average time per frame [Archive] - Overclockers Forums

belorsch

01-16-02, 11:32 AM

Originally posted by JetMech
We need to do a time study on the Duron vs the Tbird. I have seen some very fast times posted for Durons and unless I'm mistaken they are not the same core as Thunderbird. What is making the Durons Fold so much faster. Should we all be buying Durons for Folding boxes? If Durons are faster then those who are economically challenged, but want to fold competitively, can achieve that goal sooner.

On top of building a new pc I'm wanting to upgrade my p3 machine with a ECS mb and 900tbird but if a 1100 duron would fold faster at about the same or better price than I might be able to afford a 1200 duron. Do they make those?:beer:

lennytiger

01-16-02, 11:48 AM

i've never heard of a 1.2 Duron. Athlons like 1GHz AXIA can be oc'ed to 1.3 in a decent board, so u can get one of those for roughly the same price!! :)

belorsch

01-16-02, 12:16 PM

Originally posted by lennytiger
i've never heard of a 1.2 Duron. Athlons like 1GHz AXIA can be oc'ed to 1.3 in a decent board, so u can get one of those for roughly the same price!! :)
I've heard the K7S5A doesn't overclock too well. The P3 has a 256MB stick of cas2 crucial I want to be able to reuse. If there was a reliable/cheap KT133a mb out there that can do most of the things the ECS can and be a good overclocker I hope someone lets me know;)

lennytiger

01-16-02, 12:21 PM

Asus are all good make sure its a KT133A or above though...

SickBoy

01-16-02, 01:28 PM

Let's see...

If you can find an AXIA 1 GHz/200 chip, it should be able to tolerate an overclock to 1.33 by upping FSB to 133 and penciling in your L7's... My AXIA-Y does 1333 at 1.775 volts (which is just above stock).

lennytiger

01-16-02, 02:33 PM

yes SB, use a KT133A or better and then a 1GHz Athlon/200FSB, should work great!! :)

JetMech

01-17-02, 02:39 AM

Originally posted by belorsch

On top of building a new pc I'm wanting to upgrade my p3 machine with a ECS mb and 900tbird but if a 1100 duron would fold faster at about the same or better price than I might be able to afford a 1200 duron. Do they make those?:beer: 1.2 Duron Retail - Googlegear.com - $88.00. Sixteen dollars less for oem.

1100 Duron = $70.00

900 Athlon = $58.00

1.4 Athlon Tbird = $100.00

Again the question. Which will give faster fold times?

KFB

01-17-02, 03:32 AM

I am glad to see this has turned into an interesting thread. JetMech, you brought up a good question.
Diatribe follows:
I gave myself the name KFB before I bought my should-be-on-a-keychain 1.33 TBird. I guess I presumed I would end up frying the poor 'Bird. I unwittingly tried to by running my system with NO FANS AT ALL with win98 in standby while I slept one night. (yeah i was dumb!!!) My dog must have bumped the desk or something and awakened the sleeping beast b/c I woke up to the smell of melting plastic as the fan on the generic hs melted! The screen was black except for 1 line of text that said fatal exception or something like that. I shutdown immediately and cleaned up. Much to my surprise, the tough Bird actually still worked when I got a CoolerMaster hsf a few days later. But, I soon crushed the core as I tried to attach my first homemade water block. So, I bought a Duron750 a few days (weeks?) later. The point is that I have since bought 2 more Durons b/c
1. I am a poor student (economically challenged!! Ha!)
2. I noticed hardly much difference in performance between the Duron750 and the TBird850 that I bought after my voltmod fried the Duron.
So, JetMech, I agree, we should analyze the folding performance difference between Tbirds and Durons with the same clock speeds. Check my sig, I have a Duron750@1000 (it may be capable of more) and my mom's comp. is a Duron600@800 - check system specs here (http://forums.overclockers.ws/vb/showthread.php?threadid=58662) So, if someone can provide a similarly-clocked Athlon or 2, we could compare. We could set the memory timings to the same level so as to minimize variables and all that science stuff. The only diffs would probably be HD, mobo. Let me know what you think (if anyone reads this)

edit: oh yeah, the time per frame depends on the protein, so we would have to take that into account, too

JetMech

01-17-02, 04:05 AM

s

KFB thanks for the pm. My real interest in all this is to open up the possibility of increasing our folding farms by more efficiently spending our resourses on equipment that will do the mostest for the leastest. I have dumped quite a bit into this farm but when I think about the fact that by possibly using durons instead of Athlons I could have a fourth machine folding, giving me a higher output for the same amount of money. My P4 is the work machine because it is the slowest folder and any lost cycles are negligible. The amount I paid for the XP could have bought 3 OEM Duron 1.2's. The motherboard in that rig could have bought 2 ECS boards with enough left over for two sticks of ram. This thing is only folding. I think that as well as recruit new folders we can assist the team by helping them make better Farm system decisions.

KFB

01-17-02, 05:02 AM

i agree with your philosophy of efficiently spending our resourses on equipment that will do the mostest for the leastest
that's why i run durons, although that's not saying i wouldn't like to have about 5 XP2000+ rigs, though!;)

Ploaf

01-17-02, 06:40 AM

Clock for clock durons and tbirds are almost identical for folding. The differences in times are usually due to different clock speeds. I have matched a Duron at 1000 with a 1000 tbird and they ran pretty much neck and neck. The setups were pretty much the same except for the cpu. Afterward the tbird was put right back up to 1.4 and now it kicks the snot out of the duron.
If you are planning to purchase more than one machine then somthing else to consider in the bang/buck model is that you may be able to hook up with a single XP2000 and OC it for all it's worth for less than it would cost to get 2 durons up and running. Remember that you are not just buying a cpu, but also a mobo, ram, hd, vid, nic(maybe), various drives, psu/case, cables, electricity to run all that extra stuff. There are advantages to having more than one system, though. so either way it works.

belorsch

01-17-02, 11:11 AM

Duron 750 8.77 min/f contrib KFB
P3 733 11.41 min/f
P3 @600 15.08 min/f belorsch
Tbird @1600 3.99 min/f S_Wilson
Tbird @1133 5.47 min/f
Tbird @1430 4.5 min/f res0r9lm
Tbird 1400 4.63 min/f JetMech

XP1800 @1750 3.54 min/f S_Wilson
XP1800 @1670 3.73 min/f JetMech

If any of this wrong let me know. I'm hoping these were all done with villins to kind of standardize things. @= "overclocked to". We need more for comparison.

lennytiger

01-17-02, 11:29 AM

Originally posted by JetMech
1.2 Duron Retail - Googlegear.com - $88.00. Sixteen dollars less for oem.

1100 Duron = $70.00

900 Athlon = $58.00

1.4 Athlon Tbird = $100.00

Again the question. Which will give faster fold times?

THE 1.4 TBIRD!! :) LOL
ATHLON=FASTER..............

res0r9lm

01-17-02, 03:25 PM

on pricewatch they have 1300 duron $95 eta febuary

jbslow

01-17-02, 06:19 PM

Originally posted by belorsch
We need more for comparison.

Ok here you go

duron 800@1050 = 5.95min per frame (villan atom 392)
athlon 1200@1444 = 4.35min per frame (villan atom 392)

Loud

01-17-02, 08:25 PM

Originally posted by jbslow

Ok here you go

duron 800@1050 = 5.95min per frame (villan atom 392)
athlon 1200@1444 = 4.35min per frame (villan atom 392)

Every way I do the math, the Duron is working more efficiently. It is, however, a VERY SLIGHT margin. And, of course, the bottom line shows that the Athlon is accomplishing more work in real time.

What was the point of all this again? :cool:

belorsch

01-17-02, 09:22 PM

What was the point of all this again? :cool: [/B]

To help edumacate some of us on how to build a more cost efficient folding machine. I think one of my proteins hath done gone awry or maybe its just the :beer:

JetMech

01-18-02, 03:23 AM

Originally posted by Loud

Every way I do the math, the Duron is working more efficiently. It is, however, a VERY SLIGHT margin. And, of course, the bottom line shows that the Athlon is accomplishing more work in real time.

What was the point of all this again? :cool: The "POINT" was to try to determine whether one expensive Athlon could do the same or more work than two Durons whose cost totals or is less than the cost of the Athlon. Ploaf made a good point in including the full cost of all hardware necessary to build the sys. In my case I have enough spare parts, but everyone does not fall into that category. I guess the whole thing is relative. If you can build two Durons folding at 75% of the speed of the fastest Athlon then do so. Or settle for the one fast system .

Athlon XP1800+ - 1.53@1.75 = 5.9hrs per villin / 24hrs. = 4.06WU/day

Duron 800@1050 = 9.91hrs per villin / 24hrs. = 2.42WU/day x 2 = 4.84WU/day

res0r9lm

01-18-02, 05:12 AM

yea but then you have to consider the extra on power bill. if 2 athlons then you would have the folding power of 4 durons

JetMech

01-18-02, 05:28 AM

Originally posted by res0r9lm
yea but then you have to consider the extra on power bill. if 2 athlons then you would have the folding power of 4 durons Point well taken. I think it safe to say that each members situation is unique. My power comes with the rent. There may be others in the same boat.

Ploaf

01-18-02, 07:44 AM

This isn't always an easy thing to estimate since there are a lot of variables and value is dictated by today's price and today's price only. Sometimes it breaks down to one specific CPU that offers the most bang for the buck. For me, at the time, it was the 1G AYHJA Y athlons. The were about $70, but they OC like mad and were an overall great bargain. The 650-800 durons were great bargains too because they were very cheap and also OC'd fairly well. At the very least you could kick the bus up to 133 and get a decent OC out of them. Durons create less heat and are easier to cool. However you do it, it will be right as long as it works. I'd be less worried about cooking a duron too. at least they don't cost very much. I don't know as much about the newer cpu's although I, personally, wouldn't even consider a 2000XP at the prices they are asking. There may be an 1800 in the future or I may give it a month and see how things 'unfold'. ;)

KILLorBE

01-18-02, 07:49 AM

Native villin 10-11 minutes per frame, PIII600@888

KFB

01-18-02, 03:38 PM

5.53 min/frame for duron750@1.1 GHz (see sig)

JetMech

01-18-02, 05:25 PM

Originally posted by Ploaf
This isn't always an easy thing to estimate since there are a lot of variables and value is dictated by today's price and today's price only. Sometimes it breaks down to one specific CPU that offers the most bang for the buck. For me, at the time, it was the 1G AYHJA Y athlons. The were about $70, but they OC like mad and were an overall great bargain. The 650-800 durons were great bargains too because they were very cheap and also OC'd fairly well. At the very least you could kick the bus up to 133 and get a decent OC out of them. Durons create less heat and are easier to cool. However you do it, it will be right as long as it works. I'd be less worried about cooking a duron too. at least they don't cost very much. I don't know as much about the newer cpu's although I, personally, wouldn't even consider a 2000XP at the prices they are asking. There may be an 1800 in the future or I may give it a month and see how things 'unfold'. ;) I agree. The whole folding thing is just as much a science as overclocking or close anyway. I've been thinking about such things as slaving in a cdrom as needed instead of dedicating it to a particular machine. That way you could have say five puters and 1 cdrom. No sound, minimum video. I have a dead laptop with a good 2.5" HD. saw somewhere that you can adapt those for use in PC's. Ram is on the way back up so that's going to become a bottleneck again. But then what is the minimum amount needed to give folding it's full head? So many questions when you're doing your homework.:( I have my 1800+ OC'd to 2000+. I haven't stress tested it yet. It runs the fold proggy and so I'm leaving it alone. Trying to sqeeze every bit of juice from this lemon.

