![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Just got a Project 2684
- Thread starter Wesleynator
- Start date
- Joined
- Oct 23, 2007
- Messages
- 6,105
Both of mine are VM (as mentioned earlier
), specifically linuxrouter w/VMware Player 3.0.I haven't had any flush the last few times but the TPF and ETA is a turn-off. I'd rather drop some PPD and run SMP2 at this point, rather than let a 2684 chug for 3.5 days and potentially lose all the work.
Anyone know how to check the success rate of -bigadv/A3 WUs to determine percentage of completed WUs? I think I'm under the 80% threshold. I uploaded a -bigadv WU earlier and received a lot less credit than usual. If so, the 2684s are the cause because of all the EUE strings.
If you are no longer recieving a bonus, my understanding is that all you will recieve are the base points for completing the work unit. For A2, the base points would be just 25k or so, and for the 2684 A3's it would be 8955pts. For me it is rather easy to calculate the % completed as i've only done 76+ WU's, but I wonder about you since you've been at it for a while. Sorry to hear about this as this was one of my concerns that got me back to SMP after that last 2684 failed on me
Last edited:
True, perhaps that was just a low production period when the WU happen to upload. I may have been doing them since last fall, but they were only -bigadv WUs until this spring, and they accumulated slowly for many months. The EUE strings I experienced lately occurred across several -bigadv clients and I had tons of crashes (way above any possibility to estimate).
Last edited:
- Joined
- Oct 23, 2007
- Messages
- 6,105
Don't know if you fellow SMP migrators have noticed or not but there's some very large (regular) A3's out in the pipeline that are helping push PPD back up closer to bigadv levels (at least they are for me). I've seen several 10k credit ones now. (can't recall the WU # off the top of my head though)
sirmonkey1985
[H]ard|DCer of the Month - July 2010
- Joined
- Sep 13, 2008
- Messages
- 22,412
Don't know if you fellow SMP migrators have noticed or not but there's some very large (regular) A3's out in the pipeline that are helping push PPD back up closer to bigadv levels (at least they are for me). I've seen several 10k credit ones now. (can't recall the WU # off the top of my head though)
if you are using HFM go to tools/benchmark viewer and look through the WU's til you find the one that shows the point total or avg ppd you saw..
Don't know if you fellow SMP migrators have noticed or not but there's some very large (regular) A3's out in the pipeline that are helping push PPD back up closer to bigadv levels (at least they are for me). I've seen several 10k credit ones now. (can't recall the WU # off the top of my head though)
From what I understand, you may be looking at the 6040 and 6041's that are about 2/3rd the size of a typical bigadv. Those end up clocking in as a 67mb upload while your typical bigadv will clock in as a 94-101mb upload. There are also the new 6701's that clock in around 45mb or so, although these appear to be causing a drop in points across the board for everyone.
I have indeed noticed them but the PPD varies a lot between architectures. My systems are almost all exclusively Core2 based plus one Opteron machine (going to retire this not too long). Even the P2684 outproduces the best A3 WUs by thousands of PPD. There is no comparison. I will lose on average 1000-1500k PPD on a P2684 compared to an A2 -bigadv WU, and gain at most roughly the same amount with the bigger SMP2 WUs over the older units. So, these don't come close even compared to the slowest -bigadv. From what I've been gathering about i7, there is a big drop working on a P2684 and big gain with the newer standard A3 WUs. YMMV big time based on the system architecture.Don't know if you fellow SMP migrators have noticed or not but there's some very large (regular) A3's out in the pipeline that are helping push PPD back up closer to bigadv levels (at least they are for me). I've seen several 10k credit ones now. (can't recall the WU # off the top of my head though)
- Joined
- Oct 23, 2007
- Messages
- 6,105
Yeah it must have been the 6041 (thanks for the info on how to find that, sirmonkey ).
Maybe I'm spoiled from A2's and just jonesing for that time again, knowing it's coming to an end. I like the points of the 2684's but I hate the downtime that they create....sucking up 3.5 days (on average) on 1 WU, it's like playing with a stick of dynamite. Either you get rid of it in time, or it blows up in your hand and you're left with nubs. I'm not saying my systems are temperamental like the WU, I mean my systems stay up under whatever power I have. (ie other than a power outage or something like that - they don't go down, they crunch all night long) I just hate that they take so long to finish yet they flush under any instability/circumstance if you aren't able to suspend the VM.
).
Yeah, it's a drop, but some of these WU's make it a lesser drop. A drop is a drop, no two bones about it.... I'm torn, man/guys, I really am. I hate losing the PPD but the 2684's just really piss me off for some reason.APOLLO said:
Maybe I'm spoiled from A2's and just jonesing for that time again, knowing it's coming to an end. I like the points of the 2684's but I hate the downtime that they create....sucking up 3.5 days (on average) on 1 WU, it's like playing with a stick of dynamite. Either you get rid of it in time, or it blows up in your hand and you're left with nubs. I'm not saying my systems are temperamental like the WU, I mean my systems stay up under whatever power I have. (ie other than a power outage or something like that - they don't go down, they crunch all night long) I just hate that they take so long to finish yet they flush under any instability/circumstance if you aren't able to suspend the VM.
