Folding Farm Spreadsheet

Complaint withdrawn. It seems that my motherboard was at fault, not the power supply. The board, despite appearances to the contrary in the BIOS, will not support memory dividers, thereby making my Corsair value ram run faster than it will boot with. Have to find another motherboard, it seems.
 
Bill Clo said:
I'm posting to provide some feedback on the choice of power supply listed in the AMD Socket-939 spreadsheet. I tried it out with an X2 3800, and while it is barely adequate for this processor, I'd strongly suggest anyone contemplating even a mild overclock to skip it. At stock speed, 1 hard drive, and a PCI video card (10 watts, max), and 2 instances of FAH, the 12v rails are providing 11.55v, barely within the 5% max ATX-spec tolerance. As soon as I pushed the FSB past 210 (2100 mhz), the system will no longer even boot up.

So while this power supply may do for a bare bones dual core system that will do nothing but FAH, don't bother if you had planned to overclock at all. I'd even hesitate to recommend it for one of the 4200/4400/4800 X2 processors.

Your mileage may vary, but I thought everyone ought to hear this.

Are you using the BIOS, or some other sort of software to get your voltages? Having any stability issues? Stop by and see me in the PSU forum because that doesn't sound right.
 
Spectre said:
Are you using the BIOS, or some other sort of software to get your voltages? Having any stability issues? Stop by and see me in the PSU forum because that doesn't sound right.

When running 2 instances of FAH, and stock speed, I was using Abit's EQ utility program to determine voltage. But the voltage strangely enough agreed with what the BIOS was showing .

While at stock speed, I ran FAH for 5 days with no crashes/stability issues.

As an experiment, I used Clockgen (for Nforce4) to try and see was going on. After I pushed beyond 210FSB, I noticed the voltage as reported by Abit EQ to be 11.4v. Any further raising of FSB resulted in a crash.

I had memory set to CAS 3, default settings, memory speed set to DDR400. I don't know if my board has a "divider" or "memory speed lock"; don't see one explicitly mentioned in the manual. I'm using an Abit KN8.

Seeing the voltage so low, made me think PSU issue, so I ordered up a better one. I'll report back with the results.
 
Bill Clo said:
When running 2 instances of FAH, and stock speed, I was using Abit's EQ utility program to determine voltage. But the voltage strangely enough agreed with what the BIOS was showing .

While at stock speed, I ran FAH for 5 days with no crashes/stability issues.

As an experiment, I used Clockgen (for Nforce4) to try and see was going on. After I pushed beyond 210FSB, I noticed the voltage as reported by Abit EQ to be 11.4v. Any further raising of FSB resulted in a crash.

I had memory set to CAS 3, default settings, memory speed set to DDR400. I don't know if my board has a "divider" or "memory speed lock"; don't see one explicitly mentioned in the manual. I'm using an Abit KN8.

Seeing the voltage so low, made me think PSU issue, so I ordered up a better one. I'll report back with the results.


Software...BIOS or otherwise......... is unrealiable on the best days. To get an accurate voltage reading you need to check it with a volmeter/multimeter/DMM.
 
Spectre said:
Software...BIOS or otherwise......... is unrealiable on the best days. To get an accurate voltage reading you need to check it with a volmeter/multimeter/DMM.

Is there a foolproof/noobproof way of doing this, without frying something? I do have a voltmeter, but I'd be real reluctant to try it and risk $600 worth of psu, board, and cpu...

Preferably something with easy to see pics... [:)]
 
Stick the voltmeter in one of the molex plugs not currently (hah) in use; red in red and black in black for +5V check, and then red in yellow and black in black for +12V. That should tell you everything you need to know. And all of the black wires should (IIRC) be continuous to each other; there's no difference which one the black lead goes in.

 
I just updated the Spreadsheet for all those interested... Lots of prices changes... most of which are lower than before.... only a few actually went up.....

As always, let me know what you guys think!


Keep on Folding!! For the [H]orde!!

 
On the Socke 775 (single and Dual core) and the Socket 478, won't you need two sticks of RAM? Running in a pair to get the 800 MHz FSB?
 
Maybe?? I haven't ventured forth into the realm of anything beyond a 2.4C, so I'm not sure..... I've got a meeting here in a few minutes, I'll see what I can dig up after it....


Keep on Folding!! For the [H]orde!!

