OpenCL BitCrack (Experimental)

OpenCL code has been embedded into the executable. No longer required to keep .cl files with the .exe.

CUDA error checking fix

-Some CUDA runtime errors were not being checked for

Minor fixes

-Some CUDA errors were not being caught
-Replace 'unsigned long long' with 'uint64_t'
-Comment out self-test, uncomment if required for debugging
-Enforce limit on max 1024 threads per block (This is the limit in CUDA)

Core detection bug fix

-Fixed #31, proper number of CUDA cores not detected on CUDA 6.1, 6.2 devices

6.1 devices: Nvidia TITAN Xp, Titan X, GeForce GTX 1080 Ti, GTX 1080, GTX 1070 Ti, GTX 1070, GTX 1060, GTX 1050 Ti, GTX 1050, GT 1030, MX150

6.2 devices: Jetson TX2, DRIVE PX 2

Critical bloom filter bug fix

Fixes a bug where the small bloom filter (< 2^32 bits) was not looking up the correct values.

0.17

Merge branch 'master' of https://github.com/brichard19/BitCrack

CPU usage fix, CUDA 9.2 and 64-bit only

-Lowers the CPU usage while kernels are running. The CPU usage should be near 0%.
-Upgraded to Visual Studio 2017 and CUDA Toolkit 9.2
-Because of CUDA 9.2, devices with compute 2.x and lower are no longer supported
-Also note: this project is 64-bit only now. 32-bit Windows executables will not be released.

Initialize on GPU

This enhancement performs the initialization step on the GPU instead of the CPU. It is much faster and allows more points to be initialized on the device in a short amount of time.

The benefit of having more points on the GPU is that each thread can process more points in parallel (because of batch inversion) which increases performance. Set the number of points per thread using the -p option e.g. -p 256. The current default is 32, which is too low.

Note that a higher value will increase kernel run time, and may cause a crash if the GPU is also used for display.

Larger bloom filters, better output

-Can handle arbitrarily large bloom filters, provided there is enough memory.
-Added more detailed program output

Dynamic bloom filter

-Bloom filter automatically resizes according to how many addresses there are. Chance of false-positives is 1 in 1 billion
-Display GPU memory usage