arno/hashcat
1
0
Fork 0
mirror of https://github.com/hashcat/hashcat.git synced 2025-01-09 15:20:59 +00:00
Code Issues Releases Wiki Activity

Reduce some register allocation in SCRYPT based kernels to improve speed on GPU with low resources

Browse Source
This commit is contained in:
Jens Steube 2021-04-22 00:26:09 +02:00
parent c444b42290
commit 2b069a94fd
3 changed files with 58 additions and 36 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
OpenCL/m08900-pure.cl
View File

@ -132,11 +132,11 @@ DECLSPEC void salsa_r (uint4 *TI)
uint4 R2 = TI[STATE_CNT4 - 2]; uint4 R2 = TI[STATE_CNT4 - 2];
uint4 R3 = TI[STATE_CNT4 - 1]; uint4 R3 = TI[STATE_CNT4 - 1];
uint4 TO[STATE_CNT4]; uint4 TT[STATE_CNT4 / 2];
int idx_y = 0; int idx_y = 0;
int idx_r1 = 0; int idx_r1 = 0;
int idx_r2 = SCRYPT_R * 4; int idx_r2 = 0;
for (int i = 0; i < SCRYPT_R; i++) for (int i = 0; i < SCRYPT_R; i++)
{ {
@ -152,10 +152,10 @@ DECLSPEC void salsa_r (uint4 *TI)
SALSA20_8_XOR (); SALSA20_8_XOR ();
TO[idx_r1++] = R0; TI[idx_r1++] = R0;
TO[idx_r1++] = R1; TI[idx_r1++] = R1;
TO[idx_r1++] = R2; TI[idx_r1++] = R2;
TO[idx_r1++] = R3; TI[idx_r1++] = R3;
Y0 = TI[idx_y++]; Y0 = TI[idx_y++];
Y1 = TI[idx_y++]; Y1 = TI[idx_y++];
@ -164,18 +164,24 @@ DECLSPEC void salsa_r (uint4 *TI)
SALSA20_8_XOR (); SALSA20_8_XOR ();
TO[idx_r2++] = R0; TT[idx_r2++] = R0;
TO[idx_r2++] = R1; TT[idx_r2++] = R1;
TO[idx_r2++] = R2; TT[idx_r2++] = R2;
TO[idx_r2++] = R3; TT[idx_r2++] = R3;
} }
idx_r1 = 0;
idx_r2 = SCRYPT_R * 4;
#ifdef _unroll #ifdef _unroll
#pragma unroll #pragma unroll
#endif #endif
for (int i = 0; i < STATE_CNT4; i++) for (int i = 0; i < SCRYPT_R; i++)
{ {
TI[i] = TO[i]; TI[idx_r2++] = TT[idx_r1++];
TI[idx_r2++] = TT[idx_r1++];
TI[idx_r2++] = TT[idx_r1++];
TI[idx_r2++] = TT[idx_r1++];
} }
} }