 
One recommendation?

socket 754 mobos. How about a Chaintech VNF3-250. They are about $10 cheaper at the egg and o/c quite well.

 
magnusvir said:
One recommendation?

socket 754 mobos. How about a Chaintech VNF3-250. They are about $10 cheaper at the egg and o/c quite well.


The only thing it lacks is a onboard video.... I was aiming for all-in-one solutions... but if you have a spare vid card, it would definately be the cheaper solution...


Keep on Folding!! For the [H]orde!!

 
What is a good overclocking mATX socket 754 motherboard? I'm looking for 3. I've bee looking at the Biostart TForce 6100...
 
brycejones said:
What is a good overclocking mATX socket 754 motherboard? I'm looking for 3. I've bee looking at the Biostart TForce 6100...

Usually onboard video limits OCing but the guys in SFF have had some luck with theat board.
 
Will all 3 of those boards support the X2s? or just this one....


Keep on Folding!! For the [H]orde!!

 
Just the 939 board would support X2s since there is no socket 754 X2 that I am aware of.
 
OSUguy98 said:
Will all 3 of those boards support the X2s? or just this one....


Keep on Folding!! For the [H]orde!!



Well the last board is a skt 754 so no.....but according to Biostar's US website the other two do support X2's.
 
oops... didn't catch that the last one was a s754.... the first 2 were 939s...... I assumed the 3rd was too :D

edit-----Updated the pages...


Keep on Folding!! For the [H]orde!!

 
Bill Clo said:
Excellent guide. I'll check voltage this evening and report back. I think we ought to continue this in private mail so as to not hijack the message thread.

I checked the voltage, and it was 12.01v, not 11.55 as reported by BIOS/software. After further tinkering, it appears that the motherboard is not supporting memory dividers, so I can't run my Corsair Value Ram at 200 mhz, while the CPU is faster. Have to get another motherboard, apparently. Withdrew my complaint about the power supply.
 
Bill Clo said:
I checked the voltage, and it was 12.01v, not 11.55 as reported by BIOS/software. After further tinkering, it appears that the motherboard is not supporting memory dividers, so I can't run my Corsair Value Ram at 200 mhz, while the CPU is faster. Have to get another motherboard, apparently. Withdrew my complaint about the power supply.


Sorry it is the mobo.
 
Thanks Moose!
I'm not sure how useful all that is... but it's there none-the-less ;)


Keep on Folding!! For the [H]orde!!

 
Nice work. I noticed a couple things to ask about.

1) Is there a plan to populate the "estimated points per year" column? I realize this can be wildly inaccurate (I get twice as many ppd on a 600-point Gromacs than on the <100-point ones), so I was wondering.

that leads to my second question:

2) Do you plan to add a column for $$/point or points/dollar? Yes, this goes back to predicting a largely unpredictable number, but I think it's a good benchmark for value.

This leads to yet another question, and maybe I should start a new thread for this. Do we have a database for performance of various processors for various projects? With the number of people here, I get the feeling we could get a fairly-thorough table of how each processor does on each project. For example, my AthlonXP 1800 gets around 150ppd on a 600-point Gromacs, and my Pentium-M 1.4 (at 100%) gets about 206 ppd on the same kind of WU.

 
1. Yes, there's a plan... I've given the numbers for processors that I have... I need alot more info for that.... so if you guys out there have any data, feel free to post it... rkhands linked me to his page a few months back, and part of that info is his... I intended that column to be an average, to account for the differences between WUs/etc...

2. $$/point or points/$ would have to be something added once I gathered more info from the above question....


The power consumption is something else that's subjective, that $0.076/kwh is a number I got from the average of all the numbers that we gave in a thread a few months back... Gnewbury has helped with the power consumption side of things, but more data is always good ;)


I don't think we really have any performance stats on processors for different projects... you'd almost need to focus on a family of WUs, like the 600pointers:

Code:
Protien                                Points
p1140_RIBO_FSpeptide_EXT_nospring      600.00
p1141_RIBO_FSpeptide_HEL_nospring      600.00
p1144_RIBO_nopeptide                   600.00
p1149_RIBO_semihelixfrom1140           600.00
p1150_RIBO_semihelixfrom1141           600.00

or p1710 thru p1719.... they're all similar WUs, and have the same point values (and/or bonuses)..... you can't really compare a gromac average to an amber average.... you'd have to keep it as close to "apples to apples" as possible..... even if it's "granny smiths to red delicious"...... in other words, I might need a little help doing that lol

As it stands right now, I just have a big Excel file that I "save as web page" to each sheet to make it presentable... there are easier, more effective,smaller ways to do this, but I've never found time to learn how... If we were to expand it much further, I'd almost have to get some help in that area.... You guys have helped out quite a bit in finding mistakes and/or suggesting other/cheaper hardware (or correcting me when I've been WAY off :) ).... I appreciate all the help I can get....

