SPCR Folds Team Blog
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
-
- Friend of SPCR
- Posts: 1439
- Joined: Tue Dec 14, 2004 4:06 pm
- Location: New Hampshire, US
- Contact:
I just had a WU randomly freeze at 69%...power consumption went from 111W to 97W, so I checked the processes...the FAH cores were working, and using the proper amount of cpu power, but after an hour, the console didn't show any changes, still at 69%. So, I restarted it, and instead of continuing, it said "missing work files"....restarting again had no effect, I had to eventually delete the 69% of the work. I think it might be related to the system clock. I run a "dual boot", I've got ubuntu on an external hard drive as I experiment with undervolting in linux. It's not a true dual boot, instead, I just tell the BIOS which hard drive to treat as primary, and it boots accordingly. Well, a strange side-effect of this is that the system clock in windows is off by +4 hours when I boot back to it. I changed it back and after almost 20%, no issues. The system is stable, does this imply that having the software clock differ from the BIOS clock cause F@H instability? Doesn't seem like it would.
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
Both my SMP folding finished a WU yesterday. I got 3283 points (an odd number?) and each checkpoint took longer than the 1760 pointers.
I'm assuming I got a 1760 now because its running quicker.
We will drop another place in a couple of months, it's inevitable as OcUk is averaging twice as many PPD as our team.
Our team numbers though are up nicely as bolek rushes to fill the void that vg30et left.
Updated for shpelling. D'Oh!
I'm assuming I got a 1760 now because its running quicker.
We will drop another place in a couple of months, it's inevitable as OcUk is averaging twice as many PPD as our team.
Our team numbers though are up nicely as bolek rushes to fill the void that vg30et left.
Updated for shpelling. D'Oh!
Last edited by aristide1 on Thu Nov 01, 2007 12:49 pm, edited 1 time in total.
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
HardOCP tested one of Intel's new 45nm quad processors. The thing was totally stable at 4GHz, I think they ran it up to about 4.25GHz.
Also - I am not building anymore AMD folders, the Intel E2140 is a much better bang for the buck CPU. And one can put it on a P35 board and OC it's brains out while everything else runs at stock speeds.
Also - I am not building anymore AMD folders, the Intel E2140 is a much better bang for the buck CPU. And one can put it on a P35 board and OC it's brains out while everything else runs at stock speeds.
-
- Friend of SPCR
- Posts: 1439
- Joined: Tue Dec 14, 2004 4:06 pm
- Location: New Hampshire, US
- Contact:
I wish I could have gone the E21xx route...but I was around 50$ outside of my budget, and when you have no income, it matters. Hopefully the Phenoms will be amazing at performance/watt, considering the B2 phenom at 2.3ghz is only using .976w, that's pretty promising.aristide1 wrote: Also - I am not building anymore AMD folders, the Intel E2140 is a much better bang for the buck CPU. And one can put it on a P35 board and OC it's brains out while everything else runs at stock speeds.
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
I don't recall if it was out when I bought my 2nd system, only that it set me back all of $180 for cpu, ramm, mobo. There was something amazing about getting a dual core for $70. I realize its all in my head.ryboto wrote:I wish I could have gone the E21xx route...but I was around 50$ outside of my budget, and when you have no income, it matters. Hopefully the Phenoms will be amazing at performance/watt, considering the B2 phenom at 2.3ghz is only using .976w, that's pretty promising.
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
Is this your AMD X2 box? My idles 65 watts and folds at 95watts and that's with a 230MHz FSB, about 2.4GHz.ryboto wrote:I just had a WU randomly freeze at 69%...power consumption went from 111W to 97W,....
Really interesting was I went over to use the extreme power supply calculator, and even it said I didn't need more than a 250 watt power supply.
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
I'd like to setup a couple of quad cores, but yes electricity rates are a factor, so I suspect I will add one more dual core in the future, and only if I get a tax refund next year.ryboto wrote:I wish I had the money to afford the energy required to overclock this computer...or even better, buy a second dedicated to folding. Oh well, it'll have to wait until I get a job.
When is Intel's next price drop? All the fun ceased when the Q6600 reached $280. Now it's just sitting there.
Project : 2605
Core : SMP Gromacs
Frames : 100
Credit : 1760
Min. Time / Frame : 11mn 41s - 2169.24 ppd
Avg. Time / Frame : 19mn 30s - 1299.69 ppd
Cur. Time / Frame : 11mn 55s - 2126.77 ppd
R3F. Time / Frame : 11mn 54s - 2129.75 ppd
... Avg is from the old E2140 @ 2.4GHz, rest is with E6600 @ 3GHz
Project : 2609
Core : SMP Gromacs
Frames : 100
Credit : 1523
-- titan -- E6600 @ 3GHz
Min. Time / Frame : 13mn 53s - 1579.68 ppd
Avg. Time / Frame : 13mn 55s - 1575.89 ppd
Cur. Time / Frame : 13mn 55s - 1575.89 ppd
R3F. Time / Frame : 13mn 54s - 1577.78 ppd
Project : 2653
Core : SMP Gromacs
Frames : 100
Credit : 1760
-- Workstation -- E6600 @ 3.2GHz
Min. Time / Frame : 3mn 08s - 8088.51 ppd <- obviously wrong.
