Just got a Project 2684

[zero2dash]

[sigh]
Got another one.
A2 just finished, grabbed a 2684.
The other 2684 finishes tomorrow, I'm sure it'll grab another one...

At this point (0%) on the 1st new one, I'm switching over. I'm going to take my sys1 offline for a few hours tomorrow night (HD swap, HSF swap, dusting/maintenance/cable mgmt OCD ), I don't feel like having another flush and/or losing a ton of PPD just to futz around with these things.

Once the other VM finishes on sys2, if it grabs another A3 I will switch it over as well. I hate to 'cut' these WU's like that, but why even bother. They'll eventually find their way back to the assignment servers anyway.

 

[musky]

I picked up 2 new A2s this morning, so don't lose faith just yet.

 

[APOLLO]

The A2s are still hanging around, it's roughly a 50/50 spread from what I can see. Unfortunately, it appears the 2684s did one of my systems in. It had crash after crash for over a week, now it won't work for more than a few minutes without freezing. I'll try a few things like memory swap and BIOS flash but I'm not holding my breath. Sigh.

 

[Kendrak]

I had my first error on Nexus. I used the downtime to install windows on another HDD and here I am getting about 2k less ppd running normal A3 WU and I don't have to worry about it going belly up.

2:50 TPF on a p6063... I'll take it. 19k ppd on 220 watts, even more awesome sauce.

 

[Kendrak]

The A2s are still hanging around, it's roughly a 50/50 spread from what I can see. Unfortunately, it appears the 2684s did one of my systems in. It had crash after crash for over a week, now it won't work for more than a few minutes without freezing. I'll try a few things like memory swap and BIOS flash but I'm not holding my breath. Sigh.

Now for full irony... the i7 930 is still pulling A2 and is doing 22k ppd.

/shurg

 

[amdgamer]

The A2s are still hanging around, it's roughly a 50/50 spread from what I can see. Unfortunately, it appears the 2684s did one of my systems in. It had crash after crash for over a week, now it won't work for more than a few minutes without freezing. I'll try a few things like memory swap and BIOS flash but I'm not holding my breath. Sigh.

Wow, i'm sorry to hear that. Do you think something in the OS got corrupted during the repeated crashes, or do you think it is hardware that is getting flaky? I hope you get it worked out as stuff like this sucks.

 

[APOLLO]

Wow, i'm sorry to hear that. Do you think something in the OS got corrupted during the repeated crashes, or do you think it is hardware that is getting flaky? I hope you get it worked out as stuff like this sucks.

No, it freezes on both Windows and Linux and they're on separate disks using different interfaces. The system was working flawlessly since December when I put it together. I'm hoping it's just a bad module or two, but it seems like something else because it sometimes doesn't POST. I hard restarted it so many times lately, must have screwed something up.

 

[zero2dash]

I had my first error on Nexus. I used the downtime to install windows on another HDD and here I am getting about 2k less ppd running normal A3 WU and I don't have to worry about it going belly up.

2:50 TPF on a p6063... I'll take it. 19k ppd on 220 watts, even more awesome sauce.

Yeah I'm pulling 3:19 on a 6050 yielding 15,325.7 ppd vs the VM still chugging the 2684 48:58 for 17,973.6 (mhz difference though obviously). With bonus I'm looking at 61,119 for the VM vs. pulling roughly the same in various WU's floating throughout the day. (figuring conservatively 3.5-4 days for each...A3 pulled 2d 11hr ago, with 17hr 8min remaining)

I mean when I look at it that way, really what's the advantage of the VM again? [*now*]
Even if (best case) it's 50/50 on pulling an A2 vs an A3, you're breaking even. You get lucky, you finish with equal points 1.5 days earlier. Worst case, you crawl, you break even, and you sneeze and your VM and days of work get flushed.