Does anyone know if there is a way to verify the quality of your folding production. I need to make sure my methods aren't resulting in contaminated product.

BTW. That's one heck of an overclock KFB. :beer:

Ploaf

01-18-02, 08:19 PM

Originally posted by JetMech
I agree. The whole folding thing is just as much a science as overclocking or close anyway. I've been thinking about such things as slaving in a cdrom as needed instead of dedicating it to a particular machine. That way you could have say five puters and 1 cdrom. No sound, minimum video. I have a dead laptop with a good 2.5" HD. saw somewhere that you can adapt those for use in PC's. Ram is on the way back up so that's going to become a bottleneck again. But then what is the minimum amount needed to give folding it's full head? So many questions when you're doing your homework.:( I have my 1800+ OC'd to 2000+. I haven't stress tested it yet. It runs the fold proggy and so I'm leaving it alone. Trying to sqeeze every bit of juice from this lemon.

Does anyone know if there is a way to verify the quality of your folding production. I need to make sure my methods aren't resulting in contaminated product.

BTW. That's one heck of an overclock KFB. :beer:

I don't know how to verify the quality, although I think the server may reject bad wu's and I also think that an overdone OC will result in corrupted or lost wu's. I was losing some on one machine till I backed it off. As for sharing devices? I do that. I have 2 working CD Rom drives. One is my main burner the other is my feed drive. I disconnect it and use it to work on other boxen. Same goes with monitor/kb/mouse/floppy. RIght now I don't have a floppy installed in any system. I don't use them very often anyway. Folding on any OS would be fine at 128, probably fine at 64 with win98. RAM doesn't matter very much at all anyway. You just need enough to keep things running. The speed of the RAM matters little and it's all going way up in price anyway. I have a ton of pc133 and some ddr. No upgrading the rest till the price drops. That XP@2000 is pretty good.

KFB

01-18-02, 09:14 PM

Thanks JetMech, but it was the easiest OC i have ever done!:)
I originally thought 1Gig would be pushing that chip, but it is rock stable in winXP, even through the prime95 torture test.
It's good to hear that other people are into Folding production as much as I am, even if I am too lazy to setup the 2 p200's I have sitting around...yet.

lennytiger

01-19-02, 12:22 AM

nice oc kfb!! :)

JetMech

01-19-02, 02:47 AM

Originally posted by Kentucky Fried Bird
Thanks JetMech, but it was the easiest OC i have ever done!:)
I originally thought 1Gig would be pushing that chip, but it is rock stable in winXP, even through the prime95 torture test.
It's good to hear that other people are into Folding production as much as I am, even if I am too lazy to setup the 2 p200's I have sitting around...yet. It's fast becoming an obsession with me. I feel guilty about each cpu cycle I waste and am constantly trying to measure everything I do. I only do work with the P4 because it is the slowest.

Something that I have noticed is that on the machine that is running Win98 the HD light flashes constantly and memory becomes completely used up over time. I have a program called WinRamBooster that I downloaded free off the net and use everyday to recover leaks. F@H, Fold Monitor and MBM are the only programs running. One of them leaks? I see the same problem on the P4/WinME, and the XP/ME. The same proggies are running on those systems. The P4 of course is RDRAM, the 1.4 is SDRAM, and the XP is DDR. The wierd part is that even when you see the XP and P4 almost out of ram the HD is not working. I've chalked it up to a characteristic of Win98.:confused:

BTW. I specify my own virtual memory in Permanent swap file of size 2.5 x ram amount.

res0r9lm

01-19-02, 08:00 AM

Originally posted by JetMech
It's fast becoming an obsession with me. I feel guilty about each cpu cycle I waste and am constantly trying to measure everything I do. I only do work with the P4 because it is the slowest.

Something that I have noticed is that on the machine that is running Win98 the HD light flashes constantly and memory becomes completely used up over time. I have a program called WinRamBooster that I downloaded free off the net and use everyday to recover leaks. F@H, Fold Monitor and MBM are the only programs running. One of them leaks? I see the same problem on the P4/WinME, and the XP/ME. The same proggies are running on those systems. The P4 of course is RDRAM, the 1.4 is SDRAM, and the XP is DDR. The wierd part is that even when you see the XP and P4 almost out of ram the HD is not working. I've chalked it up to a characteristic of Win98.:confused:

BTW. I specify my own virtual memory in Permanent swap file of size 2.5 x ram amount.
had same problem with winme with only 64mb sdram.since upgrading doesn't go vitural memory. 320mb 64 of it is being used for vcache right now have 110mb free

funnyperson1

01-19-02, 08:21 AM

It takes 10 minute to do a Villin on the system in my sig....about 5-6mins on my dads Celly Tualatin @ 1.4....

Ploaf

01-19-02, 10:39 AM

Well here are some avg frame times

Duron@1000-winXP Pro *bba5_build_ext* 153 seconds/frame 2.55 minutes
Athlon@1.4 with DDR=WinXP Pro GNRA Hairpin 271 seconds/frame 4.52minutes
PIII@1000-Win2k Pro GNRA Hairpin 585 seconds/frame 9.74 minutes
Athlon@1.410 with PC133 on old KT7Raid(certainly not the fastest chipset)- WinXP Pro - 264 seconds/frame 4.4 minutes

KFB

01-19-02, 06:31 PM

Wow the bb5_build_ext proteins are much folded much faster! My sig system (Duron750@1104) running winXP pro gets a frame in an average 2.28 minutes! Thats more than twice as fast as a villin or GNRA hairpin.

Ploaf

01-19-02, 09:04 PM

The bba5's are only worth .6 points while the GNRA hairpins are worth 2 points. I like the 2 pointers better. :) Of course wu's that don't get sent back..... well that's another thread.

Oh here's an update. bba5 on the Duron I mentioned earlier is now taking 151 seconds/frame average using the command line client. It was actually down to an average of 146 seconds/frame with the gui console as it came closer to the end of the protein. I'll see what it is after it has come closer to finishing this time. That's winxp with the eye candy disabled with PC133 RAM. I am betting that in the end both clients will perform the work in the exact same amount of time because the part of the client that does the work is completely separate from the front end and it is exactly the same for the cmd line version as it is for the gui version.

KFB

01-19-02, 10:25 PM

that's good info, ploaf.
I have always used the console version b/c i figured it would be faster - no eye candy to slow the comp down - but if the graphical version turns out to be faster i will switch in a second.

RustyCannon

01-20-02, 03:56 AM

Hmmm I think something must be really wrong with my 2 folding computers.

My PII 400 is getting 411.79 M per frame. Very very slow compared to others I have seen. That is doing a native villin.

My other one a P4 1.3 is getting 32.78 M on a GNRA Hairpin. That still seems extremely slow for the clock speed of my cpu. Whats up with this? Does the fold monitor not work properly in win2k or does the F@H program not work as fast in win2k?

~Rusty Cannon

JetMech

01-20-02, 08:34 AM

Originally posted by RustyCannon
Hmmm I think something must be really wrong with my 2 folding computers.

My PII 400 is getting 411.79 M per frame. Very very slow compared to others I have seen. That is doing a native villin.

My other one a P4 1.3 is getting 32.78 M on a GNRA Hairpin. That still seems extremely slow for the clock speed of my cpu. Whats up with this? Does the fold monitor not work properly in win2k or does the F@H program not work as fast in win2k?

~Rusty Cannon How much work are you doing on the machines. Your P4 times seems extremely high. As a matter of fact Stanford would not even be able to use the PII's data because it computes to 28 days per WU. Six is the maximum. Your P4 times computes to 2.27days/WU. What I believe is that you are reading the average time per frame figure which is not accurate once you start to work while folding. It is reliable if the machine is idle but once you start using the machine F@H starts to wait for cpu cycles and the average goes up. Set your refresh rate to a low number (I use 60 sec) so that you can see some progress and time how long it takes to do 1 frame using the Frame Time Elapsed: number. I think you will find that the number is much smaller than the average time per frame number. Do this while Folding only. That way it won't take as long to finish a frame.

Do you shut your machine off or fold 24/7. If you shut it off before a WU is completed the Frame Time Elapsed keeps computing and when you restart the average will be off the wall. For those who start and stop the fold monitor has very limited uses. Hope this helps.:)

JetMech

01-20-02, 08:42 AM

s

bba5_build_ext times:

Athlon 1.4 = 1.41/ frame

P4 1.7@1.95 = 1.90/ frame

Athlon XP 1800+ @ 2000+ = 1.19/ frame

Duron 1.2@ 1350 = 1.70/frame

GNRA Hairpin

1.4 = 4.55/frame

P4 = 7.50/frame

XP = 3.72/frame

Duron 1.2 @ 1.35 = 4.68/frame

Native Bba5

P4 = 2.29/frame

XP = 1.38/frame

1.4 = 1.83/frame

RustyCannon

01-20-02, 06:34 PM

Those times are when the computer is idle.
I watched the monitor for several minutes and found that its actually only getting to about the 47% mark before it changes to the next frame. And I timed it for how long its really taking and its doing 7.82 minutes. Guess the monitor just isn't showing me the correct time. And i'm folding about 21 hours a day.

~Rusty Cannon

Ploaf

01-20-02, 07:27 PM

Check the refresh rate on foldmonitor too. The default is every 600 seconds or something like that.

