cuda-z Activity
Brought to you by:
andrew_golovnia
Activity for cuda-z
-
Andriy Golovnya posted a comment on discussion Help
Hi weasleyma, I do not have an access to such a exotic HW like mentioned A100 SMX GPU though I can't test CUDA-Z for this kind of modules ;) I might try to update CUDA SDK to a newer version and hope it will work better with your SMX GPU. No promises though. Best regards, AG
-
weasleyma posted a comment on discussion Help
Dear Developer. I have a 8 * NVIDIA A100 SMX2 GPU card, and install CUDA-11.6 witb NVIDIA Linux driver 510.73.08, when I run ./CUDA-Z-0.10.251-64bit.run I got error of No compatiable CUDA device found, please update your nvidia driver and try again. [root@gpu01 ~]# ./CUDA-Z-0.10.251-64bit.run CUDA-Z Started! CUDA Error: 0000001e unknown error CUDA Present: 1 CUDA Error: 0000001e unknown error CUDA Devices found: 0 Thanks!
-
jshen28 modified a comment on discussion Help
Hello community, I am using cuda-z to get a baseline benchmark of GPU/vGPU. I am trying to figure out if it is possible to get result in text format so it will be a little easier to write a script and automate benchmark process. Any idea? Thank you. Best, jshen28
-
Hello community, I am using cuda-z to get a baseline benchmark of GPU/vGPU. I am trying to figure out if it is possible to get result in text format so it will be a little easier to write a script and automate benchmark process. Any idea? Thank you. Best, jshen28
-
Hello community, I am using cuda-z to get a baseline benchmark of GPU/vGPU. I am trying to figure out if it is possible to get result in text format so it will be a little easier to write a script and automate benchmark process. Any idea? Thank you. Best, jshen28
-
Hello community, I am using cuda-z to get a baseline benchmark of GPU/vGPU. I am trying to figure out if it is possible to get result in text format so it will be a little easier to write a script and automate benchmark process. Any idea? Thank you. Best, jshen28
-
ixfd64 modified a comment on discussion Open Discussion
Version 0.10.251 of CUDA-Z has a very different appearance than previous releases. However, I prefer the original interface as it more closely resembles GPU-Z and is much more readable. Another issue is that the hyperlinks on the "About" tab are hard to see against the blue moon as shown in the attached screenshot. Is there a way to change the application's theme?
-
Version 0.10.251 of CUDA-Z has a very different appearance than previous releases. However, I prefer the original interface as it more closely resembles GPU-Z and is much more readable. Another issue is that the hyperlinks on the "About" tab are hard to see against the blue moon as shown in the attached screenshot. Is there a way to change the application's theme?
-
Version 0.10.251 of CUDA-Z has a very different appearance than previous releases. However, I prefer the original interface as it more closely resembles GPU-Z and is much more readable. Also, the hyperlinks on the "About" tab are hard to see against the blue moon as shown in the attached screenshot. Is there a way to change the application's theme?
-
sefsf posted a comment on discussion Help
Actually crossing borders seems to be a possibility : https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__INTEROP.html I managed to use JCUDA in my app, I'm retrieving the textures real fast but they do not contain the full image.. It's a promising path but so far I can't see what I did wrong.
-
You are right. The direction of data transfer does not metter. If you map GPU memory block to CPU memory you will have the abstraction code to translate your calls in backgroud. I do not know if this is the only limiting factor here, but it shurely a significant one. If you use OpenGL, the use of cudaMemcpy() in your code will be most probably useless. OpenGL and CUDA are two non-crossing subsystems, as I know. You have to find something like this in OpenGL itself.
-
Hello, Using cuda-z I can see that "Device to Host" memory bandwidth is around 2.5 GiB/s on my eGPU Thunderbolt 3 GTX 1080. On an opengl app I am developing, I'm retrieving into CPU memory each frame/texture using http://spout.zeal.co/ My textures are 3840x2160 RGBA, right now it takes around 50 ms to retrieve the texture into client memory. If one compares with the statistics in cuda-z, I should be able to transfer the 31MB in less than 15 ms which is what I need at 60 fps. Spout uses an improved...
