New hope for C2D Overclocking?

sleepeeg3

Supreme [H]ardness
Joined
Mar 4, 2004
Messages
5,406
EDIT: Setting your clocks this high on this board is apparently STUPID and will probably start killing components! DON'T DO IT! :D

This is the P5L-MX I just got.



Specs:
*P5L-MX (G945)
VCORE: None
VDIMM: 2.00v
FSB: 450

It can boot into Windows as high as 425 and I am sure the board is capable of much higher, but my CPU starts requiring more voltage after 415fsb and ram 1:1 after 410fsb.

This is with VDIMM @ 2.0, timings manually configured to stock and everything else on AUTO. I use a PCI-E x1 SATA controller for my drives and connect to the net via a wireless USB stick. The PCI bus would be running at 53 MHz, so there has to be a lock on this board.

So it is not bad, but what is really promising is another similar board that has a PCI/PCI-E lock:

P5LD2-VM R2.0 - Intel 945G chipset

1. Vcore to 1.7
2. FSB to 500mhz
3. vdimm to 1.9
4. pci-e freq lock
5. pci freq lock
6. cpu multi adjustable
7. cpu/fsb freq support to 1066
8. ddr2 400/533/667
9. vMCH to 1.7!!
10. CPU to FSB Ratio adjustment (Strap setting!? bios does not elaborate)

I am not sure how Lestat knew there was a lock, but Dgephri and my results confirm:
http://www.hardforum.com/showthread.php?t=1124087
Now if he would do some testing on his board like I was asking, we might be in business...

What is really exciting, is I think the only thing holding his overclock back is his RAM. Based on this, he is hitting a wall about where he should be, especially with 4 sticks of DDR2:
http://www.xtremesystems.org/forums/showthread.php?t=110193&highlight=P5LD2*

/whine "But with only 1.9v, won't I have to run a divider?" RAM bandwidth has at most 2-5% impact on performance. The amount of extra overclocking headroom it will gain you blows this out of the water. Plus at lower speeds, you will be able to run tighter timings to negate the difference. Just make sure you have at least PC6400 or can run your PC5300 higher with looser timings. Besides, if you can run a 475fsb, you will already be running over DDR750.

I don't know about you, but I know I sure underestimated the 945's potential!
 
Very good result but you have to used a pciex1 controller to use your SATA drives and no PCIE graphic card mounted on the MB which reduce strongly its interest...
 
I used an external PCI-E x1 card, because onboard would have capped my max transfer rate and it makes transferring RAID from one board to the next a whole lot easier.

If there was no lock, my CDROM would not be working, my USB would probably be unrecognized and my X-Fi would be going nuts!

Can you try to push your board higher? Try disabling Hyperpath 3 in the Chipset section of the bios, set PCI-E to AUTO. run one stick of RAM, run a divider and loosen up the timings. Try jumping to 325 or 350. If Dgephri can get it up to 353, something must be holding back your OC (at least to that point).
 
I could but I don't want to fry my rig...
Can you detailled yours? Thxs!
 
I can guarantee unless you set the VCORE above 1.55v, you will not kill any hardware in your rig.

Well I can try, but it would probably be easier for me to buy the board and play with it! :D I will give it a shot:

-Use 1 stick of RAM.
-Under "Chipsets" find the Hyperpath 3 setting and disable it

Set PCI-E Frequency * PCI Clock back to "AUTO" / Memory voltage to 1.904v / MCH to Auto / VCORE to 1.45v
http://img.photobucket.com/albums/v160/Dgephri/P5LD2-VM/BIOS2.jpg

Set this to the top/lowest setting:
http://img.photobucket.com/albums/v160/Dgephri/P5LD2-VM/BIOS3.jpg

Set DRAM timings to 5 - 5 - 5 - 18 - 5
http://img.photobucket.com/albums/v160/Dgephri/P5LD2-VM/BIOS10.jpg

Then raise the fsb to something higher. I jumped from 300, 325, 350, 360, 375, 400, 425, 450. The only one that did not work was the last one, and that was because I ran out of VCORE! Your board won't have that problem...

Highly likely your RAM is holding you back.
 
Yes, my video card is mounted/installed. I was playing the BF2142 demo on it last night with the settings in my first post. :)
 
your x1900xt?
can you post your cpuz details (motherboard, memory...) Thxs
 
Yes, my X1900XT.

cpu3280dh1.jpg
mainboard3280vd6.jpg
memory3280lg4.jpg
 
You know, now that I think back on it, I posted a thread that never really got answered a while back. It was about the 8800s and the 680i chipset and how the 680i could boost the PCIe clockspeed 25% and increase performance on certain cards. Currently, its a feature that only the 8800s supported.

