Ahh, the part arrival has started!

Vindicator · Post by **Vindicator** » Thu Sep 30, 2004 12:41 am

Why thank you Stress. Btw, hows your Koolance project coming along? Any big overclocking news to report?

The ground can be anywhere on your case, as long as its touching metal. Theres an unused screw hole for an extra long mobo in my case, so i just screwed the ground wire to that.

STRESSTEST · Post by **STRESSTEST** » Thu Sep 30, 2004 11:11 am

I got tied up with 3 other systems. They should be done by tomorrow. "Red" will make a debuit later today in the case mod gallery

Avder · Post by **Avder** » Thu Sep 30, 2004 8:46 pm

So what are you naming your Dual Xeon rig? Is that Red?

STRESSTEST · Post by **STRESSTEST** » Thu Sep 30, 2004 8:51 pm

no name. netbios is "Dually". Creative eh?

It's in one of those SGI rack mount cases, rememeber?

Avder · Post by **Avder** » Fri Oct 01, 2004 9:16 am

Ohhhhh yeahhhhhhh, now I remember.

Well hurry up and get it working!@~!!!!! *bashes head and jumps around*

Vindicator · Post by **Vindicator** » Sun Oct 03, 2004 9:49 pm

My decision to use the huge heatpipe coolers may have bitten me. The VRM and NB heatsinks get freakin HOT. For the past few days I've had a 80mm high-speed fan stuck between the heatpipes and my IDE cables, aimed so that the VRM and NB gets most of the air and the CPUs get the rest. With the fan in place the VRM temp (according to MBM5) stays around 50 degrees C. Removing that fan for less than 5 minutes makes the VRM temp soar to 70 degrees C.

Needless to say, I'm gonna have to rethink the airflow in my case in order to keep everything cool.

Avder · Post by **Avder** » Mon Oct 04, 2004 3:50 am

Thats part of the reason I decided to go with a Liquid cooling solution instead.

Instig8 · Post by **Instig8** » Mon Oct 04, 2004 5:07 am

Peltier. Is there a use for it? Either to cool the water or CPU?

Just wondering.

STRESSTEST · Post by **STRESSTEST** » Fri Oct 22, 2004 7:35 pm

Just a radiator Joe.

Okay guys. I really didn't want to go through the voltage mods on this mobo to get my 2.75 volt hot shot ram to work, so I opted to put some cheap-ass 3200 in there that is happy @ 2.6 volts and see what would happen. Well so far so good. 2.8 for each chip works fine.

I am working on getting it all going now and after I hit submit on this I'll be working on the 4 directories for folding. There is a bit of a family thing going on here. Let me explain

As some of you remember, Bullgod is my cousin. His mother, and mine are sisters. Bullgod configured his mothers machine to fold, which is a good thing, but did it for his own username. So now with that added extra push, he has passed my mothers points (diannnoeora) and left her in the dust. Here's where I derive fun in this lol...

I have decided that this 5.6GHz of dually system is going to be configured for mom

So, Bullgod.. Watchout! lol

More later. Oh yea, watch Diannnoeora's 24 hour average leave the 120ish mark soon

STRESSTEST · Post by **STRESSTEST** » Fri Oct 22, 2004 8:09 pm

Damn, it's just purdy idn't it

heh, for it's very first WU's it pulled
1 x 1308 = 320 points
3 x 1310 = 316 points ea.

Avder · Post by **Avder** » Fri Oct 22, 2004 8:16 pm

What OS is that? XP?

STRESSTEST · Post by **STRESSTEST** » Fri Oct 22, 2004 8:21 pm

MS Server 2003 DataCenter edition

STRESSTEST · Post by **STRESSTEST** » Fri Oct 22, 2004 10:50 pm

Been running for a few hours now. Been 100% stable too (knock on wood). Temps sit right at 45C and 46C for CPU 1 and 2 respectivly.

Not to shabby

fliptw · Post by **fliptw** » Sat Oct 23, 2004 12:14 am

doesn't the FAHcore_x.exe processes show task list, or is that a RDC connection?

BULLGOD · Post by **BULLGOD** » Sat Oct 23, 2004 12:52 am

ROFLMAO

Why am I not surprised.

STRESSTEST · Post by **STRESSTEST** » Sat Oct 23, 2004 11:19 am

Hrmm... Going through the logs this morning. No reboots, but machine ID #1 and #2 F@H clients have terminated early a few times and sent back uncomplete work.

