Kept my eye on the first of the new tasks my computer got and it to showed the same behavior. Almost got to 1% and the restarted from 0.0x-something. Did seem odd so I kept watching and it happened a few more times but every time it got a bit longer before resetting to a smaller number again. Let it go for a while and saw that progress was being made so thought it was a small issue between the app and Boinc.
Edit: Just saw it again when a task had just passed 74%, reset to 73.2xx% twice and then got to almost 75% and reset to 74.1xx%. They do seem to finish eventually.
It would seem the GC tasks run a bit longer than we planned for. For the time being I raised the flops estimation and credit for newly generated tasks by a factor of 1.6.
It would seem the GC tasks run a bit longer than we planned for. For the time being I raised the flops estimation and credit for newly generated tasks by a factor of 1.6.
BM
I was just going to post that these WU are taking my i7-920 twice as long but give the same points as STSP ... and while points really have nothing to do with anything it just looks wacky :-)
It would seem the GC tasks run a bit longer than we planned for.
Hmmm, I don't see much of an increase in run time on either my AMD XP 2200+ (uses the SSE app) or my P4 3.0Ghz (uses the SSE2 app).
If anything, the Hierarchical S5 all-sky GW search #6 tasks took even longer at times on both of my systems. I'm now in the 13.5 - 15 hours range, whereas before I was in the 19 to 20s.
If the workunit's contents are quantized and the computing power required is well off the expectations, it sounds like there's a plausible explanation. Presumably three coders and a gerbil are now frantically trying to figure out why the tasks take 1.6-1.8 times longer than expected to complete (possibly platform/architecture-dependent?).
If it takes almost twice as long between new tasks, the servers can probably breathe a sigh of relief.
It would seem the GC tasks run a bit longer than we planned for.
Hmmm, I don't see much of an increase in run time on either my AMD XP 2200+ (uses the SSE app) or my P4 3.0Ghz (uses the SSE2 app).
If anything, the Hierarchical S5 all-sky GW search #6 tasks took even longer at times on both of my systems. I'm now in the 13.5 - 15 hours range, whereas before I was in the 19 to 20s.
I'm getting mixed results here. My Pentium-IV with Windows XP is completing them in about the same amount of time as the 'R6 units. But, my Core 2 Quad Extreme with Linux is taking way longer.
Kept my eye on the first of
)
Kept my eye on the first of the new tasks my computer got and it to showed the same behavior. Almost got to 1% and the restarted from 0.0x-something. Did seem odd so I kept watching and it happened a few more times but every time it got a bit longer before resetting to a smaller number again. Let it go for a while and saw that progress was being made so thought it was a small issue between the app and Boinc.
The task did finish and validate. Think it was this one
Edit: Just saw it again when a task had just passed 74%, reset to 73.2xx% twice and then got to almost 75% and reset to 74.1xx%. They do seem to finish eventually.
Same here! My first
)
Same here!
My first S5GCE-WU (http://einsteinathome.org/workunit/71137248)
got up to 0.8xx% and than back to 0.000%.
After I stopped and restarted the WU it got up to 1.5% and than back to 1%.
Sascha
It would seem the GC tasks
)
It would seem the GC tasks run a bit longer than we planned for. For the time being I raised the flops estimation and credit for newly generated tasks by a factor of 1.6.
BM
BM
RE: It would seem the GC
)
I was just going to post that these WU are taking my i7-920 twice as long but give the same points as STSP ... and while points really have nothing to do with anything it just looks wacky :-)
--------------------------
- Crunch, Crunch, Crunch -
--------------------------
RE: It would seem the GC
)
Hmmm, I don't see much of an increase in run time on either my AMD XP 2200+ (uses the SSE app) or my P4 3.0Ghz (uses the SSE2 app).
If anything, the Hierarchical S5 all-sky GW search #6 tasks took even longer at times on both of my systems. I'm now in the 13.5 - 15 hours range, whereas before I was in the 19 to 20s.
GCE jobs follow a sawtooth
)
GCE jobs follow a sawtooth curve both in progress % and time to completion.
Tullio
If the workunit's contents
)
If the workunit's contents are quantized and the computing power required is well off the expectations, it sounds like there's a plausible explanation. Presumably three coders and a gerbil are now frantically trying to figure out why the tasks take 1.6-1.8 times longer than expected to complete (possibly platform/architecture-dependent?).
If it takes almost twice as long between new tasks, the servers can probably breathe a sigh of relief.
RE: RE: It would seem the
)
I'm getting mixed results here. My Pentium-IV with Windows XP is completing them in about the same amount of time as the 'R6 units. But, my Core 2 Quad Extreme with Linux is taking way longer.
RE: My Pentium-IV with
)
Optimized for older hardware. :P
RE: RE: My Pentium-IV
)
I'll have to get that Celeron MMX of mine back into working order, then ;-)