static void init(struct fmt_main *self) {
char *kpc;
global_work_size = MAX_KEYS_PER_CRYPT;
opencl_init("$JOHN/md4_kernel.cl", ocl_gpu_id, platform_id);
crypt_kernel = clCreateKernel(program[ocl_gpu_id], "md4", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel. Double-check kernel name?");
if( ((kpc = getenv("LWS")) == NULL) || (atoi(kpc) == 0)) {
create_clobj(MD4_NUM_KEYS);
opencl_find_best_workgroup(self);
release_clobj();
}else {
local_work_size = atoi(kpc);
}
if( (kpc = getenv("GWS")) == NULL){
max_keys_per_crypt = MD4_NUM_KEYS;
create_clobj(MD4_NUM_KEYS);
} else {
if (atoi(kpc) == 0){
//user chose to die of boredom
max_keys_per_crypt = MD4_NUM_KEYS;
create_clobj(MD4_NUM_KEYS);
find_best_kpc();
} else {
max_keys_per_crypt = atoi(kpc);
create_clobj(max_keys_per_crypt);
}
}
fprintf(stderr, "Local work size (LWS) %d, Global work size (GWS) %d\n",(int)local_work_size, max_keys_per_crypt);
self->params.max_keys_per_crypt = max_keys_per_crypt;
}
static void init(struct fmt_main *self)
{
cl_ulong maxsize;
size_t selected_gws;
opencl_init_opt("$JOHN/kernels/pwsafe_kernel.cl", ocl_gpu_id, NULL);
init_kernel = clCreateKernel(program[ocl_gpu_id], KERNEL_INIT_NAME, &ret_code);
HANDLE_CLERROR(ret_code, "Error while creating init kernel");
crypt_kernel = clCreateKernel(program[ocl_gpu_id], KERNEL_RUN_NAME, &ret_code);
HANDLE_CLERROR(ret_code, "Error while creating crypt kernel");
finish_kernel = clCreateKernel(program[ocl_gpu_id], KERNEL_FINISH_NAME, &ret_code);
HANDLE_CLERROR(ret_code, "Error while creating finish kernel");
local_work_size = cpu(device_info[ocl_gpu_id]) ? 1 : 64;
global_work_size = 0;
opencl_get_user_preferences(CONFIG_NAME);
//Initialize openCL tuning (library) for this format.
opencl_init_auto_setup(STEP, ROUNDS_DEFAULT/8, 8, split_events,
warn, &multi_profilingEvent[3], self, create_clobj, release_clobj,
sizeof(pwsafe_pass), 0);
self->methods.crypt_all = crypt_all_benchmark;
selected_gws = global_work_size;
/* Note: we ask for the kernels' max sizes, not the device's! */
maxsize = get_current_work_group_size(ocl_gpu_id, init_kernel);
maxsize = MIN(get_current_work_group_size(ocl_gpu_id, crypt_kernel),
maxsize);
maxsize = MIN(get_current_work_group_size(ocl_gpu_id, finish_kernel),
maxsize);
while (local_work_size > maxsize)
local_work_size >>= 1;
self->params.max_keys_per_crypt = (global_work_size ? global_work_size: MAX_KEYS_PER_CRYPT);
if (!local_work_size) {
create_clobj(self->params.max_keys_per_crypt, self);
find_best_lws(self, ocl_gpu_id);
release_clobj();
}
global_work_size = selected_gws;
if (global_work_size)
create_clobj(global_work_size, self);
else
//user chose to die of boredom
find_best_gws(self, ocl_gpu_id);
self->params.min_keys_per_crypt = local_work_size;
self->params.max_keys_per_crypt = global_work_size;
self->methods.crypt_all = crypt_all;
if (options.verbosity > 2)
fprintf(stderr, "Local worksize (LWS) %d, Global worksize (GWS) %d\n", (int)local_work_size, (int)global_work_size);
}
static void fmt_ssha_init(struct fmt_main *pFmt)
{
char *kpc;
opencl_init("$JOHN/ssha_opencl_kernel.cl", gpu_id);
// create kernel to execute
crypt_kernel = clCreateKernel(program[gpu_id], "sha1_crypt_kernel", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel. Double-check kernel name?");
if( (kpc = getenv("LWS")) == NULL){
create_clobj(SSHA_NUM_KEYS);
find_best_workgroup();
release_clobj();
}else {
local_work_size = atoi(kpc);
}
if( (kpc = getenv("KPC")) == NULL){
max_keys_per_crypt = SSHA_NUM_KEYS;
create_clobj(SSHA_NUM_KEYS);
} else {
if (atoi(kpc) == 0){
//user chose to die of boredom
max_keys_per_crypt = SSHA_NUM_KEYS;
create_clobj(SSHA_NUM_KEYS);
find_best_kpc();
} else {
max_keys_per_crypt = atoi(kpc);
create_clobj(max_keys_per_crypt);
}
}
printf("Local work size (LWS) %d, Keys per crypt (KPC) %d\n",(int)local_work_size,max_keys_per_crypt);
pFmt->params.max_keys_per_crypt = max_keys_per_crypt;
}
static void fmt_ssha_init(struct fmt_main *self)
{
char *temp;
cl_ulong maxsize;
global_work_size = 0;
opencl_init("$JOHN/ssha_kernel.cl", ocl_gpu_id, platform_id);
// create kernel to execute
crypt_kernel = clCreateKernel(program[ocl_gpu_id], "sha1_crypt_kernel", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel. Double-check kernel name?");
HANDLE_CLERROR(clGetKernelWorkGroupInfo(crypt_kernel, devices[ocl_gpu_id], CL_KERNEL_WORK_GROUP_SIZE, sizeof(maxsize), &maxsize, NULL), "Query max work group size");
if ((temp = cfg_get_param(SECTION_OPTIONS, SUBSECTION_OPENCL, LWS_CONFIG)))
local_work_size = atoi(temp);
if ((temp = cfg_get_param(SECTION_OPTIONS, SUBSECTION_OPENCL, GWS_CONFIG)))
global_work_size = atoi(temp);
if ((temp = getenv("LWS")))
local_work_size = atoi(temp);
if ((temp = getenv("GWS")))
global_work_size = atoi(temp);
if (!local_work_size) {