I would assume that 8800s wouldn't fry until 125 mhz.
 
Wow. Sorry in the other thread, I made it sound like I was doubting the 945g's abilities. I did, but only because I didn't read your posts correctly, and thought you were basing it off speculation instead of actual success!! From what you just posted, I think I'm going to blow another $70 and get a decent psu, the enhance 500w.

I know back in the day, when I ran 112fsb on an old via board, and effectively had 37/75 pci/agp, my onboard agp died in a few months. They could be unrelated though, since it was a piece of shit pcchips board, and was just a coincidence.

I'm guessing the ram dividers have to be in some form flawed, since it won't let me stable at 711ddr at 266fsb, but I can boot into windows at 310fsb and 8xx ddr for a longer period before it crashes.

It's a good thing with the p5l-mx that they put the pci-e in front of the pcie16 so the fan won't block it up. However, you mentioned youre using external sata for your raid. Are you saying that the sata works fine beyond 360ish still? I'm going to try 1:1 cpu:mem and see if I can bring it up just for kicks. The lan dying out is still a bummer. I guess I'll go up to best buy and see if they have any <$10 cards like they used to. It'd sit right up against my gpu fan though, but so it goes.

Guess I'll do some upping and see where it likes to post at. Promising stuff for sure man. Kudos to you!!
 
Update: I went in for some raw fsb action. Set mch up to 1.65 for safe measures. Dropped multiplier to 6x also. Kept it at 1:1. Got it to post up to 433fsb. SATA drive long gone though. Tried 450, but no post. Tried lowering ram below 1:1 but no post either. Regardless, if 433 was a usable setting, it'd be well good enough :) 3.46ghz goodness with 8x

I wonder how usable the ide opticals are even up there. It's really a shame. I thought pci locking was figured out by springdale/canterwood. Guess the mix of pcie threw things off again.

So if the higher end 945g asus does turn out a winner, then hell yeah I'm jumping ship. Just might need to figure out a way to voltmod dimms in that event.
 
ziddey, are you sure you lost your SATA drives? What happens? Could it be your chip holding you back? Did it just crash when it tried to load Windows?

If you were bottlenecking @ 300 x 8 = 2.4GHz and you bump it to 433 x 6 = 2.6GHz, that could either be A. more than you chip can handle on stock voltage or B. a crappy Ultra power supply. :)

Lower the MCH, maybe you will get higher. I am at stock so I can definitely say you do not need 1.65v and the extra power could also be limiting your OC, if it's a PSU issue.

Look 37-40fsb is the MAX I have ever seen the PCI go out of spec without the board going unstable. 53fsb... not possible.
 
sleepeeg3 said:
ziddey, are you sure you lost your SATA drives? What happens? Could it be your chip holding you back? Did it just crash when it tried to load Windows?

If you were bottlenecking @ 300 x 8 = 2.4GHz and you bump it to 433 x 6 = 2.6GHz, that could either be A. more than you chip can handle on stock voltage or B. a crappy Ultra power supply. :)

Lower the MCH, maybe you will get higher. I am at stock so I can definitely say you do not need 1.65v and the extra power could also be limiting your OC, if it's a PSU issue.

Look 37-40fsb is the MAX I have ever seen the PCI go out of spec without the board going unstable. 53fsb... not possible.
Hey, glad you're still talking to me :)
At 300fsb, post fails to detect my hdd. At 310ish, they detect again though. Same with 400 when I tried. POST detected my dvdrw but failed to find any sata drives. So when it tried to boot, it checked my cd, and then gave the no os error hit any key to try again. I tried stock voltage too for mch (surprised it posted at all in such high fsb!!) but to no avail. I'm going to hit the trigger on the highly acclaimed enhance psu tomorrow hopefully, but that shouldn't help the fsb o/c
 
I think this thread should be stickied, finally someone has been able to take a Conroe mATX over 400 and stable.