My times with both the text client and the gui were so close that it is really hard to tell which would be faster.
Another observation. I don't know how much of a role DDR plays in this or how much of it is the OS, but win98 used to be the king of the hill for folding but....

2 systems.
1- ECSK7S5A, 1.0@1.4, 512 DDR, WinXP Pro w/half of the eye candy enabled. System is being used and has been burning cd's through part of this run too. GNRA Hairpin - 274 seconds avg.

1- ECS K7S5A, 1.0@1.4, 128 PC133, Win98, no eye candy, only other thing running in background is zonealarm. System is dedicated machine. GNRA Hairpin - 272 seconds avg.

The DDR could have played a role in this or it could be that WinXP Pro is better at staying out of the way of the folding client than Win2k was.

res0r9lm

01-20-02, 07:51 PM

Originally posted by JetMech
s

bba5_build_ext times:

Athlon 1.4 = 1.41/ frame

P4 1.7@1.95 = 1.90/ frame

Athlon XP 1800+ @ 2000+ = 1.19/ frame

GNRA Hairpin

1.4 = 4.57/frame

P4 = 7.63/frame

XP = 3.72/frame
ok why is my 1.4 so slow doing the bba's (1.85). I can do a hairpin
in under (4.5)best time 4.44 avg. for whole wu

JetMech

01-21-02, 03:55 AM

Originally posted by RustyCannon
Those times are when the computer is idle.
I watched the monitor for several minutes and found that its actually only getting to about the 47% mark before it changes to the next frame. And I timed it for how long its really taking and its doing 7.82 minutes. Guess the monitor just isn't showing me the correct time. And i'm folding about 21 hours a day.

~Rusty Cannon What you are seeing is the wait time for the three hours you are down. If you don't fold 24/7 you will see correct times in fold monitor only when you wait for a frame to finish before shutting down. Even then if you let the new unit download you'll go into wait time while shutdown. After starting up ignore the Fold Monitor until a new WU starts. It should read correctly the rest of the day. You should only see a problem at start unless you do some work that is cpu intensive (benchmarking, gameplaying, or internet browsing where a lot of animation is present).

JetMech

01-21-02, 04:07 AM

Originally posted by res0r9lm

ok why is my 1.4 so slow doing the bba's (1.85). I can do a hairpin
in under (4.5)best time 4.44 avg. for whole wu I'm not sure about that except to say (and I should have pointed it out in the post) that the two Athlon machines are dedicated folders. If you are using your machine for other work besides folding your times will be higher. The OS on the 1.4 is Win98. It doesn't have the parasite programs like, System File Checker and Restore running in the background the way WinME does. Like Ploaf pointed out the best home PC OS for Folding is WIN98. Your overclock on my system would fold at about 1.38/frame.

res0r9lm

01-21-02, 07:17 AM

Originally posted by JetMech
I'm not sure about that except to say (and I should have pointed it out in the post) that the two Athlon machines are dedicated folders. If you are using your machine for other work besides folding your times will be higher. The OS on the 1.4 is Win98. It doesn't have the parasite programs like, System File Checker and Restore running in the background the way WinME does. Like Ploaf pointed out the best home PC OS for Folding is WIN98. Your overclock on my system would fold at about 1.38/frame.
didn't think about that. going to turn restore off

KFB

01-21-02, 01:12 PM

Is win98 really the fastest? 2K/XP is a faster kernel. Have you tested the same config on each OS? If not, I will probably attempt the experimant soon, just to be sure. I was running 98SE for years (only started Folding religiously recently, though)but I have switched to XP Pro now and it seems at least as fast if not faster for Folding.

Ploaf

01-21-02, 05:10 PM

Originally posted by Kentucky Fried Bird
Is win98 really the fastest? 2K/XP is a faster kernel. Have you tested the same config on each OS? If not, I will probably attempt the experimant soon, just to be sure. I was running 98SE for years (only started Folding religiously recently, though)but I have switched to XP Pro now and it seems at least as fast if not faster for Folding.

It used to be about 5% faster on win98 than 2000. I have tried it on both XP and 98 and they are very similar.

Boxen tested-
ECS & AMD@1.4 DDR winxp. ECS & AMD 1.4 pc133 win98. ABit kt7Raid&AMD@1410 with a buttload of pc133 and winxp.

As you can see there is some discrepency in the hardware configuration, but the XP boxes have been slightly faster. It's also trued that the hardware that XP is running on is faster than the hardware that win98 is running on. either a faster cpu, or faster ram, but in both cases it's faster. They are within about 15 seconds of each other per frame. It's very hard to tell because no box is the same. The fastest times of the systems above is the KT7-raid with amd@1410.

muddocktor

01-21-02, 05:50 PM

GNRA Hairpin
T-bird @ 1.4 - 4.41 min./frame
P3 1000E @ 1260 - 7.05 min./frame
Dell P3 1000E laptop - 8.81 min./frame
K6-2+ @ 550 - 22.12 min./frame

nil_esh

01-21-02, 08:19 PM

Here are some times:
K6-2+ 450@550 - GNRA Hairpin - 28.39m
Athlon TBird 1400 - GNRA Hairpin - 4.45m
Athlon TBird 1400 - bba5_build_ext - 1.63m
P2 266 - bba5_build_ext - 13.27m

JetMech

01-22-02, 03:22 AM

Originally posted by Kentucky Fried Bird
Is win98 really the fastest? 2K/XP is a faster kernel. Have you tested the same config on each OS? If not, I will probably attempt the experimant soon, just to be sure. I was running 98SE for years (only started Folding religiously recently, though)but I have switched to XP Pro now and it seems at least as fast if not faster for Folding. The omission of 2K/XP was due to my non consideration of them as Home OS's and the fact that I don't have a platform to test them on. Ploaf's acknowledgement that XP is faster proves I spoke too soon. I knoe nothing about the Home version of XP therefore I will rely on you all for that info. There is an Athlon @1.4 under 2K time posted by muddocktor which shows a faster time than I am getting. It is hardly fast enough for me to consider changing OS's though.

The big difference in times is seen when you compute long term, say 2 months even a year (although power users will have reconfigured by that time).

GNRA Hairpin

JetMech
4.55/frame=7.58/WU=3.166/day=96.25/mon=192.5/2mon=2,304/yr.

muddocktor
4.41/frame=7.35/WU=3.26/day=99.26/mon=198.5/2mon=2,382/yr

1 month=365/12=30.4days

KFB

01-22-02, 02:57 PM

Cool, I will stick with XP then.

res0r9lm

01-23-02, 07:57 PM

Originally posted by belorsch

On top of building a new pc I'm wanting to upgrade my p3 machine with a ECS mb and 900tbird but if a 1100 duron would fold faster at about the same or better price than I might be able to afford a 1200 duron. Do they make those?:beer:

Just came out with 1.3 duron so price of 1.2 will be dropping.
the durons 1000 and up are based on same core as xp so will
run cooler than t-bird

res0r9lm

01-23-02, 08:04 PM

ThePunkGeek

01-24-02, 04:04 PM

around 3.89

Malakai

01-25-02, 01:38 AM

i dont get it???
my avg time is 7 min for the system in my sig. whys it slower than durons and slower tbirds?
im folding a nativevillan#2

JetMech

01-25-02, 04:08 AM

Originally posted by Malakai
i dont get it???
my avg time is 7 min for the system in my sig. whys it slower than durons and slower tbirds?
im folding a nativevillan#2 Hard to say Malakai. I would think you should be able to do about 5.0 to 5.25 at least. Do you use your sys while folding?

You're very welcome resOr9lm

Kendan

01-25-02, 04:22 AM

Originally posted by Malakai
i dont get it???
my avg time is 7 min for the system in my sig. whys it slower than durons and slower tbirds?
im folding a nativevillan#2

Did you turn FAH off? if you did fold monitor will still calculate this time. For an acurate reading let it monitor a WU from the beginning without turning it off and you will get a good idea of your time per frame. Or you can download the beta 2.17 of F@H2 which includes time stamping. For this just look at the log in foldmonitor click clients then click log. Hope this helps. Fold on!!!

res0r9lm

01-25-02, 08:31 AM

must be using screensaver try using console

res0r9lm

01-25-02, 09:15 AM

got new wu nativebba5. first time getting one of these.
avg. 1.51 min per frame

JetMech

01-25-02, 03:23 PM

Originally posted by res0r9lm
got new wu nativebba5. first time getting one of these.
avg. 1.51 min per frame Right now I'm showing 1.80 on the 1.4 for that protein. Just had to restart though. Will update later.

Update: It's been folding only all day is is now showing 1.67. That will be my official 1.4 time. Nativebba5 on Athlon 1.4=1.67mins/frame.

P4=2.34

XP=1.39

I had to turn the XP frontside down to 142 because of some stability problems. Case is too cramped and when the AC cycles off temp goes too high.

NedClocker

01-25-02, 03:34 PM

Hey, heads up. Today's Fry's ad (in Houston) shows an AMD Duron 1GHz (oem) with an ECS K7S5A for $99.00 !!!

Does that sound like a good deal to anybody else???

I think I'll go get one right after work.

From what I've read, the Duron 1GHz will be a pretty good folder.
:D

Ploaf

01-25-02, 05:22 PM

Originally posted by NedClocker
Hey, heads up. Today's Fry's ad (in Houston) shows an AMD Duron 1GHz (oem) with an ECS K7S5A for $99.00 !!!

Does that sound like a good deal to anybody else???

I think I'll go get one right after work.

From what I've read, the Duron 1GHz will be a pretty good folder.
:D

Sounds good to me for a brick and mortar store. Nice find.

nil_esh

01-25-02, 05:35 PM

Originally posted by NedClocker
Hey, heads up. Today's Fry's ad (in Houston) shows an AMD Duron 1GHz (oem) with an ECS K7S5A for $99.00 !!!

Nice.. That's exactly the kind of deal I'm looking for... Mobo and proc for $100.. I'm gonna do my buying in March when I get back from India. I'm gonna buy a nice low budget system. Most likely something along those lines. A Duron 1Ghz will put out some WU for sure...

KFB

01-25-02, 09:01 PM

that deal for a mobo AND Duron 1G is sweeter than a southern belle dipped in sugar. scoop it up quick Ned.

NedClocker

01-29-02, 05:24 PM

KFB,

Got it! It's been running since Fri. night; well, very early Sat. morning, actually.

They had the Thermaltake Volcano 5 HSF for $6.99, also.
I scooped 2 of those.