-
Hello! Thanks for your help. You cannot copy data from CPU memory to GPU memory with normal memcpy() routine directly. I am not copying from CPU to GPU. I am doing the reverse. From GPU to CPU. I am reading from the GPU memory memcpy is acheived thanks to https://www.khronos.org/registry/OpenGL-Refpages/gl2.1/xhtml/glMapBuffer.xml memory is mapped thanks to the fact we have virtual memory in our OSes/hardware but I guess this doesn't change you point Try to use native bus transfer calls in your code...
-
mahmood naderan posted a comment on discussion Help
Hi and thanks for the tool. Regarding the integer performance, I see this kernel: __global__ void CZCudaCalcKernelInteger32( void *buf /*!<[in] Data buffer. */ ) { int index = blockIdx.x * blockDim.x + threadIdx.x; int *arr = (int*)buf; int val1 = index; int val2 = arr[index]; int i; for(i = 0; i < CZ_CALC_BLOCK_LOOPS; i++) { CZ_CALC_IMAD32_256(val1, val2); CZ_CALC_IMAD32_256(val1, val2); CZ_CALC_IMAD32_256(val1, val2); CZ_CALC_IMAD32_256(val1, val2); CZ_CALC_IMAD32_256(val1, val2); CZ_CALC_IMAD32_256(val1,...
-
Hi! You cannot copy data from CPU memory to GPU memory with normal memcpy() routine directly. The reason for this is that beween CPU and GPU you have a PCIe bus. You have to use a special bus transfer routines like cudaMemcpy(). If you ARE able to use normal memcpy() routine to transfer data to GPU memory, you are clearly dealing with some emulation layer which detects change you made in the CPU memory page and translates them (probably not efficient) to bus transfer calls. Try to use native bus...
-
I could find by reading the code that cuda-z uses cudaMemCpy (http://horacio9573.no-ip.org/cuda/group__CUDART__MEMORY_g48efa06b81cc031b2aa6fdc2e9930741.html) Does anyone know if I should be achieving the same performance with a PBO transfer + memcpy sse2?
-
Hello, Using cuda-z I can see that "Device to Host" memory bandwidth is around 2.5 GiB/s on my eGPU Thunderbolt 3 GTX 1080. On an opengl app I am developing, I'm retrieving into CPU memory each frame/texture using http://spout.zeal.co/ My textures are 3840x2160 RGBA, right now it takes around 50 ms to retrieve the texture into client memory. If one compares with the statistics in cuda-z, I should be able to transfer the 31MB in less than 15 ms which is what I need at 60 fps. Spout uses an improved...
-
Anton posted a comment on ticket #37
Hi, Andriy! Unfortunately, the workaround you suggested didn't help, and the problem remains. AK
-
Andriy Golovnya modified ticket #36
CUDA-Z not opening
-
Felipe Bachian posted a comment on ticket #36
It took so long for this answer. I already made a clean format on my PC and now everything is working fine. Thanks anyway.
-
No CUDA Devices found (sudo required) Mac
-
Hi! I can't offer my support here because I have no real mac (and no hackintosh) in my possession for now. I'll try to fix CUDA-Z for mac or I'll drop it completly. WBR, Ag
-
Anonymous posted a comment on ticket #36
Do they use CUDA or DirectX for their calculations? I have no information on that.
