New bigadv project - 2685

Good news: I just started bigadv on my SR-2 and the first WU is a 2685. I'm seeing seeing 69K PPD while also GPU folding. For reference, running standard SMP I was seeing about 35K with GPU folding (four clients) and 50K with idle GPUs. I'm running two standard clocked X5650s.

Bad news: the mobo has a bad DIMM slot, so I'm shipping it to EVGA on Monday for RMA.
 
Schnikies I'm glad I got my other "dud" boxen oc raised. :eek:
I thought I was stuck at 3.6. I've settled on 3990 (19x210) now that I've gotten it higher. (got bclk all the way to 214 but I can't get 20 or 21 multi to work without bumping Vcc)
2685 is ridiculous now...after 4% I'm getting 32,795 PPD and 74,499 credit on this bad boy.
050.gif

musky wasn't kidding when he said that bigadv loves the ram speed. :D
 
So they are giving these units out again?
Posted via [H] Mobile Device
 
I've gotten one 2685 since they came out. It's been either 2684 or some A3 WU
 
Since resetting one machine to accept -bigadv WUs, all I've received are P6701s. Makes me wonder what is the point of setting the -bigadv flag.
 
Guess I'm picking the perfect time to take my other i7 down then to re-tweak my oc...just added -oneunit. Might as well take advantage now while the bigadv pickin's are slim. :eek:
 
I've gotten two 2685s in a row on my Sr2. I threw myself a small party each time.
Posted via [H] Mobile Device
 
My rig just finished its first 2685 and unloaded 68837 points for the entire work unit with a TPF of 39 minutes. Not bad as this is as good as the old A2 bigadv days.

My rig just picked up another 2685!
 
I think my sr2 has three 2685s in a row now. my 920 keeps getting 2684 though.
 
so I have not seen a bigadv on any of my linux smp's in quite some time. Did they disable this or what? Is bigadv only avail on windows now for some reason? I'd really hate to have to go in and re-image all these machines to windows :( any idea when they'll be turning Linux bigadv back on?
 
so I have not seen a bigadv on any of my linux smp's in quite some time. Did they disable this or what? Is bigadv only avail on windows now for some reason? I'd really hate to have to go in and re-image all these machines to windows :( any idea when they'll be turning Linux bigadv back on?

They turned it off, unfortunately. There is no ETA on when they'll be back either.
 
so I have not seen a bigadv on any of my linux smp's in quite some time. Did they disable this or what? Is bigadv only avail on windows now for some reason? I'd really hate to have to go in and re-image all these machines to windows :( any idea when they'll be turning Linux bigadv back on?

Yeah, unfortunately there was serious bug with the Linux A3 core that affects the stability of bigadv work units. This was why we were having so many crashes and flushes once the 2684 A3 bigadv's came out. They are supposedly working on it, but do you have any copies of Windows around? The 2685's that just came out are yielding some incredible points.
 
I wish they would fix bigadv for Linux...Having all of these boxen running Windows make me feel...dirty...
 
sigh :( what a waste of 2min frames with no bigadv

will have to dig up some windows licenses, too bad you can't farm the F@H workload out to LInux or Windows HPC, would love to hook all these.

Is there a thread somewhere with the progress of the linux bigadv fix?
 
Also another thing I think bears mentioning concerning the OS aspect in regards to -bigadv. Unless my eyes are deceiving me, I think Win 7 produces significantly better performance with A3 than XP. So, if you guys need to migrate or revert back to using Windows, you might see superior performance in Win 7. Again, I could be wrong about the difference, but I swear I am seeing better TPFs in Win 7..
 
I've gotten two or three A3 units on my bigadv box in a row now.
 
yeah exactly.

I've continued to get regular smp units all through the night too. anybody else?
 
Man, this is such an annoying crap shoot. From what I'm seeing, the Windows bigadv units other than the 2685s get lower PPD than regular A3 units. But the 2685s are putting out the highest PPD I've ever seen.
 
yeah exactly.

I've continued to get regular smp units all through the night too. anybody else?

My new folding rig got a couple 6072's last night before it finally drew a 2685 bigadv this morning. My original folding rig just finished a 2685 this morning and managed to draw another 2685.

I have to admit that while I don't mind too much, it would be nice if Stanford made it a little more predictable like during the Linux bigadv days. Back then, getting a standard A3 mixing in didn't happen too often.

For the record, my Gulftown@3.63ghz is netting a TPF of exactly 26 minutes.
 
I ended up going through probably 5 regular A3s including a 6071 before finally grabbing a 2685 yesterday for the first time in several days, so they definitely must be having shortages right now.
 