Today is the last day of the sale. Pay day isn't until Friday. It actually hits the bank on Thurs., though. 2 days from now. Hummm....
How long will a check float?

res0r9lm

01-30-02, 10:20 PM

just setup my mom pII 400 it's take 10 minutes per a frame doing native bba5 hate to see it do a villin guessing something like 30 min per frame

JetMech

01-31-02, 02:28 AM

Hey it's all folding. You could always hook it up to Yatta.

KFB

01-31-02, 06:23 AM

Ned, sounds cool. I hope the check clears OK as I know what a headache bounced checks can be!
:(
I got a used retail Tbird AXIA 1Gig off ebay and it arrived yesterday. It is folding @1050 (140x7.5) as I type. I am going to see How High I can get it to later today (hopefully 1.4 at least! - on air alone). I also discovered that the Tbird 850 that I thought was ruined by chippage actually still works! Now I just need a mobo for it!
:D

NedClocker

01-31-02, 05:34 PM

KFB,

I had enough to get it. How can you pass up a new 1GHz 'puter for $99.00? Oh, yeah, plus $6.99 for the Volcano 5, plus tax.
I already had everything else I needed.

Tell me this, though. I put 128 meg of PC100 ram in it. How much difference would it make to put in PC133 or SDR? Assume I am running f@h, of course. It probably would not make a noticable difference normally, but would it running f@h? I suspect so. I'll explain why later. Tell me what you think. I've got to go.

JetMech

02-01-02, 03:12 AM

s

I really don't think it would affect anything more than your ability to overclock. Sounds like a good project though. I'm running my 1.4 on SDR and also wonder if an investment in DDR would significantly improve my times.

NedClocker

02-01-02, 09:52 AM

Identical systems, except for the ram. One has 256M PC133. The other has 128M PC100. The pc133 box was getting a little better frame times on proteins of the same name - BBA something or the other. About .2 minutes per frame.

BUT, now this morning, I noticed the pc100 box is getting slightly better frame times on proteins with identical names.

The only way to know for sure is to copy a protein, work unit, what ever you want to call it, from one box to the other, right after downloading. That way you would know it was truly identical work and the comparison would be valid.

Of course, that would be a "waste" of cpu cycles, because I doubt that you would get credit for the same wu twice. So, pick a short wu, for sure not a villin.

It would be interesting to see if you would get credit for it twice. That would be a dirty tricks way of upping your score. (Copying the a wu to other machines right before upload, and have them all upload it.)

res0r9lm

02-01-02, 10:38 AM

Had a disk error new core was downloaded by it's self I think I got credit for wu twice but wouldn't do something like that on purpose whole reason you are folding is to help the cause right?
if you have to cheat to get higher in the ranking the next step would be hack other compters.

NedClocker

02-01-02, 11:03 AM

Dude,

Don't get excited. The first thing I said was I doubt that you would get credit for the same work unit twice.

Then, I said it would be "interesting" to see if you could.

I acknowledged that it would be a "dirty tricks" way to up your score.
Actually, it would be a pain in the butt to copy those files around to your other computers and wouldn't be worth the effort, at least, not to me.

It would be kind of like cheating at dominos. "What's the point?"

res0r9lm

02-01-02, 11:10 AM

I'm sure there are some that would try

NedClocker

02-01-02, 11:14 AM

No doubt. In fact, I'm sure there are some who have already tried. Someone has probably already thought of my idea. I hope it would not work. I hope the people at Stanford have already thought of that and fixed it to where you can't do it.

NedClocker

02-01-02, 12:37 PM

Hey, maybe I was wrong; you can get double credit.

Look at this.

http://forums.overclockers.ws/vb/showthread.php?s=&threadid=58591

res0r9lm

02-01-02, 01:24 PM

you need to have different id's for each cpu or computer or the first wu would be assigned to someone else thinking wu was incompleted after your second cpu or computer is assigned a wu. so then there will be 2 people work on same wu.
The way I see it you could get credit twice by copying folder @99% and deleting id from config on second folder. don't see how they can fix this problem other than limiting one computer per ip address that would be a big hit for someone with dual cpu board and people on network though

NedClocker

02-01-02, 03:10 PM

Originally posted by res0r9lm
you need to have different id's for each cpu or computer or the first wu would be assigned to someone else thinking wu was incompleted after your second cpu or computer is assigned a wu. so then there will be 2 people work on same wu.

That's not the way I read it.

"The consequence of not doing this is that F@H will think that they are all the same processor - when one requests a WU, the server will think that it hasn't returned the old WU back yet, and will reassign the old WU to do again."

Sounds like you'll get the same work unit twice.

Either way, you're probably right, there's no way for them to fix it.

JetMech

02-01-02, 04:36 PM

Originally posted by res0r9lm
Had a disk error new core was downloaded by it's self I think I got credit for wu twice but wouldn't do something like that on purpose whole reason you are folding is to help the cause right?
if you have to cheat to get higher in the ranking the next step would be hack other compters. That would take all the fun out of this. Kind of like leaving a bad taste in you mouth. That is of course if you've got a conscience.:(

NedClocker

02-01-02, 04:48 PM

Please. I hope you don't mean to imply that I don't have a conscience. Just because one wants to learn whether something can be done, doesn't mean one is going to do it.

JetMech

02-01-02, 04:53 PM

Originally posted by NedClocker

That's not the way I read it.

"The consequence of not doing this is that F@H will think that they are all the same processor - when one requests a WU, the server will think that it hasn't returned the old WU back yet, and will reassign the old WU to do again."

Sounds like you'll get the same work unit twice.

Either way, you're probably right, there's no way for them to fix it. That's about the way I read it. Folding itself would be the loser because cycles that could have been used toward progress are just spinning their wheels and getting nothing accomplished. The Folder gets credit but Stanford gets nothing. If we called credits "payment" it would be like getting paid for work not done.

JetMech

02-01-02, 04:57 PM

Originally posted by NedClocker
Please. I hope you don't mean to imply that I don't have a conscience. Just because one wants to learn whether something can be done, doesn't mean one is going to do it. No NO NED. I read you loud and clear. What you were saying was theoretical and I just added my statement to say that "anyone " who did try the trick would be deviant if the act was deliberately done to deceive the system. I have total confidence in you integrity.:) :beer:

NedClocker

02-01-02, 05:03 PM

Thanks. I don't fold for the points; at least not just for the points. My wife has terminal breast cancer. It spread to her brain and spinal cord. She is paralized from the bottom of her ribs on down.

NedClocker

02-01-02, 05:09 PM

Yes, yes. I agree.
That was not the point with which I didn't agree.
The issue was whether you would get the same wu twice or someone else would get it after you started it.

res0r9lm said,
"you need to have different id's for each cpu or computer or the first wu would be assigned to someone else thinking wu was incompleted after your second cpu or computer is assigned a wu. so then there will be 2 people work on same wu."

The link said,
"...when one requests a WU, the server will think that it hasn't returned the old WU back yet, and will reassign the old WU to do again."

I thought it meant reassign the wu to you. But, now that I look closer, it does not specify, so res0r9lm may be right.

But, yes, either way it doesn't matter; it's a waste of cpu cycles.

NedClocker

02-01-02, 05:15 PM

Hey, look, guys!

I'm not even the first one to post the idea.

http://forums.overclockers.ws/vb/showthread.php?s=&threadid=52468

lennytiger said it on 12-17-2001.

"Interesting Idea why then don't we do one fold to 99/100 frames then re submit the same data by re loading the data from the disks, that way we could really catch the aussies!!"

Ebola

02-01-02, 07:46 PM

back to the thread. I'm doing about nine and a half minutes on the GNRA hairpin per frame. But then again I'm doing two at the same time.

I think the OS has a ton to do with the speed. 2k just isnt that fast for folding.

oops thats avg time. i did some resetting. the time is acutally dropping. closer to 9 minutes per frame.

JetMech

02-01-02, 09:49 PM

Originally posted by Ebola
back to the thread. I'm doing about nine and a half minutes on the GNRA hairpin per frame. But then again I'm doing two at the same time.

I think the OS has a ton to do with the speed. 2k just isnt that fast for folding.

oops thats avg time. i did some resetting. the time is acutally dropping. closer to 9 minutes per frame. Are the two clients totally independent of one another as with two seperate machines? Wouldn't it be great if a manufacturer like ECS built us folders a good cheap dually board. I'm cramped for space and two cpu's in one case is great.

res0r9lm

02-01-02, 10:03 PM

you build a dual folder cheaper than two single systems all ready

Ebola

02-01-02, 10:05 PM

Yup they run totally independent of each other. it like two machines in one. btw my avg dropped below 9 minutes per frame while i was away so i'm thinking when the machine is not being used its doing the gnra hairpins arround 8 minutes.

if your really serious about folding and have some pocket change then look into a dual processor athlon mp board. i bet the good mp's do twice what i do since my machine is doing about the same as a single processor xp 2000 overclocked.

Originally posted by JetMech
Are the two clients totally independent of one another as with two seperate machines? Wouldn't it be great if a manufacturer like ECS built us folders a good cheap dually board. I'm cramped for space and two cpu's in one case is great.

KaHNZa

02-01-02, 10:06 PM

Originally posted by Ebola
back to the thread. I'm doing about nine and a half minutes on the GNRA hairpin per frame. But then again I'm doing two at the same time.

I think the OS has a ton to do with the speed. 2k just isnt that fast for folding.

oops thats avg time. i did some resetting. the time is acutally dropping. closer to 9 minutes per frame.

