Skip to content

Commit

Permalink
Added x13mod. Bugfixes.
Browse files Browse the repository at this point in the history
  • Loading branch information
@lasybear
lasybear committed May 28, 2014
1 parent 16fa21c commit b86c26f
Show file tree
Hide file tree
Showing 7 changed files with 1,340 additions and 48 deletions.
1 change: 1 addition & 0 deletions configure.ac
View file
Original file line number Diff line number Diff line change
Expand Up @@ -357,6 +357,7 @@ AC_DEFINE_UNQUOTED([SIFCOIN_KERNNAME], ["sifcoin"], [Filename for Sifcoin optimi
AC_DEFINE_UNQUOTED([TWECOIN_KERNNAME], ["twecoin"], [Filename for Twecoin optimised kernel])
AC_DEFINE_UNQUOTED([MARUCOIN_KERNNAME], ["marucoin"], [Filename for MaruCoin optimised kernel])
AC_DEFINE_UNQUOTED([X11MOD_KERNNAME], ["x11mod"], [Filename for X11mod optimised kernel])
AC_DEFINE_UNQUOTED([X13MOD_KERNNAME], ["x13mod"], [Filename for X13mod optimised kernel])

AC_SUBST(OPENCL_LIBS)
AC_SUBST(OPENCL_FLAGS)
Expand Down
226 changes: 214 additions & 12 deletions driver-opencl.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -251,6 +251,8 @@ static enum cl_kernels select_kernel(char *arg)
return KL_TWECOIN;
if (!strcmp(arg, MARUCOIN_KERNNAME))
return KL_MARUCOIN;
if (!strcmp(arg, X13MOD_KERNNAME))
return KL_X13MOD;

return KL_NONE;
}
Expand Down Expand Up @@ -774,6 +776,8 @@ void pause_dynamic_threads(int gpu)
}
}

static _clState *clStates[MAX_GPUDEVICES];

#if defined(HAVE_CURSES)
void manage_gpu(void)
{
Expand Down Expand Up @@ -931,6 +935,7 @@ void manage_gpu(void)
gpus[selected].deven = DEV_DISABLED;
goto retry;
} else if (!strncasecmp(&input, "i", 1)) {
struct cgpu_info *cgpu;
int intensity;
char *intvar;

Expand Down Expand Up @@ -961,14 +966,46 @@ void manage_gpu(void)
wlogprint("Invalid selection\n");
goto retry;
}

bool tdymanic = gpus[selected].dynamic;
int tintensity = gpus[selected].intensity;
int txintensity = gpus[selected].xintensity;
int trawintensity = gpus[selected].rawintensity;

gpus[selected].dynamic = false;
gpus[selected].intensity = intensity;
gpus[selected].xintensity = 0; // Disable xintensity when enabling intensity
gpus[selected].rawintensity = 0; // Disable raw intensity when enabling intensity
wlogprint("Intensity on gpu %d set to %d\n", selected, intensity);
pause_dynamic_threads(selected);

if ((gpus[selected].kernel == KL_X11MOD) || (gpus[selected].kernel == KL_X13MOD)) {
for (i = 0; i < mining_threads; ++i) {
thr = get_thread(i);
cgpu = thr->cgpu;
if (cgpu->drv->drv_id != DRIVER_opencl)
continue;
if (dev_from_id(i) != selected)
continue;
if (allocateHashBuffer(selected, clStates[thr->id])) {
wlogprint("Intensity on gpu %d set to %d\n", selected, intensity);
applog(LOG_DEBUG, "Pushing sem post to thread %d", thr->id);
cgsem_post(&thr->sem);
pause_dynamic_threads(selected);
}
else {
gpus[selected].dynamic = tdymanic;
gpus[selected].intensity = tintensity;
gpus[selected].xintensity = txintensity;
gpus[selected].rawintensity = trawintensity;
}
}
}
else {
wlogprint("Intensity on gpu %d set to %d\n", selected, intensity);
pause_dynamic_threads(selected);
}
goto retry;
} else if (!strncasecmp(&input, "x", 1)) {
struct cgpu_info *cgpu;
int xintensity;
char *intvar;