I really don't see the benefit going to A3 here. They need to up those bonuses or something, man. I hope the native Win A3 pull some ridiculous #s that just blow all our minds (compared to what we're seeing in *nix/VM's) but I just don't see that happening. [shrug] I'll probably swap HDs in my boxen anyway and finally put a SATA in there; while I'm at it, I might as well spend the 2-3hrs installing and updating 7x64 on it as well and just run native.

 

[amdgamer]

LOL, I am currently crunching a 6050 on my laptop right now as i'm only going to be loading up excel and pdf files for the next several days. My TPF's are currently at 29-30min!

Yeah I'm pulling 3:19 on a 6050 yielding 15,325.7 ppd vs the VM still chugging the 2684 48:58 for 17,973.6 (mhz difference though obviously). With bonus I'm looking at 61,119 for the VM vs. pulling roughly the same in various WU's floating throughout the day. (figuring conservatively 3.5-4 days for each...A3 pulled 2d 11hr ago, with 17hr 8min remaining)

I mean when I look at it that way, really what's the advantage of the VM again? [*now*]
Even if (best case) it's 50/50 on pulling an A2 vs an A3, you're breaking even. You get lucky, you finish with equal points 1.5 days earlier. Worst case, you crawl, you break even, and you sneeze and your VM and days of work get flushed.

I really don't see the benefit going to A3 here. They need to up those bonuses or something, man. I hope the native Win A3 pull some ridiculous #s that just blow all our minds (compared to what we're seeing in *nix/VM's) but I just don't see that happening. [shrug] I'll probably swap HDs in my boxen anyway and finally put a SATA in there; while I'm at it, I might as well spend the 2-3hrs installing and updating 7x64 on it as well and just run native.

 

[APOLLO]

I really don't see the benefit going to A3 here. They need to up those bonuses or something, man. I hope the native Win A3 pull some ridiculous #s that just blow all our minds (compared to what we're seeing in *nix/VM's) but I just don't see that happening. [shrug]

I'm getting the feeling from what people have posted lately that the credit for Windows A3 -bigadv might be higher but it's not going to be a night and day difference. Perhaps if we're lucky will see similar bonuses to the original Linux A2 WUs, but I'm not betting on it.

 

[zero2dash]

Yeah I think it'll be a few K higher than the *nix A3, and the TPF may drop a few minutes...but yeah, definitely not a big difference.

Pour a forty out guys...the blessing of A2's is coming to an end.

 

[amdgamer]

Yeah I think it'll be a few K higher than the *nix A3, and the TPF may drop a few minutes...but yeah, definitely not a big difference.

Pour a forty out guys...the blessing of A2's is coming to an end.

Here Here............

It is a shame that I started folding so late and was only able to enjoy A2 bigadv for a short while.........

A3 bigadv won't be so bad though once we are virtually garenteed of its stability. I also want it to come to Windows soon so I can ditch this VM. It looks like they are studying the FLU virus with these bigadv projects, and that is a noble cause in itself considering the dangers of this highly mutagenic virus.

 

[musky]

My dual E5530 machine has run 4 of the 2684 units in a row without error, and has started its 5th. It may just be that overclocked I7s are not going to handle these things anymore.

 

[zero2dash]

It's not that they won't, it's the instability. If at any point you have to quit the VM, just assume the WU will flush (has been my experience).

Oh well...the end of an era.
i7-2 finished it's A3 and grabbed another, so I switched it over to SMP2 as well.

IDK what the ratio of A2 to A3 is now, but it's gotta be more than 50/50...I've gotten 1 A2 in the last week vs 4 or 5 A3's. Bigadv to SMP2 announcement should come any day now...[crosses fingers]

 

[APOLLO]

Oh well...the end of an era.

Well, it was not much of an era. For many if not most it has been less than half a year, for me just under three quarters. Anyway, it need not end like this. Stanford wants to change things, that's their prerogative. Let's hope Windows -bigadv is all it's cracked up to be - and that there are no 2684s!

