• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

How to build a 3+ghz, 200+mhz FSB dual Xeon thread...for cheap ;)

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Albigger said:
In regards to the PSU, i am running dual 2.4 LV xeons and when i crank them up too high (3.3-3.4 range) I get random reboots. I also get other weird USB errors and USB devices malfunctioning and I am thinking it might be my PSU. I have a dual 12V rail (18 amps each) sparkle/fortron.


are you sure the random reboots aren't chips that are just not stable at 3.3-3.4ghz? this is a high overclock for the D1 stepping chips, and you may have reached one or both of your chip's limits...

i actually had some D1, 2.4ghz SL6VL chips that could only do 3.1ghz stable, even with the 1.6v pin mod...

try to run 1 instance of prime95 on each individual cpu, and see if one cpu fails faster than the others (indicative of a less overclockeable cpu)...also after finding the "4x Prime95" stable speed, try using a less power hungry video card, less RAM, fewer hard drives, etc...to see if your overclock increases, if not, then it's less likely the psu is having trouble with producing enough power for a higher overclock...
 
dustybyrd said:
are you sure the random reboots aren't chips that are just not stable at 3.3-3.4ghz? this is a high overclock for the D1 stepping chips, and you may have reached one or both of your chip's limits...

i actually had some D1, 2.4ghz SL6VL chips that could only do 3.1ghz stable, even with the 1.6v pin mod...

try to run 1 instance of prime95 on each individual cpu, and see if one cpu fails faster than the others (indicative of a less overclockeable cpu)...also after finding the "4x Prime95" stable speed, try using a less power hungry video card, less RAM, fewer hard drives, etc...to see if your overclock increases, if not, then it's less likely the psu is having trouble with producing enough power for a higher overclock...


True, but these are default 1.3 volt chips, which some have had success at running 3.4ghz plus.

The thing is, as I tried upping the clockspeed slowly, its not like it began failing prime, then I upped some more and it started the rebooting thing. Its as soon as I find any prime errors, at that same clockspeed it starts the rebooting thing. So normally, wouldn't you see stability errors from the CPU like prime95 failing, then hard lock-ups, THEN maybe rebooting (as clockspeed increases)?

I have already run all my peripherals and fans off a separate PSU, with no improvement. However if I am exceeding the 18amps somehow then adding another PSU wouldn't have helped anyway ( I need to try one with a higher amperage rating).

But I will try to clock it down and see if it is overall speed related or just related to the fsb (I know the USB problems only occur at high fsb regardless of clockspeed - which would suggest the cpu's themselves as you said, not sure about the rebooting issue though).
 
Having a weak 12v line will cause reboots.Get a power supply with a high 12v line and life will be a lot easier in the long run..
 
diehrd said:
Having a weak 12v line will cause reboots.Get a power supply with a high 12v line and life will be a lot easier in the long run..


while that may be true, i believe he may already have a psu that is fine...

i think thorough trouble shooting should be done before a new psu is purchased...

i actually opened up the sparkle FSP550-60plg that i have (that claims 36A on one 12v line) and found that there are two separate places from the pcb's in the psu that the 12v cpu power (8 pin connector) and the 12v mobo power leads originate from...

in this case, it's unlikely that the psu is at fault...
 
dustybyrd said:
while that may be true, i believe he may already have a psu that is fine...

i think thorough trouble shooting should be done before a new psu is purchased...

i actually opened up the sparkle FSP550-60plg that i have (that claims 36A on one 12v line) and found that there are two separate places from the pcb's in the psu that the 12v cpu power (8 pin connector) and the 12v mobo power leads originate from...

in this case, it's unlikely that the psu is at fault...


Thanks for all the help guys. I have already ordered the PSU, so I guess I will try it when it gets here, unless I have enough time to track down the problem before then.

