Jens Steube
d38608b9bc
Add get_kernel_preferred_wgs_multiple() for future use
2018-02-11 16:22:24 +01:00
Jens Steube
1f442b5b0c
Add get_kernel_local_mem_size() for future use
2018-02-11 16:09:50 +01:00
jsteube
ce8f445ebe
Fix a few warnings and divide by zero (speed_pos)
2018-02-11 14:25:55 +01:00
Jens Steube
d656e9c3a4
OpenCL Kernels: Use the kernel local buffer size as additional reference in order to limit the thread-count
2018-02-11 10:56:08 +01:00
jsteube
a7a15df911
OpenCL Kernels: Improve performance preview accuracy in --benchmark, --speed-only and --progress-only mode
2018-02-10 20:44:13 +01:00
Jens Steube
fd89cac052
No longer need to warm-up OpenCL device for benchmark with new thread settings
2018-02-09 23:41:30 +01:00
Jens Steube
75c4d09301
CPU threads still work better on a single local work item
2018-02-09 23:40:52 +01:00
jsteube
5951207365
Get rid of some old volatiles
2018-02-09 19:18:30 +01:00
Jens Steube
fa379074b6
Fixed a logic error in storing temporary progress for slow hashes,leading to invalid speeds in status view
2018-02-09 18:30:53 +01:00
Jens Steube
b0ff625844
OpenCL Kernels: Thread-count is switched from native to maximum - as a consequence we assume host memory pool of 2GB per GPU
2018-02-09 16:42:28 +01:00
jsteube
088c45040b
Remove some redundant constants
2018-02-09 10:48:19 +01:00
Jens Steube
9b1e66d87c
Fixed a race condition in combinator- and hybrid-mode where the same scratch buffer was used by multiple threads
2018-02-09 09:50:13 +01:00
jsteube
05a01d3843
fix some datatypes
2018-02-08 19:13:29 +01:00
jsteube
98abea9a0f
Fix one more technical teething trouble with compressor kernels
2018-02-08 15:41:19 +01:00
jsteube
9daaafc255
Update changes with the latest missing changes
2018-02-08 13:45:48 +01:00
jsteube
d5153539e2
Some syntax error
2018-02-08 09:49:59 +01:00
jsteube
786384664e
DECLSPEC for CPU
2018-02-08 09:42:59 +01:00
jsteube
3dcacb5f55
Fix hc_clEnqueueReadBuffer buffer size in gidd_to_pw_t()
2018-02-08 08:26:24 +01:00
jsteube
1c25cf5490
Fix compiler warning about rc variable which shadows a previous local
2018-02-07 22:32:01 +01:00
jsteube
819b955d8a
Merge branch 'master' of https://github.com/hashcat/hashcat
2018-02-07 22:29:04 +01:00
jsteube
4cbd0eb812
Fix missing compressor kernel in --stdout mode
2018-02-07 22:28:52 +01:00
Jens Steube
4111ce29c9
Merge pull request #1511 from Chick3nman/master
...
Update rule syntax in best64.rule
2018-02-07 18:52:10 +01:00
jsteube
bb401c9cec
Fixed a missing check for returncode in case hashcat.hcstat2 was not found
2018-02-07 18:47:55 +01:00
jsteube
bee0d75a3e
Merge branch 'master' of https://github.com/hashcat/hashcat
2018-02-07 18:13:15 +01:00
jsteube
512fb5f6fb
No inline keyword for rules
2018-02-07 15:02:58 +01:00
jsteube
e4e1c1d515
We can't mix inline functions with static constants
2018-02-07 14:16:27 +01:00
jsteube
ff4d5208be
Merge branch 'master' of https://github.com/hashcat/hashcat
2018-02-06 22:05:32 +01:00
Jens Steube
8273bb8376
NV JiT doesn't like static inline keywords
2018-02-06 22:05:15 +01:00
jsteube
f3865d9d0d
Merge branch 'master' of https://github.com/hashcat/hashcat
2018-02-06 19:12:44 +01:00
jsteube
3e08750900
OpenCL Kernels: Add general function declaration keyword (static inline)
2018-02-06 19:12:24 +01:00
jsteube
bd9c17e49a
Add some missing function declarations
2018-02-06 13:06:10 +01:00
jsteube
e59d5fd25e
sometimes device_global_mem and device_maxmem_alloc reportedback from the opencl runtime are a bit inaccurate.