 

[amdgamer]

Well, it was not much of an era. For many if not most it has been less than half a year, for me just under three quarters. Anyway, it need not end like this. Stanford wants to change things, that's their prerogative. Let's hope Windows -bigadv is all it's cracked up to be - and that there are no 2684s!

Considering that 2684 is the first A3 bigadv work unit, I am sure we are going to see plenty of them when Windows bigadv comes out.

Now that more and more powerful systems are coming out, I wonder whether they plan to go even bigger anytime soon. Maybe a -DollyPartonbigadv?

 

[amdgamer]

Woo-hoo! I just finished a 2681 A2 bigadv and my machine just grabbed another 2683 A2 bigadv! This makes me happy.

 

[APOLLO]

One of my systems is working on a P2684 and the PPD is only ~1000 PPD lower than if it was processing an A2 -bigadv WU, yet it's 10 minutes longer per TPF. Has there been a credit adjustment, or am I just fortunate with this WU?

 

[nitrobass24]

It has lower base points, but a 25 K factor. So depending on the system these could potentially get you better ppd.

 

[APOLLO]

It has lower base points, but a 25 K factor. So depending on the system these could potentially get you better ppd.

I haven't heard of anyone getting higher but I could believe in similar PPD. Problem is that at over 52 min TPF, it is dangerously close to the cutoff for the bonus.

 

[amdgamer]

I haven't heard of anyone getting higher but I could believe in similar PPD. Problem is that at over 52 min TPF, it is dangerously close to the cutoff for the bonus.

That indeed makes me extremely nervous. With my system running so close to the deadline when doing an A3 bigadv, all it takes is for a power outage or internet outage(not rare in my neighborhood in the ghetto) to totally miss the deadline. I hear that Cincinnati is finally getting FIber optic internet service in some neighborhoods and am really hoping it comes to mine.

 

[zero2dash]

So I came back and farted around with VM's this morning and grabbed all A3's (2684), no A2's at all.
Ironically two things occured - 1) crash and flush on VM shutdown and reboot (big surprise ) and 2) I noticed PPD is only 1k (give or take a hundred or two) on a 2684 vs normal A3's. Back to SMP2 for me, for good.

 

[APOLLO]

That indeed makes me extremely nervous. With my system running so close to the deadline when doing an A3 bigadv, all it takes is for a power outage or internet outage(not rare in my neighborhood in the ghetto) to totally miss the deadline.

In my case, there is an additional obstacle - GPU. This system has 2 GPU clients, and when they're working on 353s, the TPF is nearly an hour making the bonus unattainable if 353s continue to be received. Fortunately, we don't see 353s all the time, otherwise this VM would only be producing ~2500 PPD when working on P2684s. I know most people would have switched over to the Windows SMP by now if they had this machine, but it would only be producing ~8000 PPD at best, because I'd have to use the -smp 7 flag and dedicate one core to the GPU clients.

So I came back and farted around with VM's this morning and grabbed all A3's (2684), no A2's at all.

It's probably the luck of the draw. Some days see more A3 WUs while other days see more A2s. It depends what the servers are sending out at any particular time. Unfortunatelty, it appears your clients happened to download new work when Stanford was sending A3s.

 

[amdgamer]

So I came back and farted around with VM's this morning and grabbed all A3's (2684), no A2's at all.
Ironically two things occured - 1) crash and flush on VM shutdown and reboot (big surprise ) and 2) I noticed PPD is only 1k (give or take a hundred or two) on a 2684 vs normal A3's. Back to SMP2 for me, for good.

Don't become a stranger now! I wonder whether Windows A3 bigadv will be any more stable. It seems like lots of people have these issues with A3 bigadv flushing when they stop the client for whatever reason. However, I think people have had trouble with A2 bigadv flushing after they stop the client for whatever reason, although certainly not as widespread.