Last edited:
Parja
[H]F Junkie
- Joined
- Oct 4, 2002
- Messages
- 12,671
AMD Overdrive
I totally agree with you. What I was saying is for those running older hardware, the new Windows A3 WUs are not much different than what we've already seen based on my experience, and the P2684 isn't much worse than A2 -bigadv. For Nehalem system architecture owners, the situation is quite different. I know that the wait for completion is tough, and the very high risk of failure is completely unacceptable. Stanford needs to get a handle on the issue or just revert these back to beta only until the problem is resolved.Yeah, it's a drop, but some of these WU's make it a lesser drop. A drop is a drop, no two bones about it.... I'm torn, man/guys, I really am. I hate losing the PPD but the 2684's just really piss me off for some reason.
I finally got a P2684 in a client with the new v2.22 core and I honestly can't see a performance increase that Stanford announced. Perhaps this is for Nehalem processors only but it doesn't seem to have helped this particular system.
I'm seeing a very minor improvement (maybe 1K PpD) on my sig rig with a 2684 and the new core.
Yep, as I expected, It's probably optimized for Nehalem.
- Joined
- Oct 23, 2007
- Messages
- 6,105
(Man I know I'm gonna look like I'm bitching just for the sake of bitching...)
All I'm getting in SMP2 is 6701s and they suck worse than the 2684s, really.
They drop PPD by average of 5k on my i7's, a couple K on my q9450 and 1k on my e8400.
[sigh] I feel like we can't win for losing here. LMAO
Guess I'll switch back on the VMs, why not at this point????????
)All I'm getting in SMP2 is 6701s and they suck worse than the 2684s, really.
They drop PPD by average of 5k on my i7's, a couple K on my q9450 and 1k on my e8400.
[sigh] I feel like we can't win for losing here.
LMAOGuess I'll switch back on the VMs, why not at this point????????
The type of WUs we receive appear to come and go in spates and it varies from user to user. I haven't seen a P6701 in many days and I have 4 Win SMP clients running. The P2684s likewise have been comparatively less common. If you are running i7 machines, it will probably be better for you to switch to -bigadv again, provided you keep your systems running for long periods and do not need to restart them frequently, or your systems have demonstrated solid stability. Then you will be OK. The additional minor boost in performance should be nice as well.All I'm getting in SMP2 is 6701s and they suck worse than the 2684s, really.
They drop PPD by average of 5k on my i7's, a couple K on my q9450 and 1k on my e8400.
[sigh] I feel like we can't win for losing here.
Guess I'll switch back on the VMs, why not at this point????????
tjmagneto
[H]ard DCOTM x2
- Joined
- Aug 6, 2008
- Messages
- 3,299
Today I just finished a 2681 and was rewarded by getting a 2684. This is the first one I'm going to let run since the last one popped up when I was running the dual bigadvs and I did a mercy killing on that WU.
Frame times are running shy of 36 minutes with the single VM on the 980x. Points per day dropped from 35k for the 2681 to 28k for the 2684. I get to wave bye bye to 7k ppd for the next 2 days.
This is the first one I'm going to let run since the last one popped up when I was running the dual bigadvs and I did a mercy killing on that WU.Frame times are running shy of 36 minutes with the single VM on the 980x. Points per day dropped from 35k for the 2681 to 28k for the 2684. I get to wave bye bye to 7k ppd for the next 2 days.
- Joined
- Oct 23, 2007
- Messages
- 6,105
Switching on VMs, both so far have done the 0% flush thing. Seems like one finally started chugging a 2684, the other is in the process of figuring out if it wants to 0% flush or continue on. [sigh]
Seems like one finally started chugging a 2684, the other is in the process of figuring out if it wants to 0% flush or continue on. [sigh]Well, it's better than running 48 min a frame and waiting over 3 days. Fortunately in my case, the drop is less than 2k PPD. Sometimes it pays to run older hardware, LOL.
Just leave it try to DL, it should eventually get an A2 WU if it follows the same pattern I've seen with my systems. /crosses fingersSwitching on VMs, both so far have done the 0% flush thing.
- Joined
- Oct 23, 2007
- Messages
- 6,105
Yeah - they both finally grabbed ones that worked. thanks man
thanks man - Joined
- Apr 7, 2006
- Messages
- 10,477
(Man I know I'm gonna look like I'm bitching just for the sake of bitching...
All I'm getting in SMP2 is 6701s and they suck worse than the 2684s, really.
They drop PPD by average of 5k on my i7's, a couple K on my q9450 and 1k on my e8400.
[sigh] I feel like we can't win for losing here.
Guess I'll switch back on the VMs, why not at this point????????
6701's are not supposed to get as much ppd as 2684's
Theya re just regular A3's no bigadv
tjmagneto
[H]ard DCOTM x2
- Joined
- Aug 6, 2008
- Messages
- 3,299
Well, it's better than running 48 min a frame and waiting over 3 days. Fortunately in my case, the drop is less than 2k PPD. Sometimes it pays to run older hardware, LOL.