int temp = global_work_size;
local_work_size = maxsize;
global_work_size = global_work_size ? global_work_size : 4 * maxsize;
create_clobj(global_work_size, self);
opencl_find_best_workgroup_limit(self, maxsize);
release_clobj();
global_work_size = temp;
}
if (local_work_size > maxsize) {
fprintf(stderr, "LWS %d is too large for this GPU. Max allowed is %d, using that.\n", (int)local_work_size, (int)maxsize);
local_work_size = maxsize;
}
if (!global_work_size)
find_best_gws(getenv("GWS") == NULL ? 0 : 1, self);
if (global_work_size < local_work_size)
global_work_size = local_work_size;
fprintf(stderr, "Local worksize (LWS) %d, Global worksize (GWS) %d\n", (int)local_work_size, (int)global_work_size);
create_clobj(global_work_size, self);
atexit(release_clobj);
}
static void find_best_kpc(void){
int num;
cl_event myEvent;
cl_ulong startTime, endTime, tmpTime;
int kernelExecTimeNs = 6969;
cl_int ret_code;
int optimal_kpc=2048;
int i = 0;
cl_uint *tmpbuffer;
printf("Calculating best keys per crypt, this will take a while ");
for( num=SSHA_NUM_KEYS; num > 4096 ; num -= 16384){
release_clobj();
create_clobj(num);
advance_cursor();
queue_prof = clCreateCommandQueue( context[gpu_id], devices[gpu_id], CL_QUEUE_PROFILING_ENABLE, &ret_code);
for (i=0; i < num; i++){
memcpy(&(saved_plain[i*PLAINTEXT_LENGTH]),"abacaeaf",PLAINTEXT_LENGTH);
}
clEnqueueWriteBuffer(queue_prof, mysalt, CL_TRUE, 0, SALT_SIZE, saved_salt, 0, NULL, NULL);
clEnqueueWriteBuffer(queue_prof, buffer_keys, CL_TRUE, 0, (PLAINTEXT_LENGTH) * num, saved_plain, 0, NULL, NULL);
ret_code = clEnqueueNDRangeKernel( queue_prof, crypt_kernel, 1, NULL, &global_work_size, &local_work_size, 0, NULL, &myEvent);
if(ret_code != CL_SUCCESS){
printf("Error %d\n",ret_code);
continue;
}
clFinish(queue_prof);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT, sizeof(cl_ulong), &startTime, NULL);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END , sizeof(cl_ulong), &endTime , NULL);
tmpTime = endTime-startTime;
tmpbuffer = malloc(sizeof(cl_uint) * num);
clEnqueueReadBuffer(queue_prof, buffer_out, CL_TRUE, 0, sizeof(cl_uint) * num, tmpbuffer, 0, NULL, &myEvent);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT, sizeof(cl_ulong), &startTime, NULL);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END , sizeof(cl_ulong), &endTime , NULL);
tmpTime = tmpTime + (endTime-startTime);
if( ((int)( ((float) (tmpTime) / num) * 10 )) <= kernelExecTimeNs) {
kernelExecTimeNs = ((int) (((float) (tmpTime) / num) * 10) ) ;
optimal_kpc = num;
}
free(tmpbuffer);
clReleaseCommandQueue(queue_prof);
}
printf("Optimal keys per crypt %d\n(to avoid this test on next run, put \""
KPC_CONFIG " = %d\" in john.conf, section [" SECTION_OPTIONS
SUBSECTION_OPENCL "])\n", optimal_kpc, optimal_kpc);
max_keys_per_crypt = optimal_kpc;
release_clobj();
create_clobj(optimal_kpc);
}
static void find_best_kpc(void){
int num;
cl_event myEvent;
cl_ulong startTime, endTime, tmpTime;
int kernelExecTimeNs = 6969;
cl_int ret_code;
int optimal_kpc=2048;
int i = 0;
cl_uint *tmpbuffer;
fprintf(stderr, "Calculating best keys per crypt, this will take a while ");
for( num=MD4_NUM_KEYS; num > 4096 ; num -= 4096){
release_clobj();
create_clobj(num);
advance_cursor();
queue_prof = clCreateCommandQueue( context[ocl_gpu_id], devices[ocl_gpu_id], CL_QUEUE_PROFILING_ENABLE, &ret_code);
for (i=0; i < num; i++){
memcpy(&(saved_plain[i * (PLAINTEXT_LENGTH + 1)]), "abcaaeaf", PLAINTEXT_LENGTH + 1);
saved_plain[i * (PLAINTEXT_LENGTH + 1) + 8] = 0x80;
}
clEnqueueWriteBuffer(queue_prof, data_info, CL_TRUE, 0, sizeof(unsigned int)*2, datai, 0, NULL, NULL);
clEnqueueWriteBuffer(queue_prof, buffer_keys, CL_TRUE, 0, (PLAINTEXT_LENGTH + 1) * num, saved_plain, 0, NULL, NULL);
ret_code = clEnqueueNDRangeKernel( queue_prof, crypt_kernel, 1, NULL, &global_work_size, &local_work_size, 0, NULL, &myEvent);
if(ret_code != CL_SUCCESS) {
HANDLE_CLERROR(ret_code, "Error running kernel in find_best_KPC()");
continue;
}
clFinish(queue_prof);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT, sizeof(cl_ulong), &startTime, NULL);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END , sizeof(cl_ulong), &endTime , NULL);
tmpTime = endTime-startTime;
tmpbuffer = malloc(sizeof(cl_uint) * num);
clEnqueueReadBuffer(queue_prof, buffer_out, CL_TRUE, 0, sizeof(cl_uint) * num, tmpbuffer, 0, NULL, &myEvent);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT, sizeof(cl_ulong), &startTime, NULL);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END , sizeof(cl_ulong), &endTime , NULL);
tmpTime = tmpTime + (endTime-startTime);
if( ((int)( ((float) (tmpTime) / num) * 10 )) <= kernelExecTimeNs) {
kernelExecTimeNs = ((int) (((float) (tmpTime) / num) * 10) ) ;
optimal_kpc = num;
}