In my case, there is an additional obstacle - GPU. This system has 2 GPU clients, and when they're working on 353s, the TPF is nearly an hour making the bonus unattainable if 353s continue to be received. Fortunately, we don't see 353s all the time, otherwise this VM would only be producing ~2500 PPD when working on P2684s. I know most people would have switched over to the Windows SMP by now if they had this machine, but it would only be producing ~8000 PPD at best, because I'd have to use the -smp 7 flag and dedicate one core to the GPU clients.

It's probably the luck of the draw. Some days see more A3 WUs while other days see more A2s. It depends what the servers are sending out at any particular time. Unfortunatelty, it appears your clients happened to download new work when Stanford was sending A3s.

Not being able to make the bonus deadline would be rough, especially since you already had 4 days put in on a single work unit. Perhaps Windows SMP bigadv will give you just the little extra you need to make it happen even if you ge those GPU units. One could hope at least.

 

[APOLLO]

Not being able to make the bonus deadline would be rough, especially since you already had 4 days put in on a single work unit. Perhaps Windows SMP bigadv will give you just the little extra you need to make it happen even if you ge those GPU units. One could hope at least.

The way it works is something like this: a few hours of 353s the TPF goes up and I see the PPD plummet below the bonus threshold. Then, there is half a day to a full day of other WUs and the VM makes up for lost time. Once the VM client makes headway, even if there is a spate of 353s that last most of the day, the client will still meet the deadline because of the headway it achieved earlier. It's a bit hard to explain but that's how I see things.

Now, this is true providing the client does not crash, which we all know can very well occur with the 2684s. As of now, the biggest problem by far is the crashing, not so much the higher TPFs. If Stanford could fix that, I'd like these WUs almost as much as I like the A2s.

 

[zero2dash]

@amd

Well according to HFM I'm only losing a K PPD on SMP2 for the most part but otherwise trust me I'm not going anywhere. Regarding A2 v A3 stability, I've only had a few A2's flush and I've had some even keep working even after a system crash and reboot (meaning the VM wasn't shut down properly). I've had A3's fail after a routine VM shutdown though...at this point I'm guessing the only way to quit an A3 VM but keep it from flushing would be to Suspend it (instead of Control+C and then shutting it down).

I may try that out...wait for my SMP's to finish, fire up the VM's again and suspend if I need to shut them down. Maybe that will keep them from flushing.

 

[APOLLO]

Regarding A2 v A3 stability, I've only had a few A2's flush and I've had some even keep working even after a system crash and reboot (meaning the VM wasn't shut down properly). I've had A3's fail after a routine VM shutdown though...

Same here. The A2s rarely fail even if there is a system-wide crash or spontaneous reboot. The 2684s on the contrary, are very fragile and will crash at the drop of the hat.

...at this point I'm guessing the only way to quit an A3 VM but keep it from flushing would be to Suspend it (instead of Control+C and then shutting it down).I may try that out...wait for my SMP's to finish, fire up the VM's again and suspend if I need to shut them down. Maybe that will keep them from flushing.

I totally agree about suspending. I tried that once or twice and it worked fine as expected. Unfortunately, suspension cannot be done in a native Linux environment but at least there is that option with the VM installations.

 

[zero2dash]

Yeah I went ahead and re-fired the VM after a regular A3 finished in Win...VM grabbed a 2684 of course but I'll let it run. When I shutdown later, I'll suspend and pray that it keeps it from flushing.

Apollo's right though - it is real sad for native *nix users though because if they have to reboot or power down for anything, their work is practically all for naught.

 

[amdgamer]

Haha! I can't believe my luck. I just finished a 2683 A2 bigadv and my machine just grabbed another 2683 A2 bigadv. That is 3 A2's in a row it has grabbed.

 

[APOLLO]

Haha! I can't believe my luck. I just finished a 2683 A2 bigadv and my machine just grabbed another 2683 A2 bigadv. That is 3 A2's in a row it has grabbed.