Man, this is such an annoying crap shoot. From what I'm seeing, the Windows bigadv units other than the 2685s get lower PPD than regular A3 units. But the 2685s are putting out the highest PPD I've ever seen.
The only other bigadv unit is 2684, and it consistently produces significantly higher PPD than normal A3 units (about 17-20k PPD compared to 13-15k with normal SMP and ~10k with P6701/6702).
 
Parja, your refering to your SR2 right? On mine I've noticed this..

There are a few Normal units that make just a hair more than a 2684

there are like 2-3 units however that score much lower than anything, even my i7 lol

running big adv and only getting 2684s gave me a stable and usualy higher ppd over extended periods of time due to the drastic drop I'd see if I ran normal and got a slow unit

now with 2685 out, worst case is you'll pull a constant ppd and get a huge boost any time you snag an '85
Posted via [H] Mobile Device
 
Also, it is my finding that running bigadv keeps your draw fluctuating between normal A3 levels and about 30 watts below that, this leads to a slightly lower power draw over 24 hours vs regular units

if I can figure out how to get my killawatt to do a weekly average, I could compare first hand between reg and big, but I feel safe saying bigadv draws less
Posted via [H] Mobile Device
 
The only other bigadv unit is 2684, and it consistently produces significantly higher PPD than normal A3 units (about 17-20k PPD compared to 13-15k with normal SMP and ~10k with P6701/6702).

This is completely accurate for Core I7s running GPU and SMP A3 units, but I find on dedicated Core I7 boxes, the difference between P2684 and P60xx is only between 1 and 2K PPD, and 5 to 8K PPD if comparing P670x units. I can't really figure this one out, as I'd be assuming the impact of GPU clients running at equal priority should be, well, equal between different SMP type units too.

My 3876mhz Core I7 box (204x19, Ram at 1632mhz) shows very little difference between p60xx units and p2684 units. It has no GPU clients running.

For other architectures that use full cores I don't think the difference would be as high. I will try my GTX 470 on my Magny Cours box and see how that works out.
 
My i7-920@ 200X19 1600mem 7-8-7-20-1t get 21,000ppd on a 2684, 31,000ppd on a 2685. 19-20k on my regular A3's 13-14k on 6701-2 with as many windows services turned off as i can get away with.
 
Okay, i'll update this thread for people who use it for reference.

My Gulftown@3.63ghz just dumped its first 2685 and netted 84284 points for the entire work unit. The work unit completed with 70% of time to deadline remaining, and needed 1.8 to 1.9 days to complete thus yielding me a ppd average of 46454.

Thankfully, it picked up another 2685 right away after finishing this one. My Bloomfield is still working on its 2685 and gets a TPF of 39min thus netting around 25k ppd average.
 
Once you get that 980X dialed in you should be able to 22 minute time frams or lower on a 2685. depending how hard you wanna push it and your cooling configuration. I like the theralright venemous X over the true as the fan attchment allows a better choice of fans. http://www.frozencpu.com/products/8...WM_Fan_-_11311_CFM_AFC1212D.html?tl=g36c15s69 something like this can be controlled to allow for a temp noise balance. In my antec 300 3X120 intake, 120 and 140 exhaust I lost the cpu fan and didnt know it for days until i checked my temps and saw 87C instead of 75c
 
Last edited:
Once you get that 980X dialed in you should be able to 22 minute time frams or lower on a 2685. depending how hard you wanna push it and your cooling configuration. I like the theralright venemous X over the true as the fan attchment allows a better choice of fans. http://www.frozencpu.com/products/8...WM_Fan_-_11311_CFM_AFC1212D.html?tl=g36c15s69 something like this can be controlled to allow for a temp noise balance. In my antec 300 3X120 intake, 120 and 140 exhaust I lost the cpu fan and didnt know it for days until i checked my temps and saw 87C instead of 75c

I'm actually waiting for Kyle's full review of the Corsair H70 to see what i'm going to do. I'm so busy right now that I just don't have the time to fiddle with any of my rigs for a while. Folding1 is running with a dead fan and another fan that is spinning way too slow right now, but it will be left that way unless I see temps hit the mid 70's or higher(currently low 70's). The Cogage is nice, but I can see something like the H70 really working for me, and I really like the preliminary results that Kyle is getting so far.

For now, i'm just going to try and see how far I can take my Gulftown on stock voltage. I am at 3.63ghz(upped the multi 2x) and am going to see if I can take it up one more without boosting voltage. I'm not really an extreme overclocker as I just want to get more than what I paid for. My Gulftown is capable of overclocking as I did take it up past 4.1ghz@1.335 volts the night I first bought it for fun, but took it back to stock voltage since I wouldn't have had time to fine tune for a while.
 
You guys are killing me with your 2685 talk as I'm working on my third 2684 in a row this week. I wish I knew what I did to deserve this sort of punishment.
 
Back
Top