• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Um Wow. P4 3.73EE

Oldie

Oldie

Mean Old Administrator
2FA
Joined
Jan 12, 2004
Messages
21,915
A friend of mine just got a 3.73EE (yeah, he's a rich bastard), and we loaded FAH console on it, and he pulled this monster:

p1140_RIBO_FSpeptide_EXT_nospring

That's a 600 pointer right there. He d'lded it about 12 hours ago and it's at 91%. Looks like my stats are going up :D. That's some serious power. My main rig is a P4 3.0 Prescott and that would have taken me at least 2 days@24/7.

So, just out of curiosity, what's the most badass system you have folding?

 
bastard.... I mean... umm... Congrats on the add!! :D


Keep on Folding!!

 
Most badass system = p4 2.4c at stock :(

I'm working on it, though...

 
good borg!!

most kick ass system hmm either my a64 3000+ with 2 gigs o ram or my dads 3.2p4 with 1 gig ram
 
3.0C @ 3.6 is as good mine get ><

does FTW mean "for teh win"? for some reason that's the first thing that comes to mind when i see it...and i haven't seen it before..

 
gwai lo said:
3.0C @ 3.6 is as good mine get ><

does FTW mean "for teh win"? for some reason that's the first thing that comes to mind when i see it...and i haven't seen it before..

Click to expand...

LoL for teh win. It means "for the win" :)
 
yeah old joke on a forum i was on...eh close nuff, teh win +1!!11

 
I'd have to say my main rig is my biggest producer, although the hardbowl kind of has it taken hostage. 24/7 3GHz of OpterOWNage tends to be fairly nice (it's just under, but no damn time to tweak it to not overvolt 3). :)

Full Folding Ahead!

 
unhappy_mage said:
Most badass system = p4 2.4c at stock :(

I'm working on it, though...

Click to expand...

Holy crap, I totally missed that too. A 2.4C at stock??!?!?! :eek: That thing should do 3.4GHz without a sweat IIRC, those 2.4C's were mean machines!

Full Folding Ahead!

 
My biggest one would be the 2.8 Preshott as long as it's pulling QMD's or my XP-m@2.5. If the P4 gets normal WU's, my XP will tear it a new asshole. I still prefer AMDs for overall production. One of these days I'll have a dual core AMD which will make the P4 it's bitch no matter what. :D
 
my most badass system is my main box, dual opteron 246 2.0 GHz, folding 24/7. all i get are 600 pointers. prob 19 out of every 20 WUs it gets are 600 pointers. dual cpu box means two 600 pointers at a time, all the time :) also just got an athlon xp 3000+ and a sempron 3300+, both 24/7.
 
3.6@3.8 GHz. Yeah, I know, I am already in shorts in my den. Any more heaters and I will have to break out the towels for the sauna.
 
Most badass would have to be my humble Opteron 252 @ 2.6GHz. Someday I'll pick up a second 252... when prices come down.

Puts out about 340ppd on average. Ranges from 100-480ppd. Lately I've been getting nothing but the small 40-50 pointers..... pulling in 100ppd........ this sucks!

~Towert7
 
tsuehpsyde said:
Holy crap, I totally missed that too. A 2.4C at stock??!?!?! :eek: That thing should do 3.4GHz without a sweat IIRC, those 2.4C's were mean machines!

Full Folding Ahead!

Click to expand...
Well, it does 3.0 okay, but then I have to run my RAM with a divider, and it doesn't end up going any faster. Apparently this thing is starved enough for memory bandwidth that clock speed doesn't make much difference. My theory, anyways.

rayman2k2: You want it? I may be looking into selling it in the relatively new future. Chip only or the whole set?

 
towert7 said:
Most badass would have to be my humble Opteron 252 @ 2.6GHz. Someday I'll pick up a second 252... when prices come down.

Puts out about 340ppd on average. Ranges from 100-480ppd. Lately I've been getting nothing but the small 40-50 pointers..... pulling in 100ppd........ this sucks!

~Towert7
Click to expand...

Yeah I've had my opty 170 running for a couple weeks and I still haven't pulled a 600 pointer with either core, but it still puts out a nasty amount of points at 2.63ghz.

Oldie, too bad your friend didn't wait for the FX-60, now THAT woulda been a bad ass folder!
 
