BitBOINC
Thread 'Sphinx error?'
Questions and Answers : Getting started & support : Sphinx error?
Message board moderation
| Author | Message |
|---|---|
 dskagcommunityNew member  Send message Joined: 29 Sep 26 Posts: 14 Credit: 494,000 RAC: 78,545    
|
Tested one shinx between the other applications and it errord out: https://bitboinc.athena.org.tr/result.php?resultid=1081206 ---------- 24/7 Crunching since 2011 ----------- DSKAG Austria: http://www.dskag.at 
|
|
New member Send message Joined: 27 Sep 26 Posts: 82 Credit: 877,800 RAC: 67,962      
|
Since I couldn't resolve the issue for now, I disabled the applications that were causing problems. 
|
|
New member Send message Joined: 28 Sep 26 Posts: 3 Credit: 1,085,900 RAC: 482,434   
|
My 3060m, Mint 22 returned a few valid results which I also got credit for but the gpu appears to mainly just sit there almost at idle. Could you check the following task if it produced a meaningful result: sphinx_eccp131_1790800503_7983_1 Workunit: 1231506 Created: 1 Oct 2026, 11:10:07 UTC Sent: 1 Oct 2026, 15:38:47 UTC Deadline: 15 Oct 2026, 15:38:47 UTC Received: 1 Oct 2026, 16:19:13 UTC Server status: Completed Result: Successful Client status: Finished Computer ID: 89 Exit status: 0 (0x00000000) Run time: 27 min 28 sec CPU time: 27 min 28 sec Credit: 128.77 Validation status: Valid Priority: 0 Device peak FLOPS: 2,408.41 GFLOPS Application version: Sphinx (Certicom ECC Challenge discrete-log search) v0.02 (cuda_turing_plus) Platform: x86_64-pc-linux-gnu Peak working set size: 149.52 MB Peak disk usage: 17.22 KB Stderr: <core_client_version>8.3.0</core_client_version> <![CDATA[ <stderr_txt> Sphinx GPU (Certicom ECCp-131 discrete-log search, CUDA) v1 bitboinc -- https://bitboinc.athena.org.tr/ coded by Alperen Yavuz 2026-10-01 18:16:05 (345217): called boinc_finish(0) </stderr_txt> ]]> |
|
New member Send message Joined: 27 Sep 26 Posts: 82 Credit: 877,800 RAC: 67,962
|
The result is valid. I will correct such strange behaviors.
|
|
New member Send message Joined: 29 Sep 26 Posts: 22 Credit: 0 RAC: 0 |
Found the actual cause of "the provided PTX was compiled with an unsupported toolchain" from the original report: the old build embedded a PTX fallback (for forward compatibility with future GPU architectures) compiled by a newer CUDA toolkit than some installed drivers could JIT-compile -- a toolkit/driver mismatch, not a real hardware incompatibility. Rebuilt Sphinx's CUDA app tonight targeting Turing through Hopper (sm_75-sm_90) as real compiled machine code only, with no PTX fallback embedded at all -- so there's nothing for an older driver to fail to JIT in the first place. Trade-off: no PTX forward-compat path for GPUs newer than Hopper (Blackwell etc.), but those are rare among current volunteers and can be added back with a dedicated newer-toolchain build later if that changes. Live now; let us know if you still see this. |
|
dskagcommunity New member Send message Joined: 29 Sep 26 Posts: 14 Credit: 494,000 RAC: 78,545
|
Running just fine ^^ great ---------- 24/7 Crunching since 2011 ----------- DSKAG Austria: http://www.dskag.at
|
|
dskagcommunity New member Send message Joined: 29 Sep 26 Posts: 14 Credit: 494,000 RAC: 78,545
|
ohhh..it runs on a 3060 fine, but 5070ti errors out O.o Still brakes down like my result on my start of the thread. Somethin to do with the 5xxx series? https://bitboinc.athena.org.tr/result.php?resultid=3358386 ---------- 24/7 Crunching since 2011 ----------- DSKAG Austria: http://www.dskag.at
|