I did reboot at a low clockspeed at 133 x 12, and upped the fsb to 200-205 with clockgen and was still getting instabilities. I relaxed my mem timings to 2.5-3-3-7 (they SHOULD do that at 250 fsb) and still failed the OCCT test (and hence probably prime95, although I didn't get a chance to run it).
 
Albigger said:
Thanks for all the help guys. I have already ordered the PSU, so I guess I will try it when it gets here, unless I have enough time to track down the problem before then.

I did reboot at a low clockspeed at 133 x 12, and upped the fsb to 200-205 with clockgen and was still getting instabilities. I relaxed my mem timings to 2.5-3-3-7 (they SHOULD do that at 250 fsb) and still failed the OCCT test (and hence probably prime95, although I didn't get a chance to run it).


it's looking more and more like the memory is the problem to me...

this patriot xbl stuff is very weird in my ncch-dl...i have only gotten 2,2,2,6 to work with the ram at 200+mhz...i've tried cas 2.5, 3 and 2,2,2,5 (spd settings) and none of those work at 200+mhz
 
Well, I have the PQI turbo platinum PC4000 (I don't really think that is TCCD). Anyway its rated for 2.5-3-3-7 at 250.

I have Vdimm set to 2.8. It runs 2-2-2-5 all day long at 190 fsb, I mean priming, folding, crunching matlab programs, CAD modeling etc.... Just seems weird it won't run just 10mhz faster with the looser timings. I'll try to keep adjusting and testing them.

so yours runs 2226 up to what - 210 (as in your sig)?

I remember back in the K6 days it seemed like you could just slap a stick in and go. Now every motherboard seems to be picky about the RAM, and even RAM that is rated at some timings won't do what they should.

The kicker is I already RMA'd one of these PQI sticks (See Here)
 
Albigger said:
Well, I have the PQI turbo platinum PC4000 (I don't really think that is TCCD). Anyway its rated for 2.5-3-3-7 at 250.

I have Vdimm set to 2.8. It runs 2-2-2-5 all day long at 190 fsb, I mean priming, folding, crunching matlab programs, CAD modeling etc.... Just seems weird it won't run just 10mhz faster with the looser timings. I'll try to keep adjusting and testing them.

so yours runs 2226 up to what - 210 (as in your sig)?

I remember back in the K6 days it seemed like you could just slap a stick in and go. Now every motherboard seems to be picky about the RAM, and even RAM that is rated at some timings won't do what they should.

The kicker is I already RMA'd one of these PQI sticks (See Here)

is this your ram? if so then it's tccd...

http://www.newegg.com/app/ViewProductDesc.asp?description=20-141-138&depa=1

my patriot xbl runs dual channel at least up to 215mhz at cas 2,2,2,6 with 2.73v in bios (2.8v setting)...

also, you can check if you RAM is tccd by looking on the chips...should say -TCCD on them...just gently lift off the heatspreaders...you can still put them back on if you want...although, i don't think they're needed...

not all tccd has the same o/cing properties
 
Last edited:
dustybyrd said:
updated directions to use Speedfan 4.21 on the NCCH-DL to monitor TEMPS, Voltages and Fan speeds....

Does it work? I tried it on my MB, with BIOS 1003 and no luck at all.
 
{PMS}fishy said:
Does it work? I tried it on my MB, with BIOS 1003 and no luck at all.


it should work if you follow all of the directions...

you need to install other things first...one of which you can uninstall later...

and then it works great!


what i really like about speed fan are:

1) don't have to be connected to the web

2) uses almost no cpu resources compared with ~10% every couple of seconds with the asus program

3) i can see all the items i'm interested in on one screen, not like the asus program where you have to go to different links and wait to update
 
Albigger said:
Well, I have the PQI turbo platinum PC4000 (I don't really think that is TCCD). Anyway its rated for 2.5-3-3-7 at 250.

