Windows BIGADV is Here

Maybe just get some crappy xp copies off the fs/ft? I guess you could even go 32bit now that the ram req's are so low, I seen a 24 core rig using less than 1.5 so 2 gigs could in theory power a dedicated box, but I'd go3-4 just to be safe
Posted via [H] Mobile Device
 
Well I'm the crash test dummy on that. My i7 930 box has only 2gb.
 
Folding box in my sig. Oh crap, need a cool one word name for it. Umm...Unicron!

TPF is 17:03 for a PPD of 87.5k.

Maybe just get some crappy xp copies off the fs/ft? I guess you could even go 32bit now that the ram req's are so low, I seen a 24 core rig using less than 1.5 so 2 gigs could in theory power a dedicated box, but I'd go3-4 just to be safe
Posted via [H] Mobile Device

Yup, I'm only seeing about 1.4GB of memory used for Folding a 2684, so you could technically get away with just 2GB. Not sure I'd be too stingy on the memory, though. Who knows what future Windows -bigadv units have in store.
 
Still, 19K ppd for Windows bigadv versus 18K ppd for normal A3 through Linux = not even close to worth the expense of Windows licenses.

My next try is using a WinXP 32 bit VM image in Linux...ain't that backward? :)
 
Has anyone tried adjusting the number of cores mid-unit?

I know this would fail in VM world with -bigadv, wondering about Windows. I use my rig to game in the evenings after work so I like to cut back to 6 cores for a couple of hours, then bump back to 8. I switch to 6 and reboot VM or restart SMP then switch back to 8 later with another reboot and no problems at all while not crunching a -bigadv.

If this is only going to use a couple gigs of RAM I would like to bump down to 6 cores and continue to work on -bigadv instead of halting everything while I game.
 
I'll add my numbers here as well

i7 860 @ 3.9GHz TPF of 52:18 with a PPD of 16199, or can just look here for my updated stats for this WU.
 
i7 920 @ 3.89 (21 x 185) : 6GB @ 1476
TPF: 51:13 (16.7K PpD) [not bad considering it is also driving 3 GTX285's on GPU2)

i7 920 @ 3.6 (20 x 180) : 6GB @ 1442
TPF: 51:07 (16.3K PpD) [not bad considering it is driving a 9800GX2)

It is still down 12 to 15K PpD overall for both rigs from A2 Big Adv, but better than normal A3.
 
So this is what I miss when I'm fighting off the EUEs popping up on my GPUs. This morning I deleted the 2.19 core, got the 2.22 core and added the bigadv flag.

Now this is weird, I just saw the ppd on the 6014 I'm working on jump from 22k to 28k. Just from changing cores in mid workunit?
 
That means that their coders have been doing their jobs.
 
Has anyone tried adjusting the number of cores mid-unit?


I did yesterday on an i7 and a dual 5506 machine...no issues with changing from smp to smp 7. I stopped and started the clients multiple times to do other things with no issues either.
 
Has anyone tried adjusting the number of cores mid-unit?

I know this would fail in VM world with -bigadv, wondering about Windows. I use my rig to game in the evenings after work so I like to cut back to 6 cores for a couple of hours, then bump back to 8. I switch to 6 and reboot VM or restart SMP then switch back to 8 later with another reboot and no problems at all while not crunching a -bigadv.

If this is only going to use a couple gigs of RAM I would like to bump down to 6 cores and continue to work on -bigadv instead of halting everything while I game.
why mess with it?

My recommendation, now that we're in the native Windows world, is just to leave FAH running in the background, assuming you have enough RAM. The CPU usage should scale back appropriately to allow your game to perform well.
 
Kendrak, you notice your times coming down? I'm in mid 28's now, havnt changed a thing

so these are good to shut down right? I need to try and up this OC... Gotta shoot for 45k atleast
Posted via [H] Mobile Device
 
Kendrak, you notice your times coming down? I'm in mid 28's now, havnt changed a thing

so these are good to shut down right? I need to try and up this OC... Gotta shoot for 45k atleast
Posted via [H] Mobile Device

Correct @ the bolded - Restarted my system a few times last night to tweak a few things in the BIOS, stability seems to be excellent.

And you may have mentioned this in another post earlier Vaulter, but what are you running that makes you want to get 45k? This is my first time running any bigadv and I'm a tad disappointed in the PPD from it, I thought it would be higher.
 
Glad to see all you -bigadv guys are back earning the big cheese..I really wish I had opted for an i7 now...
Anyone wanna trade?:D

On a positive note, I'm now a lifetime member of GenMay and I've broken into the top 300!
 
Correct @ the bolded - Restarted my system a few times last night to tweak a few things in the BIOS, stability seems to be excellent.