OMG. Why are your times so high? With my Athlon 1.33G my frame times are never more than 2m20sec. :eek:

res0r9lm

02-01-02, 10:11 PM

depends on the protien your folding. you must be folding bba5

Ebola

02-01-02, 10:33 PM

yup its a gnra hairpin. they take forever. a few other people with 1ghz p3's posted the same results on protein

KaHNZa

02-01-02, 10:42 PM

Originally posted by res0r9lm
depends on the protien your folding. you must be folding bba5

Yea thats possible. I've been getting alot of those up until lately. I've got a PHE Sidechain - OPLSAA right now. Fold Monitor says over 1000 seconds per frame. But as it finishes a frame the time goes down a bit. :eh?:

res0r9lm

02-01-02, 10:56 PM

definitely dual amd is the way to go. would be able to get twice as many wu's done with 2 xp 1600+. I think that might be what I build next just can't decide between 760mp & 760 mpx. with 760mp you can use tbirds but have to use registered ddr and 760mpx you have to use xp's,mp's,or morgans can use unbuffed ddr. I heard that you can use one ddr if your first stick is registered. I can get 760mp for $186 & 760mpx for $216.
what I have right now is 1 t bird 1.4 and 256 ddr. just wondering which would be a better choice?

res0r9lm

02-01-02, 10:59 PM

Originally posted by KaHNZa

Yea thats possible. I've been getting alot of those up until lately. I've got a PHE Sidechain - OPLSAA right now. Fold Monitor says over 1000 seconds per frame. But as it finishes a frame the time goes down a bit. :eh?:
I get a little less than 10 minutes per a frame with those luckly they are only 10 frames per a wu

Ebola

02-01-02, 11:32 PM

isnt abit working on a new dual athlon board? sorry i dont know much about amd stuff.

res0r9lm

02-01-02, 11:39 PM

haven't heard anything, I know epox should have one out pretty soon

res0r9lm

02-01-02, 11:41 PM

Right now the only ones I know of is tyan, msi, asus, and epox

KFB

02-03-02, 03:13 AM

TBird @ 1.4 [140x10] GNRA Hairpin - 04:20 (mm:ss)
*edit* = 4.34 minutes = 260 seconds

res0r9lm

02-13-02, 09:46 AM

got new motherboard for tbird and put duron in old a7v. now I can finally oc the tbird. both are folding gnra hairpins.
1.4 tbird@1.61 3.88 min per a frame.
800 duron@959 6.54 min per a frame
wierd. same temp on both. would have thought tbird would have been alot hotter

NedClocker

02-13-02, 09:54 AM

Originally posted by res0r9lm
wierd. same temp on both. would have thought tbird would have been alot hotter

Yeah, me too! But my durons run as hot or hotter. That one duron @1.24 got up to 131 F a couple of days ago.

res0r9lm

02-13-02, 10:20 AM

Originally posted by NedClocker

Yeah, me too! But my durons run as hot or hotter. That one duron @1.24 got up to 131 F a couple of days ago.
Thats kinda toasty but I have seen hotter. theres shop here local that custom builds systems. I buy parts from them sometimes anyway I was in there one day and their display system was running some type of demo from comcast cable internet checked their system temps against mine because we had the same config except for case and video card and their's was running 138F heck mine was actually running 20F cooler with a 166mhz oc.That goes to show you that lapping is well worth the effort.

NedClocker

02-13-02, 10:27 AM

Wow, I didn't think lapping would make that much difference. I'll have to try it.

There may be something wrong with that case or position in the room. The other processor/board I had in there ran hotter there than it does now in another case in another place in the room.

It did'nt stay that high.

That's the wall that catches the evening sun and it was evening. Mobo monitor went off; that's how I knew. So I reset it for 135 and it hasn't gone off again.

res0r9lm

02-13-02, 11:26 AM

I think they might have been using those pads instead of grease.
the heatsink that was beening used was a globalwin fop32. I now got it on k-2 550@600 before I switch hsf it wouldn't go over 560. a 50 mhz isn't much but every little bit helps

NedClocker

02-13-02, 11:29 AM

Yes, it does. And I sometimes think of it this way. My first computer was a 486 sx 25. 50mhz is twice as fast as my first computer.

doer

02-13-02, 11:40 AM

tualatin celeron 1.2@1.6ghz, 256mb pc133 cl3 sdram, win2k, graphical version, no screensaver active gives about this:

res0r9lm

02-13-02, 12:08 PM

DOER thinks that a Celeron is going to best xp folding. ned's durons would give your celly a run it's money.if not beat it

NedClocker

02-13-02, 12:18 PM

I'll have to remember to take a look at that when I get home.

robertm

02-13-02, 12:27 PM

Here the time from my AMD system at work.

AMD XP at 1.58Ghz 256 DDR CAS2 Windows XP Pro.

doer

02-13-02, 12:29 PM

Originally posted by res0r9lm
DOER thinks that a Celeron is going to best xp folding. ned's durons would give your celly a run it's money.if not beat it

crap. <flame on>res0r9lm, don't you whole time forge what I'm saying!! :mad: I've never said "a Celeron is going to best xp folding." nor even compared celeron vs xp or celeron vs duron, HUH!<flame off> I just posted results for my cpu, peace.:p

res0r9lm

02-13-02, 12:36 PM

Originally posted by doer

crap. <flame on>res0r9lm, don't you whole time forge what I'm saying!! :mad: I've never said "a Celeron is going to best xp folding." nor even compared celeron vs xp or celeron vs duron, HUH!<flame off> I just posted results for my cpu, peace.:p
didn't you post in a different thread that celly is faster than duron, if not my bad
each celly and duron have their good and bad points

NedClocker

02-13-02, 08:10 PM

All I've gotten in the last day are NativeBBA5 and bba5_build_ext. Those are all I've had since Monday at 10:00 PM.

THIS IS THE ONE @ 1.24:
Client : DURON ver 2.12: //DURON/C/Program Files/Folding@Home
Core : Version 2.41 (Jan 22, 2002)
Protein : GNRA Hairpin
Frames : 100
Workunit : 01
Atoms : 389
Steps : 500000
Server IP : 115
Completed : Between Mon Feb 11 15:31:02 2002 and Mon Feb 11 15:41:02 2002
Refresh at : 600 sec.
Frame Time : 5.01 min. ( 301 sec. )
Total Time : 495.30 min. ( 29718 sec. )

THIS IS THE ONE AT 1.12:
Client : My Client ( 1GHz@1.12 GHZ ): //DURONDUDE/C/Program Files/Folding@Home
Core : Version 2.41 (Jan 22, 2002)
Protein : GNRA Hairpin
Frames : 100
Workunit : 07
Atoms : 389
Steps : 500000
Server IP :
Completed : Between Mon Feb 11 16:31:03 2002 and Mon Feb 11 16:41:03 2002
Refresh at : 600 sec.
Frame Time : 5.56 min. ( 334 sec. )
Total Time : 552.92 min. ( 33175 sec. )

Kendan

02-13-02, 08:13 PM

I have been getting all kinds of new proteins. I like this better than always getting the same ones.

res0r9lm

02-13-02, 08:14 PM

thats funny for the last 72 hours all I gotten is gnra hairpins.

NedClocker

02-13-02, 08:21 PM

Originally posted by robertm
Here the time from my AMD system at work.

AMD XP at 1.58Ghz 256 DDR CAS2 Windows XP Pro.

Don't tell me DDR memory makes this much difference!!!

Client : new one for 1.4 @ 1.538 ver 2.15: C:/Program Files/Folding@Home
Core : Version 2.41 (Jan 22, 2002)
Protein : GNRA hairpin
Frames : 100
Workunit : 07
Atoms : 389
Steps : 500000
Server IP :
Completed : Between Fri Feb 8 13:23:49 2002 and Fri Feb 8 13:33:49 2002
Refresh at : 600 sec.
Frame Time : 4.29 min. ( 258 sec. )
Total Time : 425.52 min. ( 25531 sec. )

Client : new one for 1.4 @ 1.538 ver 2.15: C:/Program Files/Folding@Home
Core : Version 2.41 (Jan 22, 2002)
Protein : GNRA Hairpin
Frames : 100
Workunit : 03
Atoms : 389
Steps : 140000
Server IP :
Completed : Between Mon Feb 11 21:50:56 2002 and Mon Feb 11 22:00:56 2002
Refresh at : 600 sec.
Frame Time : 4.38 min. ( 263 sec. )
Total Time : 430.90 min. ( 25854 sec. )

:D :burn: :D

res0r9lm

02-13-02, 08:42 PM

it might have a little to do with it. he's also clocked a little higher.
I'm getting 3.88 min per frame doing hairpins. I noticed like a .05 difference by increasing fsb from 140 to 153 while keeping cpu same speed so I think it's only mhz that make any signifcate difference

belorsch

02-13-02, 08:43 PM

HIE SIDECHAIN - AMBER
127.6 mins/frame on 366 celery

JetMech

02-14-02, 02:44 AM

Originally posted by doer
tualatin celeron 1.2@1.6ghz, 256mb pc133 cl3 sdram, win2k, graphical version, no screensaver active gives about this: 5.50 min/frame for GNRA Hairpin That's 1.28 min/frame faster than a P4@2.1mhz:eek:

NedClocker

02-14-02, 09:43 AM

Originally posted by res0r9lm
it might have a little to do with it. he's also clocked a little higher.
I'm getting 3.88 min per frame doing hairpins. I noticed like a .05 difference by increasing fsb from 140 to 153 while keeping cpu same speed so I think it's only mhz that make any signifcate difference

Yes, once you have enough ram, it's the cpu speed that's most important. Like you wont notice a difference going from 128 meg to 256 meg of ram.

But he's getting .36 minutes/frame better time (or more), and that's 36 minutes per wu faster! I wouldn't think 47 mhz would make that much difference.

Is the XP faster than the regular tbird at any given clock speed?

res0r9lm

02-14-02, 11:17 AM

Originally posted by NedClocker

Yes, right once you have enough ram, it's the cpu speed that's most important. Like you wont notice a difference going from 128 meg to 256 meg of ram.

But he's getting .36 minutes/frame better time (or more), and that's 36 minutes per wu faster! I wouldn't think 47 mhz would make that much difference.

Is the XP faster than the regular tbird at any given clock speed?
Yea your right I wouldn't think 47mhz would be that much faster eithier must have to do with the system as a whole. some mb are just faster than others. look at jetmech's times for xp@1.67 and compare that to mine @1.61 don't think there is that much difference if I had another 60mhz it would close. I knocked .65 offmy times by switching mb, ram, and adding 200 mhz. I think it was the mb that made the difference

robertm

02-14-02, 12:18 PM

The XP Processer does perform better per clock cycle than a t-bird do to better caching so they say. Also it has SSE instruction set that might make a little differace depedning on how they do there program.

Also DDR has the bandwith to help keep a processor cache full at times SDRAM might not be able to keep up. Thats why the new AMD chips will have on chip DDR memory controlers to help with memory bandwith. (Thats what I heard anyway)

I think you combine the 40Mhz, DDR, processor maybe even the OS, motherboard, other software that might be running you could easy get the second per frame differance we are talking about.

Thats what I think anyway:burn:

I want spellcheck back :)

res0r9lm

02-14-02, 12:24 PM

Originally posted by robertm
The XP Processer does perform better per clock cycle than a t-bird do to better caching so they say. Also it has SSE instruction set that might make a little differace depedning on how they do there program.

