OC - Xeon SL74T 2.4ghz advice

Nucleo

Nucleo

Nucleo

Associate
Joined
26 Jul 2004
Posts
389
hi all

im a total noob at overclocking but i know the basics, just not had the experience.

I have a server type machine:
Asus NCCH-DL
Intel Xeon SL74T 2.4ghz (x2)
2 Thermaltake HSF's
1GB DDR400 Geil Value Ram

Xeon Spec
http://www.cpu-world.com/sspec/SL/SL74T.html

I want to OC the xeons to see what I can get out of them without killing them ;)

They are currently running 133x18 to get the 2.4ghz. they did boot at 166x18 when i first put them in and it worked okay lol, but i soon lowered so i dont kill them.

what do you think is a reasonable overclock for them?

I have a pentium 4 3ghz (HT) machine and was considering replacing it with these xeons.

thanks all, any advice would be cool

cheers
 
Dont worry about killing them quick unless you have increased the voltage by a noticable amount mate. A processors lifetime is something like 8-9 years unclocked. So even if you shorten it by a few years i dont think youll notice :p.
 
i was about to start overclocking, upped the fsb by 10 and got a BSOD :confused: ... rebooted and a config file in windows was damaged so im gonna have to reformat...

stupid machine lol.

it needed reformatting anyway :(
 
I've a pair of 1.6 GHz LV D1 stepping Xeons and an Asus PC-DL Deluxe that are running at 16x205 (3.2 GHz). I find that in order to get decent overclocks you need to raise the CPU core voltage. You might not have to do that becasue yours are high voltage right off the bat. Try lowering the multi as low as it'll go and raising the FSB to 166. That mobo should be pretty good for overclocking.

If that doesn't work, try raising the vDIMM to about 2.8v. The wire mod necessary to raise the vCore is very very easy so that should be your next step. Write back and tell us how it goes.
 
Stock Intel Wind tunnel copper heatsinks and stock 60mm x 38mm fans. They run at about 42 degrees at full load. Full load in this case is 4 threads of Folding@Home 24 hours per day.

looks like so:
xeon15vu.jpg
 
the only voltage changed i can make are to the chipset vcore and dram vcore.

seems to be a bit dodgey booting up, getting an overclock boot fail error at post,

'f1 to continue or del to enter setup'
 
core voltage on both cpu's is at about 1.27~1.29

at the moment running @ 166x18 for 2.98ghz.

gonna run some superPI to make sure it is stable and check temps
 
gah, this is doing my nut in.

I set it to 166x18 get into windows, check cpuz and it has dropped the clock back to 133x18, resulting in 2.39ghz.
 
right

did the U-wire mod, took longer than 20 mins because im butter fingers ;)

only did the 2 wires to see how it is, boots @ 3.0ghz with ease now.

I know why it wouldn't boot before, but i wont say because the reason is so nooby lol
(the jumper was set to 166, had to change it to 166 it seems i can change the fsb by using the jumpers, but the editing in the bios does nothing?... i can change the multiplier in the bios, but the alterations to fsb do nothing)

ran superPI, 2 instances for both cpu's and all is well, new volts rewgistered and temps @ 40c

what kind of temps are too high?
what kind of speeds should i expect?
 
Huzzah! I thought I'd scared you away for a bit there. :p
Nucleo said:
(the jumper was set to 166, had to change it to 166 it seems i can change the fsb by using the jumpers, but the editing in the bios does nothing?... i can change the multiplier in the bios, but the alterations to fsb do nothing.)
Click to expand...
It should be that the jumpers let you set 100, 133, 166, and 200 MHz front side busses. Between these you (at least on a PC-DL) are able to specify an exact FSB. So if the jumper is at 133 I can tell it to be 140. I, however, cannot make it 166 becasue to do that I'd have to move the jumper again. Get it? I don't know if that's what you're experiencing or not. If you find that it won't let you raise it higher than 200 in the BIOS, you'll have to raise it through software.
Nucleo said:
What kind of temps are too high?
what kind of speeds should i expect?
Click to expand...
With my 1.6 GHz chips running at 205 x 16, I see CPU temps in the area of 40-42 degrees C in the CPU and around 48-52 degrees C. I'm fairly certain that if I were to better cool the voltage regulators and chipset on the 'board I could get a much higher FSB speed. As a first order of business, place a high-flowing fan directly over the area of the 'board bounded by the graphics card to the south, the CPUs to the west, the memory at the east, and the edge of the board at the north. This is the area where all the magic of the MOSFETS takes place and they get very warm in still air.

I actually have no idea what sort of speeds to expect. I would try to keep all temps under 50C as a first order of business. I think you should be able to get 200 x 16. I'm under the impression that it's better to raise the FSB and lower the multiplier than to do the opposite. See if you can get it to run at your lowest multi (13 maybe) and a 200 MHz FSB. From there you could creep up the multiplier. After (if) you find a good solid stable 200 MHz setting, benchmark it versus your best 166 MHz setting. Go with the best one. Since I run F@H I find that memory bandwidth is the most limiting factor. If I can increase the bandwidth I can get much more speed.

Tell us how you do. :)
 
You must log in or register to reply here.
Back
Top Bottom