I'd like this to be a useful guide for those of us out there that want to build another boxen or don't have enough time to learn about the latest and greatest (or can't afford it)...


Keep on Folding!! For the [H]orde!!

 
OSUguy98 said:
I don't think we really have any performance stats on processors for different projects... you'd almost need to focus on a family of WUs, like the 600pointers:

<snip> 1140-1150 </snip>

or p1710 thru p1719.... they're all similar WUs, and have the same point values (and/or bonuses)..... you can't really compare a gromac average to an amber average.... you'd have to keep it as close to "apples to apples" as possible..... even if it's "granny smiths to red delicious"...... in other words, I might need a little help doing that lol
I agree. Here's my idea. It's just a simple spreadsheet. Each row is a processor, each column is a project or family of projects. Just a simple matrix. Perhaps with color coding to indicate level of PPD.

Should I start assembling such a graph? I don't know if it has already been done by anyone, and you could incorporate some of the data into your spreadsheets.

 
I don't know that it's been done before.... maybe on a private scale, but not a whole team's info.....

Feel free to play around with some ideas.... I'll see what I can come up with too... I have over a year and a half worth of data from EMIII that I use to make this.... I'll play with those numbers/porjects and see what I can come up with.... next week, everyone here at work will be off for hunting season, with the exception of 3 or 4 of us (out of 13 or 14).... so I'll start today and probably have something worth looking at next week.... Let me know what you come up with.... then we can compare/combine/etc


Keep on Folding!! For the [H]orde!!

 
I updated the Spreadsheet this morning... made a couple changes you guys might want to look over... I changed the RAM for the Socket 479 setups (and in turn the Socket 775 Dual Core setups as well).... and I changed the motherboard for the Socket 775 setups..... Let me know what you guys think!


Keep on Folding!! For the [H]orde!!

 
Man, those pages are malformatted like a bitch in firefox. They show up fine in IE though. Clack another one up for that craptacular code MS generates from excel! I really wish they would generate html 4.01 instead of xml. Oh well.

Nice work on the speadsheet, but it might be time to update the kWh price, as mine has jumped by 50%(14c/kWh...ouch) in the past few months, I'm not sure if it's a trend everywhere.
 
sandmanx said:
Man, those pages are malformatted like a bitch in firefox. They show up fine in IE though. Clack another one up for that craptacular code MS generates from excel! I really wish they would generate html 4.01 instead of xml. Oh well.

Not sure what you mean malformated they display fine in Firefox for me.
 
Spectre said:
Not sure what you mean malformated they display fine in Firefox for me.

Hmmm, it's displaying fine for me now as well. It was displaying the columns all messed up earlier. Weird, I haven't rebooted either.
 
sandmanx said:
Hmmm, it's displaying fine for me now as well. It was displaying the columns all messed up earlier. Weird, I haven't rebooted either.

Had to do the View in IE from Firefox.... Maybe not... :(


 
sandmanx said:
Man, those pages are malformatted like a bitch in firefox. .... Clack another one up for that craptacular code MS generates from excel!
Whenever I'm forced to export something from office into a web document, I go into the "html"/XML afterward and do a lot of global find-and-replace (typically replacing with nothing!). It reduces the size of the file by at least 75%.


 
Ran across an interesting thread and figured I'd share it with you guys.... this makes things MUCH easier to update... I'm sure it's not a perfect solution... but it works... so far at least....

I should have the update done sometime today... so check back here to see the new changes.... I've got everything updated.... just trying to figure out how to have it say something other than #Value if the website doesn't exist......... I'll work on that this afternoon.... work calls... lol

edit---I updated the Socket 479 and 478 sheets so you guys could see the changes


Keep on Folding!! For the [H]orde!!

 
Back
Top