![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Drought over? 300,000+ PPD, 2 boxen,1200 watts. Goal Achieved.
- Thread starter may i be worthy
- Start date
Yeah, this time I happened to catch the thread over there and it did get really ugly. It is almost like those two are paranoid and bipolar. I'm sure that many at eVGA are not upset by them leaving for good this time, although I enjoyed the drama they caused since i'm not on their team.
zero2dash
Supreme [H]ardness
- Joined
- Oct 23, 2007
- Messages
- 6,100
I didn't see them that way, I saw them (from their comments) as people who think they're better than everybody else, ie. stuck up/snobbish etc. Now granted I'm on the outside looking in (because I post here, not there)....but I've seen enough evidence to merit that theory.
Oh well...onto more pertinent business....FOLD ON!
and CONGRATS MIBWThat's the triple-word I was looking for!
My mistake, it was rcranfield's thread
http://www.evga.com/forums/tm.aspx?m=614120
appears all the emo drama has been edited out
that's how we do it around here. start a thread about tweaking GPU3 performance and somebody disables NUMA and next thing you know we're comparing the size of our comic book collections.
keeps things lively
Yes, but now they have their own 'team' and fold for none but themselves. I guess they don't need to deal with other people this way.
Well I finally managed to get a bigadv on each SR-2, but they're both P2684s, so 140K PPD total. My 980X is of course folding a P6701. With the current mix of WUs, I won't feel too bad powering them down to work on overclocks. At least I'm back in the top 10 and closing in on the top 5.
I have a complete set of G.I. Joe comics (the Marvel ones from the '80s).
My 980X is of course folding a P6701. With the current mix of WUs, I won't feel too bad powering them down to work on overclocks. At least I'm back in the top 10 and closing in on the top 5.I have a complete set of G.I. Joe comics (the Marvel ones from the '80s).
that's so perfect. Searching for "bwm" on the EOC stats database tells a revealing story.......
lolWell I finally managed to get a bigadv on each SR-2, but they're both P2684s, so 140K PPD total.
I have a complete set of G.I. Joe comics (the Marvel ones from the '80s).
Ahh!
You must have missed the good stuff then.
Yeah, they went straight after Afterburner's throat and were practically foaming at the mouth as they were screaming at him to ban them. Afterburner didn't even make any mention of the twin cancers in that thread, but I guess the shoe fit and being the paranoid people they were, they thought that everyone was secretly talking about them.
This time I happened to be browsing your forums when it started going down. I don't know what happened afterwards, but it appears they left.
ooh good stuff. maybe you have the page cached in your browser?
may i be worthy
[H]ard|DCer of the Month - December 2010
- Joined
- Aug 17, 2010
- Messages
- 1,186
Good lord, I go to bed and the thread explodes
Excuse me while I catchup:
Sorry I mean the FAH performance fraction that is calculated when you return a unit - but good tip - I will disable that silly windows thing.
I had to google that, and that IS Amsome!
Sorry you are getting smashed - but do try finishing a unit, changing your machine id to a higher number and see if that does anything. I am dying to know if the great run I am getting is just dumb luck, or because I reset some performance metric the influences what units I get sent. I am between jobs right now, so both machines are dedicated since Saturday - and this is my run since since I reset the machine id:
SR2-1 SR2-2
2685 2685
2686 2685
6058
6021 2692
2685 2684
6702
2686 6701 - Tuesday
The line by line timing is a bit off, but this run was proceeded by 5 straight 6701s each..
No problem - every good thread needs some drama... see below.
I think it uses a rolling 7 day average? I never get too high on that, because I can almost never fold on both machines for a full week, I tend to go in bursts around work.
Make that top 3 for the week so far. You are back.
$0.27c kwh! so yes, I am sure there will be a reassessment when I get my next quarterly bill. I am thinking of renting these out to [H] at cost.
I am sure nobody here would be so cheap, so classless, so tacky as to record that unfortunate incident...
I am sorry - it was like slowing down to look at a truck smash - I could not help it.
Excuse me while I catchup:
Sorry I mean the FAH performance fraction that is calculated when you return a unit - but good tip - I will disable that silly windows thing.
That's just Amsome
I had to google that, and that IS Amsome!
Sorry you are getting smashed - but do try finishing a unit, changing your machine id to a higher number and see if that does anything. I am dying to know if the great run I am getting is just dumb luck, or because I reset some performance metric the influences what units I get sent. I am between jobs right now, so both machines are dedicated since Saturday - and this is my run since since I reset the machine id:
SR2-1 SR2-2
2685 2685
2686 2685
6058
6021 2692
2685 2684
6702
2686 6701 - Tuesday
The line by line timing is a bit off, but this run was proceeded by 5 straight 6701s each..
