Jens Steube
cace799187
Move more no unrolling exceptions from inc_vendor.cl to specific modules
6 years ago
jsteube
dc9279c95c
New Strategy: Instead of using volatile, mark the mode as unstable. Remove all volatiles
6 years ago
jsteube
7c1120b784
Migrate unroll handling for -m 15900 from OpenCL/inc_vendor.cl to module_jit_build_options()
6 years ago
jsteube
1714382daa
Migrate unroll handling for -m 14100 from OpenCL/inc_vendor.cl to module_jit_build_options()
6 years ago
jsteube
b0a83b28d1
Migrate unroll handling for -m 12300 from OpenCL/inc_vendor.cl to module_jit_build_options()
6 years ago
jsteube
1937b67595
Migrate unroll handling for -m 8200 from OpenCL/inc_vendor.cl to module_jit_build_options()
6 years ago
jsteube
fee37ac4c3
Migrate unroll handling for -m 8000 from OpenCL/inc_vendor.cl to module_jit_build_options()
6 years ago
jsteube
88a051629c
Support module specific JiT compiler build options
6 years ago
jsteube
4359418918
OpenCL Runtime: Not using amd_bytealign (amd_bitalign is fine) on AMDGPU driver drastically reduces JiT segfaults
6 years ago
jsteube
540b405e3a
Replace IS_ROCM with HAS_VPERM and HAS_VADD3
6 years ago
Jens Steube
ae577410d0
OpenCL Device: Do a real query on OpenCL local memory type instead of just assuming it
6 years ago
jsteube
6469357c74
Remove SCR_TYPE macro from OpenCL code
...
Disable REAL_SHM access to AMD platform devices
6 years ago
jsteube
dad05d9f69
Testing: Workaround some AMD OpenCL runtime segmentation faults
6 years ago
Jens Steube
a43d3ad176
Rename some hashcat specific OpenCL functions to avoid conflicts with existing OpenCL functions from OpenCL runtime
6 years ago
jsteube
c1622d6593
Fixed detection of AMD_GCN version in case the rocm driver is used
6 years ago
Jens Steube
ad50883080
Allow unroll for DES based algorithms but not bitsliced versions
7 years ago
Jens Steube
aa82d8d34d
Re-enable CPU optimizations and some CPU case in thread management
7 years ago
Jens Steube
2a19f19904
Drop -m 6800 from no-unroll list in inc_vendor.cl
7 years ago
jsteube
d5153539e2
Some syntax error
7 years ago
jsteube
786384664e
DECLSPEC for CPU
7 years ago
jsteube
e4e1c1d515
We can't mix inline functions with static constants
7 years ago
Jens Steube
8273bb8376
NV JiT doesn't like static inline keywords
7 years ago
jsteube
3e08750900
OpenCL Kernels: Add general function declaration keyword (static inline)
7 years ago
jsteube
00abb849e1
Update inc_vector.cl for ideal performance with reference GTX1080 for NV and Vega64 for AMD
7 years ago
Fist0urs
a6294537fd
Splitted DPAPI kernel in 2 to increase performances
7 years ago
jsteube
1963b12fdc
According to AMD docs, GCN 3 and 4 are the same
7 years ago
jsteube
ac9f1da747
Add fine-tuned AMD GCN control macros
7 years ago
Jens Steube
b58aa445b4
Do not use __local memory for -m 7500 if running on a device without physical shared memory
7 years ago
Jens Steube
63f6ca5114
Do not use __local memory for whirlpool if running on a device without physical shared memory
7 years ago
jsteube
00e38cc2c6
Add VEGA specific inline assembly to improve all MD4, MD5, SHA1 and SHA256 based kernels
7 years ago
jsteube
51372438fe
Allow OpenCL kernel inline assembly if ROCm drivers was detected
7 years ago
jsteube
bf299fe043
Optimized 3DES for rocm
7 years ago
jsteube
9ee5da40e0
Workaround rocm compiler error for -m 15300
7 years ago
jsteube
772441448a
Optimized -m 8000 for ROCm
7 years ago
jsteube
3125a756d9
Remove some AMD _unroll restrictions no longer required with ROCm
7 years ago
jsteube
5e34ec348e
Optimize kernels for ROCm 1.6
...
- Remove inline keywords
- Remove volatile keywords where it causes ROCm to slow down
- Replace DES functions (looks like bitselect somehow is no longer mapped to BFI_INT)
7 years ago
jsteube
cefd2ddb94
Tune AMD unroll settings for AMD-GPU-PRO 17.40
7 years ago
jsteube
8257883ec1
Added hash-mode 14600 = LUKS
8 years ago
jsteube
899413f88b
Allow unroll of Android FDE (Samsung DEK) on NV
8 years ago
jsteube
ac02c60b3f
Do not allow unroll of Drupal7, but for NV only
8 years ago
jsteube
ee2e8ba598
Do not allow unroll of Drupal7, increases performance on CL1.2
8 years ago
jsteube
0de908905f
Allow unroll of PBKDF2-HMAC-SHA256 on NV
8 years ago
jsteube
0206142a69
Bcrypt on NVIDIA works better when unrolled
8 years ago
Gabriele Gristina
246336c49c
Fix CL_BUILD_PROGRAM_FAILURE on OSX ( #412 )
8 years ago
jsteube
91a69fa635
Experiment to fix https://github.com/hashcat/hashcat/issues/412
8 years ago
jsteube
92f7a560a7
Optimize DES/LM cracking speed boost on AMD
8 years ago
jsteube
31a616e00e
Add a shared section in inc_vendor.cl
8 years ago
jsteube
306fc8a59a
Optimize unrolling for AMD based on AMDGPU-Pro 16.40
8 years ago
jsteube
ac44087c83
Optimize unrolling for AMD based on AMDGPU-Pro 16.40
8 years ago
jsteube
f58c9baade
Do not unroll -m 6800 based algorithm on NV, latest windows driver can't compile them otherwise
8 years ago