(foundNounce < 0xffffffff)
         {

diff --git a/heavy/heavy.cu b/heavy/heavy.cu
@@ -137,7 +137,8 @@ int scanhash_heavy(int thr_id, uint32_t *pdata,
     const uint32_t first_nonce = pdata[19];
     // CUDA will process thousands of threads.
     uint32_t throughput = opt_work_size ? opt_work_size : (1 << 19); // 128*4096
-    throughput = min(throughput, max_nonce - first_nonce);
+    apiReportThroughput(thr_id, (uint32_t) throughput);
+    throughput = min(throughput, (int)(max_nonce - first_nonce));
 
     int rc = 0;
     uint32_t *hash = NULL;

diff --git a/lyra2/lyra2RE.cu b/lyra2/lyra2RE.cu
@@ -65,9 +65,10 @@ extern "C" int scanhash_lyra2(int thr_id, uint32_t *pdata,
 	const uint32_t first_nonce = pdata[19];
 	int intensity = (device_sm[device_map[thr_id]] > 500) ? 256 * 256 * 25 : 256 * 256 * 14;
 	if (device_sm[device_map[thr_id]] < 320) intensity = 256 * 256 * 6;
-
+	int throughput = opt_work_size ? opt_work_size : (1 << intensity); // 18=256*256*4;
 	uint32_t throughput = opt_work_size ? opt_work_size : intensity;
 	throughput = min(throughput, max_nonce - first_nonce);
+	apiReportThroughput(thr_id, (uint32_t) throughput);
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0x000f;

diff --git a/miner.h b/miner.h
@@ -378,6 +378,7 @@ extern int scanhash_x17(int thr_id, uint32_t *pdata,
 
 /* api related */
 void *api_thread(void *userdata);
+void apiReportThroughput(int thr_id, uint32_t throughput);
 
 struct cgpu_info {
 	uint8_t gpu_id;
@@ -386,7 +387,7 @@ struct cgpu_info {
 	int rejected;
 	int hw_errors;
 	double khashes;
-	uint8_t intensity;
+	uint8_t intensity_int;
 	uint8_t has_monitoring;
 	float gpu_temp;
 	uint16_t gpu_fan;
@@ -407,6 +408,8 @@ struct cgpu_info {
 
 	char gpu_sn[64];
 	char gpu_desc[64];
+	float intensity;
+	uint32_t throughput;
 };
 
 struct thr_api {

diff --git a/myriadgroestl.cpp b/myriadgroestl.cpp
@@ -44,8 +44,9 @@ extern "C" int scanhash_myriad(int thr_id, uint32_t *pdata, const uint32_t *ptar
 
 	uint32_t start_nonce = pdata[19]++;
 
-	uint32_t throughPut = opt_work_size ? opt_work_size : (1 << 17);
-	throughPut = min(throughPut, max_nonce - start_nonce);
+	uint32_t throughput = opt_work_size ? opt_work_size : (1 << 17);
+	apiReportThroughput(thr_id, throughput);
+	throughput = min(throughput, max_nonce - start_nonce);
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0x0000ff;
@@ -55,7 +56,7 @@ extern "C" int scanhash_myriad(int thr_id, uint32_t *pdata, const uint32_t *ptar
 	{
 #if BIG_DEBUG
 #else
-		myriadgroestl_cpu_init(thr_id, throughPut);
+		myriadgroestl_cpu_init(thr_id, throughput);
 #endif
 		cudaMallocHost(&(h_found[thr_id]), 4 * sizeof(uint32_t));
 		init[thr_id] = true;

diff --git a/pentablake.cu b/pentablake.cu
@@ -462,7 +462,8 @@ extern "C" int scanhash_pentablake(int thr_id, uint32_t *pdata, const uint32_t *
 	uint32_t endiandata[20];
 	int rc = 0;
 	uint32_t throughput = opt_work_size ? opt_work_size : (128 * 2560); // 18.5
-	throughput = min(throughput, max_nonce - first_nonce);
+	apiReportThroughput(thr_id, (uint32_t) throughput);
+	throughput = min(throughput, (int)(max_nonce - first_nonce));
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0x000F;

diff --git a/quark/animecoin.cu b/quark/animecoin.cu
@@ -165,7 +165,8 @@ extern "C" int scanhash_anime(int thr_id, uint32_t *pdata,
 {
 	const uint32_t first_nonce = pdata[19];
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << 19); // 256*2048
-	throughput = min(throughput, max_nonce - first_nonce);
+	apiReportThroughput(thr_id, (uint32_t) throughput);
+	throughput = min(throughput, (int)(max_nonce - first_nonce));
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0x00000f;

diff --git a/quark/quarkcoin.cu b/quark/quarkcoin.cu
@@ -142,6 +142,7 @@ extern "C" int scanhash_quark(int thr_id, uint32_t *pdata,
 	int intensity = 128 * 256 * 30;
 	if (device_sm[device_map[thr_id]] == 520)  intensity = 256 * 256 * 20;
 	uint32_t throughput = opt_work_size ? opt_work_size : intensity; // 256*4096
+	apiReportThroughput(thr_id, (uint32_t) throughput);
 	throughput = min(throughput, max_nonce - first_nonce);
 
 	if (opt_benchmark)

diff --git a/qubit/deep.cu b/qubit/deep.cu
@@ -60,7 +60,8 @@ extern "C" int scanhash_deep(int thr_id, uint32_t *pdata,
 	const uint32_t first_nonce = pdata[19];
 	uint32_t endiandata[20];
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << 19); // 256*256*8
-	throughput = min(throughput, max_nonce - first_nonce);
+	apiReportThroughput(thr_id, (uint32_t) throughput);
+	throughput = min(throughput, (int)(max_nonce - first_nonce));
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0x0000f;