ICE_9 said:
3.6@3.8 GHz. Yeah, I know, I am already in shorts in my den. Any more heaters and I will have to break out the towels for the sauna.
Click to expand...
:eek: IN ARIZONA !! :eek: You could probably teach those 'roos a thing or 2 about cooling, but I hope you don't !!

 
I've had my opty 170 running for a couple weeks and I still haven't pulled a 600 pointer with either core
Click to expand...
that is strange. 90% of the time, i'm running 600 pointers on each cpu. i need to replace these single core chips with dual cores, 4x 600 pointers on the same machine. folding paradise!

dualdrop[H]enia
 
Are you setting both the advanced methods and the Big WU flags? I'm just setting the advanced and it only comes up with the smaller cores. I just don't want to get one of those QMD cores -- since they run slow on AMD systems.
 
Technically you should not be getting QMDs on any AMD system. They do not have it AMD compatible yet. There was an issue when some AMD system received QMDs but that is rare. I dont think Stanford has a timeline on when the QMDs will come on AMD based systems.

 
i set the big WU flag, but not the advmethods. i tried advmethods once before (a few weeks ago), hoping for only 600 pointers, and i got none. looking at my weekly stats, you can see when this was done. i set advmethods to off, and i got 600 pointers again.

dualdrop[H]enia
 
santaliqueur said:
i set the big WU flag, but not the advmethods. i tried advmethods once before (a few weeks ago), hoping for only 600 pointers, and i got none. looking at my weekly stats, you can see when this was done. i set advmethods to off, and i got 600 pointers again.

dualdrop[H]enia
Click to expand...

Yeah that might be my problem, I've got both the big work unit and the advanced methods flags turned on. Maybe I'll try them without it and see what happens.
 
600-pointers are not advmethods WUs. They are just Big WUs. There are basically 5 types of WUs:

Normal: No special flags set, <20MB RAM taken all the time
Timeless: enabled in client.cfg. Same as Normal, except there's no deadline for getting the WU done.
Big WUs: bigpackets=yes in client.cfg, but no -advmethods flag on executable. Allows 600-pointers, which take ~110MB RAM
Advmethods: -advmethods flag on the executable, but no bigpackets=yes in client.cfg. Allows for beta proteins that don't take a lot of RAM
(original) QMD: bigpackets=yes, -advmethods flag when run. Allows the monster WUs that take up to 300+ MB RAM. Great on P4's for lots of PPD
 
Mohonri said:
600-pointers are not advmethods WUs. They are just Big WUs. There are basically 5 types of WUs:

Normal: No special flags set, <20MB RAM taken all the time
Timeless: enabled in client.cfg. Same as Normal, except there's no deadline for getting the WU done.
Big WUs: bigpackets=yes in client.cfg, but no -advmethods flag on executable. Allows 600-pointers, which take ~110MB RAM
Advmethods: -advmethods flag on the executable, but no bigpackets=yes in client.cfg. Allows for beta proteins that don't take a lot of RAM
(original) QMD: bigpackets=yes, -advmethods flag when run. Allows the monster WUs that take up to 300+ MB RAM. Great on P4's for lots of PPD
Click to expand...

A couple of corrections/clarifications:

Timeless are not the same as normal. Until recently they were Tinker cores only. People have said Gromacs core WU's have made their way to this now. It's been a while since I've grabbed any new ones so I'm not totally sure. These were also great fold older CPU's that don't have SSE like the old T-Birds and Spitfire core Durons since AMDs chewed through Tinker core proteins in no time flat compared to P4's. (I have a T-Bird 1.4 that will chew through Tinkers at 110 PPD whereas my P4 2.8 will do the same)

Bigpackets only will get you 600 and 364 point proteins. On my AMD machines the 364 pointers get better PPD. 275 PPD for 600 pointers and around 300-320 PPD (I think) on my Athlon XP-m@2.5Ghz. Also, you must have at least 256 meg of RAM to receive these units.

I'm not toally sure about advmethods. I haven't run it on an AMD machine in a long while. You can get beta proteins when running this. This means the proteins may not complete completely before being sent back. Getting Early_Unit_Ends is somewhat common when running these. I'm not sure about the points value. I thought at least some of the beta units were given a little higher points values due to their beta nature. Someone else will have to chime in on this.

bigpackets & advmethods. As I said, I haven't run this on an AMD machine in a long time so I'm not sure what you will get out of this. Right now I think this may get you monster WU's that take a long time to process. They are large point value proteins but there is no points bonus for cruching them. They get the same PPD as regular small Gromacs.
On a P4, these settings can get you QMD's if you have at least 512 meg of RAM. They are RAM and bandwidth monsters.I'm currently running one on my work machine here. It's chewed through a lot of these things. They will give some of the highest PPD numbers you will see. My P4 2.8 will average 405-410 PPD on these. BIG WARNING: do not run these on a borg or production machine that requires speed. They can easily bog down your system. My work machine crawls since it only has 512 meg of RAM. That's fine for me as I don't need anything to run fast in most cases and it does no gaming whatsoever.