-
Build error
-
Build error
-
Hi! This is a bug in CUDA Toolset sinse version 8.0 or so. It seams they refuse to fix it till now. I happens because of size of sible linear test blocks like this: for(i = 0; i < CZ_CALC_BLOCK_LOOPS; i++) { CZ_CALC_FMAD_256(val1, val2); CZ_CALC_FMAD_256(val1, val2); CZ_CALC_FMAD_256(val1, val2); CZ_CALC_FMAD_256(val1, val2); CZ_CALC_FMAD_256(val1, val2); CZ_CALC_FMAD_256(val1, val2); CZ_CALC_FMAD_256(val1, val2); CZ_CALC_FMAD_256(val1, val2); } If you reduce them to 2 lines it works (or at least...
-
Build error
-
Faceworks from Nvidia is working normaly. https://www.geforce.com/games-applications/pc-applications/faceworks
-
Do other CUDA-enabled program work at your PC?
-
Anonymous posted a comment on ticket #36
Thanks Andriy, but it seems all version has the same issue for me. I tried v0.9.231.
-
Thanks Andriy, but it seems all versions has the same issue. I tried v0.9.231.
-
Try any other version from here: https://sourceforge.net/projects/cuda-z/files/cuda-z/ Have a good day!
-
I'm running on Windows 10 Pro (build 17134.228). I just downloaded the CUDA-Z version that's available on the website (v0.10.251 64bits I believe). Do you recommend any other version?
-
I can't reproduce this behavior. What OS bitness you run it on? What exact version and bitness of CUDA-Z you trying to run? Do other versions of CUDA-Z are starting ok?
-
I only have Windows Defender. And yes, I tried to run CUDA-Z with Defender disabled.
-
It can be an antivirus issue. Did you try to disable your antivirus before strating CUDA-Z?
-
CUDA-Z not opening
-
Anonymous created ticket #35
No CUDA Devices found (sudo required) Mac
-
It seams you are right! WBR, AG
-
Performance / Device To Host value decreasing
-
Anonymous posted a comment on ticket #34
Sorry, found the possible reason: #Bug10
-
Anonymous created ticket #34
Performance / Device To Host value decreasing
-
drop information to stdout instead of stderr.
-
update upx for windows and linux
-
fix generation of resource file for windows
-
improve html5 compatibility
-
improve information output
-
cli development
-
fix the badge display based on device name
-
print and export of info
-
preparing for print/export of info
-
cli development
-
cli development
-
beginning of cli development
-
Further separation textual information from CZDialog class with a help of decoder.
-
Begin a separation textual information from CZDialog class with a help of decoder.
-
False positive The CUDA-Z binary is compressed with UPX utility. Perhaps some viruses use this utility too. My antivirus does not react like this. --AG
-
Alan Insley posted a comment on discussion Help
https://svwh.dl.sourceforge.net/project/cuda-z/cuda-z/0.10/CUDA-Z-0.10.251-64bit.exe
-
That's what I meant above. Thanks for the link!
-
Alexander Stohr posted a comment on discussion Help
see the very top dotted list of this chapter for a match between CUDA SDK version and the supported platforms. https://en.wikipedia.org/wiki/CUDA#GPUs_supported
-
DRIVER DLL is delivered within a driver package. If you want to try to trick your minor you may copy nvcuda.dll from older driver to minor's program folder but I do believe it will not work properly with new driver. -- Andriy
-
Greg Iverson posted a comment on discussion Help
Thanks for the quick reply Andriy! That clears things up! I will just pay attention to the DLL version :) It appears when I load an older CUDA SDK (like 6.5), it installs fine, but your program still shows 9.10 as active, and the miner software doesn't seem to know it's there or be able to find the newer version.it's possible that I could drop the DLL's into the mining directory, but I wouldn't know which one's it wanted :) (I could also totally remove the entire video driver package (to get rid...
-
Hi Greg! You may look at DRIVER DLL as the CUDA DLL version your NVIDIA DRIVER delivered with. In contrary to this the RUNTIME DLL is the DLL CUDA-Z COMPILED WITH and comes from nvidia toolkit version used. DRIVER DLL = 9.10 means your NVIDIA DRIVER supports CUDA (up to) 9.10 RUNTIME DLL = 6.50 means the CUDA-Z APP compiled with compiler version 6.50 As I remember nvidia obsoletes some old CUDA HW with each new driver release. For example CUDA HW < 2.0 does not work since CUDA driver 7.0 (or 7.5)....