diff --git a/qubit/doom.cu b/qubit/doom.cu
@@ -41,6 +41,7 @@ extern "C" int scanhash_doom(int thr_id, uint32_t *pdata,
 	const uint32_t first_nonce = pdata[19];
 	uint32_t endiandata[20];
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << 22); // 256*256*8*8
+	apiReportThroughput(thr_id, (uint32_t) throughput);
 	throughput = min(throughput, (max_nonce - first_nonce));
 
 	if (opt_benchmark)

diff --git a/qubit/qubit.cu b/qubit/qubit.cu
@@ -79,7 +79,8 @@ extern "C" int scanhash_qubit(int thr_id, uint32_t *pdata,
 	uint32_t endiandata[20];
 	const uint32_t first_nonce = pdata[19];
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << 19); // 256*256*8
-	throughput = min(throughput, max_nonce - first_nonce);
+	apiReportThroughput(thr_id, (uint32_t) throughput);
+	throughput = min(throughput, (int)(max_nonce - first_nonce));
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0x0000ff;

diff --git a/x11/fresh.cu b/x11/fresh.cu
@@ -77,6 +77,7 @@ extern "C" int scanhash_fresh(int thr_id, uint32_t *pdata,
 	uint32_t endiandata[20];
 
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << 19); // 256*256*8;
+	apiReportThroughput(thr_id, (uint32_t) throughput);
 	throughput = min(throughput, (int) (max_nonce - first_nonce));
 
 	if (opt_benchmark)

diff --git a/x11/s3.cu b/x11/s3.cu
@@ -62,7 +62,8 @@ extern "C" int scanhash_s3(int thr_id, uint32_t *pdata,
 	intensity--;
 #endif
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << intensity);
-	throughput = min(throughput, max_nonce - first_nonce);
+	apiReportThroughput(thr_id, (uint32_t) throughput);
+	throughput = min(throughput, (int)(max_nonce - first_nonce));
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0xF;

diff --git a/x11/x11.cu b/x11/x11.cu
@@ -152,6 +152,7 @@ extern "C" int scanhash_x11(int thr_id, uint32_t *pdata,
 	uint32_t throughput = opt_work_size ? opt_work_size : intensity; // 20=256*256*16;
 
 	throughput = min(throughput, max_nonce - first_nonce);
+	apiReportThroughput(thr_id, (uint32_t) throughput);
 
 
 	if (opt_benchmark)

diff --git a/x13/x13.cu b/x13/x13.cu
@@ -158,6 +158,8 @@ extern "C" int scanhash_x13(int thr_id, uint32_t *pdata,
 	uint32_t endiandata[20];
 	int intensity = (device_sm[device_map[thr_id]] > 500) ? 256 * 256 * 20 : 256 * 256 * 10;
 	uint32_t throughput = opt_work_size ? opt_work_size : intensity; // 20=256*256*16;
+	apiReportThroughput(thr_id, (uint32_t) throughput);
+	throughput = min(throughput, (int)(max_nonce - first_nonce));
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0xf;

diff --git a/x15/whirlpool.cu b/x15/whirlpool.cu
@@ -58,7 +58,8 @@ extern "C" int scanhash_whc(int thr_id, uint32_t *pdata,
 	const uint32_t first_nonce = pdata[19];
 	uint32_t endiandata[20];
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << 19); // 256*256*8;
-	throughput = min(throughput, max_nonce - first_nonce);
+	apiReportThroughput(thr_id, (uint32_t) throughput);
+	throughput = min(throughput, (int)(max_nonce - first_nonce));
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0x0000ff;

diff --git a/x15/x14.cu b/x15/x14.cu
@@ -158,9 +158,8 @@ extern "C" int scanhash_x14(int thr_id, uint32_t *pdata,
 	int intensity = 256 * 256 * 9;
 	if (device_sm[device_map[thr_id]] == 520)  intensity = 256 * 256 * 15;
 	uint32_t throughput = opt_work_size ? opt_work_size : intensity; // 256*256*8;
-
-
-	 throughput = min(throughput, max_nonce - first_nonce);
+	apiReportThroughput(thr_id, (uint32_t) throughput);
+	throughput = min(throughput, max_nonce - first_nonce);
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0x000f;

diff --git a/x15/x15.cu b/x15/x15.cu
@@ -171,6 +171,8 @@ extern "C" int scanhash_x15(int thr_id, uint32_t *pdata,
 	int intensity = 256 * 256 * 9;
 	if (device_sm[device_map[thr_id]] == 520)  intensity = 256 * 256 * 15;
 	uint32_t throughput = opt_work_size ? opt_work_size : intensity; // 256*256*8;
+	apiReportThroughput(thr_id, (uint32_t) throughput);
+	throughput = min(throughput, (int)(max_nonce - first_nonce));
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0x00F;

diff --git a/x17/x17.cu b/x17/x17.cu
@@ -186,7 +186,8 @@ extern "C" int scanhash_x17(int thr_id, uint32_t *pdata,
 {
 	const uint32_t first_nonce = pdata[19];
 	uint32_t throughput = opt_work_size ? opt_work_size : (1 << 19); // 256*256*8;
-	throughput = min(throughput, max_nonce - first_nonce);
+	apiReportThroughput(thr_id, (uint32_t) throughput);
+	throughput = min(throughput, (int)(max_nonce - first_nonce));
 
 	if (opt_benchmark)
 		((uint32_t*)ptarget)[7] = 0x00ff;