Expand All @@ -990,14 +1027,46 @@ void manage_gpu(void)
wlogprint("Invalid selection\n");
goto retry;
}

bool tdymanic = gpus[selected].dynamic;
int tintensity = gpus[selected].intensity;
int txintensity = gpus[selected].xintensity;
int trawintensity = gpus[selected].rawintensity;

gpus[selected].dynamic = false;
gpus[selected].intensity = 0; // Disable intensity when enabling xintensity
gpus[selected].rawintensity = 0; // Disable raw intensity when enabling xintensity
gpus[selected].intensity = 0; // Disable intensity when enabling intensity
gpus[selected].xintensity = xintensity;
wlogprint("Experimental intensity on gpu %d set to %d\n", selected, xintensity);
pause_dynamic_threads(selected);
gpus[selected].rawintensity = 0; // Disable raw intensity when enabling intensity

if ((gpus[selected].kernel == KL_X11MOD) || (gpus[selected].kernel == KL_X13MOD)) {
for (i = 0; i < mining_threads; ++i) {
thr = get_thread(i);
cgpu = thr->cgpu;
if (cgpu->drv->drv_id != DRIVER_opencl)
continue;
if (dev_from_id(i) != selected)
continue;
if (allocateHashBuffer(selected, clStates[thr->id])) {
wlogprint("Experimental intensity on gpu %d set to %d\n", selected, xintensity);
applog(LOG_DEBUG, "Pushing sem post to thread %d", thr->id);
cgsem_post(&thr->sem);
pause_dynamic_threads(selected);
}
else {
gpus[selected].dynamic = tdymanic;
gpus[selected].intensity = tintensity;
gpus[selected].xintensity = txintensity;
gpus[selected].rawintensity = trawintensity;
}
}
}
else {
wlogprint("Experimental intensity on gpu %d set to %d\n", selected, xintensity);
pause_dynamic_threads(selected);
}
goto retry;
} else if (!strncasecmp(&input, "a", 1)) {
struct cgpu_info *cgpu;
int rawintensity;
char *intvar;

Expand All @@ -1019,12 +1088,43 @@ void manage_gpu(void)
wlogprint("Invalid selection\n");
goto retry;
}

bool tdymanic = gpus[selected].dynamic;
int tintensity = gpus[selected].intensity;
int txintensity = gpus[selected].xintensity;
int trawintensity = gpus[selected].rawintensity;

gpus[selected].dynamic = false;
gpus[selected].intensity = 0; // Disable intensity when enabling raw intensity
gpus[selected].xintensity = 0; // Disable xintensity when enabling raw intensity
gpus[selected].rawintensity = rawintensity;
wlogprint("Raw intensity on gpu %d set to %d\n", selected, rawintensity);
pause_dynamic_threads(selected);
gpus[selected].intensity = 0; // Disable intensity when enabling intensity
gpus[selected].xintensity = 0; // Disable xintensity when enabling intensity
gpus[selected].rawintensity = rawintensity;

if ((gpus[selected].kernel == KL_X11MOD) || (gpus[selected].kernel == KL_X13MOD)) {
for (i = 0; i < mining_threads; ++i) {
thr = get_thread(i);
cgpu = thr->cgpu;
if (cgpu->drv->drv_id != DRIVER_opencl)
continue;
if (dev_from_id(i) != selected)
continue;
if (allocateHashBuffer(selected, clStates[thr->id])) {
wlogprint("Raw ntensity on gpu %d set to %d\n", selected, rawintensity);
applog(LOG_DEBUG, "Pushing sem post to thread %d", thr->id);
cgsem_post(&thr->sem);
pause_dynamic_threads(selected);
}
else {
gpus[selected].dynamic = tdymanic;
gpus[selected].intensity = tintensity;
gpus[selected].xintensity = txintensity;
gpus[selected].rawintensity = trawintensity;
}
}
}
else {
wlogprint("Raw ntensity on gpu %d set to %d\n", selected, rawintensity);
pause_dynamic_threads(selected);
}
goto retry;
} else if (!strncasecmp(&input, "r", 1)) {
if (selected)
Expand Down Expand Up @@ -1057,7 +1157,6 @@ void manage_gpu(void)
}
#endif

