Weird error! Initialization interactions? Different results when running from terminal

Developer Tools Nsight Eclipse Edition
1

Hello ,

I have this problem.When I am running the code through terminal ,it gives me sometimes another value for my result.

In the nsight debugger I am not receving any errors (except when running the code at some part it shows

warning: Unmapped DWARF Register #83886081 encountered.
cuda-gdb/7.2/gdb/regcache.c:181: internal-error: register_size: Assertion `regnum >= 0 && regnum < (gdbarch_num_regs (gdbarch) + gdbarch_num_pseudo_regs (gdbarch))' failed.
A problem internal to GDB has been detected,
further debugging may prove unreliable.
Quit this debugging session? (y or n) [answered Y; input not from terminal]
cuda-gdb/7.2/gdb/regcache.c:181: internal-error: register_size: Assertion `regnum >= 0 && regnum < (gdbarch_num_regs (gdbarch) + gdbarch_num_pseudo_regs (gdbarch))' failed.
A problem internal to GDB has been detected,
further debugging may prove unreliable.
Create a core file of GDB? (y or n) [answered Y; input not from terminal]

)

but the code executes just fine!

If I run cuda-memcheck from terminal it shows me 30 errors:

Program hit error 29 on CUDA API call to cudaFree
Program hit error 29 on CUDA API call to cudaEventDestroy 
Program hit error 29 on CUDA API call to cudaThreadSynchronize

Any ideas??

2

Hmm , I noticed that all the errors produced by memcheck are related to a CULA function call.If I ommit it all the errors are gone.But I still have problem with the results.

Also , once in a while I am receiving wrong results from nsight too…

3

Any ideas?

4

Hi ggeo,

Which version of the CUDA toolkit are you using? Are you able to post the steps and code that reproduce this issue?