free(tmpbuffer);
clReleaseCommandQueue(queue_prof);
}
fprintf(stderr, "Optimal keys per crypt %d\n(to avoid this test on next run do \"export GWS=%d\")\n",optimal_kpc,optimal_kpc);
max_keys_per_crypt = optimal_kpc;
release_clobj();
create_clobj(optimal_kpc);
}
static void find_best_kpc(void){
int num;
cl_event myEvent;
cl_ulong startTime, endTime, tmpTime;
int kernelExecTimeNs = INT_MAX;
cl_int ret_code;
int optimal_kpc=2048;
int i = 0;
cl_uint *tmpbuffer;
fprintf(stderr, "Calculating best keys per crypt, this will take a while ");
for( num=MAX_KEYS_PER_CRYPT; num >= 4096 ; num -= 4096){
release_clobj();
create_clobj(num);
advance_cursor();
queue_prof = clCreateCommandQueue( context[ocl_gpu_id], devices[ocl_gpu_id], CL_QUEUE_PROFILING_ENABLE, &ret_code);
for (i=0; i < num; i++){
strcpy(&(saved_plain[i * keybuf_size]), tests[0].plaintext);
}
clEnqueueWriteBuffer(queue_prof, buffer_keys, CL_TRUE, 0, keybuf_size * num, saved_plain, 0, NULL, NULL);
ret_code = clEnqueueNDRangeKernel( queue_prof, crypt_kernel, 1, NULL, &global_work_size, &local_work_size, 0, NULL, &myEvent);
if(ret_code != CL_SUCCESS){
fprintf(stderr, "Error %d\n",ret_code);
continue;
}
clFinish(queue_prof);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT, sizeof(cl_ulong), &startTime, NULL);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END , sizeof(cl_ulong), &endTime , NULL);
tmpTime = endTime-startTime;
tmpbuffer = mem_alloc(sizeof(cl_uint) * num);
clEnqueueReadBuffer(queue_prof, buffer_out, CL_TRUE, 0, sizeof(cl_uint) * num, tmpbuffer, 0, NULL, &myEvent);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT, sizeof(cl_ulong), &startTime, NULL);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END , sizeof(cl_ulong), &endTime , NULL);
tmpTime = tmpTime + (endTime-startTime);
if( ((int)( ((float) (tmpTime) / num) * 10 )) <= kernelExecTimeNs) {
kernelExecTimeNs = ((int) (((float) (tmpTime) / num) * 10) ) ;
optimal_kpc = num;
}
MEM_FREE(tmpbuffer);
clReleaseCommandQueue(queue_prof);
}
fprintf(stderr, "Optimal keys per crypt %d\n(to avoid this test on next run do export GWS=%d)\n",optimal_kpc,optimal_kpc);
global_work_size = optimal_kpc;
release_clobj();
create_clobj(optimal_kpc);
}
void sha1_init(size_t user_kpc)
{
kpc = user_kpc;
load_source();
createDevice();
createkernel();
create_clobj();
}
/* --
This function could be used to calculated the best num
of keys per crypt for the given format
-- */
static void find_best_gws(struct fmt_main * self, int sequential_id) {
//Call the common function.
common_find_best_gws(
sequential_id, ROUNDS_DEFAULT, 0,
(cpu(device_info[ocl_gpu_id]) ? 500000000ULL : 2400000000ULL)
);
create_clobj(global_work_size, self);
}
static void fmt_ssha_init(struct fmt_main *pFmt)
{
char *temp;
opencl_init("$JOHN/ssha_kernel.cl", gpu_id, platform_id);
// create kernel to execute
crypt_kernel = clCreateKernel(program[gpu_id], "sha1_crypt_kernel", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel. Double-check kernel name?");
if ((temp = cfg_get_param(SECTION_OPTIONS, SUBSECTION_OPENCL,
LWS_CONFIG)))
local_work_size = atoi(temp);
if ((temp = getenv("LWS")))
local_work_size = atoi(temp);
if (!local_work_size) {
create_clobj(SSHA_NUM_KEYS);
find_best_workgroup();
release_clobj();
}
if ((temp = cfg_get_param(SECTION_OPTIONS, SUBSECTION_OPENCL,
KPC_CONFIG)))
max_keys_per_crypt = atoi(temp);
else
max_keys_per_crypt = SSHA_NUM_KEYS;
if ((temp = getenv("KPC")))
max_keys_per_crypt = atoi(temp);
if (max_keys_per_crypt) {
create_clobj(max_keys_per_crypt);
} else {
//user chose to die of boredom
max_keys_per_crypt = SSHA_NUM_KEYS;
create_clobj(SSHA_NUM_KEYS);
find_best_kpc();
}
printf("Local work size (LWS) %d, Keys per crypt (KPC) %d\n",(int)local_work_size,max_keys_per_crypt);
pFmt->params.max_keys_per_crypt = max_keys_per_crypt;
}
static void find_best_gws(int do_benchmark, struct fmt_main *self)
{
cl_event myEvent;
cl_ulong startTime, endTime, tmpTime;
int kernelExecTimeNs = INT_MAX;
cl_int ret_code;
int optimal_kpc=2048;
int gws, i = 0;
cl_uint *tmpbuffer;
for(gws = local_work_size << 2; gws <= 8*1024*1024; gws <<= 1) {
create_clobj(gws, self);
advance_cursor();
queue_prof = clCreateCommandQueue( context[ocl_gpu_id], devices[ocl_gpu_id], CL_QUEUE_PROFILING_ENABLE, &ret_code);
for (i=0; i < gws; i++) {
memcpy(&(saved_plain[i*PLAINTEXT_LENGTH]),"abacaeaf",PLAINTEXT_LENGTH);
}
clEnqueueWriteBuffer(queue_prof, mysalt, CL_TRUE, 0, SALT_SIZE, saved_salt, 0, NULL, NULL);
clEnqueueWriteBuffer(queue_prof, buffer_keys, CL_TRUE, 0, (PLAINTEXT_LENGTH) * gws, saved_plain, 0, NULL, NULL);
ret_code = clEnqueueNDRangeKernel( queue_prof, crypt_kernel, 1, NULL, &global_work_size, &local_work_size, 0, NULL, &myEvent);
if(ret_code != CL_SUCCESS) {
// We hit some resource limit so we end here.