More so with #1 then #2. #3 and 4 have been working on the same work unit downloaded yesterday and continue to do so. Both 1 and 2 are working on cores now that have been going for some time. Much longer then the ones that terminated early.

Any ideas why this would happen other than the obvious reasons of overclocking Jim? Do dually systems see more of this kind of thing? It's happened on other machines here that are not overclocked, so Im just wondering.

Vindicator · Post by **Vindicator** » Sat Oct 23, 2004 1:49 pm

I was getting a few early ends with the 130x units I was working on, but I turned off -forceasm and havent had any since.

STRESSTEST · Post by **STRESSTEST** » Sat Oct 23, 2004 2:34 pm

Are you using a dually system?

Vindicator · Post by **Vindicator** » Sat Oct 23, 2004 5:10 pm

Yes, I'm referring to my dual Xeon/PCDL combo that you also have. None of my other machines have problems with using -forceasm. Its currently running 2.65ghz (one of my sticks of RAM is flaky so for the time being thats what I'm stuck at). Its also running 4 instances of Folding.

STRESSTEST · Post by **STRESSTEST** » Sat Oct 23, 2004 6:56 pm

hrmmm

I'll give that a try next time one kills over

STRESSTEST · Post by **STRESSTEST** » Sun Oct 24, 2004 1:23 pm

welp, I removed that switch and they seem to all be working now without core crashes.. I'll give it another day to see how it does. Thanks for the pointer BTW Vindi.

Avder · Post by **Avder** » Sun Oct 24, 2004 2:57 pm

How come youre running them at 2.8 instead of 3.2, stress?

STRESSTEST · Post by **STRESSTEST** » Mon Oct 25, 2004 1:57 am

that's 200 FSB and it will stay there for a week maybe to make sure the sytems is perfectly stable. What better way to determine that then to fold

I'll dick around with more fsb settings after all that.

Avder · Post by **Avder** » Mon Oct 25, 2004 10:45 pm

So 200x14? I assume you mean youre going to bump them up to 200x15 and then 200x16 later on?

Vindicator · Post by **Vindicator** » Mon Oct 25, 2004 11:44 pm

The Xeons love higher FSB (as with the P4), so ideally rather than increase the multi you'd set the FSB at around 220-230.

STRESSTEST · Post by **STRESSTEST** » Tue Oct 26, 2004 12:03 am

we'll see. This ram may nit run that high so I may have to work with the multi too.. been busy on re-fitting the desk here for the last couple of days.

Avder · Post by **Avder** » Tue Oct 26, 2004 1:14 am

Well from what Ive read, whatever you set the FSB at, the CPU has to be able to at least post at 16x that speed until the bios can drop the multi back down. Thats why when I get the watercooling all set up next year im shooting for just 200x16.

STRESSTEST · Post by **STRESSTEST** » Tue Oct 26, 2004 10:06 am

no, it doesn't quite work like that

XEONs are first off, not multi locked. The rub is that you can only go down in multiplier, not up. So for example these at 16x100 cannot go 17x+ etc.

The way you have to do it is to set the multi at 14 and leave the fsb at 100 (1.4GHz) and then save and exit the BIOS. Once it reboots and you see that the 1.6s have posted at 1.4, you're ready to get into the BIOS and play with the FSB since the multi stuck. Youa re free to play with the fsb without worry until you get a hard lock and have to use the clear cmos jumper. At that point you have just reset the BIOS back to 16x and have to start all over again. This goes on till you find the sweet spot.

Maybe this is the most I can get out of mine, I don't know. I do know that mine only get 1.55 volts with the 1.6 mod (the highest you can go). Others get 1.57 1.58 volts with the 1.6 mod. That little bit could be the difference between 2.8 and 3.0~3.2 GHz for me

Vindicator · Post by **Vindicator** » Tue Oct 26, 2004 11:02 am

Avder's referring to the 'feature' on the PCDL (unsure if all Xeon boards do this) where the Xeons have to POST at default multiplier for a split second before the BIOS sets it at whatever you chose. So if you set it at 14x200 its actually running 16x200 for a split second when the system boots.

And I've never had to clear the CMOS, since it resets itself if overclocking fails

Avder · Post by **Avder** » Wed Oct 27, 2004 12:02 am

Yeah, that feature is what I was referring to.