You're lucky. I had yet another 2684 failure prompting me to shut down my first -bigadv client yesterday. Today, I had an A2 -bigadv in another machine complete only to be followed by a 2684. If this WU fails, that will force me to shut down another -bigadv client. I'm using this to determine whether I'll continue folding at all. I could accept suddenly raising the hardware requirements up to a point, but not ridiculous instability. Beta WUs and cores should remain beta until there is absolutely no doubt in regards to their reliability.

 

[zero2dash]

Haha! I can't believe my luck. I just finished a 2683 A2 bigadv and my machine just grabbed another 2683 A2 bigadv. That is 3 A2's in a row it has grabbed.

Stop hogging all the damn A2's man!!!!!! j/k

I still haven't gotten one. Both of mine are crunching 2684's and have been since midday yesterday.
Oddly enough I had to do a hard reset/power off last night on i7-1 without being able to shut the 2684 VM down and it actually didn't flush itself when I restarted it.

edit
I hate to switch back *again*, but I did just a little while ago. I'm seeing (under VM 2684) HFM reporting roughly 65k PPD. Under A3 in Windows, it bounces between 61-62k PPD.
Given that I'm planning on some TF2 this weekend (which is a easy way to suck away time), I'd have to shutdown the VM's anyway, possibly catch a flush (worst case) or lose some bonus (best case), I might as well do normal A3's for the time being....I'm still personally not finding incentive to crank out 2684's instead of switching to SMP2 natively. (especially since amd here is hogging all the damn A2's j/k again)

 

[musky]

Ya'all will hate me, but I currently have 4 A2's processing on my 4 bigadv machines. My 5th machine is having major memory problems and is forced to normal A3 duties for a while...and is sucking at that no less (random reboots.)

 

[APOLLO]

Just lost another one due to a forced reboot...

 

[Mr. Pedantic]

Wow. So the problems are still big...I clocked my i7 up to 3.5GHz to do these when they come onto Windows, but the way they're going now, I'll probably put it back to 3GHz undervolted and stick with normal SMP2. I'm not sure that random restarts and WU losses are worth turning the fans up an extra 500rpm and still having the temps go through the roof...

 

[amdgamer]

Haha! My machine just finished that 2683 bigadv an hour ago, and I noticed that it just grabbed yet another 2683 A2 bigadv! That is 4 A2 bigadv WU's in a row. Can't believe my luck. The bad thing about being lucky, is that your luck will run out............

EDIT.......shows you how oblivious I have been this weekend. It looks like I have had 5 A3's in a row and not 4.

 

[APOLLO]

Haha! My machine just finished that 2683 bigadv an hour ago, and I noticed that it just grabbed yet another 2683 A2 bigadv! That is 4 A2 bigadv WU's in a row. Can't believe my luck. The bad thing about being lucky, is that your luck will run out............

I've also been getting more A2 than A3 -bigadvs but that can just be the luck of the draw, or Stanford is running low on 2684s for the last little while..

 

[amdgamer]

I've also been getting more A2 than A3 -bigadvs but that can just be the luck of the draw, or Stanford is running low on 2684s for the last little while..

Since I won't have to worry about possibly drawing an A3 bigadv for the next 2.5 days, i'm hoping that Stanford is making sure that they are addressing the problems we have seen. Maybe they are checking for stability? I was under the impression that A2's were going to be gone completely, but they are still hanging around.

 

[capreppy]

I've got an A2 & an A3 in progress. I do miss the PpD so I may keep the VMs running for a bit. This will be especially true if I continue to get A2's.

 

[zero2dash]

i7-1 finished an A3 and surprisingly grabbed an A2. i7-2 just finished an A3 and now I'm getting "attempt to get work failed" on 171.67.108.22. I was going to keep the VM's up as long as I grabbed an A2 instead, otherwise I'd shut them down and go back to native.... I may shut down -2 now since it's not grabbing work anyway.

 

[APOLLO]

One of my -bigadv machines downloaded a new version of the A3 core, v2.21. My standard SMP2 clients are all still using v2.19. Anyone else receive a new A3 core?