As a final note, you are not guaranteed any certain type of work unit no matter the settings. Based on the way the client is setup, Stanford will send whatever they have at the moment that matches the settings of your client. I have this P4 setup with bigpackets and advmethods. 95% of the time I get QMD's. I can and have gotten regular gromacs, tinkers, and amber core proteins. I've also gotten bigpacket 600 point proteins.

If you have an AMD setup with the bigpacket options only, you will never get a QMD (as long as Stanford did fix that screwup) until and unless Stanford is able to use Intel's compiler for SSE2 & 3 to get the client optimized for A64's. I'm not going to get into either issue right now.

 
SmokeRngs said:
A couple of corrections/clarifications:

Timeless are not the same as normal. Until recently they were Tinker cores only. People have said Gromacs core WU's have made their way to this now.
Click to expand...
You are correct. I guess I should have made it clearer when I originally posted it. I meant that currently, running timeless vs normal won't necessarily make a difference in the type of WUs you get--you can get Tinkers and Gromacs either way. The timeless Gromacs WUs are a fairly recent development. In practice, the only difference between running timeless WUs and normal ones is that the timeless WUs have no deadline.

 
here at work we just got some new dells. I just wish I could run this thing overnight and let it go, but we can't do that. I also just reinstalled FAH on my home machine, as the electricity isn't really the costly part of my energy bill this time of year. That one will be the better producer for me since it can run 24/7.
 
SmokeRngs said:
A couple of corrections/clarifications:
Snip.......

bigpackets & advmethods. As I said, I haven't run this on an AMD machine in a long time so I'm not sure what you will get out of this. Right now I think this may get you monster WU's that take a long time to process. They are large point value proteins but there is no points bonus for cruching them. They get the same PPD as regular small Gromacs.

snip.............
Click to expand...

The moster DGromacs are optimized to run under SSE2 boost.
If you can run them under that boost they will give you a +150% boost on PpD as running without.

Luck........... :D
 
SmokeRngs said:
A couple of corrections/clarifications:

Timeless are not the same as normal. Until recently they were Tinker cores only. People have said Gromacs core WU's have made their way to this now. It's been a while since I've grabbed any new ones so I'm not totally sure. These were also great fold older CPU's that don't have SSE like the old T-Birds and Spitfire core Durons since AMDs chewed through Tinker core proteins in no time flat compared to P4's. (I have a T-Bird 1.4 that will chew through Tinkers at 110 PPD whereas my P4 2.8 will do the same)

Bigpackets only will get you 600 and 364 point proteins. On my AMD machines the 364 pointers get better PPD. 275 PPD for 600 pointers and around 300-320 PPD (I think) on my Athlon XP-m@2.5Ghz. Also, you must have at least 256 meg of RAM to receive these units.

I'm not toally sure about advmethods. I haven't run it on an AMD machine in a long while. You can get beta proteins when running this. This means the proteins may not complete completely before being sent back. Getting Early_Unit_Ends is somewhat common when running these. I'm not sure about the points value. I thought at least some of the beta units were given a little higher points values due to their beta nature. Someone else will have to chime in on this.

bigpackets & advmethods. As I said, I haven't run this on an AMD machine in a long time so I'm not sure what you will get out of this. Right now I think this may get you monster WU's that take a long time to process. They are large point value proteins but there is no points bonus for cruching them. They get the same PPD as regular small Gromacs.
On a P4, these settings can get you QMD's if you have at least 512 meg of RAM. They are RAM and bandwidth monsters.I'm currently running one on my work machine here. It's chewed through a lot of these things. They will give some of the highest PPD numbers you will see. My P4 2.8 will average 405-410 PPD on these. BIG WARNING: do not run these on a borg or production machine that requires speed. They can easily bog down your system. My work machine crawls since it only has 512 meg of RAM. That's fine for me as I don't need anything to run fast in most cases and it does no gaming whatsoever.