Keep us posted.
 
ziddey, what bios are you on? Stock #202 or something? Does your board say <GREEN> over the parallel port? I think that just means Conroe compatable, but I want to rule out all possibilities. Which acclaimed Enhance PSU? :)

Nice job, pvhk! Did you really bump the VCORE, because it looks like it is not working on that board... Will it post with stock PCI-E on auto?
 
yes it works cause when i set vcore to default, windows reports 1.28v
when I set vcore to 1.425 bios, it reports 1.34-1.38v in windows ( probably voltage drop as seen with asus motherboards in general)
I did not set PCIE to auto, i prefer to set it manually (so that i know what i'm doing)
 
I'm running the latest that was posted in november. I think it's 301. I noticed you were running an older one in the screenshots.

edit: http://www.ewiz.com/detail.php?name=PS-E5150GH
I hear that's a great one for the price and has the same internals as one of the seasonic models. I'd rather not spend that much though, but was told there's nothing "good" under it
 
pvhk, Dgephri is reporting the same thing. I just wanted to confirm. He said he gets no post if he sets the PCI-E on AUTO so I guess I am wrong about the lock! I asked him to try one more thing, just to confirm. I may still buy the board, just to fool around with it myself. ;) Dgephri sounded like he was very thorough though. Try setting the PCI-E to 100 and see if it will boot. Wondering if the bios version has anything to do with it, because mine sure seems to have a lock...

ziddey, ya I saw David Hammock highly recommending it and he knows his stuff. It should be good. ;) What are the 12v amps though, I don't see them posted?
 
sleepeeg3 said:
pvhk, Dgephri is reporting the same thing. I just wanted to confirm. He said he gets no post if he sets the PCI-E on AUTO so I guess I am wrong about the lock! I asked him to try one more thing, just to confirm. I may still buy the board, just to fool around with it myself. ;) Dgephri sounded like he was very thorough though. Try setting the PCI-E to 100 and see if it will boot. Wondering if the bios version has anything to do with it, because mine sure seems to have a lock...

ziddey, ya I saw David Hammock highly recommending it and he knows his stuff. It should be good. ;) What are the 12v amps though, I don't see them posted?
I'm not sure, I'll have to look them up at some point. Meanwhile, I had a buddy at XXXX loan me one of the tech center-- antec 650w psu under the table for tonight to do some testing to rule mine out. let's see how it goes
 
Well an interesting thing I tried... I set the PCI-E to 100 and guess what? It didn't post! :D I also tried 105 and 120... still nothing @ 410fsb. So do you think I had the PCI-E running as high as 160MHz? :eek: Right now I have the fsb @ 350 and PCI-E locked at 120.

Good luck with the OC, I bet it will help. ;)
 
sleepeeg3 said:
Well an interesting thing I tried... I set the PCI-E to 100 and guess what? It didn't post! :D I also tried 105 and 120... still nothing @ 410fsb. So do you think I had the PCI-E running as high as 160MHz? :eek: Right now I have the fsb @ 350 and PCI-E locked at 120.

Good luck with the OC, I bet it will help. ;)
Hehe, I tried 100pcie a bunch of times and it never worked on the p5l-mx. I'm guessing auto doesn't float it all the way up. It must still be doing something in terms of keeping it real. I'm assuming auto is more or less like manually defining 116pcie or something like that, although without an oscilliscope to find the right spots to measure, it's hard to say. I gotta say though, this antec neo he 550 is doing pretty well. I remember reading that big power supplies, especially modular ones, and long dvd drives like the nec 3500 are a nono in the microfly, but mine's fitting fine right now.

Voltages came back way down. 3.3 was at around 3.5ish before iirc is now 3.28. +5 was 5.2 maybe is now 5.04. +12 which was 12.5 is now 12.20. VCore seems similar to before, although I can't account for droop. Time to orthos and 3dmark and see what happens

update: Too bad no way to monitor voltages while 3dmark is running. I only get that glimpse between benchmarks. But it's been half an hour running orthos and 3d06 (overclocked 7900gs @ 580/800) and it's chugging along just fine at the moment. Guess the ultra 500w was the culprit.
 
sorry I had missed this thread before.

Glad to see some more people giving this board a chance.

On the PCIe: the only real symptom I have been able to detect, is what appears to be a limited OC on my 8800GTX, or I have a dunce card. :)

Much over 675/995 (11,156 3dMark06) and it would fail the benchmark.

That is on DD waterblock, and stable temps of 44C on the GPU (doesn't move at all during load, according to NVtune eventlogger).
 
