I recently converted one of my dedicated crunchers to linux and noticed that it's only getting v8.10 opencl_nvidia_SoG for x86_64-pc-linux-gnu -- is there something I'm missing that I need for v8.19 like windows_intelx86?

Ask petri33 for his binaries....;)

---

Always refer to the Applications Page to see what the current version numbers are for any particular platform.

Any application listed on that page, and any third-party application, should produce the same result file as any other, and validate against it.

Ok let me rephrase the question: why are there no official 8.19 binaries for linux? I'm guessing it's a matter of resources? I know my way around build-essential; if it's a matter of gitting, making and testing I'd be willing to try to help.

Ok let me rephrase the question: why are there no official 8.19 binaries for linux? I'm guessing it's a matter of resources? I know my way around build-essential; if it's a matter of gitting, making and testing I'd be willing to try to help.

One of the first steps would be having the apps deployed at Beta and having them undergo a round of testing. Then if all went well they would get deployed to main. In another thread Raistmer had mentioned (Urs said debugging finished so some new test binaries could be generated).
That may or may not be related to a 8.19 release. You would have to get into the loop with Urs to find out exactly what the current plans are.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

There's nothing magical about BOINC version numbers - source code SVN revision numbers (r3566 and counting) are far more significant.

We've only reached such high version numbers for Windows because Raistmer has been actively developing his SoG code to run Breakthrough Listen VLARs more efficiently on NVidia GPUs. Every time we put another version up for wider Beta testing, we use up another version number - and every time we find another bug or improvement, we throw it away again.

Raistmer, TBar and Urs are trying to produce a single coherent codebase which compiles, runs, and produces valid results on Windows, OS X and Linux respectively. I wouldn't expect more than one of those to be fast-tracked from Beta to Main before all the bugs are squashed: compilation and running can be confirmed at Beta, but validation testing requires exposure to the full range of tasks at Main. And Windows seems to have been assigned that role, hence the inflated version number.

I noticed it crashed my drivers (NVIDIA-Linux-x86_64-375.20) this afternoon

Nov 19 16:34:03 blue kernel: [16513.904279] NVRM: Xid (PCI:0000:01:00): 32, Channel ID 00000031 intr 00040000
Nov 19 16:34:03 blue kernel: [16513.904646] NVRM: Xid (PCI:0000:01:00): 32, Channel ID 00000031 intr 00040000
Nov 19 16:35:16 blue kernel: [16587.022960] NVRM: Xid (PCI:0000:01:00): 13, Graphics Exception: MISSING_INLINE_DATA
Nov 19 16:35:16 blue kernel: [16587.023245] NVRM: Xid (PCI:0000:01:00): 13, Graphics Exception: ESR 0x404600=0x80000002
Nov 19 16:35:16 blue kernel: [16587.023543] NVRM: Xid (PCI:0000:01:00): 13, Graphics Exception: ChID 0012, Class 0000c1c0, Offset 000001b4, Data 00fffc80
Nov 19 16:37:26 blue kernel: [16716.772541] NVRM: Xid (PCI:0000:01:00): 32, Channel ID 0000000a intr 00040000

NVidia's XID Errors page suggests MISSING_INLINE_DATA means is "Invalid or corrupted push buffer stream."

I hope this doesn't mean I'm to far off the beaten path with Linux -- I'd really like to spend the cost of a Windows license on more hardware.

Evidently 8.10 SoG on Linux is r3430 and 8.19 on Windows is r3528. Looking at this SVN I see 3522 as the only change that jumps out as potentially impacting stability.

I noticed I was on the older version because instances_per_device wasn't showing up in the stdout so my analysis think I'm only running one job at a time. It's really too bad I can't sniff out the app_config.xml gpu_usage scalar if set so I can analyze better without depending on this command-line parameter.