As a final note, you are not guaranteed any certain type of work unit no matter the settings. Based on the way the client is setup, Stanford will send whatever they have at the moment that matches the settings of your client. I have this P4 setup with bigpackets and advmethods. 95% of the time I get QMD's. I can and have gotten regular gromacs, tinkers, and amber core proteins. I've also gotten bigpacket 600 point proteins.

If you have an AMD setup with the bigpacket options only, you will never get a QMD (as long as Stanford did fix that screwup) until and unless Stanford is able to use Intel's compiler for SSE2 & 3 to get the client optimized for A64's. I'm not going to get into either issue right now.

Click to expand...

So basically you're saying that I should take off the -advmethods flag off my opty consoles to get 600 pointers? I definitely have the big WU flag on, we'll see how it goes. And yes, QMD's are monsters for both RAM and PPD. The 3.2ghz P4 I'm sitting at here at work shows a noticable decrease in performance with one running, but at about 420PPD I don't really notice that my compile times take about 2 seconds longer than normal.
 
Imitation said:
So basically you're saying that I should take off the -advmethods flag off my opty consoles to get 600 pointers? I definitely have the big WU flag on, we'll see how it goes. And yes, QMD's are monsters for both RAM and PPD. The 3.2ghz P4 I'm sitting at here at work shows a noticable decrease in performance with one running, but at about 420PPD I don't really notice that my compile times take about 2 seconds longer than normal.
Click to expand...
Precisely. It's still possible to get 600-pointers if you use -advmethods, but it's far less likely. Using only the bigpackets will give you the greatest probability of getting 600-pointers. In order to get the juicy QMDs, you have to have both bigpackets and -advmethods.

 
It's been my experience that running WITHOUT the -advmethods flag, greatly decreases folding speed. On some projects, I've seen ~50% improvement using the -adv flag. (on amd cpus at least, not sure about intel)

What good is getting a 600pt protein, when it'll take 60hrs to crunch withouth -advmethods, but only 40 hrs with -advmethods?

I'd just leave bigpackets = yes and -advmethods on. Try using -forceasm as well (may or may not give any performance boost, but if folding crashes, it won't restart with -advmethods off)
 
Mohonri said:
Precisely. It's still possible to get 600-pointers if you use -advmethods, but it's far less likely. Using only the bigpackets will give you the greatest probability of getting 600-pointers. In order to get the juicy QMDs, you have to have both bigpackets and -advmethods.

Click to expand...

Not entirely so. I've had a good run on 2 of my home boxes, where all I got for 2 months was 600ptrs. And they both use the -advmethods / bigpackets=yes configs.
I still think it's more "luck of the draw", and connecting to the same wu server at stanford than anything.
 
Mohonri said:
Precisely. It's still possible to get 600-pointers if you use -advmethods, but it's far less likely. Using only the bigpackets will give you the greatest probability of getting 600-pointers. In order to get the juicy QMDs, you have to have both bigpackets and -advmethods.

Click to expand...

Exactly.

 
600-pointers are not advmethods WUs. They are just Big WUs
Click to expand...
good to know, i wasnt aware. the only thing i knew is that i was getting many 600 pointers before and after i set the flag, and none during it.

dualdrop[H]enia
 
Oldie said:
My main rig is a P4 3.0 Prescott and that would have taken me at least 2 days@24/7.
Click to expand...

Wow, My Celeron D can crank out a 600 pointer in about 2 1/2 days. I guess thats not too shabby. I was thinking about upgrading to a P4 for the same pc but maybe It would'nt make that much of a difference.
 
I just picked up a 600 pointer, 26.5 minutes a frame so far, not too shabby! Considering my XP-M was over 30 mins a frame. To you guys that have the advmethods flag on, when I had it on I got mostly tinkers and double gromacs. The doubles are pretty quick cause of the SSE2 boost but I think the 600 pointers are still gonna give me more ppd.
 
I'm working a couple of QMD cores right now, and EMIII computes it out to 36 hours. Those will get 450 points each (p1911.)

I did get a p2078 with my BigWU off and ADV on -- on the Opty. p2078 must be Beta as it doesn't even show up in the summary stats. It's 244 points.

UPDATE: On my Quad-core machine, I just picked up a 722! p2363. EMIII projects 62 hours to completion! :eek:

em3.bmp
 
You must log in or register to reply here.
Back
Top