E5310 and -bigadv?

[RAD_MAN]

I'm trying to get my folding game back on after a recent break...

I have a dual xeon E5310 (quad core 1.6 GHz. x2) server running server 2008 R2. I tried running -bigadv on it and she seems to want to come in just over 4 days. which i'm pretty sure isn't gunna cut it. Should i just run -advmethods on that guy instead? Thanks.

 

[Tobit]

-advmethods tells the servers you want public beta work units. Unfortunately, there aren't any at the moment. I'd just remove the -bigadv flag and run the client with -smp so it will crunch the regular SMP work units.

 

[APOLLO]

You won't get -bigadv completing within the preferred deadline if you happen to receive a P2684 WU. I have a dual E5310 machine OCed to ~2.4GHz. It cannot make the bonus deadline processing a P2684. Minimum tentative frequency for a dual Clovertown system to successfully complete all -bigadv WUs within the preferred deadline would be around 2.9-3.0GHz. Approximately 2.8GHz minimum for dual Harpertown processors. This is assuming your system has a quad channel memory capable chipset (5000X, 5000P, etc). If not, then you will need to increase the minimum frequency requirement I estimated by at least 100MHz..

 

[Tobit]

You won't get -bigadv completing within the preferred deadline if you happen to receive a P2684 WU. I have a dual E5310 machine OCed to ~2.4GHz. It cannot make the bonus deadline processing a P2684. Minimum tentative frequency for a dual Clovertown system to successfully complete all -bigadv WUs within the preferred deadline would be around 2.9-3.0GHz. Approximately 2.8GHz minimum for dual Harpertown processors. This is assuming your system has a quad channel memory capable chipset (5000X, 5000P, etc). If not, then you will need to increase the minimum frequency requirement I estimated by at least 100MHz..

Apollo confirming what I should have said...

 

[RAD_MAN]

hehe.. i figured, i somehow figure i am not going to be able OC those CPUs to almost double speed on an Intel branded mainboard (S5000PSLsata). i've already switched back to regular SMP and i'll let it ride. it has what i think are 2 x16 PCIe slots (8x throughput)... would it be worth while to put some GPU crunching video cards in there? dunno if the powersupply in that server chassis would support those though.

 

[Tobit]

it has what i think are 2 x16 PCIe slots (8x throughput)... would it be worth while to put some GPU crunching video cards in there? dunno if the powersupply in that server chassis would support those though.

fire those retron torpedoes! Well, there is only one way to find out.

 

[APOLLO]

Yes, I have most of my systems running GPU clients. Whether the PSU can handle the video cards will depend as much on the type and number of cards as the PSU's capacity to provide adequate power.

 

[10e]

You won't get -bigadv completing within the preferred deadline if you happen to receive a P2684 WU. I have a dual E5310 machine OCed to ~2.4GHz. It cannot make the bonus deadline processing a P2684. Minimum tentative frequency for a dual Clovertown system to successfully complete all -bigadv WUs within the preferred deadline would be around 2.9-3.0GHz. Approximately 2.8GHz minimum for dual Harpertown processors. This is assuming your system has a quad channel memory capable chipset (5000X, 5000P, etc). If not, then you will need to increase the minimum frequency requirement I estimated by at least 100MHz..

Thanks for the info APOLLO. It's nice to see that F@H projects antiquated "minimum requirements" are now made redundant.

Some of the FAH fanboys on foldingforum have been educated that the 8 true core, 2.4Ghz minimum is no longer good enough.

 

[APOLLO]

Thanks for the info APOLLO. It's nice to see that F@H projects antiquated "minimum requirements" are now made redundant.

All thanks to the venerable P2684.

Some of the FAH fanboys on foldingforum have been educated that the 8 true core, 2.4Ghz minimum is no longer good enough.

The P2684 is solely responsible for this. It is deleterious on older systems that they cannot meet the preferred deadline because of it. Were it not for that fact and we had only the P2685, P2686 and P2692 to fold -bigadv, we could easily have reduced the minimum system requirements a couple of hundred MHz. That is why I term the P2684 a *control* WU, and I do not mean that in any way metaphorically but rather euphemistically.

 

[APOLLO]

hehe.. i figured, i somehow figure i am not going to be able OC those CPUs to almost double speed...