No problem - every good thread needs some drama... see below.
I think it uses a rolling 7 day average? I never get too high on that, because I can almost never fold on both machines for a full week, I tend to go in bursts around work.
Make that top 3 for the week so far. You are back.
$0.27c kwh! so yes, I am sure there will be a reassessment when I get my next quarterly bill. I am thinking of renting these out to [H] at cost.
I am sure nobody here would be so cheap, so classless, so tacky as to record that unfortunate incident...
I am sorry - it was like slowing down to look at a truck smash - I could not help it.
Last edited:
Don't be sorry, you are more bold than I am and this dross needs immortalization to provide an example of the extreme negative side of the project, and all DC, FTM.
Unfortunately I set my browser to clear the cache at the end of every session, but it got even nastier after that if I remember things correctly. Let me tell you that it was kind of like one of those Jerry Springer moments. Pathetic, but strangely you cannot compel yourself to change channels at the same time
zero2dash
Supreme [H]ardness
- Joined
- Oct 23, 2007
- Messages
- 6,100
I eagerly await the day when Stanford's client (or their assignment servers) have more robust IP logging and they start refusing connections from people who purposely flush and delete WUs (as shown by repeated WU assignment within a certain period, say 6-12 hours or even 24 hours).
Like one of the blue collar comedy guys once said "you can't fix stupid".
Like one of the blue collar comedy guys once said "you can't fix stupid".
sirmonkey1985
[H]ard|DCer of the Month - July 2010
- Joined
- Sep 13, 2008
- Messages
- 22,414
I eagerly await the day when Stanford's client (or their assignment servers) have more robust IP logging and they start refusing connections from people who purposely flush and delete WUs (as shown by repeated WU assignment within a certain period, say 6-12 hours or even 24 hours).
Like one of the blue collar comedy guys once said "you can't fix stupid".
no its actually far worse then that. if they catch your deleting WU's. say 10-20 WU's get sent to a passkey. but only 1 WU is returned, they will just zero out your points.
may i be worthy
[H]ard|DCer of the Month - December 2010
- Joined
- Aug 17, 2010
- Messages
- 1,186
no its actually far worse then that. if they catch your deleting WU's. say 10-20 WU's get sent to a passkey. but only 1 WU is returned, they will just zero out your points.
Hmmm, I am glad that was not the rules when I was fiddling with -smp 23 flags - which when you get a non bigadv will instant EUE and I think that killed a couple dozen units before I noticed.
Hence their 80% success rate rule for bigadv. Update on the run since Saturday. Not as shiny but if they alternate with 2686 I don't mind, that is still 140,000+ppd per machine.
SR2-1 SR2-2
2685 2685
2686 2685
6058
6021 2692
2685 2684
6702
2686 6701 - Tuesday
6701 2686 - Tuesday night
And how is this for a a nice demo of the problem in one screenie :
102,000 PPD variation!
sirmonkey1985
[H]ard|DCer of the Month - July 2010
- Joined
- Sep 13, 2008
- Messages
- 22,414
eue's are fine because the WU gets sent back to them then gets sent back out to some one else. if your deleting WU's and not returning the WU then it disapears for a couple weeks then its re-released.
may i be worthy
[H]ard|DCer of the Month - December 2010
- Joined
- Aug 17, 2010
- Messages
- 1,186
Oh, I did not know that. Bugger - I ran SMP for a day or two to requalify 10 units just in case.
But my one day with WCG opened my eyes a bit - oh look they have an abort button, so presumably the servers know this unit can be reissued right away.
For a ten year old project, FAH is not very sophisticated.
Xilikon
[H]ard|DCer of the Year 2008
- Joined
- Oct 12, 2004
- Messages
- 15,011
Be careful, there are EUE instances which doesn't get sent back because the work files are corrupted or there is something else which make the client think it's wiser to delete without sending.
Nope
I've gotten nothing but regular A3 (including a metric shit-ton of 6701's) for the last 36-48 hours.
Regular are fine, and a 670x here and there is fine too, but seriously, when it's like 70 - 80% of what I'm folding lately, my GTX 295 is looking pretty good right about now.
B&C Folding FTW! What a great boon for folding these two are.
B&C Folding FTW!
B&C Folding FTW!
Oops I broke the record.