Avg. Time / Frame : 25mn 51s - 980.43 ppd
Cur. Time / Frame : 15mn 44s - 1610.85 ppd
R3F. Time / Frame : 15mn 44s - 1610.85 ppd
-- //titan/wibla/folding/ -- E2140 @ 2.4GHz
Min. Time / Frame : 19mn 06s - 1326.91 ppd
Avg. Time / Frame : 20mn 39s - 1227.31 ppd
No Cur. Time / Frame
No R3F. Time / Frame
I'm running debian amd64 on the folding box, winXP on the workstation...
Core : SMP Gromacs
Frames : 100
Credit : 1760
Min. Time / Frame : 11mn 41s - 2169.24 ppd
Avg. Time / Frame : 19mn 30s - 1299.69 ppd
Cur. Time / Frame : 11mn 55s - 2126.77 ppd
R3F. Time / Frame : 11mn 54s - 2129.75 ppd
... Avg is from the old E2140 @ 2.4GHz, rest is with E6600 @ 3GHz
Project : 2609
Core : SMP Gromacs
Frames : 100
Credit : 1523
-- titan -- E6600 @ 3GHz
Min. Time / Frame : 13mn 53s - 1579.68 ppd
Avg. Time / Frame : 13mn 55s - 1575.89 ppd
Cur. Time / Frame : 13mn 55s - 1575.89 ppd
R3F. Time / Frame : 13mn 54s - 1577.78 ppd
Project : 2653
Core : SMP Gromacs
Frames : 100
Credit : 1760
-- Workstation -- E6600 @ 3.2GHz
Min. Time / Frame : 3mn 08s - 8088.51 ppd <- obviously wrong.
Avg. Time / Frame : 25mn 51s - 980.43 ppd
Cur. Time / Frame : 15mn 44s - 1610.85 ppd
R3F. Time / Frame : 15mn 44s - 1610.85 ppd
-- //titan/wibla/folding/ -- E2140 @ 2.4GHz
Min. Time / Frame : 19mn 06s - 1326.91 ppd
Avg. Time / Frame : 20mn 39s - 1227.31 ppd
No Cur. Time / Frame
No R3F. Time / Frame
I'm running debian amd64 on the folding box, winXP on the workstation...
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
I went over to Tom's Hardware, I decided to do some comparisions:ryboto wrote:I wish I could have gone the E21xx route...but I was around 50$ outside of my budget, and when you have no income, it matters. Hopefully the Phenoms will be amazing at performance/watt, considering the B2 phenom at 2.3ghz is only using .976w, that's pretty promising.aristide1 wrote: Also - I am not building anymore AMD folders, the Intel E2140 is a much better bang for the buck CPU. And one can put it on a P35 board and OC it's brains out while everything else runs at stock speeds.
1. At the same clock speeds the Intel 4300 vs. the 2160, the only difference being cache and cash. About $45 (the higher one is actually a 4500 on NewEgg). All the tests I looked at (no games) had them neck and neck.
2. I then compared E2160 ($83) to the X2 4400+ ($85) They were very close in a lot of tests, the Intel winning in the multimedia integer and multimedia floating points test, and the AMD winning the memory integer and memory floating point test. I don't know which if either is more relevent to FAH.
3. The $130 comparision was much more interesting, pitting the E4300 against the AMD 5000+ Black Edition. Here the AMD way the leader most of the time, and the Intel would need some OCing just to catch up. It's not a real comparision however when you consider the E2xxx's.
4. Here comes "take advantage of your strengths time." The Intel E2xxx performs almost as well as the E4xxx's for about $45 less. However consider this, to get mobo that OC's well, a P965 or a P35, you will spend more than AMD T-Force board like the Biostar 7050. And the Intel board will need some video whereas the 7050 has it builtin. For the price penalty on the Intel side you can move up from an X2 4400+ to a 5000+ and gain two things. First is a substantial head start in performance, and the second is an unlocked multiplier. For those that do not OC the AMD would be the winner. For those that live on the edge (yeah right) the Intel would probably win, but at least it would be a much more interesting horserace comparing these 2 setups.
As if AMD doesn't cause enough confusion with their naming and numbering scheme (or lack thereof) the Black Brisbane here has a 13X stock multiplier, no .5 multiplier like the rest of the Brisbane line. Duh!
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
Wibla,
Well it's freezing in the apartment now and it's going to remain that way so I did my first Intel OC last night. I took the FSB from 266 straight to 333 with no increase in any voltages. Now folding at 2.66GHz.
My checkpoints used to be every 22.5 minutes now they are 16.75 minutes. I decided to run the memory synch (it's actually running at 333Mhz as well, DDR2-667). I just happened to find out that the CAS was set to 6 be default, so I changed it to manual and it's now running at CAS 4. I didn't see the command rate anywhere.