3.5ish?? :eek: Your RAM was probably freaking out. I was going to recommend the NeoHE, but I didn't remember it being so expensive. Upgraded my parent's system awhile back with one. Must be doing fine, because I am still getting e-mails from them... nuts. ;)

I hear that about the oscilliscope, but I just can't imagine that my sound card still works with the PCI bus nearly twice as high as normal. AUTO must be doing something.

ziddey and I are actually on the little brother P5L-MX, but they seem to share the same symptoms. Main difference is the lack of VCORE for us. I was looking at jumping ship. :) Ah the 8800GTX just came out, so they won't have much headroom, not to mention all the wattage they throw into the loop. Oh and it is a SFF. :cool: Those look like nice clocks though, you can probably run BF2 in 16xAA with it? Will be interesting to see what AMD springs in January. Maybe I will upgrade when Crysis hits...
 
Damn... Looks like 3dmark still crashes the system, even with the antec neo he. Must be the video card then?? It ran for about an hour and a half at 580/800 (never a single artifact). Then I ran it at stock evga ko's 500/690 and it crashed after 2 minutes. Tried again and it lasted about a half hour. It's going again but I'm willing to bet it'll crash again no problem. :(

As for the p5l-mx not supporting vcore, I'll have to check the datasheet for it on intel's website. There should still be a way to overvolt, even with the speedstep bullshit.

Oh yeah and by the way, sleepeeg, do ram dividers other than 1:1 work right for you?


update: AGH, just no stability. Looks like 3dm2k3 can crash my computer quicker. I'm guessing it's due to the fact that the intermediate screen between testing isn't back to the desktop. But earlier, I was running stock 7900gs (450/660 as opposed to evga ko overclocked 500/690 from the factory) and it died a few times. This is all with the new he550 powersupply, so that's ruled out. I ram memtest through the night two nights ago and 0 errors at 333 5-5-5-15-5 2.0v (and damn right it shouldnt since it's rated 400 5-5-5-15-5 2.1v). Ran orthos last night and it didn't crash at all. Looks like that one time it did crash on me was because I had recently ran 3dmark or something. Ohh man. Guess I'm going to have to rma either to evga or newegg. Oooooof, at least I'd be dealing with the best, eh? The strangest thing about it all is that there are absolutely zero artifacts, whether I run at 450/660, 500/690, or 580/800. Load temps rise to around 70'c for the gpu, which is well inline from what I'm told to expect on stock cooling. I've tried two windows installs now
 
I am going to try going back to stock setting on all CPU OC, then seeing if the GPU will OC further.

Then see how my BM scores change as a result.

Maybe a middle ground gets me over 12k on BM.
 
It seems like the p5l-mx is following in the footsteps of it's older brother the P5LD2-vm (ver 1). The PCIe lock on the board was basically nonexistant. Overclocks crapped out at like 23MHz over stock FSB unless you manually set the PCIe to like 116 and the PCI to 36.5 (or whatever the setting was for that).
 
Alright, so I finally decided to see if my ram is functional, since I'd like to rma now instead of wait months to find out if it will or will not work at rated speeds. So I have fsb at 401 at the moment, 1:1, 2.0v. Obviously, no onboard sata, but it managed to pick up my cdrom so it's running memtest at 800ddr2 spd (I assume it's 5-5-5-15-5 since that's what the last programmed spd is at 333) and 2.0v instead of 2.1v. It's been going for about 25 minutes and is 85% of the way done with one pass, and no errors so far. So it looks like some of the dividers might be faulty and that's all. I'll try other dividers at different fsb's later on to make sure of the fact (although it seems well proven since 266fsb 711ddr (3:4) bsod's into windows everytime).

I wonder about the additional sata controller on the higher end asus 945g? Is it jmicron? I hear those don't respond well to high fsb's either? If it does work, then 400+ using that controller should be the solution for now

Update: hmm 401fsb. I forgot to drop the multiplier, so it's running 3.2ghz. It ran 4 passes and registered 1 error at one point. Bah. Reset it to 400fsb, 6x multiplier, and manually defined ram timings to 5-5-5-15-5 @ 2v.

update: dropped multi to 6x and fsb to 400 even instead of 401 (not that it should make a difference). It's almost done with pass3 and still no errors now :). Guess that means something is up with my e6400 not wanting stable 8x400fsbat stock volts which reads 1.35ish, at least on this motherboard. Looks more promising for the ram though, since it had one failure in test2 when at 8x. Can't ever be sure though, so overnighting and hope if any errors do arise, it's due to ram and not motherboard.
 