release_clobj();
break;
}
clFinish(queue_prof);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT, sizeof(cl_ulong), &startTime, NULL);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END , sizeof(cl_ulong), &endTime , NULL);
tmpTime = endTime-startTime;
tmpbuffer = malloc(sizeof(cl_uint) * gws);
clEnqueueReadBuffer(queue_prof, buffer_out, CL_TRUE, 0, sizeof(cl_uint) * gws, tmpbuffer, 0, NULL, &myEvent);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT, sizeof(cl_ulong), &startTime, NULL);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END , sizeof(cl_ulong), &endTime , NULL);
tmpTime = tmpTime + (endTime-startTime);
if (do_benchmark)
fprintf(stderr, "%10d %10llu c/s %-.2f us\n", gws, gws * 1000000000ULL / tmpTime, tmpTime / 1000.0);
if( ((int)( ((float) (tmpTime) / gws) * 10 )) <= kernelExecTimeNs) {
kernelExecTimeNs = ((int) (((float) (tmpTime) / gws) * 10) ) ;
optimal_kpc = gws;
}
MEM_FREE(tmpbuffer);
clReleaseCommandQueue(queue_prof);
release_clobj();
}
global_work_size = optimal_kpc;
}
static void init(struct fmt_main *self)
{
cl_ulong maxsize;
opencl_init("$JOHN/kernels/pbkdf2_hmac_sha512_unsplit_kernel.cl", ocl_gpu_id);
set_lws_gws(DEFAULT_LWS,DEFAULT_GWS);
crypt_kernel = clCreateKernel(program[ocl_gpu_id], KERNEL_NAME, &cl_error);
HANDLE_CLERROR(cl_error, "Error creating kernel");
create_clobj(global_work_size, self);
/* Note: we ask for the kernels' max sizes, not the device's! */
HANDLE_CLERROR(clGetKernelWorkGroupInfo(crypt_kernel,
devices[ocl_gpu_id], CL_KERNEL_WORK_GROUP_SIZE,
sizeof(maxsize), &maxsize, NULL), "Query max workgroup size");
while (local_work_size > maxsize)
local_work_size >>= 1;
self->params.min_keys_per_crypt = local_work_size;
self->params.max_keys_per_crypt = global_work_size;
}
cl_ulong gws_test(int gws)
{
cl_ulong startTime, endTime, run_time;
cl_command_queue queue_prof;
cl_event myEvent;
cl_int ret_code;
int i;
int num = VF * gws;
create_clobj(gws);
queue_prof = clCreateCommandQueue(context[ocl_gpu_id], devices[ocl_gpu_id], CL_QUEUE_PROFILING_ENABLE, &ret_code);
for (i = 0; i < num; i++)
set_key(rar_fmt.params.tests[0].plaintext, i);
HANDLE_CLERROR(clEnqueueWriteBuffer(queue_prof, cl_salt, BLOCK_IF_DEBUG, 0, 8, saved_salt, 0, NULL, NULL), "Failed transferring salt");
HANDLE_CLERROR(clEnqueueWriteBuffer(queue_prof, cl_saved_key, BLOCK_IF_DEBUG, 0, UNICODE_LENGTH * num, saved_key, 0, NULL, NULL), "Failed transferring keys");
HANDLE_CLERROR(clEnqueueWriteBuffer(queue_prof, cl_saved_len, BLOCK_IF_DEBUG, 0, sizeof(int) * num, saved_len, 0, NULL, NULL), "Failed transferring lengths");
ret_code = clEnqueueNDRangeKernel(queue_prof, crypt_kernel, 1, NULL, &global_work_size, &local_work_size, 0, NULL, &myEvent);
if (ret_code != CL_SUCCESS) {
fprintf(stderr, "Error: %s\n", get_error_name(ret_code));
clReleaseCommandQueue(queue_prof);
release_clobj();
return 0;
}
HANDLE_CLERROR(clFinish(queue_prof), "Failed running kernel");
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT, sizeof(cl_ulong), &startTime, NULL);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END, sizeof(cl_ulong), &endTime, NULL);
run_time = endTime - startTime;
HANDLE_CLERROR(clEnqueueReadBuffer(queue_prof, cl_aes_iv, BLOCK_IF_DEBUG, 0, 16 * num, aes_iv, 0, NULL, &myEvent), "Failed reading iv back");
HANDLE_CLERROR(clEnqueueReadBuffer(queue_prof, cl_aes_key, BLOCK_IF_DEBUG, 0, 16 * num, aes_key, 0, NULL, &myEvent), "Failed reading key back");
HANDLE_CLERROR(clFinish(queue_prof), "Failed reading results back");
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT, sizeof(cl_ulong), &startTime, NULL);
clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END, sizeof(cl_ulong), &endTime, NULL);
clReleaseCommandQueue(queue_prof);
release_clobj();
return (run_time + endTime - startTime);
}
/* ------- Initialization ------- */
static void init(struct fmt_main * self) {
char * tmp_value;
char * task = "$JOHN/sha256_kernel.cl";
opencl_init_dev(ocl_gpu_id, platform_id);
source_in_use = device_info[ocl_gpu_id];
if ((tmp_value = getenv("_TYPE")))
source_in_use = atoi(tmp_value);
opencl_build_kernel(task, ocl_gpu_id);
// create kernel(s) to execute
crypt_kernel = clCreateKernel(program[ocl_gpu_id], "kernel_crypt", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel. Double-check kernel name?");
cmp_kernel = clCreateKernel(program[ocl_gpu_id], "kernel_cmp", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel_cmp. Double-check kernel name?");
global_work_size = get_task_max_size();
local_work_size = 0;
if (source_in_use != device_info[ocl_gpu_id]) {
device_info[ocl_gpu_id] = source_in_use;
fprintf(stderr, "Selected runtime id %d, source (%s)\n", source_in_use, task);
}
if ((tmp_value = cfg_get_param(SECTION_OPTIONS,
SUBSECTION_OPENCL, LWS_CONFIG)))
local_work_size = atoi(tmp_value);
if ((tmp_value = getenv("LWS")))
local_work_size = atoi(tmp_value);
//Check if local_work_size is a valid number.
if (local_work_size > get_task_max_work_group_size()){
fprintf(stderr, "Error: invalid local work size (LWS). Max value allowed is: %zd\n" ,
get_task_max_work_group_size());
local_work_size = 0; //Force find a valid number.