Also DDR has the bandwith to help keep a processor cache full at times SDRAM might not be able to keep up. Thats why the new AMD chips will have on chip DDR memory controlers to help with memory bandwith. (Thats what I heard anyway)

I think you combine the 40Mhz, DDR, processor maybe even the OS, motherboard, other software that might be running you could easy get the second per frame differance we are talking about.

Thats what I think anyway:burn:

I want spellcheck back :)
I'm not sure that F@H2.0 takes advantage of the added instruction in the xp.

doer

02-14-02, 01:23 PM

Originally posted by JetMech
That's 1.28 min/frame faster than a P4@2.1mhz:eek:

:) I guess f@h would need optimizations for Intel processors anyway... especially in the case of p4, but for p3 family too, cause Amd is clearly running f@h faster than Intel.

I mailed Intel (that f@h www-page) and asked for code optimazion help for f@h team LOL :D , but I doubt they do anything. Maybe some1 should mail the f@h technical team or something...

robertm

02-14-02, 01:46 PM

Intel is helping the F@H team alot infact I think intel is given as contacts for tech support :) There using an intel compiler so am sure they have intel optimizations turned on and SSE the new instruction set to the XP is one of these. But what I dont know is how much SSE would help in there software as I think SSE is interger math instructions. Does F@H use alot of interger math or is it all float? I would think its mostly float.

res0r9lm

02-14-02, 02:49 PM

Originally posted by robertm
Intel is helping the F@H team alot infact I think intel is given as contacts for tech support :) There using an intel compiler so am sure they have intel optimizations turned on and SSE the new instruction set to the XP is one of these. But what I dont know is how much SSE would help in there software as I think SSE is interger math instructions. Does F@H use alot of interger math or is it all float? I would think its mostly float.
If that is the fact it would explain alot

NedClocker

02-14-02, 03:14 PM

I don't know much about the SSE, but do know that the morgan core duron is supposed to have it.

I imagine that the f@h work involved floating point work more than integer.

Malakai

02-15-02, 12:12 PM

here the sys in my sig. she's goin back to 1.7ghz this weekend when i unlock:

Silversinksam

02-16-02, 12:26 PM

getting 2.05 minutes per Nativebba5 on my new rig in sig

I let it run at default last few days But I tried 2.2ghz and it had no probs booting into windows

Guess Im gonna have to Overclock it a little to keep up with you AMD ppl/ :burn:

lennytiger

02-16-02, 12:34 PM

P4 looks nice bro!! ;) keep her folding! :)

Malakai

02-16-02, 05:22 PM

cool. I get 1.56m on that same protien with the comp in my sig.
is yours a northwood? i heard the northwoods at the same speed do them like 10% faster than willamettes.
p.s. those are still awesome times:burn:

Silversinksam

02-16-02, 07:47 PM

I wish it was a Northwood :p

Got a deal on a new 2ghz Williamette for $181, I was originally going to buy a Northwood but couldn't pass on this deal, as I got the whole package (cpu, mobo and DDR) mem cheaper than the lowest price for the Cpu alone.

Its churning out bba5_build_ext's frames at 1.87 minutes per frame.....Not as efficient as XP's it appears...

My new motto is
'what the h3ll do I know, Im an Intel user' :beer:

:D

res0r9lm

02-16-02, 07:47 PM

foldmonitor is wacked. current frame 2.8?Frame 1.8 of 10?

lennytiger

02-18-02, 09:42 AM

I take it your using 2.19 client??

this is the new way the log file logs the WU, decimalisation system in the client is a new addition, and gives serious folders even more knowledge about there system speed.

res0r9lm

02-18-02, 09:59 AM

yea it started after downloading 2.19

Loud

02-18-02, 10:04 AM

CYS SIDECHAIN - OPLSAA

10 frames
50000 steps
2694 atoms

32.12 min/frame

Loud

02-19-02, 04:03 PM

ABeta17-42

100 frames
500000 steps
225 atoms

5.50 minutes per frame

Loud

02-25-02, 12:41 AM

proteinA

100 frames
500000 steps
601 atoms

31.45 m / frame
WU start Sun Feb 24 2002 20:25:50
projected finish Wed Feb 27 2002 00:50:25 :(

Are we not keeping up on this data any more? Might this be a good sticky?

belorsch

02-25-02, 07:31 AM

proteinAnat=10.38 mins/frame on 1.33tbird @1.4

doer

02-25-02, 09:41 AM

GNRA Hairpin (console 2.19 both comps)

Duron700@736Mhz, 8.27m/frame (folding only)
Celeron(tualatin)1200@1604, 5,47m/frame (best recorded)

duron will soon be replaced by 1.4Tb

1604/735,8=2,179

some1 plz check if I count correct the multiplier how much faster duron is folding or make some stupid mistakes ;) / Mhz

2,179x5,47= 11,92

8,27/11,92 = 0,693

So to beat 1604 Celeron (tualatin) in folding, you'll need only (0,693x 1604) 1111Mhz Duron, am I correct?

KFB

02-25-02, 09:19 PM

proteinAg29aNat: 9.68 min/frame Tbird@1496 (see sig for details)

edit: BEST time per frame as recorded by 2.17 text-only client output, not FoldMonitor

Preachr46

02-25-02, 09:49 PM

protienAg29aNat. 9.75m Other than two 17-42's thats all I have been getting.:(

Loud

02-25-02, 10:01 PM

Originally posted by doer
GNRA Hairpin (console 2.19 both comps)

Duron700@736Mhz, 8.27m/frame (folding only)
Celeron(tualatin)1200@1604, 5,47m/frame (best recorded)

duron will soon be replaced by 1.4Tb

1604/735,8=2,179

some1 plz check if I count correct the multiplier how much faster duron is folding or make some stupid mistakes ;) / Mhz

2,179x5,47= 11,92

8,27/11,92 = 0,693

So to beat 1604 Celeron (tualatin) in folding, you'll need only (0,693x 1604) 1111Mhz Duron, am I correct?

Okay. It took me a few minutes to walk through the process, but yes your math seems to be correct.

Of course ... :rolleyes: ... all this assumes that throughput increases in a linear fashion as CPU cycles increase. I crunched some numbers a while back when someone asked about cost / benefit of two Durons at 800MHz vs. one Athlon at 1.6 GHz. The two Durons won out by a slim margin. It's not the size of the margin that really matters here - it's the invalidation of the assumption.

So in the real world, a 1.2 or 1.3 GHz AMD should do nicely. :D

Have you considered just adding another 700MHz Duron to the fold? (sorry - I just couldn't resist the pun :D )

JetMech

02-27-02, 03:25 AM

s

Abeta17-42

Duron 1.2 @ 1.35 = 1.66/frame*

XP 1800+ @ 2000+ = 1.31/frame

P4 1.7 @ 2.13 = 2.43/frame

T-Bird 1.4 @ 1.45 = 1.56/frame

lennytiger

02-27-02, 11:11 AM

Loud

03-04-02, 08:17 AM

1UBQ

100 Frames
500000 Steps
763 Atoms

14.03m / frame
on a Celly 800

Loud

03-05-02, 01:58 AM

1APS

100 frames
500000 steps
1022 atoms

~40 mins / frame
on the Celly 500

deadhead

03-07-02, 12:10 PM

when you say time per frame do you mean the time it takes for it to go from 1/100 to 2/100 etc.?

Kendan

03-07-02, 01:38 PM

Originally posted by deadhead
when you say time per frame do you mean the time it takes for it to go from 1/100 to 2/100 etc.?

Yes that is a frame.

deadhead

03-08-02, 11:16 AM

ok, thanks for clarifying it

Loud

03-08-02, 12:09 PM

Penguin Freak posted this in a different thread, but I thought I'd put it here, too.

2SPZ

100 frames
500000 steps
604 atoms

9.23 min/frame on a Celly 800
(I think Penguin said his Tbird 1100 was folding this one at a bit over 4 min/frame. :eek: WOW!)

Are these proteins and some sort of WU "size" being gathered somewhere? ... I mean ... other than scattered throughout this thread ... :D

Ploaf

03-12-02, 07:02 PM

bump... Hey!! Move it back to the top! :D

res0r9lm

03-12-02, 07:18 PM

ok now that I have got both system up and running and have done quite a few gnra hairpins now I'm ready to post the times.

duron 800@958 6.54 min. per a frame
Tbird 1.4@1618 3.86 min per a frame

these are with gnra hairpins and not just one but several.

belorsch

03-13-02, 06:57 AM

I was folding frames on this protein at 3.29min/frame last night on the XP1600.

Dragonbud

03-13-02, 08:00 AM

1800xp@1621mhz proteinAnat = avg. time/frame = 9.06

duron @ 1456mhz protein29aNat = avt time/frame = 10.41
:beer:

NASsoccer

03-14-02, 12:50 AM

xp2000 @ 1.812 = avg time per fame at 4.28 min

doer

03-18-02, 10:29 AM

I just got my mp3-player folding... it has Pentium MMX 200Mhz cpu lol. Chipset Intel VX, 96Mb ram, Win98SE

bba5_build_ext (steps 500000, atoms 260)
1 frame takes a whooping 38,57mins!!

I think that computer is too slow for folding... 3857mins /60 /24 ~ 2,68. So it would take 2,68 days to get a WU done, hmm.

doer

03-18-02, 10:36 AM

Pentium MMX 200Mhz cpu lol. Chipset Intel VX, 96Mb ram, Win98SE

I think that computer is too slow for folding... 3857mins /60 /24 ~ 2,68. So it would take 2,68 days to get a WU done, hmm. [/B]

yeps, it looks like it's impossible to complete a WU in time. I checked f@h www-pages. Timeout is 2 days.

"Project 550 and 554 (protein name : bba5_build_ext) analyze folding of BBA5. BBA5 is a small protein with 23 amino acids with a well-known folded state. With the release of the new Folding@home software, we have adopted a new simulation algorithm which will be very useful with a massive number of CPUs. Project 550 tests this algorithm with BBA5. Stay tuned to see what Folding@home can do with your help!

Since BBA5 is relatively small, we've set the deadline for returning WUs to be 2 days."

res0r9lm

03-18-02, 12:58 PM

Originally posted by NASsoccer
xp2000 @ 1.812 = avg time per fame at 4.28 min
if those times are folding a gnra hairpin they could be improved alot.

Client : AK31 ( Tbird@ 1654): C:/WINDOWS/Desktop/Folding
Core : Version 2.41 (Jan 22, 2002)
Protein : GNRA hairpin
Frames : 100
Workunit : 02
Atoms : 389
Steps : 500000
Server IP :
Completed : Between Mon Mar 18 08:44:10 2002 and Mon Mar 18 08:44:20 2002
Refresh at : 10 sec.
Frame Time : 3.79 min. ( 227 sec. )
Total Time : 378.50 min. ( 22710 sec. )

Gator

03-18-02, 03:38 PM

Originally posted by res0r9lm

if those times are folding a gnra hairpin they could be improved alot.

