samr7
|
3bcb68878d
|
Note that oclvanitygen kernel no longer breaks with latest NVIDIA
driver on Windows.
Add advisory about kernel compile times.
|
13 years ago |
samr7
|
933a020205
|
Add explicit preprocessor unrolling.
Limit use of #pragma unroll to NVIDIA platforms.
|
13 years ago |
samr7
|
46aed119a1
|
Split base-58 encoding primitives out of pattern.c.
|
13 years ago |
samr7
|
31783f5197
|
Be more specific in error message about missing kernel code.
|
13 years ago |
samr7
|
216cb03c5c
|
Display a warning when used with low-difficulty patterns and -k.
|
13 years ago |
samr7
|
dd2f5670e6
|
Change the meaning of -w again. Now it's the number of address
calculations per thread in a work unit. The previous meaning
(thread slots per multiprocessor) is now -t.
|
13 years ago |
samr7
|
e025be9648
|
Add AMD BFI_INT optimizations to SHA-2 and RIPEMD160 hash functions.
The full effect of this optimization will come later.
|
13 years ago |
samr7
|
7ba90d3086
|
Add implementation of bignum add/subtract better suited to AMD Radeon.
|
13 years ago |
samr7
|
04dd595ce2
|
Add device quirks mechanism.
Disable stored program binaries on RV710 devices.
|
14 years ago |
samr7
|
749daa7b17
|
Bump version to 0.16.
Use stdio binary mode when reading/writing kernel binaries.
Fix up help message for oclvanitygen.
|
14 years ago |
samr7
|
a53a6e52da
|
Add safe mode option to disable problematic optimizations.
With safe mode, oclvanitygen works with AMD's CPU OpenCL device.
|
14 years ago |
jackjack
|
75246451d3
|
Add addrtype flag
|
14 years ago |
samr7
|
31ca88ab40
|
Apply some optimizations to the OpenCL kernel.
Add a new flag, VERY_EXPENSIVE_BRANCHES, for various Radeon devices.
|
14 years ago |
Erik Henriksson
|
ca372318b3
|
*Added makefile for osx
*Added INSTALL file
*Changed one include for running oclvanitygen on osx
*Added some files to .gitignore
|
14 years ago |
samr7
|
efb7cedde5
|
Warn about NVIDIA drivers that seem to produce broken device code.
Bump version number.
|
14 years ago |
samr7
|
dff09af8cb
|
Bump version numbers.
|
14 years ago |
samr7
|
a4d342e916
|
Save kernel binary after compiling.
|
14 years ago |
samr7
|
29e8b034f6
|
Change invsize to be interpreted consistently with the comments / help text.
|
14 years ago |
samr7
|
7c13c25653
|
Auto-configure work unit parameters.
|
14 years ago |
samr7
|
69004b2062
|
Optimize global memory access pattern for row buffer.
|
14 years ago |
samr7
|
ffbc9d929f
|
Optimize global memory access patterns in heap_invert.
|
14 years ago |
samr7
|
f2de8f73d8
|
Use private key type 180 for namecoin private keys, as pointed out by
jackjack-jj.
|
14 years ago |
samr7
|
317cade95f
|
Enhance reporting of OpenCL errors
|
14 years ago |
samr7
|
d64d355010
|
Checkpoint development of oclvanitygen.
|
14 years ago |
samr7
|
e328e73d6a
|
Add some loop unrolling optimizations to the OpenCL kernel.
Add GPU idle time reporting in verbose mode.
|
14 years ago |
samr7
|
a5ff3bd4ac
|
Checkpoint some optimizations to oclvanitygen.
|
14 years ago |
samr7
|
a8bf7cff08
|
Use two output buffers so that the GPU doesn't trample on the hash
list while the CPU is checking for matches.
|
14 years ago |
samr7
|
19db491cb8
|
Fix rekey synchronization between OpenCL master and CPU search threads.
|
14 years ago |
samr7
|
2dcd8fdff6
|
Display hint message when given impossible prefix.
Split source into multiple files.
Add experimental, unoptimized OpenCL implementation.
Bump version to 0.13
|
14 years ago |