...
let's add some extra space just to be sure.
2018-02-06 10:11:06 +01:00
jsteube
cdc4f3828b
Reduce max kernel threads from 256 to 64 in -w 4 to save host and GPU memory
2018-02-06 09:38:54 +01:00
jsteube
5391edca0d
Weird macOS JiT likes this more
2018-02-05 19:39:20 +01:00
Chick3nman
b22839c481
Merge pull request #2 from hashcat/master
...
OpenCL Kernels: Add a decompressing kernel and a compressing host cod…
2018-02-05 10:25:23 -06:00
jsteube
a3a16f676f
OpenCL Kernels: Add a decompressing kernel and a compressing host code in order to reduce PCIe transfer time
...
For details see https://hashcat.net/forum/thread-7267.html
2018-02-05 17:18:58 +01:00
Chick3nman
69414400e3
Updated 'x' to 'O'
...
Rule function for 'x' has changed since creation of this rule set. Replaced 'x' with 'O' to bring back originally intended rule functionality.
2018-02-05 10:18:31 -06:00
Chick3nman
93951b5ac5
Merge pull request #1 from hashcat/master
...
Merge out to bring fork up to date
2018-02-05 10:07:27 -06:00
jsteube
2f30e5e929
Fixed invalid support for SIMD in -m 400
2018-02-05 15:57:36 +01:00
jsteube
c99de3c22a
Fix missing function export in test.pl
2018-02-04 14:01:11 +01:00
Jens Steube
665a514977
Use a more conservative mechanic in the last step of the autotune
2018-02-03 19:01:11 +01:00
Jens Steube
fb00b528c4
Improve autotune and synchronize results on same OpenCL devices
2018-02-03 13:49:39 +01:00
Jens Steube
3d2c0018fc
Limit the maximum OpenCL thread count to 256
...
If an OpenCL device allows a very high thread count (for example 1024 on nvidia), the host memory required is 32 times as high with 32 (It jumps from 128MB to 4GB device memory requirement). since there's no device with that much device memory (because of 1/4 memory rule) it has to limit the kernel_accel_max to be a very low number because the pws buffer will be so large otherwise. therefore autotune will be unable to calculate a good kernel_accel multiplier. currently there's no OpenCL device known that needs result in a better performance with 1024 threads compared to 256. as a result, we limit the number of threads to 256, which turns out to be a general good value.
2018-02-03 12:28:00 +01:00
jsteube
517062849e
Save some space in speed display to avoid line breaks
2018-02-03 11:21:28 +01:00
jsteube
11b18512c7
Terminal: Add workitem settings to status display (can be handy for debugging)
2018-02-02 23:22:21 +01:00
jsteube
53e2b40bad
Fixed a uninitialized value in OpenCL kernels 9720, 9820 and 10420 leading to absurd benchmark performance
2018-02-02 14:02:33 +01:00
jsteube
76a3e3f7af
OpenCL Runtime: Add current timestamp to OpenCL kernel source in order to force OpenCL JiT compiler to recompile and not use the cache
2018-02-01 16:57:59 +01:00
jsteube
3272e29497
Returncode: Enforce returncode 0 in case the user selected --speed-only or --progress-only and no other error occured
2018-02-01 11:18:40 +01:00
jsteube
a8ca6862bf
OpenCL Runtime: Enforce to use OpenCL version 1.2 to restrain OpenCL runtimes to make use of the __generic address space qualifier
2018-02-01 10:48:24 +01:00
jsteube
13a79cf942
Fix unoptimized Kerberos 5 TGS-REP etype 23 kernel for use on macOS
2018-01-31 21:28:46 +01:00
