Cuda on F11 x86_64

[kramulous]

Does anybody have it running?

I get:
/usr/local/cuda/bin/../include/math_functions.h:404: error: inline function ‘int __signbit(double)’ cannot be declared weak

I've tried
gcc (GCC) 4.4.0 20090506 (Red Hat 4.4.0-4)
gcc34 (GCC) 3.4.6 20060404 (Red Hat 3.4.6-13)
icc (ICC) 11.0 20081105

uname -a:
Linux hahah.haha.ha 2.6.29.5-191.fc11.x86_64 #1 SMP Tue Jun 16 23:23:21 EDT 2009 x86_64 x86_64 x86_64 GNU/Linux

[kramulous]

AARRRGGHHHH!!

Gotta stop finding the solution after posting. Getting sick of looking stupid.

Found the solution here:
http://forums.nvidia.com/index.php?s...t=#entry555518

Essentially:
To compile the sample-projects on F11 w/gcc34 you can change the NVIDIA_CUDA_SDK/common/common.mk lines:
CXX := g++34
CC := gcc34
LINK := g++34 -fPIC
and
NVCCFLAGS := --compiler-bindir /tmp

Then create a symbolic link in /tmp:
ln -s /usr/bin/gcc34 gcc

[scottro]

Firstly, thanks for coming back and *posting* the solution. You'd be amazed how many people just delete their thread with a never mind fixed it type message.

Secondly, don't worry. It's one of those things--for some reason, it's far more likely that you'll find the solution after posting the question.

[pawelk]

kramulous,

I recently tried installing cuda on my p4 but didn't get past installing the NVIDIA Driver. (I made sure to download the Linux 32-bit version.) I got a bunch of suggestions, from the install program, as to the cause but nothing that seemed particularly straight forward to remedy. How was it for you installing/building the drivers? I also made sure to read and follow the relevant README provided by nVidia.

thanks,
pawelk

[kramulous]

Yeah, I had issues installing. In the end, I removed everything *nividia* from my system (including all of that kmod stuff - not sure if that was a smart move - but it worked) and downloaded the latest linux_64 driver for my card (8800GTX). I had to reboot a few times given that when I ctrl-alt F% I got blank screens. At least boot times are pretty good.

I then installed everything:
1) CUDA Driver: NVIDIA Driver for Linux (RHEL 5.3) 185.18.14 (which is exactly the same name for the previous driver I installed)
2) CUDA Toolkit: CUDA Toolkit 2.2 for Linux (RHEL 5.3)
3) CUDA SDK: CUDA SDK 2.2.1 code samples for Linux (RHEL 5.3)

The SDK required those changes above to build.

[pawelk]

Thanks for that. In the meantime I made a tactical retreat and downgraded back to fc10. In my case I don't have a cuda enabled card but understood that emulation is supported. I had assumed I needed the drivers for emulation (thought I read that in the nvidia docs somewhere) but then found on the cuda forum that the driver is not needed for emulation only coding. So I have now (no driver) but have installed the toolkit and sdk. Running a "make emu=1" compiled two of the example programs. Another had errors that are probably due as per some of the release notes.. Executing the two examples works. - For the record.

pawelk