The command rate on my AMD X2 was 2T, I changed it to 1T and cut off 30 seconds per checkpoint, which adds up to a 50 minute reduction to complete a WU. Nice.
Well it's freezing in the apartment now and it's going to remain that way so I did my first Intel OC last night. I took the FSB from 266 straight to 333 with no increase in any voltages. Now folding at 2.66GHz.
My checkpoints used to be every 22.5 minutes now they are 16.75 minutes. I decided to run the memory synch (it's actually running at 333Mhz as well, DDR2-667). I just happened to find out that the CAS was set to 6 be default, so I changed it to manual and it's now running at CAS 4. I didn't see the command rate anywhere.
The command rate on my AMD X2 was 2T, I changed it to 1T and cut off 30 seconds per checkpoint, which adds up to a 50 minute reduction to complete a WU. Nice.
Last edited by aristide1 on Thu Nov 08, 2007 10:09 am, edited 1 time in total.
ok, i've only had 2653 WU's so far on my E6600 so that looks pretty close at 15mins
Project : 2653
Core : SMP Gromacs
Frames : 100
Credit : 1760
-- gaming rig --
Min. Time / Frame : 14mn 19s - 1770.24 ppd
Avg. Time / Frame : 21mn 20s - 1188.00 ppd
Cur. Time / Frame : 14mn 40s - 1728.00 ppd
R3F. Time / Frame : 14mn 26s - 1755.94 ppd
Eff. Time / Frame : 14mn 42s - 1724.08 ppd
don't know how far back the average looks, it probably includes 70minute frames whilst gaming
Project : 2653
Core : SMP Gromacs
Frames : 100
Credit : 1760
-- gaming rig --
Min. Time / Frame : 14mn 19s - 1770.24 ppd
Avg. Time / Frame : 21mn 20s - 1188.00 ppd
Cur. Time / Frame : 14mn 40s - 1728.00 ppd
R3F. Time / Frame : 14mn 26s - 1755.94 ppd
Eff. Time / Frame : 14mn 42s - 1724.08 ppd
don't know how far back the average looks, it probably includes 70minute frames whilst gaming
running FSB at 333 (stock is 266)
CPU x9 at 1.4v (reports 1.425) at 3Ghz (stock voltage was 1.325 i think, though CPU-Z shows an unchanging 1.264V in the processor info section)
4gb pc8500 at 1:1 (667mhz) at 5-5-5-15 1T dual channel
had some ram problems at 533Mhz using the EPP profiles with asus striker mobo. OCZ recommend NOT using profiles with this motherboard (mobo gives wrong voltages/what's set in BIOS isn' what is reported etc) so when i have time i'll notch it up using manual settings to see if any sticks are actually damaged (again) have had two replaced already and OCZ said they'll be happy to replace the rest if the mobo has blown them (which is very decent of them) but I'll hang off till I know which sticks, if any, are bad.
CPU x9 at 1.4v (reports 1.425) at 3Ghz (stock voltage was 1.325 i think, though CPU-Z shows an unchanging 1.264V in the processor info section)
4gb pc8500 at 1:1 (667mhz) at 5-5-5-15 1T dual channel
had some ram problems at 533Mhz using the EPP profiles with asus striker mobo. OCZ recommend NOT using profiles with this motherboard (mobo gives wrong voltages/what's set in BIOS isn' what is reported etc) so when i have time i'll notch it up using manual settings to see if any sticks are actually damaged (again) have had two replaced already and OCZ said they'll be happy to replace the rest if the mobo has blown them (which is very decent of them) but I'll hang off till I know which sticks, if any, are bad.
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
Oddly, after I changed the FSB to 333 I went into Windows and CPU-Z shows 2.67GHz, but every few seconds the cpu multiplier switches from 8 to 6 and back again.CPU x9 at 1.4v (reports 1.425) at 3Ghz (stock voltage was 1.325 i think, though CPU-Z shows an unchanging 1.264V in the processor info section)
When I switch cases I want to bump it up to 400MHz FSB for the winter.
Some team stats:
1. We're hovering around 130 active users.
2. Today everybody on the top 20 folders is bright red, I've never seen that before. We even had 3 colors up there.
3. While we stayed at an average of about 518 ppd our current average ppd is 681. Even when vg30et was here our average was around 600 ppd, so for those concerned about filling the gap vg30et left, we've done that and then some.
Last edited by aristide1 on Thu Nov 08, 2007 10:07 am, edited 1 time in total.
2 sticks OCZ sli PC8500 and 2 sticks OCZ reaper pc8500. I know it can do better, but the whole lot did crash not long ago. I'm not gonna push it till i have time to do it slowly and test properly. I've had both lots of sticks runnin at 560mhz 5-5-5-15 2T in the past, but didnt see a great change in frame time. Maybe 3-3-3-9 will make a bigger difference in F@H than clock speed