}
self->params.max_keys_per_crypt = global_work_size;
if (!local_work_size) {
local_work_size = get_task_max_work_group_size();
create_clobj(global_work_size, self);
find_best_workgroup(self);
release_clobj();
}
if ((tmp_value = cfg_get_param(SECTION_OPTIONS,
SUBSECTION_OPENCL, GWS_CONFIG)))
global_work_size = atoi(tmp_value);
if ((tmp_value = getenv("GWS")))
global_work_size = atoi(tmp_value);
//Check if a valid multiple is used.
global_work_size = get_multiple(global_work_size, local_work_size);
if (global_work_size)
create_clobj(global_work_size, self);
else {
//user chose to die of boredom
global_work_size = get_task_max_size();
find_best_gws(self);
}
fprintf(stderr, "Local work size (LWS) %d, global work size (GWS) %zd\n",
(int) local_work_size, global_work_size);
self->params.max_keys_per_crypt = global_work_size;
}
/* --
This function could be used to calculated the best num
of keys per crypt for the given format
-- */
static void find_best_gws(struct fmt_main * self) {
size_t num = 0;
cl_ulong run_time, min_time = CL_ULONG_MAX;
int optimal_gws = local_work_size, step = STEP;
int do_benchmark = 0;
unsigned int SHAspeed, bestSHAspeed = 0;
unsigned long long int max_run_time = 1000000000ULL;
char *tmp_value;
if ((tmp_value = getenv("STEP"))){
step = atoi(tmp_value);
do_benchmark = 1;
}
step = get_multiple(step, local_work_size);
if ((tmp_value = cfg_get_param(SECTION_OPTIONS, SUBSECTION_OPENCL, DUR_CONFIG)))
max_run_time = atoi(tmp_value) * 1000000000UL;
fprintf(stderr, "Calculating best global work size (GWS) for LWS=%zd and max. %llu s duration.\n\n",
local_work_size, max_run_time / 1000000000ULL);
if (do_benchmark)
fprintf(stderr, "Raw speed figures including buffer transfers:\n");
for (num = get_step(num, step, 1); num; num = get_step(num, step, 0)) {
//Check if hardware can handle the size we are going to try now.
if (sizeof(sha256_password) * num * 1.2 > get_max_mem_alloc_size(ocl_gpu_id))
break;
if (! (run_time = gws_test(num, self)))
continue;
if (!do_benchmark)
advance_cursor();
SHAspeed = num / (run_time / 1000000000.);
if (run_time < min_time)
min_time = run_time;
if (do_benchmark) {
fprintf(stderr, "gws: %8zu\t%12lu c/s %8.3f ms per crypt_all()",
num, (long) (num / (run_time / 1000000000.)),
(float) run_time / 1000000.);
if (run_time > max_run_time) {
fprintf(stderr, " - too slow\n");
break;
}
} else {
if (run_time > min_time * 20 || run_time > max_run_time)
break;
}
if (((long) SHAspeed - bestSHAspeed) > 10000) {
if (do_benchmark)
fprintf(stderr, "+");
bestSHAspeed = SHAspeed;
optimal_gws = num;
}
if (do_benchmark)
fprintf(stderr, "\n");
}
fprintf(stderr, "Optimal global work size %d\n", optimal_gws);
fprintf(stderr, "(to avoid this test on next run, put \""
GWS_CONFIG " = %d\" in john.conf, section [" SECTION_OPTIONS
SUBSECTION_OPENCL "])\n", optimal_gws);
global_work_size = optimal_gws;
create_clobj(optimal_gws, self);
}
//Do the proper test using different sizes.
static cl_ulong gws_test(size_t num, struct fmt_main * self) {
cl_event myEvent;
cl_int ret_code;
cl_uint *tmpbuffer;
cl_ulong startTime, endTime, runtime;
int i;
//Prepare buffers.
create_clobj(num, self);
tmpbuffer = mem_alloc(sizeof(sha256_hash) * num);
if (tmpbuffer == NULL) {
fprintf(stderr, "Malloc failure in find_best_gws\n");
exit(EXIT_FAILURE);
}
queue_prof = clCreateCommandQueue(context[ocl_gpu_id], devices[ocl_gpu_id],
CL_QUEUE_PROFILING_ENABLE, &ret_code);
HANDLE_CLERROR(ret_code, "Failed in clCreateCommandQueue");
// Set keys
for (i = 0; i < num; i++) {
set_key("aaabaabaaa", i);
}
//** Get execution time **//
HANDLE_CLERROR(clEnqueueWriteBuffer(queue_prof, pass_buffer, CL_FALSE, 0,
sizeof(sha256_password) * num, plaintext, 0, NULL, &myEvent),
"Failed in clEnqueueWriteBuffer");
HANDLE_CLERROR(clFinish(queue_prof), "Failed in clFinish");
HANDLE_CLERROR(clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT,
sizeof(cl_ulong), &startTime, NULL),
"Failed in clGetEventProfilingInfo I");
HANDLE_CLERROR(clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END,
sizeof(cl_ulong), &endTime, NULL),
"Failed in clGetEventProfilingInfo II");
HANDLE_CLERROR(clReleaseEvent(myEvent), "Failed in clReleaseEvent");
runtime = endTime - startTime;
//** Get execution time **//
ret_code = clEnqueueNDRangeKernel(queue_prof, crypt_kernel,
1, NULL, &num, &local_work_size, 0, NULL, &myEvent);
HANDLE_CLERROR(clFinish(queue_prof), "Failed in clFinish");
HANDLE_CLERROR(clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT,
sizeof(cl_ulong), &startTime, NULL),
"Failed in clGetEventProfilingInfo I");
HANDLE_CLERROR(clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END,
sizeof(cl_ulong), &endTime, NULL),
"Failed in clGetEventProfilingInfo II");