And you may have mentioned this in another post earlier Vaulter, but what are you running that makes you want to get 45k? This is my first time running any bigadv and I'm a tad disappointed in the PPD from it, I thought it would be higher.

He is running dual Xeon's in an SR-2 board I believe..

 
Glad to see all you -bigadv guys are back earning the big cheese..I really wish I had opted for an i7 now...
Anyone wanna trade?:D

I sold my x6 setup for an i7 just so I could run bigadv, just running SMP though from my experience it's about equal. Really thought my 3.9GHz i7 would fair better then 16k PPD running a bigadv - only does 13k with SMP so a 3k increase is worth it. (although lately all I have been pulling while running SMP is the dreaded 6701s which would make my PPD plummet to like 8k, so another pro to bigadv)

He is running dual Xeon's in an SR-2 board I believe..

Ah, gotcha - that makes sense then
 
I sold my x6 setup for an i7 just so I could run bigadv, just running SMP though from my experience it's about equal. Really thought my 3.9GHz i7 would fair better then 16k PPD running a bigadv - only does 13k with SMP so a 3k increase is worth it. (although lately all I have been pulling while running SMP is the dreaded 6701s which would make my PPD plummet to like 8k, so another pro to bigadv)



Ah, gotcha - that makes sense then

I'm currently only pulling 10800K on a damn 6040..They are almost as bad as the 6701's...I usually don't dip below 10K on them, but I can't seem to break 14-15KPPD on other units @ 4.1Ghz..I would sell this combo, but I really can't afford to take a huge loss :(

Are you sure you're running the -bigadv flag? You should be getting way more then 16K PPD @ 3.9Ghz...
 
Well I'm the crash test dummy on that. My i7 930 box has only 2gb.
I'm going to cut back two of my machines from 4GB to 2GB each. Maybe I'll use the spare memory to build another box later this year depending what will be worthwhile doing at that time.

My next try is using a WinXP 32 bit VM image in Linux...ain't that backward? :)
No, not really. Most of my machines are still running XP-32. If you don't need greater than ~3GB of RAM, there's no major advantage to buying newer, more expensive OSes, IMHO.

That means that their coders have been doing their jobs.
For Nehalem at least. I updated the core on 6 machines running older architectures and I can't really notice the difference unless it's only on specific WUs.

My recommendation, now that we're in the native Windows world, is just to leave FAH running in the background, assuming you have enough RAM. The CPU usage should scale back appropriately to allow your game to perform well.
Precisely. If the client was configured to run in idle priority mode, it should be like all previous CPU clients and no need to adjust flags unless one runs concurrent GPU clients, which has proven problematic on my systems with mixed client configurations..

(although lately all I have been pulling while running SMP is the dreaded 6701s which would make my PPD plummet to like 8k, so another pro to bigadv)
Yeah, just by maintaining a moderately high production rate at a steady pace instead of the high/low fluctuations of standard A3, is alone worth running -bigadv. Too bad the P2684 is not as productive as the original A2 -bigadv WUs, but it's better than nothing. Now, if I can just find out how to receive these -bigadv WUs in my Windows clients...
 
I have the -bigadv flag set but have yet to get one. :(
Me neither and I have set the -bigadv flag as well as reconfigured the client to accept big packets and advanced methods. This was done on 2 machines about 12 hours ago... :confused:
 
Me neither and I have set the -bigadv flag as well as reconfigured the client to accept big packets and advanced methods. This was done on 2 machines about 12 hours ago... :confused:

Same here...still nothing. I'm approaching a new unit download in about 20 mins. I'll let you know if I pick anything up. :D
 
Are you sure you're running the -bigadv flag? You should be getting way more then 16K PPD @ 3.9Ghz...

Yup, can see it here (remote stat monitoring FTW!). And I was thinking the same thing in regards to the PPD, was hoping to see 20k from it running bigadv
 
Yup, can see it here (remote stat monitoring FTW!). And I was thinking the same thing in regards to the PPD, was hoping to see 20k from it running bigadv
If it was a 2681, 2682 or 2683 you would have and probably higher. There's a significant difference with the former and current P2684 -bigadv WUs.
 
Yup, can see it here (remote stat monitoring FTW!). And I was thinking the same thing in regards to the PPD, was hoping to see 20k from it running bigadv

I see you've got an old version of the A3 core (2.19). Not sure if it would make a big difference, but you might want to delete it and let your client download the new 2.22 version.
 
I see you've got an old version of the A3 core (2.19). Not sure if it would make a big difference, but you might want to delete it and let your client download the new 2.22 version.

What should I be deleting exactly? New bigadv folder (was doing SMP until it hit Windows).

