Best folding (p35?) mobo for all these new C2Q's?

[MVC]

Thanks for the testing and letting us know what works! I've got two P4s doing GPU folding that I'm thinking it's time to replace and knowing what parts play nice together before I spend any money is a very big help.

FWIW, you've already saved me a headache--I'm tending away from Asus boards since I've had nothing but problems with Ubuntu and the Marvell NICs, but knowing that the DS3R doesn't like Ballistix RAM (which I was thinking of using) will allow me to avoid that problem.

Please keep the updates coming!

 

[pscout]

bump ... preliminary update ... ds3r beats p5k vanilla for a basic folding only rig for quads. BIOS updates to both mobos could change this in the future i suppose but I am not sure how bios changes will fix the power consumption difference

I updated the OP removing the earlier warning.

Will update more detailed results later today I hope.

 

[cw823]

bump ... preliminary update ... ds3r beats p5k vanilla for a basic folding only rig for quads. BIOS updates to both mobos could change this in the future i suppose but I am not sure how bios changes will fix the power consumption difference

I updated the OP removing the earlier warning.

Will update more detailed results later today I hope.

Running my x3210 on a DS3R as I had yet another total mobo failure in a P5B-Deluxe. Found that TPF was lower at 400x8 (mem1000) than it was at 457*7 (mem1142)

 

[LandShark]

bump ... preliminary update ... ds3r beats p5k vanilla for a basic folding only rig for quads. BIOS updates to both mobos could change this in the future i suppose but I am not sure how bios changes will fix the power consumption difference

I updated the OP removing the earlier warning.

Will update more detailed results later today I hope.

told you the DS3R is a pretty good board following the DS3's trail!!

 

[pscout]

no update yet ... have been doing some RL stuff and researching ( pencil ) mods for the p35-ds3r and g33-ds2r (the giga equivalent of p5k-vm). p5k could also use a vdroop mod. If pencil mods work ok then i might see a better oc than 370x9 on the G0. And of course, still some 8 multi testing ..... but it might have to wait till i get some boat work done.

 

[pscout]

I had great plans to even include some charts with power consumption vs oc vs volts for the 2 mobos mentioned so far but i have not had the time/focus to do it yet

.... however i had tried it with a g33 mobo ... the ga g33m-ds2r and it oc's almost as well as the ds3R and should be similar to the p5k-vm. Biggest issue with it so far is it can only feed 2.2 v to memory which may limit the memory oc depending on your sticks. Otherwize it behaves much like the ds3r and is about the same price. All tests so far only with integrated video enabled.

One more mobo i am about to oc is the p35-ds3L which is a fair bit cheaper than the p35-ds3r at least partly due to using the ich9 instead of ich9r. I might have some results on it by tomorrow if i get enuf time.

 

[ffzen]

What are you using for cooling on the ds3L; TRU120E ?

 

[pscout]

What are you using for cooling on the ds3L; TRU120E ?

I have been using TT sonic towers with 2x120 fans and TT BT's with upgraded fans. They are pretty similar to each other performance wise. Also similar to a tuniq 120 i have used on a quad.

All tests on the same bench caseless - ambients in the room are high atm with 3 quads and a/c not running today

 

[pscout]

small update ... the p35-ds3l is looking like the mobo of choice for me atm.

I haven't finished with it yet but seems to oc as well as the ds3r and costs ~ 50% less. I have tested it on another G0 q6600 but this one has a much higher vid than the others so does not appear to be as 'nice' a g0 as the other 2. vid is 1.30v vs 1.2375 and 1.25 on the first 2. Quick tests were done on an arctic freezer 7 with a stock fan so a better hsf may yield higher oc, but i don't think it will reduce the vc the current oc takes to the same level as the others.

I will try to swap some of the cpu's already tested onto it to see if results are consistent.

 

[Sorin]

small update ... the p35-ds3l is looking like the mobo of choice for me atm.

I haven't finished with it yet but seems to oc as well as the ds3r and costs ~ 50% less. I have tested it on another G0 q6600 but this one has a much higher vid than the others so does not appear to be as 'nice' a g0 as the other 2. vid is 1.30v vs 1.2375 and 1.25 on the first 2. Quick tests were done on an arctic freezer 7 with a stock fan so a better hsf may yield higher oc, but i don't think it will reduce the vc the current oc takes to the same level as the others.

I will try to swap some of the cpu's already tested onto it to see if results are consistent.

Lucky SOB. The VID on mine is 1.3125.

 

[pscout]

Lucky SOB. The VID on mine is 1.3125.

... i wonder if that means the next one i get will have an even higher vid?

When i ordered the first 2 from ncix during a special (for $40 less than the most recent one) if I had know they were gonna be G0's i would have ordered more.

with my normal luck, they will probably be the last good ones i see

 

[ihrsetrdr]

I'm eyeballin' the p35-ds3l as a suitable home for my 5th quad; I'm a little leery of the p5k because I view it as the successor to the p5b-d, which I had nothing but heartburn with.