static _clState *clStates[MAX_GPUDEVICES];

#define CL_SET_BLKARG(blkvar) status |= clSetKernelArg(*kernel, num++, sizeof(uint), (void *)&blk->blkvar)
#define CL_SET_ARG(var) status |= clSetKernelArg(*kernel, num++, sizeof(var), (void *)&var)
Expand Down Expand Up @@ -1148,6 +1247,53 @@ static cl_int queue_x11mod_kernel(_clState *clState, dev_blk_ctx *blk, __maybe_u
return status;
}

static cl_int queue_x13mod_kernel(_clState *clState, dev_blk_ctx *blk, __maybe_unused cl_uint threads)
{
unsigned char *midstate = blk->work->midstate;
cl_kernel *kernel;
unsigned int num = 0;
cl_ulong le_target;
cl_int status = 0;

le_target = *(cl_ulong *)(blk->work->device_target + 24);
flip80(clState->cldata, blk->work->data);
status = clEnqueueWriteBuffer(clState->commandQueue, clState->CLbuffer0, true, 0, 80, clState->cldata, 0, NULL,NULL);

//clbuffer, hashes
kernel = &clState->kernel_blake;
CL_SET_ARG_N(0,clState->CLbuffer0);
CL_SET_ARG_N(1,clState->hash_buffer);
kernel = &clState->kernel_bmw;
CL_SET_ARG_N(0,clState->hash_buffer);
kernel = &clState->kernel_groestl;
CL_SET_ARG_N(0,clState->hash_buffer);
kernel = &clState->kernel_skein;
CL_SET_ARG_N(0,clState->hash_buffer);
kernel = &clState->kernel_jh;
CL_SET_ARG_N(0,clState->hash_buffer);
kernel = &clState->kernel_keccak;
CL_SET_ARG_N(0,clState->hash_buffer);
kernel = &clState->kernel_luffa;
CL_SET_ARG_N(0,clState->hash_buffer);
kernel = &clState->kernel_cubehash;
CL_SET_ARG_N(0,clState->hash_buffer);
kernel = &clState->kernel_shavite;
CL_SET_ARG_N(0,clState->hash_buffer);
kernel = &clState->kernel_simd;
CL_SET_ARG_N(0,clState->hash_buffer);
kernel = &clState->kernel_echo;
CL_SET_ARG_N(0,clState->hash_buffer);
kernel = &clState->kernel_hamsi;
CL_SET_ARG_N(0,clState->hash_buffer);
//hashes, output, target
kernel = &clState->kernel_fugue;
CL_SET_ARG_N(0,clState->hash_buffer);
CL_SET_ARG_N(1,clState->outputBuffer);
CL_SET_ARG_N(2,le_target);

