Error: invalid device symbol

D0ngerin0 commented 5 years ago

help Hey guys! I'm getting this error. I got a video card Gefore GTX 760, any clue why this is happening? I tried to change from default 30 ,to 60, to 50, to 90 and nothing is working out. What am I doing wrong??

https://imgur.com/x4Wj4j4

D0ngerin0 commented 5 years ago

I am also Running Windows 7 SP1 Does this somehow affect?

D0ngerin0 commented 5 years ago

i hope somebody can help! =(

bill32767 commented 5 years ago

Run on windows 10 latest version x64. Install CUDA toolkit and visual studio all the latest version and you should not face any issues.

On Fri, Sep 13, 2019, 1:29 AM D0ngerin0 notifications@github.com wrote:

i hope somebody can help! =(

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/brichard19/BitCrack/issues/194?email_source=notifications&email_token=AHWRFWLWIER7PCO4H3QNAUTQJLNGTA5CNFSM4IWICSC2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD6TUJ7A#issuecomment-531055868, or mute the thread https://github.com/notifications/unsubscribe-auth/AHWRFWL54NYYJNRYXCZRKOTQJLNGTANCNFSM4IWICSCQ .

D0ngerin0 commented 5 years ago

…

Yea, I installed Windows 10 and x64. I installed everything, still having the same issue.

D0ngerin0 commented 5 years ago

+-----------------------------------------------------------------------------+ | Processes: GPU Memory | | GPU PID Type Process name Usage | |=============================================================================| | 0 Not Supported | +-----------------------------------------------------------------------------+

bill32767 commented 5 years ago

Can you a drop a screenshot of your problem? did you check your firewall and/or anti virus?? To be sure it does not interfere with the functions of bitcrack......

On Fri, Sep 13, 2019 at 12:39 PM D0ngerin0 notifications@github.com wrote:

…

Yea, I installed Windows 10 and x64. I installed everything, still having the same issue.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/brichard19/BitCrack/issues/194?email_source=notifications&email_token=AHWRFWPRUO2BTSXM4RB5EBTQJPT5TA5CNFSM4IWICSC2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD6V77EA#issuecomment-531365776, or mute the thread https://github.com/notifications/unsubscribe-auth/AHWRFWI5RPCO3JYADRBOYCTQJPT5TANCNFSM4IWICSCQ .

superviech5000 commented 5 years ago

I have the same Problem on Ubuntu 16.04 and 18.04, opencl works though

Bildschirmfoto vom 2019-10-10 11-37-18

alkiczx commented 4 years ago

Did anyone had any progress with this? On AMD Ryzen 7 + Nvidia 1050 or 2060 everything is fine however when there is Intel CPU I have this problem on multiple cpu/gpu combinations. Edit: actually it works on celeron with 1060 on windows but same computer on ubuntu gives the error

The error is caused by line 182 of CudaDeviceKeys.cu in function CudaDeviceKeys::initializePublicKeys err = cudaMemcpyToSymbol(_xPtr, &_devX, sizeof(unsigned int *));

alkiczx commented 4 years ago

so after some research linux problem it is caused by incompatible architecture settings in Makefile. Fix is replacing line 18 in Makefile: NVCCFLAGS=-std=c++11 -gencode=arch=compute_${COMPUTE_CAP},code=\"sm_${COMPUTE_CAP}\" -Xptxas="-v" -Xcompiler "${CXXFLAGS}" with following line for 1060/1070/1080 cards: NVCCFLAGS=-std=c++11 -arch=sm_30 -gencode=arch=compute_30,code=sm_30 -gencode=arch=compute_50,code=sm_50 -gencode=arch=compute_52,code=sm_52 -gencode=arch=compute_60,code=sm_60 -gencode=arch=compute_61,code=sm_61 -gencode=arch=compute_61,code=compute_61 -Xptxas="-v" -Xcompiler "${CXXFLAGS}"

Windows problem was caused by card with lower CUDA core than 35.

bill32767 commented 4 years ago

Wow!!! Nice one. Kudos Bro. Can this work on a Linux VBox?

On Thu, Nov 21, 2019 at 11:04 AM alkiczx notifications@github.com wrote:

so after some research linux problem it is caused by incompatible architecture settings in Makefile. Fix is replacing line 18 in Makefile: NVCCFLAGS=-std=c++11 -gencode=arch=compute_${COMPUTECAP},code=\"sm${COMPUTE_CAP}\" -Xptxas="-v" -Xcompiler "${CXXFLAGS}" with following line for 1060/1070/1080 cards: NVCCFLAGS=-std=c++11 -arch=sm_30 -gencode=arch=compute_30,code=sm_30 -gencode=arch=compute_50,code=sm_50 -gencode=arch=compute_52,code=sm_52 -gencode=arch=compute_60,code=sm_60 -gencode=arch=compute_61,code=sm_61 -gencode=arch=compute_61,code=compute_61 -Xptxas="-v" -Xcompiler "${CXXFLAGS}"

Windows problem was caused by card with lower CUDA core than 35.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/brichard19/BitCrack/issues/194?email_source=notifications&email_token=AHWRFWNFX7XYFLECCLISPUDQU3LSVA5CNFSM4IWICSC2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEE3JXCA#issuecomment-557226888, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHWRFWKVN547YDOGKC2ZC7TQU3LSVANCNFSM4IWICSCQ .

alkiczx commented 4 years ago

Not for Virtualbox, it does not have any direct HW access possibility and your GPU will not be visible for guest OS. Some virtual machines have that possibility, but it is really not easy task. Use system native binary (linux or windows) instead.

Jolly-Pirate commented 3 years ago

so after some research linux problem it is caused by incompatible architecture settings in Makefile. Fix is replacing line 18 in Makefile: NVCCFLAGS=-std=c++11 -gencode=arch=compute_${COMPUTE_CAP},code=\"sm_${COMPUTE_CAP}\" -Xptxas="-v" -Xcompiler "${CXXFLAGS}" with following line for 1060/1070/1080 cards: NVCCFLAGS=-std=c++11 -arch=sm_30 -gencode=arch=compute_30,code=sm_30 -gencode=arch=compute_50,code=sm_50 -gencode=arch=compute_52,code=sm_52 -gencode=arch=compute_60,code=sm_60 -gencode=arch=compute_61,code=sm_61 -gencode=arch=compute_61,code=compute_61 -Xptxas="-v" -Xcompiler "${CXXFLAGS}"

Windows problem was caused by card with lower CUDA core than 35.

Also, without touching the Makefile, you can achieve the same by passing the compute cap like this (for example a GTX 1650):

make BUILD_CUDA=1 COMPUTE_CAP=75

TheDukeOfChutney commented 3 months ago

For future reference, you need to align the compute cap value with your cards architecture. I have a 3060, thus Ampere architecture, and 80 was the correct value. https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/index.html#gpu-compilation

brichard19 / BitCrack

Error: invalid device symbol #194