HANDLE_CLERROR(clReleaseEvent(myEvent), "Failed in clReleaseEvent");
runtime += endTime - startTime;
//** Get execution time **//
HANDLE_CLERROR(clEnqueueReadBuffer(queue_prof, hash_buffer, CL_FALSE, 0,
sizeof(uint32_t) * num, tmpbuffer, 0, NULL, &myEvent),
"Failed in clEnqueueReadBuffer");
HANDLE_CLERROR(clFinish(queue_prof), "Failed in clFinish");
HANDLE_CLERROR(clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_SUBMIT,
sizeof(cl_ulong), &startTime, NULL),
"Failed in clGetEventProfilingInfo I");
HANDLE_CLERROR(clGetEventProfilingInfo(myEvent, CL_PROFILING_COMMAND_END,
sizeof(cl_ulong), &endTime, NULL),
"Failed in clGetEventProfilingInfo II");
HANDLE_CLERROR(clReleaseEvent(myEvent), "Failed in clReleaseEvent");
runtime += endTime - startTime;
MEM_FREE(tmpbuffer);
HANDLE_CLERROR(clReleaseCommandQueue(queue_prof),
"Failed in clReleaseCommandQueue");
release_clobj();
if (ret_code != CL_SUCCESS) {
if (ret_code != CL_INVALID_WORK_GROUP_SIZE)
fprintf(stderr, "Error %d\n", ret_code);
return 0;
}
return runtime;
}
static void init(struct fmt_main *self)
{
#ifdef CL_VERSION_1_0
char *temp;
cl_ulong maxsize;
global_work_size = 0;
opencl_init("$JOHN/rar_kernel.cl", ocl_gpu_id, platform_id);
// create kernel to execute
crypt_kernel = clCreateKernel(program[ocl_gpu_id], "SetCryptKeys", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel. Double-check kernel name?");
/* We mimic the lengths of cRARk for comparisons */
if (get_device_type(ocl_gpu_id) == CL_DEVICE_TYPE_GPU) {
#ifndef DEBUG
self->params.benchmark_comment = " (6 characters)";
#endif
self->params.tests = gpu_tests;
#if defined(DEBUG) && !defined(ALWAYS_OPENCL)
fprintf(stderr, "Note: will use CPU for some self-tests, and Single mode.\n");
#endif
}
if ((temp = cfg_get_param(SECTION_OPTIONS, SUBSECTION_OPENCL, LWS_CONFIG)))
local_work_size = atoi(temp);
if ((temp = cfg_get_param(SECTION_OPTIONS, SUBSECTION_OPENCL, GWS_CONFIG)))
global_work_size = atoi(temp);
if ((temp = getenv("LWS")))
local_work_size = atoi(temp);
if ((temp = getenv("GWS")))
global_work_size = atoi(temp);
/* Note: we ask for this kernel's max size, not the device's! */
HANDLE_CLERROR(clGetKernelWorkGroupInfo(crypt_kernel, devices[ocl_gpu_id], CL_KERNEL_WORK_GROUP_SIZE, sizeof(maxsize), &maxsize, NULL), "Query max work group size");
#ifdef DEBUG
fprintf(stderr, "Max allowed local work size %d\n", (int)maxsize);
#endif
if (!local_work_size) {
if (get_device_type(ocl_gpu_id) == CL_DEVICE_TYPE_CPU) {
if (get_platform_vendor_id(platform_id) == INTEL)
local_work_size = 8;
else
local_work_size = 1;
} else {
local_work_size = 64;
}
}
if (local_work_size > maxsize) {
fprintf(stderr, "LWS %d is too large for this GPU. Max allowed is %d, using that.\n", (int)local_work_size, (int)maxsize);
local_work_size = maxsize;
}
if (!global_work_size)
find_best_gws(temp == NULL ? 0 : 1);
if (global_work_size < local_work_size)
global_work_size = local_work_size;
fprintf(stderr, "Local worksize (LWS) %d, Global worksize (GWS) %d\n", (int)local_work_size, (int)global_work_size);
create_clobj(global_work_size);
#ifdef DEBUG
{
cl_ulong loc_mem_size;
HANDLE_CLERROR(clGetKernelWorkGroupInfo(crypt_kernel, devices[ocl_gpu_id], CL_KERNEL_LOCAL_MEM_SIZE, sizeof(loc_mem_size), &loc_mem_size, NULL), "Query local memory usage");
fprintf(stderr, "Kernel using %lu bytes of local memory out of %lu available\n", loc_mem_size, get_local_memory_size(ocl_gpu_id));
}
#endif
atexit(release_clobj);
*mkpc = VF * global_work_size;
#endif /* OpenCL */
#if defined (_OPENMP)
omp_t = omp_get_max_threads();
self->params.min_keys_per_crypt *= omp_t;
#ifndef CL_VERSION_1_0 /* OpenCL gets to decide */
*mkpc = omp_t * OMP_SCALE * MAX_KEYS_PER_CRYPT;
#endif
init_locks();
#endif /* _OPENMP */
if (options.utf8)
self->params.plaintext_length = PLAINTEXT_LENGTH * 3;
unpack_data = mem_calloc_tiny(sizeof(unpack_data_t) * omp_t, MEM_ALIGN_WORD);
cracked = mem_calloc_tiny(sizeof(*cracked) * *mkpc, MEM_ALIGN_WORD);
#ifndef CL_VERSION_1_0
saved_key = mem_calloc_tiny(UNICODE_LENGTH * *mkpc, MEM_ALIGN_NONE);
saved_len = mem_calloc_tiny(sizeof(*saved_len) * *mkpc, MEM_ALIGN_WORD);
saved_salt = mem_calloc_tiny(8, MEM_ALIGN_NONE);
aes_key = mem_calloc_tiny(16 * *mkpc, MEM_ALIGN_NONE);
aes_iv = mem_calloc_tiny(16 * *mkpc, MEM_ALIGN_NONE);
#endif
/* OpenSSL init */
init_aesni();
SSL_load_error_strings();
SSL_library_init();
OpenSSL_add_all_algorithms();
#ifndef __APPLE__
atexit(openssl_cleanup);
#endif
/* CRC-32 table init, do it before we start multithreading */
{
CRC32_t crc;
CRC32_Init(&crc);
}
}
/* ------- Initialization ------- */
static void init(struct fmt_main * self) {
char * tmp_value;
char * task = "$JOHN/cryptsha512_kernel_DEFAULT.cl";
uint64_t startTime, runtime;