Ayup, how's this for size?

Client : niteowl XP@1667mHz: J:/FAHTASK1
Core : Version 2.41 (Jan 22, 2002)
Protein : GNRA hairpin
Frames : 100
Workunit : 03
Atoms : 389
Steps : 500000
Server IP :
Completed : Between Sun Feb 3 20:34:34 2002 and Sun Feb 3 20:39:34 2002
Refresh at : 300 sec.
Frame Time : 3.38 min. ( 203 sec. )
Total Time : 334.70 min. ( 20082 sec. ) :p :p

Gator

03-18-02, 03:55 PM

Most recent three proteins on my #1 machine, these "1xxx" proteins seem to be all different sizes:

Client : niteowl XP@1667mHz: J:/FAHTASK1
Core : Version 2.41 (Jan 22, 2002)
...
Protein : 1APS
Frames : 100
Atoms : 1022
Steps : 500000
Frame Time : 6.64 min. ( 399 sec. )
Total Time : 662.78 min. ( 39767 sec. )
...
Protein : 1SHF
Frames : 100
Atoms : 620
Steps : 500000
Frame Time : 2.44 min. ( 147 sec. )
Total Time : 243.02 min. ( 14581 sec. )
...
Protein : 1UBQ
Frames : 100
Atoms : 763
Steps : 500000
Frame Time : 3.63 min. ( 218 sec. )
Total Time : 363.25 min. ( 21795 sec. )
:beer:

Dragonbud

03-18-02, 05:50 PM

proteinA on

633 celeron
37.02m avg. time/frame:p

duron @1.46mhz
10.68m avg.time/frame

GNRA protein

duron @ 1.46mhz = 4.45 avg time/frame
1800xp @ 1.68mhz = 3.82 avg time/frame
celeron@ 633mhz = 15.52 avg time/frame

1APS
1800xp @ 1.68 = 6.73 avgtime/frame
duron @ 1.3 = 8.88 avgtme/frame

res0r9lm

03-18-02, 05:59 PM

Originally posted by Gator

Ayup, how's this for size?

Client : niteowl XP@1667mHz: J:/FAHTASK1
Core : Version 2.41 (Jan 22, 2002)
Protein : GNRA hairpin
Frames : 100
Workunit : 03
Atoms : 389
Steps : 500000
Server IP :
Completed : Between Sun Feb 3 20:34:34 2002 and Sun Feb 3 20:39:34 2002
Refresh at : 300 sec.
Frame Time : 3.38 min. ( 203 sec. )
Total Time : 334.70 min. ( 20082 sec. ) :p :p

pretty fast for only 1.67ghz. folding monitor is messing up though. average frame rate should be 3.34 min per a frame if the total time is right. I see you refesh rate is at 300 sec maybe that had something to do with it but I don't see how.

JetMech

03-26-02, 02:06 PM

Finally getting the big ones.

XP1800+ @ 1672
ProteinA - 8.88 min/frame
ProteinAnat - 8.85 min/frame
ProteinAg29a - 9.04 min/frame
ProteinAg29aNat - 8.86 min/frame

Duron 1.2 @ 1350
ProteinA - 11.09 min/frame
ProteinAnat - 11.06 min/frame
ProteinAg29a - 11.11 min/frame
ProteinAg29aNat - 11.13 min/frame

T-Bird 1.4 @ 1450
proteinA - 10.43 min/frame
ProteinAnat - 10.39 min/frame
ProteinAg29a - 10.48 min/frame
ProteinAg29aNat - 10.44 min/frame

P4 1.7 @ 2005
ProteinA - 17.02 min/frame
ProteinAnat - 17.00 min/frame
ProteinAg29a - 17.21 min/frame
ProteinAg29aNat - 16.09 min/frame (clock @ 2125mhz)

New Core 65
P4 1.7 @ 2125 = min/frame
XP1800+ @ 1672 = 7.70 min/frame
T-Bird 1.4 @ 1450 = 8.96 min/frame
Duron 1.2 @ 1350 = 9.63 min/frame

res0r9lm

03-26-02, 11:45 PM

Originally posted by Ploaf
Clock for clock durons and tbirds are almost identical for folding. The differences in times are usually due to different clock speeds. I have matched a Duron at 1000 with a 1000 tbird and they ran pretty much neck and neck. The setups were pretty much the same except for the cpu. Afterward the tbird was put right back up to 1.4 and now it kicks the snot out of the duron.
If you are planning to purchase more than one machine then somthing else to consider in the bang/buck model is that you may be able to hook up with a single XP2000 and OC it for all it's worth for less than it would cost to get 2 durons up and running. Remember that you are not just buying a cpu, but also a mobo, ram, hd, vid, nic(maybe), various drives, psu/case, cables, electricity to run all that extra stuff. There are advantages to having more than one system, though. so either way it works.
I think the newer duron morgan might fold slightly fast than tbirds clock for clock. I'm basing this on the fact that xp's are folding a little faster than tbird and the morgan are basicly a 200fsb xp

RickShelton

03-28-02, 12:07 AM

It looks like my P4 is getting stomped by my "Birds" even the 1.1 Athlon beats the P4 at Folding Hairpins.

Protein :GNRA Hairpins.
Frame Time : 5.53 min. ( 332 sec. ) Athlon 1ghz @ 1.1ghz, KT7-Raid, 256m Ram, WinXP.
Frame Time : 4.22 min. ( 253 sec. ) Athlon 1.4ghz@1.47ghz, KT7A, 512m Ram, WinXP
Frame Time : 5.98 min. ( 359 sec. ) P4 1.6a@2.35ghz, P4B266-C, 512m DDR Ram, WinXP

Protein : Ab28-43V
Frame Time : 2.42 min. ( 145 sec. ) 1.47 Athlon
Frame Time : 3.49 min. ( 209 sec. ) 2.35 P4
The Athlon 1.1 is only getting Hairpins so no data avail.

Protein : bba5_build_ext
Frame Time : 1.63 min. ( 98 sec. ) 1.4 Athlon
Frame Time : 1.77 min. ( 106 sec. ) 2.35 P4

Protein : NativeBBA5
Frame Time : 1.54 min. ( 92 sec. ) 1.4 Athlon
Frame Time : 1.69 min. ( 102 sec. ) 2.35 P4

Those were averages taken from Foldmon log. The Athlons are dedicated Folders with not much else running and the P4 is my main system.
Now every benchmark I have ran tells me the P4 is faster but this Folding makes me wonder if core_65 is as optimized as it can be?

Food for thought, Fold on :)
RickShelton, Team 32

res0r9lm

03-28-02, 12:24 AM

yea p4 aren't much on folding

Tweaked!

03-28-02, 05:08 AM

Tell me about it! I've got 4 rigs ranging from a p4, p3, p2, and a mmx for a combined mhz of 2932. I'm getting stomped by people w/ less rigs and mhz in Amd. Thats a little disheartening :(
Bet you can guess what my next rig is going to be....;)
(but its all good)

JetMech

03-29-02, 12:06 PM

These times were posted by Newbie_Doo in another thread.

Newbie_Doo's Rigs:

C2-600@900
ProteinA- 29.25 min/frame

P3-600@800
ProteinAg29a - 27.12 min/frame

Athlon 1000@1333
ProteinAg29aNat - 11.76 min/frame

XP1600+
ProteinAg29a - 10.58 min/frame

P3 - 1000
ProteinA - 20.63 min/frame

belorsch

03-29-02, 01:30 PM

Tbird 1330@1507
Protein l6c 1.50m/90s a frame

XP1600@1480
Protein 2ABD 6.10m/366s a frame

Newbie_Doo

03-29-02, 02:08 PM

XP1600+
proteinAg29Nat
10.46 Min / Frame

T-Bird 1000@1333
proteinAg29aNat
11.54 Min / Frame

Xp1600+
proteinAnat
11.72 Min / Frame

C2-600@900
proteinAnat
24.21 Min / Frame

P3-1000EB
proteinA
20.61 Min / Frame

C2-600@900
proteinA
28.43 Min / Frame

Something interesting:
I have 2 identical P3-600@800 on a dual processor box. Both are folding the same protein. Each protein has one processor assigned exclusively to it. The only difference is that one is the graphical client, and the other is the console version.

P3-600@800 (Graphic)
proteinAg29a
27.04 Min / Frame (1622 sec.)

P3-600@800 (Console)
proteinAg29a
26.77 Min / Frame (1606 sec.)

JetMech

03-29-02, 06:29 PM

I think you have proven the question that's been asked for a while now. Which folds faster, graphical or console. With the faster folding proteins the difference is so small that no one could really see a difference. With the monster proteins the difference has become obvious. Good find Newbie_Doo.:beer: :burn:

KFB

04-01-02, 01:51 AM

yeah definitely good work newbie_doo. i always figured the console had to be faster, since the graphical portion is bound to use the occasional cpu cycle. your data backs the theory up.:D

res0r9lm

04-03-02, 01:42 AM

finally getting some of the big ones.:clap:
Client : AK31 ( Tbird@ 1654): C:/WINDOWS/Desktop/Folding
Core : Version 2.41 (Jan 22, 2002)
Protein : proteinAg29aNat
Frames : 100
Workunit : 04
Atoms : 602
Steps : 500000
Server IP :
Completed : Between Wed Apr 3 00:24:43 2002 and Wed Apr 3 00:24:53 2002
Refresh at : 10 sec.
Frame Time : 8.87 min. ( 532 sec. )
Total Time : 887.32 min. ( 53239 sec.)