Oh, I did not know that. Bugger - I ran SMP for a day or two to requalify 10 units just in case.
But my one day with WCG opened my eyes a bit - oh look they have an abort button, so presumably the servers know this unit can be reissued right away.
For a ten year old project, FAH is not very sophisticated.
And often I find that "Folding@home", "donor friendly" and "sophisticated" should never be used in a sentence together!
sirmonkey1985
[H]ard|DCer of the Month - July 2010
- Joined
- Sep 13, 2008
- Messages
- 22,414
yep happened to me a couple times. for some reason aero completely crashed and took windows explorer with it too and would cause the smp client to think the WU was corrupt because the data didnt match and would say the WU was corrupt and delete it. but thing i noticed is that while it doesnt send the WU back it does return an error code because checking the stats it would show a WU being returned for 0 points. not sure if it does it every time but it did that time.
Oh, I did not know that. Bugger - I ran SMP for a day or two to requalify 10 units just in case.
But my one day with WCG opened my eyes a bit - oh look they have an abort button, so presumably the servers know this unit can be reissued right away.
For a ten year old project, FAH is not very sophisticated.
BOINC is light years ahead of F@H and is community driven and worked on. even the projects almost all the applications have been modified/changed by the community for the projects. milkyway@home is a good example. the ATI application they use was a modified code created by a community member. which was then later ported over to be used by collatz. this is the biggest issue with F@H is that everything is closed. all you can do is make suggestions. do they listen to them? hell no.
And often I find that "Folding@home", "donor friendly" and "sophisticated" should never be used in a sentence together!
pretty much sums it up.
like me being called a cheater by F@H dev's instead of just asking me how i found the fermi glitch on the gpu3 client for non fermi cards. which is easily fixable if they actually gave a crap but they dont.
Last edited:
Yeah, and it's still not rectified, AFAIAA.
may i be worthy
[H]ard|DCer of the Month - December 2010
- Joined
- Aug 17, 2010
- Messages
- 1,186
SR2-1 SR2-2
2685 2685 - Saturday
2686 2685
6058
6021 2692
2685 2684
6702
2686 6701 - Tuesday
6701 2686 - Tuesday night
2684 2686 - currently Wednesday
Back to back 2686's! Far out. I can't help but look at this and think that faster machines are getting special treatment since Saturday. Only ever had back to back bigadv twice before this week since I started in August. Now I have scored that 5 times in 4 days.
I am in second place for the week fer crying out loud, with two rigs! But the really funny one is the world rankings this week - I am 18th so far, but look who pips me at 17 - that is karma for me posting that screengrab earlier in the thread
2685 2685 - Saturday
2686 2685
6058
6021 2692
2685 2684
6702
2686 6701 - Tuesday
6701 2686 - Tuesday night
2684 2686 - currently Wednesday
Back to back 2686's!
Far out. I can't help but look at this and think that faster machines are getting special treatment since Saturday. Only ever had back to back bigadv twice before this week since I started in August. Now I have scored that 5 times in 4 days. I am in second place for the week fer crying out loud, with two rigs! But the really funny one is the world rankings this week - I am 18th so far, but look who pips me at 17 - that is karma for me posting that screengrab earlier in the thread
zero2dash
Supreme [H]ardness
- Joined
- Oct 23, 2007
- Messages
- 6,100
Sheesh
I'll be glad when they get caught and zeroed out, they deserve it for deleting WUs.
I'll be glad when they get caught and zeroed out, they deserve it for deleting WUs.
Last edited:
Based on Dr. Pande's latest statement about bigadv, an "investment" in a system optimized for bigadv might not be the best choice now...
2) The use of big adv depends on having very big simulations to run (i.e. lots of atoms). This will come and go as time goes on. we are working on rolling out new SMP WUs, but most of them are smaller systems at the moment, since that's what's needed right now. I think everyone would agree that big adv should help push the science and not the other way around, i.e. we shouldn't design science purely to keep big adv fed if it's not useful to our cause. With this said, I am looking to see what can work based on what's going on.
http://foldingforum.org/viewtopic.php?f=16&t=16474#p163725
2) The use of big adv depends on having very big simulations to run (i.e. lots of atoms). This will come and go as time goes on. we are working on rolling out new SMP WUs, but most of them are smaller systems at the moment, since that's what's needed right now. I think everyone would agree that big adv should help push the science and not the other way around, i.e. we shouldn't design science purely to keep big adv fed if it's not useful to our cause. With this said, I am looking to see what can work based on what's going on.
http://foldingforum.org/viewtopic.php?f=16&t=16474#p163725
That seems to be in line with what he said earlier that bigadv may not always be available. I will be happy to crunch anything, so long as it is absolutely stable. But it sounds like those of us exclusively running bigadv SMP are probably going to get hit in the near future once bigadv is not as available.
zero2dash
Supreme [H]ardness
- Joined
- Oct 23, 2007
- Messages
- 6,100
Technically we're already getting hit, it just might get worse.