Not double the speed but you know you can still OC these processors, right? No one who is running E5310, E5320, etc. Xeons should have them operating at their factory clocks. It is a travesty to do so. Your E5310s are begging to have the BSEL mod performed for a stately 2.0GHz. That's a 25% increase from the stock frequency. If your board supports the higher FSB and it does, there should be no problems and no risks.

Look here for details and specifically at this pic.

..on an Intel branded mainboard (S5000PSLsata)..

I had that board or one very similar to it briefly in my hands and did not like it (no offense). I switched to Supermicro after that experience and never looked back. I also have one Tyan S-771 board but prefer the Supermicros because they're more compact and OCing seems to be more achievable.

 

[RAD_MAN]

I keep getting crashes on this box for some reason, I had also gotten some crashes when I was trying -bigadv so that was another reason I switched to just -smp. I can run prime95 max heat / CPU for days without issue but I keep getting intermittant crashes with my E5310 box. My X5550 dual xeon box seems to be fine running whatever I throw at it. Any tips?

[04:59:28] Completed 25000 out of 500000 steps (5%)
[05:05:09] Completed 30000 out of 500000 steps (6%)
[10:30:19] - Autosending finished units... [October 18 10:30:19 UTC]
[10:30:19] Trying to send all finished work units
[10:30:19] + No unsent completed units remaining.
[10:30:19] - Autosend completed
[16:30:20] - Autosending finished units... [October 18 16:30:20 UTC]
[16:30:20] Trying to send all finished work units
[16:30:20] + No unsent completed units remaining.
[16:30:20] - Autosend completed
[22:30:21] - Autosending finished units... [October 18 22:30:21 UTC]
[22:30:21] Trying to send all finished work units
[22:30:21] + No unsent completed units remaining.
[22:30:21] - Autosend completed
[02:36:05] CoreStatus = C0000029 (-1073741783)
[02:36:05] Client-core communications error: ERROR 0xc0000029
[02:36:05] Deleting current work unit & continuing...
[02:36:19] Trying to send all finished work units
[02:36:19] + No unsent completed units remaining.
[02:36:19] - Preparing to get new work unit...
[02:36:19] Cleaning up work directory
[02:36:19] + Attempting to get work packet

 

[amdgamer]

I believe that is a similar(if not exact) error I was getting on folding2 before a dimm slot shorted out. I beileve it is tied to bad or unstable memory whether it is system ram or on the cpu cache.

Someone more knowledgable will have to give you advice, but are you doing any overclocking of the processor or the ram? Try running the Windows memory diagnostic for at least several passes.

I keep getting crashes on this box for some reason, I had also gotten some crashes when I was trying -bigadv so that was another reason I switched to just -smp. I can run prime95 max heat / CPU for days without issue but I keep getting intermittant crashes with my E5310 box. My X5550 dual xeon box seems to be fine running whatever I throw at it. Any tips?

[04:59:28] Completed 25000 out of 500000 steps (5%)
[05:05:09] Completed 30000 out of 500000 steps (6%)
[10:30:19] - Autosending finished units... [October 18 10:30:19 UTC]
[10:30:19] Trying to send all finished work units
[10:30:19] + No unsent completed units remaining.
[10:30:19] - Autosend completed
[16:30:20] - Autosending finished units... [October 18 16:30:20 UTC]
[16:30:20] Trying to send all finished work units
[16:30:20] + No unsent completed units remaining.
[16:30:20] - Autosend completed
[22:30:21] - Autosending finished units... [October 18 22:30:21 UTC]
[22:30:21] Trying to send all finished work units
[22:30:21] + No unsent completed units remaining.
[22:30:21] - Autosend completed
[02:36:05] CoreStatus = C0000029 (-1073741783)
[02:36:05] Client-core communications error: ERROR 0xc0000029
[02:36:05] Deleting current work unit & continuing...
[02:36:19] Trying to send all finished work units
[02:36:19] + No unsent completed units remaining.
[02:36:19] - Preparing to get new work unit...
[02:36:19] Cleaning up work directory
[02:36:19] + Attempting to get work packet

 

[APOLLO]

I agree 100% with amdgamer there is a very high probability it is a memory related issue. These systems use FB-DIMMs which run hotter than most other RAM, and memory as well as board manufacturers strongly recommend active cooling for these modules. Since the SMP client uses a fair amount of memory, your modules might be overheating if there is inadequate cooling for continuous full load situations. Another possibility is faulty RAM. Try swapping for other modules or testing each module one at a time to determine if there is a problem.