32
OpenCL/m15700-pure.cl
View File

@ -139,11 +139,11 @@ DECLSPEC void salsa_r (uint4 *TI)
uint4 R2 = TI[STATE_CNT4 - 2]; uint4 R2 = TI[STATE_CNT4 - 2];
uint4 R3 = TI[STATE_CNT4 - 1]; uint4 R3 = TI[STATE_CNT4 - 1];
uint4 TO[STATE_CNT4]; uint4 TT[STATE_CNT4 / 2];
int idx_y = 0; int idx_y = 0;
int idx_r1 = 0; int idx_r1 = 0;
int idx_r2 = SCRYPT_R * 4; int idx_r2 = 0;
for (int i = 0; i < SCRYPT_R; i++) for (int i = 0; i < SCRYPT_R; i++)
{ {
@ -159,10 +159,10 @@ DECLSPEC void salsa_r (uint4 *TI)
SALSA20_8_XOR (); SALSA20_8_XOR ();
TO[idx_r1++] = R0; TI[idx_r1++] = R0;
TO[idx_r1++] = R1; TI[idx_r1++] = R1;
TO[idx_r1++] = R2; TI[idx_r1++] = R2;
TO[idx_r1++] = R3; TI[idx_r1++] = R3;
Y0 = TI[idx_y++]; Y0 = TI[idx_y++];
Y1 = TI[idx_y++]; Y1 = TI[idx_y++];
@ -171,16 +171,24 @@ DECLSPEC void salsa_r (uint4 *TI)
SALSA20_8_XOR (); SALSA20_8_XOR ();
TO[idx_r2++] = R0; TT[idx_r2++] = R0;
TO[idx_r2++] = R1; TT[idx_r2++] = R1;
TO[idx_r2++] = R2; TT[idx_r2++] = R2;
TO[idx_r2++] = R3; TT[idx_r2++] = R3;
} }
idx_r1 = 0;
idx_r2 = SCRYPT_R * 4;
#ifdef _unroll
#pragma unroll #pragma unroll
for (int i = 0; i < STATE_CNT4; i++) #endif
for (int i = 0; i < SCRYPT_R; i++)
{ {
TI[i] = TO[i]; TI[idx_r2++] = TT[idx_r1++];
TI[idx_r2++] = TT[idx_r1++];
TI[idx_r2++] = TT[idx_r1++];
TI[idx_r2++] = TT[idx_r1++];
} }
} }

32
OpenCL/m22700-pure.cl
View File

@ -180,11 +180,11 @@ DECLSPEC void salsa_r (uint4 *TI)
uint4 R2 = TI[STATE_CNT4 - 2]; uint4 R2 = TI[STATE_CNT4 - 2];
uint4 R3 = TI[STATE_CNT4 - 1]; uint4 R3 = TI[STATE_CNT4 - 1];
uint4 TO[STATE_CNT4]; uint4 TT[STATE_CNT4 / 2];
int idx_y = 0; int idx_y = 0;
int idx_r1 = 0; int idx_r1 = 0;
int idx_r2 = SCRYPT_R * 4; int idx_r2 = 0;
for (int i = 0; i < SCRYPT_R; i++) for (int i = 0; i < SCRYPT_R; i++)
{ {
@ -200,10 +200,10 @@ DECLSPEC void salsa_r (uint4 *TI)
SALSA20_8_XOR (); SALSA20_8_XOR ();
TO[idx_r1++] = R0; TI[idx_r1++] = R0;
TO[idx_r1++] = R1; TI[idx_r1++] = R1;
TO[idx_r1++] = R2; TI[idx_r1++] = R2;
TO[idx_r1++] = R3; TI[idx_r1++] = R3;
Y0 = TI[idx_y++]; Y0 = TI[idx_y++];
Y1 = TI[idx_y++]; Y1 = TI[idx_y++];
@ -212,16 +212,24 @@ DECLSPEC void salsa_r (uint4 *TI)
SALSA20_8_XOR (); SALSA20_8_XOR ();
TO[idx_r2++] = R0; TT[idx_r2++] = R0;
TO[idx_r2++] = R1; TT[idx_r2++] = R1;
TO[idx_r2++] = R2; TT[idx_r2++] = R2;
TO[idx_r2++] = R3; TT[idx_r2++] = R3;
} }
idx_r1 = 0;
idx_r2 = SCRYPT_R * 4;
#ifdef _unroll
#pragma unroll #pragma unroll
for (int i = 0; i < STATE_CNT4; i++) #endif
for (int i = 0; i < SCRYPT_R; i++)
{ {
TI[i] = TO[i]; TI[idx_r2++] = TT[idx_r1++];
TI[idx_r2++] = TT[idx_r1++];
TI[idx_r2++] = TT[idx_r1++];
TI[idx_r2++] = TT[idx_r1++];
} }
} }