![]() |
|
|
#1 |
|
(loop (#_fork))
Feb 2006
Cambridge, England
2·7·461 Posts |
Has anyone seen this message, and have you any idea how to fix it?
Code:
coeff 12 specialq 1 - 13227 other 876207 - 2102897 [./b40c/radix_sort/../radix_sort/../util/kernel_props.cuh, 81] cudaFuncGetAttributes failed (CUDA error 78: a PTX JIT compilation failed) [sort_engine.cu, 159] sort engine: (CUDA error 78: a PTX JIT compilation failed) Code:
$ nvidia-smi
Mon Apr 2 11:49:19 2018
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 384.130 Driver Version: 384.130 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 108... Off | 00000000:17:00.0 On | N/A |
| 80% 83C P2 174W / 250W | 211MiB / 11169MiB | 100% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 1696 G /usr/lib/xorg/Xorg 43MiB |
| 0 18662 C ...pineapple/cuda-ecm/IVB-20160720/bin/ecm 155MiB |
+-----------------------------------------------------------------------------+
|
|
|
|
|
|
#2 |
|
Tribal Bullet
Oct 2004
5×23×31 Posts |
Which CUDA toolkit version and GPU compute capability is this? It's possible the compile arguments need tweaking for compute capability > 5.0
It's also possible I need to update the source of the CUB library for latter-day cards, there have been new version releases since February. ETA: with the 9.0 toolkit we may have to remove attempts to load the SM 2.0 PTX code, as I think support for compute capability 2.0 is removed now. Last fiddled with by jasonp on 2018-04-02 at 11:42 |
|
|
|
|
|
#3 |
|
(loop (#_fork))
Feb 2006
Cambridge, England
2·7·461 Posts |
This is toolkit version
Code:
pineapple@pineapple:/scratch/L2675A$ /usr/local/cuda-8.0/bin/nvcc --version nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2016 NVIDIA Corporation Built on Tue_Jan_10_13:22:03_CST_2017 Cuda compilation tools, release 8.0, V8.0.61 Code:
selected card has CUDA arch 6.1 Arranging to build cub only for sm=5.2 doesn't seem to have helped. |
|
|
|
![]() |
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| At least one upcoming paper on polsel | Dubslow | Msieve | 0 | 2016-03-16 06:24 |
| Feature request: multithreaded polsel | Andi47 | Msieve | 1 | 2010-02-20 01:16 |
| bad parameters for 153-digit CPU polsel ? | fivemack | Msieve | 8 | 2010-02-08 10:35 |
| Trouble Running Msieve | Sab | Msieve | 4 | 2009-07-07 06:19 |
| Trouble making msieve | CRGreathouse | Msieve | 5 | 2009-04-05 18:29 |