JetMech

04-03-02, 03:58 AM

Originally posted by res0r9lm
finally getting some of the big ones.:clap:
Client : AK31 ( Tbird@ 1654): C:/WINDOWS/Desktop/Folding
Core : Version 2.41 (Jan 22, 2002)
Protein : proteinAg29aNat
Frames : 100
Workunit : 04
Atoms : 602
Steps : 500000
Server IP :
Completed : Between Wed Apr 3 00:24:43 2002 and Wed Apr 3 00:24:53 2002
Refresh at : 10 sec.
Frame Time : 8.87 min. ( 532 sec. )
Total Time : 887.32 min. ( 53239 sec.) this thread is finally paying off. Ralph your TBird @ 1654 folds that 29aNat 0.01 sec slower than my 1800+ @ 1672. So much for XP's Folding faster at a given speed.:rolleyes: :beer: :burn:

Malakai

04-03-02, 06:48 AM

heres me with that protien, i guess that is put to rest
i just take a moment here to see that overclocking gains me a little over 2min per frame, wich in turn knocks off 200 minutes of total wu time. thats a lot!!!!! o/c'ing rules
-Malakai

res0r9lm

04-03-02, 12:23 PM

Originally posted by JetMech
this thread is finally paying off. Ralph your TBird @ 1654 folds that 29aNat 0.01 sec slower than my 1800+ @ 1672. So much for XP's Folding faster at a given speed.:rolleyes: :beer: :burn:
I don't know if it has anything to do with how good my cooling is or not but most people have a hard time keeping the tbirds cool even at default. my time might have been better than yours jetmech if I would have had refresh set to 600 sec instead of 10 sec.
ocua has a speed contest going on over there and fm has to be set to 10. my tbird has been right in the middle of thier ranking which the rest is xps. you have my wondering if I just got luck when I got this tbird maybe folds faster than the average tbird.

Loud

04-04-02, 06:00 PM

Ab28-43V

100 frames
2500000 steps (yes, 2.5 M :eek: )
125 atoms

7.70 m/frame
on the Celly 800

res0r9lm

04-04-02, 08:55 PM

ok here is another protienA

Client : AK31 ( Tbird@ 1654): C:/WINDOWS/Desktop/Folding
Core : Version 2.41 (Jan 22, 2002)
Protein : proteinAg29a
Frames : 100
Workunit : 07
Atoms : 602
Steps : 500000
Server IP :
Completed : Between Thu Apr 4 20:49:11 2002 and Thu Apr 4 20:49:21 2002
Refresh at : 10 sec.
Frame Time : 8.87 min. ( 532 sec. )
Total Time : 886.88 min. ( 53213 sec. )

belorsch

04-06-02, 08:46 AM

Client : My Client ( 1600+@1470):
Core : Version 2.41 (Jan 22, 2002)
Protein : 601_BBAW_helix
Frames : 200
Workunit : 06
Atoms : 133
Steps : 1000000
Server IP :
Completed : Between Sat Apr 6 09:37:53 2002 and Sat Apr 6 09:40:53 2002
Refresh at : 180 sec.
Frame Time : 0.48 min. ( 29 sec. )
Total Time : 96.22 min. ( 5773 sec. )

so many choices

04-12-02, 09:58 AM

proteinAnat 9.55m a frame

-I'm currently using version 2.19. Would it be faster if I used 2.17?

res0r9lm

04-12-02, 10:08 AM

Originally posted by so many choices
proteinAnat 9.55m a frame

-I'm currently using version 2.19. Would it be faster if I used 2.17?
no, but you could improve your times something is slowing you down at least compared to my tbird. I get in the high 8's per a frame

KFB

04-12-02, 11:35 AM

Originally posted by so many choices
proteinAnat 9.55m a frame

-I'm currently using version 2.19. Would it be faster if I used 2.17?

my sig system @1463 gets the same frame in 9:55 (9 minutes 55 seconds)

Gator

04-12-02, 12:05 PM

Client : waystar task 2 PIII@600mHz: K:/FAHTASK2
Core : Version 2.41 (Jan 22, 2002)
Protein : proteinA
Frames : 100
Workunit : 05
Atoms : 601
Steps : 500000
Server IP :
Completed : Between Wed Apr 10 05:06:32 2002 and Wed Apr 10 05:08:32 2002
Refresh at : 120 sec.
Frame Time : 35.42 min. ( 2125 sec. )
Total Time : 3540.17 min. ( 212410 sec. )

Client : waystar task 1 PIII@600mHz: K:/FAHTASK1
Core : Version 2.41 (Jan 22, 2002)
Protein : proteinAnat
Frames : 100
Workunit : 03
Atoms : 601
Steps : 500000
Server IP :
Completed : Between Wed Apr 10 05:12:32 2002 and Wed Apr 10 05:14:32 2002
Refresh at : 120 sec.
Frame Time : 35.44 min. ( 2127 sec. )
Total Time : 3544.10 min. ( 212646 sec. )

Note:
"waystar" is an SMP dual Pentium-III/600mHz (Supermicro P6DGU mobo) running two instances of FAH2Console v2.19.

Client : niteowl XP@1667mHz: J:/FAHTASK1
Core : Version 2.41 (Jan 22, 2002)
Protein : proteinAg29aNat
Frames : 100
Workunit : 05
Atoms : 602
Steps : 500000
Server IP :
Completed : Between Wed Apr 10 22:19:34 2002 and Wed Apr 10 22:21:34 2002
Refresh at : 120 sec.
Frame Time : 8.42 min. ( 505 sec. )
Total Time : 841.83 min. ( 50510 sec. )

Client : nitehawk Tbird@1.4gHz: F:/FAHTASK1
Core : Version 2.41 (Jan 22, 2002)
Protein : proteinAnat
Frames : 100
Workunit : 02
Atoms : 601
Steps : 500000
Server IP :
Completed : Between Thu Apr 11 08:27:36 2002 and Thu Apr 11 08:29:36 2002
Refresh at : 120 sec.
Frame Time : 10.02 min. ( 601 sec. )
Total Time : 1001.75 min. ( 60105 sec. )

Client : niteowl XP@1667mHz: J:/FAHTASK1
Core : Version 2.41 (Jan 22, 2002)
Protein : proteinAnat
Frames : 100
Workunit : 06
Atoms : 601
Steps : 500000
Server IP :
Completed : Between Thu Apr 11 12:19:37 2002 and Thu Apr 11 12:21:37 2002
Refresh at : 120 sec.
Frame Time : 8.40 min. ( 504 sec. )
Total Time : 838.95 min. ( 50337 sec. )

Matthew1001

04-12-02, 05:15 PM

My average time per frame is 2.54 mins.

Wedo

04-12-02, 06:51 PM

Here's my short list. It's nine of the 13 machine's I have running because I don't leave the Fold Monitor on since it slows down my production (and makes my actual "work" that I need to do more difficult).

Client : YellowBrickRoad ( PIII @ 933MHZ)
Protein : bba5_build_ext
Frames : 100
Atoms : 260
Steps : 500000
Completed : Between Wed Apr 3 14:12:51 2002 and Wed Apr 3 14:22:51 2002
Frame Time : 4.39 min. ( 263 sec. )
Total Time : 437.38 min. ( 26243 sec. )

Client : Over The Rainbow (PIII @ 933MHZ)
Frames : 100
Workunit : 03
Atoms : 126
Steps : 2500000
Completed : Between Wed Apr 3 14:23:17 2002 and Wed Apr 3 14:33:17 2002
Frame Time : 5.78 min. ( 347 sec. )
Total Time : 571.52 min. ( 34291 sec. )

Client : Auntie_Em (PIII @ 933MHZ)
Protein : NativeBBA5
Frames : 100
Atoms : 260
Steps : 160000
Completed : Between Wed Apr 3 18:15:22 2002 and Wed Apr 3 18:25:22 2002
Frame Time : 5.34 min. ( 320 sec. )
Total Time : 527.53 min. ( 31652 sec. )

Client : SingingTree (PIII @ 450MHZ)
Protein : NativeBBA5
Frames : 100
Atoms : 260
Steps : 500000
Completed : Between Wed Apr 3 18:35:22 2002 and Wed Apr 3 18:45:22 2002
Frame Time : 8.71 min. ( 522 sec. )
Total Time : 865.33 min. ( 51920 sec. )

Client : Poppies (Celeron @ 400MHZ )
Protein : Ab28-43V
Frames : 100
Atoms : 125
Steps : 2500000
Completed : Between Thu Apr 4 04:35:27 2002 and Thu Apr 4 04:45:27 2002
Frame Time : 12.99 min. ( 779 sec. )
Total Time : 1293.75 min. ( 77625 sec. )

Client : 32-Bit-Horse (AMD @ 1.4GHZ)
Frames : 100
Atoms : 602
Steps : 375000
Completed : Between Thu Apr 4 06:25:25 2002 and Thu Apr 4 06:35:25 2002
Frame Time : 11.37 min. ( 682 sec. )
Total Time : 1135.32 min. ( 68119 sec. )

Client : Toto-Too (AMD @ 1GHZ)
Protein : proteinAg29a
Frames : 100
Workunit : 01
Atoms : 602
Steps : 500000
Completed : Between Thu Apr 4 11:55:26 2002 and Thu Apr 4 12:05:26 2002
Frame Time : 14.06 min. ( 844 sec. )
Total Time : 1398.87 min. ( 83932 sec. )

Client : RubySlippers (PIII @ 1GHZ)
Protein : NativeBBA5
Frames : 100
Workunit : 01
Atoms : 260
Steps : 500000
Completed : Between Thu Apr 4 12:43:20 2002 and Thu Apr 4 12:53:20 2002
Frame Time : 13.61 min. ( 817 sec. )
Total Time : 1349.73 min. ( 80984 sec. )

Client : Rainbow (PIII @ 667MHZ)
Protein : bba5_build_ext
Frames : 100
Atoms : 260
Steps : 500000
Completed : Between Tue Apr 2 15:10:43 2002 and Tue Apr 2 15:20:43 2002
Frame Time : 6.10 min. ( 366 sec. )
Total Time : 601.98 min. ( 36119 sec. )

These computer's are work station's durring the day, but run all night without an end user slowing things down.

Fold On.

Wedo

Dragonbud

05-11-02, 11:49 AM

*bump* it up!

Gator

05-11-02, 06:45 PM

These P70x bag-biters may only be 10 frames each but the molecules are bloody huge...during folding, the "work" directory takes up as much 10-megabytes or more. :eek:

Client : niteowl XP@1667mHz: J:/FAHTASK1
Core : Version 2.44 (May 2, 2002)
Protein : P701_Src/Peptide
Frames : 10
Workunit : 08
Atoms : 10482
Steps : 50000
Server IP :
Completed : Between Sat May 11 18:42:48 2002 and Sat May 11 18:44:48 2002
Refresh at : 120 sec.
Frame Time : 36.74 min. ( 2205 sec. )
Total Time : 329.23 min. ( 19754 sec. )

Whomever said that only the Linux clients were getting P70x's is mistaken... "niteowl" is running Windows-2000-Pro and has been chewing on P70x's for the past couple weeks. :p

Fold on people... this bud's for you! :beer: