arno/hashcat
1
0
Fork 0
mirror of https://github.com/hashcat/hashcat.git synced 2025-05-09 10:28:48 +00:00
Code Issues Releases Wiki Activity

Allow async execution of autotune() and prepare try_run() for async kernel execution

Browse Source
This commit is contained in:
Jukka Ojanen 2021-07-26 01:51:19 +03:00
parent 79404b1ff1
commit a86b44a750
1 changed files with 36 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

38
src/autotune.c
View File

@ -37,6 +37,23 @@ static double try_run (hashcat_ctx_t *hashcat_ctx, hc_device_param_t *device_par
device_param->spin_damp = 0; device_param->spin_damp = 0;
if (device_param->is_cuda == true)
{
if (hc_cuStreamSynchronize (hashcat_ctx, device_param->cuda_stream) == -1) return -1;
}
if (device_param->is_hip == true)
{
if (hc_hipStreamSynchronize (hashcat_ctx, device_param->hip_stream) == -1) return -1;
}
if (device_param->is_opencl == true)
{
/*if (hc_clFlush (hashcat_ctx, device_param->opencl_command_queue) == -1) return -1;*/
if (hc_clFinish (hashcat_ctx, device_param->opencl_command_queue) == -1) return -1;
}
if (hashconfig->attack_exec == ATTACK_EXEC_INSIDE_KERNEL) if (hashconfig->attack_exec == ATTACK_EXEC_INSIDE_KERNEL)
{ {
if (hashconfig->opti_type & OPTI_TYPE_OPTIMIZED_KERNEL) if (hashconfig->opti_type & OPTI_TYPE_OPTIMIZED_KERNEL)
@ -60,6 +77,23 @@ static double try_run (hashcat_ctx_t *hashcat_ctx, hc_device_param_t *device_par
run_kernel (hashcat_ctx, device_param, KERN_RUN_2, 0, kernel_power_try, true, 0); run_kernel (hashcat_ctx, device_param, KERN_RUN_2, 0, kernel_power_try, true, 0);
} }
if (device_param->is_cuda == true)
{
if (hc_cuStreamSynchronize (hashcat_ctx, device_param->cuda_stream) == -1) return -1;
}
if (device_param->is_hip == true)
{
if (hc_hipStreamSynchronize (hashcat_ctx, device_param->hip_stream) == -1) return -1;
}
if (device_param->is_opencl == true)
{
/*if (hc_clFlush (hashcat_ctx, device_param->opencl_command_queue) == -1) return -1;*/
if (hc_clFinish (hashcat_ctx, device_param->opencl_command_queue) == -1) return -1;
}
device_param->spin_damp = spin_damp_sav; device_param->spin_damp = spin_damp_sav;
const double exec_msec_prev = get_avg_exec_time (device_param, 1); const double exec_msec_prev = get_avg_exec_time (device_param, 1);
@ -200,14 +234,14 @@ static int autotune (hashcat_ctx_t *hashcat_ctx, hc_device_param_t *device_param
{ {
if (device_param->is_cuda == true) if (device_param->is_cuda == true)
{ {
CU_rc = hc_cuMemcpyDtoD (hashcat_ctx, device_param->cuda_d_rules_c, device_param->cuda_d_rules, MIN (kernel_loops_max, KERNEL_RULES) * sizeof (kernel_rule_t)); CU_rc = hc_cuMemcpyDtoDAsync (hashcat_ctx, device_param->cuda_d_rules_c, device_param->cuda_d_rules, MIN (kernel_loops_max, KERNEL_RULES) * sizeof (kernel_rule_t), device_param->cuda_stream);
if (CU_rc == -1) return -1; if (CU_rc == -1) return -1;
} }
if (device_param->is_hip == true) if (device_param->is_hip == true)
{ {
HIP_rc = hc_hipMemcpyDtoD (hashcat_ctx, device_param->hip_d_rules_c, device_param->hip_d_rules, MIN (kernel_loops_max, KERNEL_RULES) * sizeof (kernel_rule_t)); HIP_rc = hc_hipMemcpyDtoDAsync (hashcat_ctx, device_param->hip_d_rules_c, device_param->hip_d_rules, MIN (kernel_loops_max, KERNEL_RULES) * sizeof (kernel_rule_t), device_param->hip_stream);
if (HIP_rc == -1) return -1; if (HIP_rc == -1) return -1;
} }