If it was a 2681, 2682 or 2683 you would have and probably higher. There's a significant difference with the former and current P2684 -bigadv WUs.

This being my luck, when the 6701s first came out I pulled nothing but those on three different SMP clients for days on end.
 
What should I be deleting exactly? New bigadv folder (was doing SMP until it hit Windows).
FahCore_a3.exe. Do it after the current WU completes to be on the safe side.

This being my luck, when the 6701s first came out I pulled nothing but those on three different SMP clients for days on end.
Unfortunately, only the P2684 WUs are available for the Windows client. The older ones were on Linux only and they have been discontinued from the looks of it.
 
Glad to see all you -bigadv guys are back earning the big cheese..I really wish I had opted for an i7 now...
Anyone wanna trade?:D

On a positive note, I'm now a lifetime member of GenMay and I've broken into the top 300!

Don't regret your Thuban one bit. Especially now that we have gone to A3 bigadv, there isn't much of a difference in points between doing bigadv and regular A3 SMP stuff. The diffference for me is only 2k ppd and that is not a bad tradeoff considering that losing a 2684 means that you may have lost out on almost 4 days worth of work. I am going to do bigadv so long as they are stable in Windows simply because I like big things.
 
What should I be deleting exactly? New bigadv folder (was doing SMP until it hit Windows).

Shut down the client, then go into the folder where you installed the client. Delete FahCore_a3.exe, then start the client back up. It'll download a new core executable automagically.

Some will caution against replacing the core file in the middle of a WU, so do it at your own risk. I've never had a problem with it, though.
 
Don't regret your Thuban one bit. Especially now that we have gone to A3 bigadv, there isn't much of a difference in points between doing bigadv and regular A3 SMP stuff. The diffference for me is only 2k ppd and that is not a bad tradeoff considering that losing a 2684 means that you may have lost out on almost 4 days worth of work. I am going to do bigadv so long as they are stable in Windows simply because I like big things.
Stanford should really revise the way they evaluate the credit for some WUs. There's got to be a better way of doing it than simply benching them on systems and assign value based on a set of numbers. Other criteria like stability and average time to completion (client commitment) in addition to TPF should weigh in somewhere, IMO.
 
Some will caution against replacing the core file in the middle of a WU, so do it at your own risk. I've never had a problem with it, though.
It happened to me on at least one occasion in the past. Granted, this was on a GPU client but I still add a disclaimer just in case someone doesn't wish to risk it. If it's near the start of a WU, say under 10 frames or thereabouts, then I'd probably do it since the risk is less.
 
Stanford should really revise the way they evaluate the credit for some WUs. There's got to be a better way of doing it than simply benching them on systems and assign value based on a set of numbers. Other criteria like stability and average time to completion (client commitment) in addition to TPF should weigh in somewhere, IMO.

Well, and they're currently benchmarking the WUs based on one data point. Depending on how well the WU scales, higher-end machines can get pretty wildly varying PPD.
 
PPD is down for me on these as well (core not the problem...every SMP2 client I'm running has 2.22 core).
Windows -bigadv seems to be affected a lot more by simultaneous gpu client(s), linuxrouter VM was not this way IME. I would average no less than 18k PPD and 61k credit on my 2x 260 machine.

Windows 2684 [email protected] I'm getting 53:41 TPF (that's at least 4 minutes higher), 15.6k PPD and just over 58k credit.
[email protected] (dedicated boxen, no gpu clients) I'm getting 48:40 TPF, 18.1k PPD and just over 61k credit (that sounds more like it).
 
Well I'm the crash test dummy on that. My i7 930 box has only 2gb.


i think the big deciding factor is the OS you use.. since windows 7 on avg uses between 1-1.4gigs and windows vista uses 1.2-1.8gigs depending on how much memory you have installed.. now if 1.4 gigs is your total memory usage including your OS and the winSMP client then thats awesome..

PPD is down for me on these as well (core not the problem...every SMP2 client I'm running has 2.22 core).
Windows -bigadv seems to be affected a lot more by simultaneous gpu client(s), linuxrouter VM was not this way IME. I would average no less than 18k PPD and 61k credit on my 2x 260 machine.

Windows 2684 [email protected] I'm getting 53:41 TPF (that's at least 4 minutes higher), 15.6k PPD and just over 58k credit.
[email protected] (dedicated boxen, no gpu clients) I'm getting 48:40 TPF, 18.1k PPD and just over 61k credit (that sounds more like it).

a lot of it is due to the windows overhead because windows has to make sure everything else is usable on the system.. so when you load a program then it chokes everything in the background so that program loads and with the windows default settings it gives resource priority to programs vs things running in the background like the SMP client when its minimized.. you could try setting your resources to background programs instead programs in your advanced settings under my computer/properties and see what happens.. ive never done it with the smp client running so i have no clue if it will work..
 