return status;
}


static void set_threads_hashes(unsigned int vectors, unsigned int compute_shaders, int64_t *hashes, size_t *globalThreads,
unsigned int minthreads, __maybe_unused int *intensity, __maybe_unused int *xintensity, __maybe_unused int *rawintensity)
Expand Down Expand Up @@ -1471,6 +1617,9 @@ static bool opencl_thread_prepare(struct thr_info *thr)
case KL_X11MOD:
cgpu->kname = X11MOD_KERNNAME;
break;
case KL_X13MOD:
cgpu->kname = X13MOD_KERNNAME;
break;
default:
break;
}
Expand Down Expand Up @@ -1502,6 +1651,9 @@ static bool opencl_thread_init(struct thr_info *thr)
case KL_X11MOD:
thrdata->queue_kernel_parameters = &queue_x11mod_kernel;
break;
case KL_X13MOD:
thrdata->queue_kernel_parameters = &queue_x13mod_kernel;
break;
case KL_ALEXKARNEW:
case KL_ALEXKAROLD:
case KL_CKOLIVAS:
Expand Down Expand Up @@ -1643,6 +1795,41 @@ static int64_t opencl_scanhash(struct thr_info *thr, struct work *work,
CL_ENQUEUE_KERNEL(echo, NULL);
}
}
else if (clState->chosen_kernel == KL_X13MOD) {
if (clState->goffset) {
size_t global_work_offset[1];
global_work_offset[0] = work->blk.nonce;

CL_ENQUEUE_KERNEL(blake, global_work_offset);
CL_ENQUEUE_KERNEL(bmw, global_work_offset);
CL_ENQUEUE_KERNEL(groestl, global_work_offset);
CL_ENQUEUE_KERNEL(skein, global_work_offset);
CL_ENQUEUE_KERNEL(jh, global_work_offset);
CL_ENQUEUE_KERNEL(keccak, global_work_offset);
CL_ENQUEUE_KERNEL(luffa, global_work_offset);
CL_ENQUEUE_KERNEL(cubehash, global_work_offset);
CL_ENQUEUE_KERNEL(shavite, global_work_offset);
CL_ENQUEUE_KERNEL(simd, global_work_offset)
CL_ENQUEUE_KERNEL(echo, global_work_offset);
CL_ENQUEUE_KERNEL(hamsi, global_work_offset);
CL_ENQUEUE_KERNEL(fugue, global_work_offset);
}
else {
CL_ENQUEUE_KERNEL(blake, NULL);
CL_ENQUEUE_KERNEL(bmw, NULL);
CL_ENQUEUE_KERNEL(groestl, NULL);
CL_ENQUEUE_KERNEL(skein, NULL);
CL_ENQUEUE_KERNEL(jh, NULL);
CL_ENQUEUE_KERNEL(keccak, NULL);
CL_ENQUEUE_KERNEL(luffa, NULL);
CL_ENQUEUE_KERNEL(cubehash, NULL);
CL_ENQUEUE_KERNEL(shavite, NULL);
CL_ENQUEUE_KERNEL(simd, NULL)
CL_ENQUEUE_KERNEL(echo, NULL);
CL_ENQUEUE_KERNEL(hamsi, NULL);
CL_ENQUEUE_KERNEL(fugue, NULL);
}
}
else {
if (clState->goffset) {
size_t global_work_offset[1];
Expand Down Expand Up @@ -1711,6 +1898,21 @@ static void opencl_thread_shutdown(struct thr_info *thr)
clReleaseKernel(clState->kernel_simd);
clReleaseKernel(clState->kernel_echo);
}
else if (clState->chosen_kernel == KL_X13MOD) {
clReleaseKernel(clState->kernel_blake);
clReleaseKernel(clState->kernel_bmw);
clReleaseKernel(clState->kernel_groestl);
clReleaseKernel(clState->kernel_skein);
clReleaseKernel(clState->kernel_jh);
clReleaseKernel(clState->kernel_keccak);
clReleaseKernel(clState->kernel_luffa);
clReleaseKernel(clState->kernel_cubehash);
clReleaseKernel(clState->kernel_shavite);
clReleaseKernel(clState->kernel_simd);
clReleaseKernel(clState->kernel_echo);
clReleaseKernel(clState->kernel_hamsi);
clReleaseKernel(clState->kernel_fugue);
}
else {
clReleaseKernel(clState->kernel);
}
Expand Down
Loading

0 comments on commit b86c26f

Please sign in to comment.