 

[drshivas]

So far so good with the Abit IP35-E and my Q6600 G0. Runs 9x333 and 8x400 on stock cooling, and moderate vcore bump. Getting 8.5min/frame on 2653 WINSMP. Temps low 60s, with ambient in mid 70F in the basement. I jumped on the rebate deal at newegg. Mwave has a similar deal. I think the IP35-E is a budget contender with the rebate. And it has PS2 ports!

CPU specs: G0, 1.2625 vid, L721A, pack date 7/31/07, ClubIT

 

[pscout]

Here is a link i just ran across which may help with reaserching best quad mobo's and has a few links to other reviews.

With such a plethora of versions of p35 mobos the variety to chose from is very large. Bear in mind looking at an XS thread, you are more likely to be reading opinions and results related to uber oc'ing borads, which are often overkill if you are looking for a good value oc'able mobo for a dedicated folding rig. If it will be your main rig, then the higher end models may be what you need/want anyway.

http://www.xtremesystems.org/forums/showthread.php?p=2347910#post2347910

I haven't digested it all but there does seem to be some common anectodal info that many of the asus models are more power hungry the the gigabytes or abits which is consistent with my test of a sample of 1 asus mobo (p5k). Again, this may not matter much if you only run 1 of them, but farmers with multiples will care.

 

[ChasR]

THough posted elsewhere, this is the proper thread for any discussion so I'm reposting here.

Pete,
I have to respectfully disagree with you on which board is the best for the following reasons:


1st, the max stable OC on the Ds3R appears to be lower, but I'm not using the same cpu to test. I took a E6600 out of a P5B that was folding stable @ 378 x 9, 3.402 GHz and can't get it to more than 3.3 GHz on the DS3R.

2nd, I don't like the voltage options in bios on the DS3R. I want to KNOW what voltage I'm setting not how much more voltage I'm applying.

3rd, it's ridiculous to have to hit crtl-F1 to have the ram timings appear on the DS3R and I'd prefer to set ram to a frequency rather than by a multiplier.

4th, ram bandwidth on DDR2 memory is greatly reduced on the DS3R. Memtest86+ v1.70 reports 3270 MB/sec (edit: retesting proved this to be 4675 MB/sec) on the DS3R @ 413 x 8, 1:1, 4-4-4-12 while a mere P5B attains 4485 MB/sec at 370 x 9, 1:1, 4-4-4-12 and 5004 MB/sec at 370 x 9, 4:5, 5-5-5-15. I'll play with the DS3R some over the weekend and see if strapping issues affect bandwidth on the DDR2 memory (apparently non native support) by attempting testing @ 370 x 9 with the same multipliers. You certainly don't want to saddle a quad with 33% less bandwidth than you'd get on a vanilla P5B. On the P5K with a quad, ram bandwith is reported to be 5220 MB/sec @ 475 x 7, 1:1, 5-5-5-15. I'm not yet prepared to spend 2 quads worth of $ on DDR3 memory to see what the DS3R can do with it.

 

[NedClocker]

Dam*! Now I don't know what to buy.

 

[pscout]

Sounds like I should do more apples to apples testing between the cpu's and mobo's here.

There is no question that I have had a bias focusing on low vc and power consumption measures lately ... due to rig density in my little rack room. I have not being going for the max out of every rig lately like I used to. I have also been trying to get rigs into production and not spending that much time trying to understand the myriad of new bios settings available with the p35's.

I haven't used memtest to measure bandwidth ... I have mostly used sandra for a bandwidth measure, and superpi for latency, but only as indicators. My standard bench for memory tuning has been a test 2610 rather than a synthetic bench tool. Maybe i should be running 2 2610 instances for timings.

More testing won't happen this weekend tho, but maybe some time next week.

And ChasR, or anyone else ... you don't need to apologize for not agreeing with my results/conclusions. I have been corrected many times and am glad for it
Thats why i started the thread ... so we could all share our learnings

I agree with your comments on the gigabyte bios ... it is a crappy interface.

 

[ChasR]

I did some testing on the DS3R and found my memory in error. The DS3R @ 413 x 8, 1:1, DDR2, 4-4-4-12 had a ram bandwidth o4 4675 MB/sec. At 370 x 9, 4:5, DDR2, 5-5-5-15 it had a bandwidth of 4819 MB/sec, almost 200 MB/sec less than a P5B using identical ram at the same speed.

 

[NedClocker]

I did some testing on the DS3R and found my memory in error.

Which memory are you talking about? Do you mean your ram or your brain?

The DS3R @ 413 x 8, 1:1, DDR2, 4-4-4-12 had a ram bandwidth o4 4675 MB/sec. At 370 x 9, 4:5, DDR2, 5-5-5-15 it had a bandwidth of 4819 MB/sec, almost 200 MB/sec less than a P5B using identical ram at the same speed.

So, what does that mean? I should buy the DS3R?

 

[ffzen]

I think he's saying ram bandwidth is slower/worse on the DS3R.
But i believe Pete and some review sites mentioned recent Asus boards
use more power than Gigabyte boards. So i guess it's pick your poison.