Last edited:
Don't regret your Thuban one bit. Especially now that we have gone to A3 bigadv, there isn't much of a difference in points between doing bigadv and regular A3 SMP stuff. The diffference for me is only 2k ppd and that is not a bad tradeoff considering that losing a 2684 means that you may have lost out on almost 4 days worth of work. I am going to do bigadv so long as they are stable in Windows simply because I like big things.

So you're telling us you like "big things?":D..Wayyyy to much info Mr. Gamer...

I just checked my stats on Kakao, and it says for the last 7 days I've done 138,796 points, so that is an average of 19828K PPD...That seems decent before you factor in that I have 2 highly o/c'd 8800 series cards churning away as well..Hell, my 8800GT is pulling in over 5100-5500K PPD depending on the WU alone...

I am going to be bring an AMD quad SMP box online next week, and it will run 24/7 as well...

At least you're lawn mower won't be getting much use in my grass;)

 
a lot of it is due to the windows overhead because windows has to make sure everything else is usable on the system.. so when you load a program then it chokes everything in the background so that program loads and with the windows default settings it gives resource priority to programs vs things running in the background like the SMP client when its minimized.. you could try setting your resources to background programs instead programs in your advanced settings under my computer/properties and see what happens.. ive never done it with the smp client running so i have no clue if it will work..

Hmm interesting thought.
I would think with the amount of ram and free disk space I have, it wouldn't choke anything but then again obviously I don't see what happens in the background like that.

I would figure I'd get better performance native vs. VMware player+a VM though....that's what baffles me. Maybe the optimizations still aren't there yet. :confused:
 
I just threw my hat into the ring and added the -bigadv tag. Be interesting to see what happens since my folding box seems to the the family Daily Driver :rolleyes:
Guess I need to upgrade everyones boxes so they stay off mine.

Seriously, I actually drain more than they do as I'm the only one that actually games on this thing, so I'm curious to see what kind of impact I get. I was going to hold off to see if anyone else was going to try gaming while these were running, but we have too many dedicated boxen here it seems :D
 
I just threw my hat into the ring and added the -bigadv tag. Be interesting to see what happens since my folding box seems to the the family Daily Driver :rolleyes:
Guess I need to upgrade everyones boxes so they stay off mine.

Seriously, I actually drain more than they do as I'm the only one that actually games on this thing, so I'm curious to see what kind of impact I get. I was going to hold off to see if anyone else was going to try gaming while these were running, but we have too many dedicated boxen here it seems :D


you could game with it running no problem, thats the advantage to running the winSMP client.. but you wont hit the deadline.. for example running MWLL(crysis mod) the game uses roughly 25-45% of my phenom II 940.. no gaming TPF on a 6013 is 5:30ish.. gaming it hits about 15 minutes.. so if you figure running a standard A3 WU adding 10 minutes or so to my TPF your looking at probably a 70-80 minute TPF on the bigadv wu's.. but thats just a rough guesstimation..
 
I know I can game with it, was just wondering about how it'd do with the deadline... guess there's only one way to find out :D

actually wont matter a whole lot, what with all the 6701s I've been getting lately, PPD wont more a whole lot even without the bonus:(
 
Still nothing....:(

You arn't getting any of these bigadv work units?

So you're telling us you like "big things?":D..Wayyyy to much info Mr. Gamer...

I just checked my stats on Kakao, and it says for the last 7 days I've done 138,796 points, so that is an average of 19828K PPD...That seems decent before you factor in that I have 2 highly o/c'd 8800 series cards churning away as well..Hell, my 8800GT is pulling in over 5100-5500K PPD depending on the WU alone...

I am going to be bring an AMD quad SMP box online next week, and it will run 24/7 as well...

At least you're lawn mower won't be getting much use in my grass;)


Yeah, I like everything big. I like big cars, I like big computers, I like my burgers thick, I like my steaks sinfully huge, and I like my folding work units Dolly Parton BIG!

I probably won't be able to mow you anytime soon, but that is okay as we fold for the science, and the competition and fun we have amongst ourselves is like the super creamy icing on top of a rich moist corn mix cake.
 
Just got home from work and I've got a Win SMP 2684 running right now.

Here are some numbers on the rig in my signature taken from HFM.NET Benchmark viewer showing the difference between the 2684 WU in the VM and Windows:
2684 in VM: avg tpf 35:32 for 29k ppd
2684 Win SMP(in progress): avg tpf 32:21 for 33.4k ppd
 
Back
Top