-
Hey all! I'm curious what the difference is between the DRIVER DLL and the RUNTIME DLL as reported by CUDA-Z...? I just loaded the latest Nvidia drivers, and I know they come with CUDA 9, so I get what the DRIVER DLL is (as it shows 9.10), but the RUNTIME shows 6.50...where did that come from? I ask because I have not done any separate CUDA SDK's installs. Do you think this RUNTIME 6.5 ver was installed via this latest Nvidia driver install along with cuda 9.10? After I installed the latest Nvidia...
-
Here is a list of CUDA enaled GPUs https://developer.nvidia.com/cuda-gpus good day, AG
-
puzl posted a comment on discussion Help
how do i check if my graphic card have cuda or not? not in the nvidia site but in cuda-z thanks
-
Switch report output from HTML4 to HTML5.
-
Add a GPU Architecture name to the output
-
Hi! It might be a cuda driver issue. Try to update the nvidia driver package and try again. If the problem persists let me know. WBR, AG
-
Hangs on 'Getting information' with GTX 1080 in TB3 enclosure
-
Anonymous created ticket #33
Hangs on 'Getting information' with GTX 1080 in TB3 enclosure
-
Add support of QString to logging
-
Adding some basic commandline mode support.
-
Windows 10 P100 Crash on Startup
-
Adware embedded in installer??
-
nVidia GTX 560 Support
-
Does not work on windows 7
-
can't load driver linux version
-
No CUDA devices detected
-
Hi! I don't see a CUDA-Z bug here. It's most probably a bug in operating system how it handles PCIe over TB3 enumeration and how it works with NVIDIA driver together. Actually you did not mention the OS in your bugreport. If you have more information on this issue please let me know, especially if CUDA-Z can be improved to handle this situation better. With best regards, AG
-
Anonymous created ticket #32
No CUDA devices detected
-
Add Volta cores count.
-
Hi! Without X11 will be hard as fer as this utility is Qt based. Perhaps it makes more sense to write a new utility which is puerly and cleanly console based.
-
Anonymous posted a comment on ticket #29
Basically, it could be useful to dump to the console what is also exported from the GUI, or to offer an --export parameter. It would be nice to support running it without X11, then.
-
Anonymous modified a comment on ticket #9
Does this mean improvement ? p/cuda-z/code - Revision 272: /trunk
-
Anonymous modified a comment on ticket #9
Do you mean this ? http://www.nvidia.com/object/mac-driver-archive.html
-
Anonymous posted a comment on ticket #9
CUDA Toolkit 8.0.61 is out for all OS's...
-
Anonymous posted a comment on ticket #9
Do you this ? http://www.nvidia.com/object/mac-driver-archive.html
-
Anonymous posted a comment on ticket #9
Does this mean improuvment ? p/cuda-z/code - Revision 272: /trunk
-
Add UPX 3.93 for Linux
-
Add UPX 3.93 for Windows
-
Make CUDA-Z compile for CUDA 8.0
-
Thanks for reminding. See commit https://sourceforge.net/p/cuda-z/code/269/ --Ag
-
Add 6.x CM architectures.
-
Nan Xiao posted a comment on discussion Open Discussion
Hi Andriy, When reading ConvertSMVer2Cores implementation: #define ConvertSMVer2Cores(major,...
-
Windows 10 P100 Crash on Startup
-
Hi! Thanks for reporting. It seem like it's a bug in nvcuda.dll. It is your NVIDIA...
-
Anonymous created ticket #31
Windows 10 P100 Crash on Startup
-
i checked those number and found it mainly mentioned for MacOS X. when going for...
-
Anonymous posted a comment on ticket #9
Version 8.0.63 is out !
1 >