Lol!! Literally 10 seconds after my last update, I got an error in memtest. This time it's at a far different end of the memory range though (236mb vs 2038mb) so I'm guessing it's on the other module. Kind of makes me hesitant to believe both sticks are bad, so I guess I'm thinking it's still the motherboard??? This is with 1:1 too, so I don't know what's up. Maybe they really need 2.1v and 2.0v doesn't cut it? Sleepeeg, you doing a-ok with your ram eh? That's good to hear. Good ol' corsair I suppose :)

And now before I submit this post, it errored out again at 418mb.

I gotta say though, my sticks are REALLY burning up. The heatspreader is real hot to the touch, and with my temp sensor put in between the two sticks, it's reading 50'c even. With there was a better way to providing circulation there more than just from the 80mm from the front blowing there, the exhaust from the cpu hsf going there, and the general 120mm up top drafting air, as well as the psu. Makes me almost wonder what actual voltage is being supplied. I managed to just slip the sensor into the heatspreader a bit and it's showing 58'c now. Hmmm
 
sleepeeg3 said:
I used an external PCI-E x1 card, because onboard would have capped my max transfer rate and it makes transferring RAID from one board to the next a whole lot easier.

sleepeeg3, can you please advise on wether or not the P5L-MX has on-board SATA RAID? The website is unclear, but I noticed instuctions for making a RAID driver disk in the pdf manual. I am looking to buy about 12 of these for an office setup. I want to have RAID mirroring on each of these. I just need to know if the mobo supports it on-board or if I would have to get a seperate RAID card.

Thank You for your help!

the_buzz_man
 
the_buzz_man said:
sleepeeg3, can you please advise on wether or not the P5L-MX has on-board SATA RAID? The website is unclear, but I noticed instuctions for making a RAID driver disk in the pdf manual. I am looking to buy about 12 of these for an office setup. I want to have RAID mirroring on each of these. I just need to know if the mobo supports it on-board or if I would have to get a seperate RAID card.

Thank You for your help!

the_buzz_man
It's the standard ich7 controller for sata, so whatever raid that allows. I'm currently finishing up pass 26 of memtest @ 320fsb 4:5 400ddr2 no errors :) and will see if i can find out more for you.

last time i setup raid was on ich7r so things were a bit different
 
Thanks for your input. I look forward to your answer. I mostly need to know if it supports RAID 1 for mirroring.

Once again, thank you.

the_buzz_man
 
the_buzz_man said:
Thanks for your input. I look forward to your answer. I mostly need to know if it supports RAID 1 for mirroring.

Once again, thank you.

the_buzz_man
I'm not really sure how it works. There is no raid bios to select from the bios. Although I think there are options available if you hit f6 during setup and install the intel ahci raid drivers and use the intel program. I've never done that before. Maybe you could try posting in general hardware and/or the disk storage section about ich7 and what type of raid it can do and how.
 
Also, I couldn't help but wonder, why are you looking at the p5l-mx? It sounds like you're looking to setup office computers or something of that nature. I tried searching for ich type r controllers in a matx form that supports c2d and pcie / ddr2 and this was the only board i came across: http://www.newegg.com/Product/Product.asp?Item=N82E16813186103
foxconn should be great for desktop use

edit: hmm doesnt list 1066 though. not sure


update: so it seems like the ram is fine. and running stock, if i dont run orthos in the background, 3dmark seems to be able to loop for a few hours just fine. guess it might be coming down to motherboard?!?! overheating something, even at stock??
 
ziddey said:
Also, I couldn't help but wonder, why are you looking at the p5l-mx? It sounds like you're looking to setup office computers or something of that nature. I tried searching for ich type r controllers in a matx form that supports c2d and pcie / ddr2 and this was the only board i came across: http://www.newegg.com/Product/Product.asp?Item=N82E16813186103
foxconn should be great for desktop use

edit: hmm doesnt list 1066 though. not sureQUOTE]


1066 was one reason, as wel as gigabit LAN. I think I may just get the P5L-MX and also get a decent PCI SATA RAID card. I appreciate your help.

the_buzz_man
 
has anyone been able to get an E6400 over 353FSB on either of these Asus boards yet?

I am considering testing a big jump, but a little wary that something like 390-420 could cause something else to bug out.

I use th onboard SATA, X-Fi Music, and run all SATA drives (including optical), so taking the PCIe higher, or bumping the PCI up to 35 makes me a tad nervous.
 
Back
Top