I have Vdimm set to 2.8. It runs 2-2-2-5 all day long at 190 fsb, I mean priming, folding, crunching matlab programs, CAD modeling etc.... Just seems weird it won't run just 10mhz faster with the looser timings. I'll try to keep adjusting and testing them.

so yours runs 2226 up to what - 210 (as in your sig)?

I remember back in the K6 days it seemed like you could just slap a stick in and go. Now every motherboard seems to be picky about the RAM, and even RAM that is rated at some timings won't do what they should.

The kicker is I already RMA'd one of these PQI sticks (See Here)



check this post out:

http://www.ocforums.com/showthread.php?t=366571

in it, the ncch-dl user, godofgorks, found that prime95 wasn't stable when using the RAM in the "blue" slots, but was stable when using the Patriot XBL TCCD ram in the "black" slots...

have you guys tried the other ram slots?
 
I just ordered a NCCH-DL from ewiz.com. You guys are a unhealthy influence. :D

I have 2 x 1.6 Xeon LV's to go in it as well. Hoping for 3.0ghz
 
dustybyrd said:
check this post out:

http://www.ocforums.com/showthread.php?t=366571

in it, the ncch-dl user, godofgorks, found that prime95 wasn't stable when using the RAM in the "blue" slots, but was stable when using the Patriot XBL TCCD ram in the "black" slots...

have you guys tried the other ram slots?

No, I haven't tried that. Ill give it a shot, as this setup has a heck of alot more headroom in it. I just wasn't able to break 200fsb. I know the RAM can do 240 easy.
 
Albigger said:
Tried it but no improvement.

Albigger,
Is this with or without your new PSU? If you still have the dual rail PSU in, that still may be the cause of your instabilities...

I'm having similar problems to you (see my post in NCCH-DL thread on 2CPU.com), but with 2.4 M0's and the -PLG Fortron supply and Patriot XBL 3200 TCCD memory. I was getting Prime errors, random reboots, BSOD's, and lockups at 3.2-3.4Ghz, so I'm starting low (12x200) and working up from there. I run my mem at 2-2-2-6 2.8v now, and it Prime'd all night at 12x200 (which really isn't an OC for a 2.4Ghz M0 chip). Now its Prime'ing at 2.8Ghz. Next up is 3.0Ghz, but before that I plan to switch the RAM to the black sockets to see if that helps.

I still suspect my PSU, even though it is the Fortron 550W -PLG that dusty raves about. I have occasionally seen the 12v dip to 11.97, and if it dipped lower than that I wonder if that caused the reboot (unless you had a multimeter hooked up, you'd never know because it would have rebooted/crashed/locked before the sensor software would report it!).

-shoek
 
shoek said:
Albigger,
Is this with or without your new PSU? If you still have the dual rail PSU in, that still may be the cause of your instabilities...

I'm having similar problems to you (see my post in NCCH-DL thread on 2CPU.com), but with 2.4 M0's and the -PLG Fortron supply and Patriot XBL 3200 TCCD memory. I was getting Prime errors, random reboots, BSOD's, and lockups at 3.2-3.4Ghz, so I'm starting low (12x200) and working up from there. I run my mem at 2-2-2-6 2.8v now, and it Prime'd all night at 12x200 (which really isn't an OC for a 2.4Ghz M0 chip). Now its Prime'ing at 2.8Ghz. Next up is 3.0Ghz, but before that I plan to switch the RAM to the black sockets to see if that helps.

I still suspect my PSU, even though it is the Fortron 550W -PLG that dusty raves about. I have occasionally seen the 12v dip to 11.97, and if it dipped lower than that I wonder if that caused the reboot (unless you had a multimeter hooked up, you'd never know because it would have rebooted/crashed/locked before the sensor software would report it!).

-shoek

I've been following the thread at 2cpu as well. My testing is still with the -PLN because the -PLG I ordered is scheduled to arrive thursday. It doesn't reassure me to know that you are having problems even with the single rail PSU.

I do have a multimeter, it is probably worth it to check out the rails I suppose. Maybe later today...
 
Back