opencl_init_dev(ocl_gpu_id, platform_id);
startTime = (unsigned long) time(NULL);
source_in_use = device_info[ocl_gpu_id];
if ((tmp_value = getenv("_TYPE")))
source_in_use = atoi(tmp_value);
if ((tmp_value = getenv("_FAST")))
fast_mode = TRUE;
if (use_local(source_in_use))
task = "$JOHN/cryptsha512_kernel_LOCAL.cl";
else if (gpu(source_in_use)) {
fprintf(stderr, "Building the kernel, this could take a while\n");
task = "$JOHN/cryptsha512_kernel_GPU.cl";
}
fflush(stdout);
opencl_build_kernel(task, ocl_gpu_id);
if ((runtime = (unsigned long) (time(NULL) - startTime)) > 2UL)
fprintf(stderr, "Elapsed time: %lu seconds\n", runtime);
fflush(stdout);
// create kernel(s) to execute
crypt_kernel = clCreateKernel(program[ocl_gpu_id], "kernel_crypt", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel. Double-check kernel name?");
if (gpu(source_in_use) || use_local(source_in_use)) {
prepare_kernel = clCreateKernel(program[ocl_gpu_id], "kernel_prepare", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel_prepare. Double-check kernel name?");
final_kernel = clCreateKernel(program[ocl_gpu_id], "kernel_final", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel_final. Double-check kernel name?");
}
global_work_size = get_task_max_size();
local_work_size = get_default_workgroup();
if (source_in_use != device_info[ocl_gpu_id])
fprintf(stderr, "Selected runtime id %d, source (%s)\n", source_in_use, task);
if ((tmp_value = cfg_get_param(SECTION_OPTIONS,
SUBSECTION_OPENCL, LWS_CONFIG)))
local_work_size = atoi(tmp_value);
if ((tmp_value = getenv("LWS")))
local_work_size = atoi(tmp_value);
//Check if local_work_size is a valid number.
if (local_work_size > get_task_max_work_group_size()){
local_work_size = 0; //Force find a valid number.
}
self->params.max_keys_per_crypt = global_work_size;
if (!local_work_size) {
local_work_size = get_task_max_work_group_size();
create_clobj(global_work_size, self);
find_best_workgroup(self);
release_clobj();
}
if ((tmp_value = cfg_get_param(SECTION_OPTIONS,
SUBSECTION_OPENCL, GWS_CONFIG)))
global_work_size = atoi(tmp_value);
if ((tmp_value = getenv("GWS")))
global_work_size = atoi(tmp_value);
//Check if a valid multiple is used.
global_work_size = get_multiple(global_work_size, local_work_size);
if (global_work_size)
create_clobj(global_work_size, self);
else {
//user chose to die of boredom
global_work_size = get_task_max_size();
find_best_gws(self);
}
fprintf(stderr, "Local work size (LWS) %d, global work size (GWS) %zd\n",
(int) local_work_size, global_work_size);
self->params.max_keys_per_crypt = global_work_size;
}
static void init(struct fmt_main *self)
{
char *temp;
cl_ulong maxsize, maxsize2;
char build_opts[64];
global_work_size = 0;
snprintf(build_opts, sizeof(build_opts),
"-DHASH_LOOPS=%u -DUNICODE_LENGTH=%u %s",
HASH_LOOPS,
UNICODE_LENGTH,
(options.flags & FLG_VECTORIZE) ? "-DVECTORIZE" :
(options.flags & FLG_SCALAR) ? "-DSCALAR" : "");
opencl_init_opt("$JOHN/office2007_kernel.cl", ocl_gpu_id, platform_id, build_opts);
// create kernel to execute
GenerateSHA1pwhash = clCreateKernel(program[ocl_gpu_id], "GenerateSHA1pwhash", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel. Double-check kernel name?");
crypt_kernel = clCreateKernel(program[ocl_gpu_id], "HashLoop", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel. Double-check kernel name?");
Generate2007key = clCreateKernel(program[ocl_gpu_id], "Generate2007key", &ret_code);
HANDLE_CLERROR(ret_code, "Error creating kernel. Double-check kernel name?");
if (options.flags & FLG_VECTORIZE) {
/* Run vectorized code */
VF = 4;
self->params.algorithm_name = "OpenCL 4x";
}
if ((temp = cfg_get_param(SECTION_OPTIONS, SUBSECTION_OPENCL, LWS_CONFIG)))
local_work_size = atoi(temp);
if ((temp = cfg_get_param(SECTION_OPTIONS, SUBSECTION_OPENCL, GWS_CONFIG)))
global_work_size = atoi(temp);
if ((temp = getenv("LWS")))
local_work_size = atoi(temp);
if ((temp = getenv("GWS")))
global_work_size = atoi(temp);
/* Note: we ask for the kernels' max sizes, not the device's! */
HANDLE_CLERROR(clGetKernelWorkGroupInfo(GenerateSHA1pwhash, devices[ocl_gpu_id], CL_KERNEL_WORK_GROUP_SIZE, sizeof(maxsize), &maxsize, NULL), "Query max work group size");
HANDLE_CLERROR(clGetKernelWorkGroupInfo(crypt_kernel, devices[ocl_gpu_id], CL_KERNEL_WORK_GROUP_SIZE, sizeof(maxsize2), &maxsize2, NULL), "Query max work group size");
if (maxsize2 < maxsize) maxsize = maxsize2;
HANDLE_CLERROR(clGetKernelWorkGroupInfo(Generate2007key, devices[ocl_gpu_id], CL_KERNEL_WORK_GROUP_SIZE, sizeof(maxsize2), &maxsize2, NULL), "Query max work group size");
if (maxsize2 < maxsize) maxsize = maxsize2;
#if 0
/* Our use of local memory sets a limit for LWS */
maxsize2 = get_local_memory_size(ocl_gpu_id) / (24 * VF);