I've pulled 1 bigadv so far this week (or rather, the last 5 days) and it was a 2684. My longterm hope for an onflux of more bigadv (like it was during the *nix bigadv a2 days), I'm starting to doubt that will ever happen again. Hell, I'm starting to doubt that they'll make more bigadv WU period. The shortage has been ongoing for....how many months now? It's been at least 2. It's been worse the last few weeks, but as a whole - even for the last few months, it's been bigadv > regular (sometimes a few regulars) > bigadv. Now it's many regulars > maybe you get lucky and get a bigadv.
I'm not saying WU distribution has ever been "guaranteed", but the odds of pulling a bigadv is definitely a lot less certain these days.
I've pulled 1 bigadv so far this week (or rather, the last 5 days) and it was a 2684. My longterm hope for an onflux of more bigadv (like it was during the *nix bigadv a2 days), I'm starting to doubt that will ever happen again. Hell, I'm starting to doubt that they'll make more bigadv WU period. The shortage has been ongoing for....how many months now? It's been at least 2. It's been worse the last few weeks, but as a whole - even for the last few months, it's been bigadv > regular (sometimes a few regulars) > bigadv. Now it's many regulars > maybe you get lucky and get a bigadv.
I'm not saying WU distribution has ever been "guaranteed", but the odds of pulling a bigadv is definitely a lot less certain these days.
yep. pursuing the "best" PPD hardware has always been a moving target. At least, it has been ever since the introduction of clients beyond the uniprocessor client.
When SMP was introduced, it was the undisputed king of PPD. Then GPU came. Then Bigadv and bonus A3 came.
Now, it truly appears GPU is back on top.
A "budget" SR2 rig will still be in the neighborhood of $1000, at a very minimum, and as much as $3000, depending on the CPUs used. And it will produce anywhere from 40k ppd (6701s on dual quad cores) and as much as 150k ppd (2686 on dual hex).
However, you can add GTS450s, at about $100 a piece, for about 10k ppd a piece. So the PPD that an SR2 rig produces for between $1000 to $3000 only costs $400 to $1500 if you use GPUs instead.
This is simplified, and not entirely accurate, but you get the point.
If you want big points today, your easiest, cheapest, and probably most consistent way to upgrade is to buy GTS450s or GTX460s.
Assuming you have the PCIe slots and PSUs to get the job done.
When SMP was introduced, it was the undisputed king of PPD. Then GPU came. Then Bigadv and bonus A3 came.
Now, it truly appears GPU is back on top.
A "budget" SR2 rig will still be in the neighborhood of $1000, at a very minimum, and as much as $3000, depending on the CPUs used. And it will produce anywhere from 40k ppd (6701s on dual quad cores) and as much as 150k ppd (2686 on dual hex).
However, you can add GTS450s, at about $100 a piece, for about 10k ppd a piece. So the PPD that an SR2 rig produces for between $1000 to $3000 only costs $400 to $1500 if you use GPUs instead.
This is simplified, and not entirely accurate, but you get the point.
If you want big points today, your easiest, cheapest, and probably most consistent way to upgrade is to buy GTS450s or GTX460s.
Assuming you have the PCIe slots and PSUs to get the job done.
rmdashrrootsplat
Gawd
- Joined
- Apr 2, 2008
- Messages
- 921
I think I'm going to -oneunit my bigadv box running on WINE. Yeah it finishes units within the preferred deadline but only gets about 30k PPD if all it gets is bigadv. Those units could be better used on boxen with better PPD. I'll switch back to native linux SMP and try to make a dent on this 670x surge.
Yep, getting sick of this roller-coaster ride myself.yep. pursuing the "best" PPD hardware has always been a moving target. At least, it has been ever since the introduction of clients beyond the uniprocessor client.
When SMP was introduced, it was the undisputed king of PPD. Then GPU came. Then Bigadv and bonus A3 came.
To some degree, it still requires a lot of cards to get into the top 10.
Yes, particularly since those cards actually top out at ~15k when OCed a substantial but attainable amount. The 450s are especially noteworthy in the $/PPD department.