This is not going to sound so good but the one 2684 I pulled down while I was running dual VMs was getting frame times of 55.5 min per frame. This is with the other VM working on an A2 bigadv. So now it does not sound so bad as I was using smp -6 when it popped up.
- Joined
- Oct 23, 2007
- Messages
- 6,105
I know...but what I mean is taking the regular A3 (incl. a 6071 or two), doing a few a day you get 15-18k PPD (at least I do), which is roughly the same I get PPD listed in HFM.net doing a 2684.
Hmm, well, it's tricky in your case because you're running dual VMs with a small number of cores allocated to each. Too bad you can't risk stopping the client to change the flags. If it was safe, you could have stopped and -oneunit the A2 -bigadv, and once that completed, stop and allocate more cores to the A3 VM to quicken its pace and ensure it made the bonus or accrue a higher bonus. Unfortunately this isn't recommended with a highly unstable series of units.
Today I just finished a 2681 and was rewarded by getting a 2684.
Frame times are running shy of 36 minutes with the single VM on the 980x. Points per day dropped from 35k for the 2681 to 28k for the 2684. I get to wave bye bye to 7k ppd for the next 2 days.
Just imagine what you could do if bigadv finally came to native windows and you could throw all 12 virtual cores at it!
That is just darn good though.
Well I finished a 2681 & 2684 (at the same damn time, let's just say that uploading was a b!tch).
I was rewarded by getting normal A3 WUs in both my VMs. I got a P6075 & P6014. PpD in the VMs with normal A3 WUs SUCKS BALLS!!!! I am currently running the P6075 with 2.22 and the P6014 with 2.21
/rant off
We'll see what I pick up after they are done.
I was rewarded by getting normal A3 WUs in both my VMs. I got a P6075 & P6014. PpD in the VMs with normal A3 WUs SUCKS BALLS!!!! I am currently running the P6075 with 2.22 and the P6014 with 2.21
/rant off
We'll see what I pick up after they are done.
- Joined
- Oct 23, 2007
- Messages
- 6,105
I have 2 2684's that will finish at nearly the same time very early Wed. morning, my DSL line is going to cccrrraaawwwllllllllllllll (even though it's 6meg down/768k up). Luckily we'll probably all be asleep anyway.
For no apparent reason, I just lost the last P2684 WU my -bigadv clients were working on. It was at over 50% completion and processed close to 2 days...
In the past 24hrs, I and others have been getting nothing but standard A3 WUs. At this point I don't really care. At least these WUs complete even if the P6701 is grossly disproportionately credited. It's the P2665 all over again for those who recall what we went through in the height of the A1 days.
In any case, I won't be running the P2684 hereon unless Stanford releases a more stable WU and/or core improvement to address stability specifically. If the recent -bigadv WU dry spell continues indefinitely, which seems like a distinct possibility considering that some people were actually advising it, I will drop my VMs and native Linux client installs.
- Joined
- Oct 23, 2007
- Messages
- 6,105
Aw WTF man...well if that's the case, once these 2684's finish if my VMs grab anything other than a 2681-2684, the VMs are getting trashed for the final time. I almost want to trash them NOW but I'm not losing the 56% of worth both of them have done.
PG is losing their minds with this crap.
- Joined
- Apr 7, 2006
- Messages
- 10,477
So what is wrong with the standard A3 units?
tjmagneto
[H]ard DCOTM x2
- Joined
- Aug 6, 2008
- Messages
- 3,299
After I finished the mobo change I decided to try one VM using -smp 8 which really simplified things when the 2684 came up. Eventually I may fire up VM number 2 and see what happens just for giggles but I don't think I'll try juggling flags since I think I'll break something.Hmm, well, it's tricky in your case because you're running dual VMs with a small number of cores allocated to each. Too bad you can't risk stopping the client to change the flags. If it was safe, you could have stopped and -oneunit the A2 -bigadv, and once that completed, stop and allocate more cores to the A3 VM to quicken its pace and ensure it made the bonus or accrue a higher bonus. Unfortunately this isn't recommended with a highly unstable series of units.
Fun times are ahead, it's just a rough patch with all these crazy WUs coming out.
I understand. Now I guess you're back to Windows since -bigadv has not been received on my clients for over a day. All I've been getting is back to back standard A3 WUs. Don't mind though, it's better than 2684s which complete only half the time if even, that much..After I finished the mobo change I decided to try one VM using -smp 8 which really simplified things when the 2684 came up. Eventually I may fire up VM number 2 and see what happens just for giggles but I don't think I'll try juggling flags since I think I'll break something.
tjmagneto
[H]ard DCOTM x2
- Joined
- Aug 6, 2008
- Messages
- 3,299
Yay! The 2684 finally finished and here's a screenie to prove it. It took 2 days 11 hrs.
It took 8 minutes from attempting to send the results to getting the next WU, a 6053
It took 8 minutes from attempting to send the results to getting the next WU, a 6053