while (maxsize > maxsize2)
maxsize >>= 1;
#endif
/* maxsize is the lowest figure from the three different kernels */
if (!local_work_size) {
if (getenv("LWS")) {
/* LWS was explicitly set to 0 */
int temp = global_work_size;
local_work_size = maxsize;
global_work_size = global_work_size ? global_work_size : 4 * maxsize;
create_clobj(global_work_size, self);
opencl_find_best_workgroup_limit(self, maxsize);
release_clobj();
global_work_size = temp;
} else {
if (cpu(device_info[ocl_gpu_id])) {
if (get_platform_vendor_id(platform_id) == DEV_INTEL)
local_work_size = MIN(maxsize, 8);
else
local_work_size = 1;
} else
local_work_size = MIN(maxsize, 64);
}
}
if (local_work_size > maxsize) {
fprintf(stderr, "LWS %d is too large for this GPU. Max allowed is %d, using that.\n", (int)local_work_size, (int)maxsize);
local_work_size = maxsize;
}
if (!global_work_size)
find_best_gws(getenv("GWS") == NULL ? 0 : 1, self);
if (global_work_size < local_work_size)
global_work_size = local_work_size;
fprintf(stderr, "Local worksize (LWS) %d, Global worksize (GWS) %d\n", (int)local_work_size, (int)global_work_size);
create_clobj(global_work_size, self);
atexit(release_clobj);
if (options.utf8)
self->params.plaintext_length = MIN(125, 3 * PLAINTEXT_LENGTH);
}
static cl_ulong gws_test(int gws, int do_benchmark, struct fmt_main *self)
{
cl_ulong startTime, endTime;
cl_command_queue queue_prof;
cl_event Event[6];
cl_int ret_code;
int i;
size_t scalar_gws = VF * gws;
create_clobj(gws, self);
queue_prof = clCreateCommandQueue(context[ocl_gpu_id], devices[ocl_gpu_id], CL_QUEUE_PROFILING_ENABLE, &ret_code);
for (i = 0; i < scalar_gws; i++)
set_key(tests[0].plaintext, i);
set_salt(get_salt(tests[0].ciphertext));
HANDLE_CLERROR(clEnqueueWriteBuffer(queue_prof, cl_saved_key, CL_TRUE, 0, UNICODE_LENGTH * scalar_gws, saved_key, 0, NULL, &Event[0]), "Failed transferring keys");
HANDLE_CLERROR(clEnqueueWriteBuffer(queue_prof, cl_saved_len, CL_TRUE, 0, sizeof(int) * scalar_gws, saved_len, 0, NULL, &Event[1]), "Failed transferring lengths");
HANDLE_CLERROR(clEnqueueNDRangeKernel(queue_prof, GenerateSHA1pwhash, 1, NULL, &scalar_gws, &local_work_size, 0, NULL, &Event[2]), "running kernel");
for (i = 0; i < 50000 / HASH_LOOPS - 1; i++)
HANDLE_CLERROR(clEnqueueNDRangeKernel(queue_prof, crypt_kernel, 1, NULL, &global_work_size, &local_work_size, 0, NULL, NULL), "running kernel");
HANDLE_CLERROR(clEnqueueNDRangeKernel(queue_prof, crypt_kernel, 1, NULL, &global_work_size, &local_work_size, 0, NULL, &Event[3]), "running kernel");
HANDLE_CLERROR(clEnqueueNDRangeKernel(queue_prof, Generate2007key, 1, NULL, &global_work_size, &local_work_size, 0, NULL, &Event[4]), "running kernel");
HANDLE_CLERROR(clEnqueueReadBuffer(queue_prof, cl_key, CL_TRUE, 0, 16 * scalar_gws, key, 0, NULL, &Event[5]), "failed in reading key back");
#if 0
HANDLE_CLERROR(clGetEventProfilingInfo(Event[2],
CL_PROFILING_COMMAND_START, sizeof(cl_ulong), &startTime,
NULL), "Failed to get profiling info");
HANDLE_CLERROR(clGetEventProfilingInfo(Event[2],
CL_PROFILING_COMMAND_END, sizeof(cl_ulong), &endTime,
NULL), "Failed to get profiling info");
fprintf(stderr, "GenerateSHA1pwhash kernel duration: %llu us, ", (endTime-startTime)/1000ULL);
#endif
HANDLE_CLERROR(clGetEventProfilingInfo(Event[3],
CL_PROFILING_COMMAND_START, sizeof(cl_ulong), &startTime,
NULL), "Failed to get profiling info");
HANDLE_CLERROR(clGetEventProfilingInfo(Event[3],
CL_PROFILING_COMMAND_END, sizeof(cl_ulong), &endTime,
NULL), "Failed to get profiling info");
if (do_benchmark)
fprintf(stderr, "%.2f ms x %u = %.2f s\t", (float)((endTime - startTime)/1000000.), 50000/HASH_LOOPS, (float)(50000/HASH_LOOPS) * (endTime - startTime) / 1000000000.);
/* 200 ms duration limit for GCN to avoid ASIC hangs */
if (amd_gcn(device_info[ocl_gpu_id]) && endTime - startTime > 200000000) {
if (do_benchmark)
fprintf(stderr, "- exceeds 200 ms\n");
clReleaseCommandQueue(queue_prof);
release_clobj();
return 0;
}
#if 0
HANDLE_CLERROR(clGetEventProfilingInfo(Event[4],
CL_PROFILING_COMMAND_START, sizeof(cl_ulong), &startTime,
NULL), "Failed to get profiling info");
HANDLE_CLERROR(clGetEventProfilingInfo(Event[4],
CL_PROFILING_COMMAND_END, sizeof(cl_ulong), &endTime,
NULL), "Failed to get profiling info");
fprintf(stderr, "Generate2007key kernel duration: %llu us\n", (endTime-startTime)/1000ULL);
#endif
HANDLE_CLERROR(clGetEventProfilingInfo(Event[0],
CL_PROFILING_COMMAND_SUBMIT, sizeof(cl_ulong), &startTime,
NULL), "Failed to get profiling info");
HANDLE_CLERROR(clGetEventProfilingInfo(Event[5],
CL_PROFILING_COMMAND_END, sizeof(cl_ulong), &endTime,
NULL), "Failed to get profiling info");
clReleaseCommandQueue(queue_prof);
release_clobj();
return (endTime - startTime);
}