Sunday all day

Now trying to get down to the business of oc'ing DreamCatcher v2.0's mobo / CPU I managed to learn several lessons that caused a lot of frustration........

I had finished flashing to Bios 15 and everything seemed to be working "normally" that was until I went into the bios to start setting up for oc'ing!

SoftMenu Setup:

Set CPU Operating Speed to User Defined. Went into CPU External Clock (MHz) and the up/down arrows don't respond nor can I change any values with the keyboard. The screen shows the min and max setting plus the line with the red box that I assumed is where the new value should appear.

Advanced Chipset Features:

HT Spread Spectrum doesn't even show (as illustrated in the manual).

It was then brought to my attention that I forgot to reset the CMOS after flash so did so and still no joy. At another site it had been mentioned by someone that they had problems flashing their nView with Winflash and things went better when Abit's FlashMenu was used, so off to do that??..still no joy.

Anyway when it was all said and done I spent several hours Sunday re-flashing (both Bios 14 & 15), pulling out what little hair I have, and being generally puzzled. Then inspiration (desperation?) hit and I was left feeling like an idiot. DOH!!!!! Anyone ever hear of the num lock key!?!? I could then type in the clock values! Also it turns out the HT Spread Spectrum is a none issue

This particular wireless k'board doesn't have an idiot light for cap or num lock so one has to be "aware". Also it doesn't help to ass_u_me because bios has num lock "set" for normal operations that it would be set IN the bios operations!

Well at least I won't be such a woose at flashing as I've done Bios 15 with Winflash and Bios 14 & 15 with Abit's FlashMenu (super simple!).

By the time I had most of it figured out, the wife wanted to do some road running, then time for a nap before work so I'll have to wait ?till later Monday to see what I can screw ? err setup and run.

So remember, clear the cmos after flashing and num lock isn't a given when changing bios settings!

Mr. Mod, why is it when I copy a word document over most of my " & ' turn into ? ??

02/25/08 15:30

After conversations with ud I decided to dump the Abit EQ utility. Also after some thought took a chance that the HT Frequency I could not find was what Abit referred to under ?Advanced Chipset Features? as K8 <-> HT Speed might be the same, so???

Changed mem voltage to 2.1v and the K8 <-> HT Speed from Auto to 3x I rebooted back to 280HTT and started again.

Right off the bat I managed to do an 8M S-Pi run so decided to try Prime again. I got impatient (what?s new) and stopped it at 1-1/2 hours to play with ClockGen.

Okay I?ve managed to get an 8M S-Pi run at ClkGen of 295 for 3.10GHz, however I noticed in SpeedFan Core #1 is doing most of the work and am a little concerned as the prior failures in Prime have been with Core #2.

Before hitting the sack for a nap before heading into work I booted to 290 (gives me 3.0GHz) and started a Prime run. When I left for work at 21:45 it had been running for 3.5 hrs, so just left it running ?til I get home. Figure if I?m still awake after getting home will attempt to boot at 295 and play with ClkGen some more!

02/26/08 early morning

The Prime session was still going when I got home from work and when I stopped it; the run had gone on for approximately 10.5 hours. Sadly I forgot to get a screenie for proof.

So it would seem that

Core: 1.40v
Memory Voltage: 2.1v
CPU Frequency: 290
Hammer Fid Control: Factory (Can?t Find)
HT Frequency: 3X (K8 <-> HT Speed in Abit)
PCIE Clock: 100
Multi: 10.5
Memclock: 667-5-5-5-15
1T/2T Memory Timing: 2T

is a confirmed bootable base from which to work. Actually it wouldn?t be a bad stopping point as it gives me 3.05ish GHz.

Below are what are probably some meaningless screen shorts from this morning oc?ing session.

This one shows the highest ClkGen setting obtained while using the v2.0 version of Stresscpu a HTT value of 298 for 3.13GHz



This comes under the ?for S & Gs? category??. the highest ?bootable? HTT value was 300 but by no means stable with both cores at 100%. However I did manage (this is the S & G part) to get a 4M run of Super Pi at this setting. When I tried for an 8M run when it got to the 18th loop it locked up with the wee-woo, wee-woo alarm.



To end the morning session I started a Prime run with HTT set at 295 for 3.097GHz. Core #2 crashed at 3 hours, 32minutes and I stopped the session at 3 hours 40.



After the Prime session for 295 I shut down the system. On a side note my ?R & D? area has been moved (banished?) to the garage. This meant the ambient temp was about 18C through the morning session with cpu temps never breaking 50C.

02/26/08 late afternoon

As I?m fighting a major cough and chest issue I felt it prudent to get some heat in the garage so the ambient moved up to around 24C. Because of this the afternoon session have core temps running in the 52 ? 58C range. This was with the hsf fan rpm raised to about 4500-4600 as opposed to 4000 in the morning. Also the twin 120s on the Tech Station went form min to max.

The main thing the afternoon session taught me is I needed another value when I refer to my ?settings?. Originally I would reference ClkGen and ?bootable??.now I think I need to add ?Cold Bootable?.

In my ignorance I decided to drop the mem from the 667 to the 554(?) divider and see if I could at least break the ClkGen barrier of 300. Well that was meaningless attempt as nothing changed. With ud?s suggestion to keep the 667 and boost the vcore I gave that a try. Now comes my reason to add ?Cold Bootable??.no matter what I did, I could not obtain any of the oc values above 290 that had been achieved in the morning session. The morning HTT bumps were done as the system was ?running? = ?warm boots?. In the afternoon I went directly to 295 at startup. Prime runs at 295 wouldn?t even last 20 minutes this time and I think it was Core #1 that crashed. This was a first as #2 up to this time had been the weak link.

So I turned everything off before my pre-work nap, and will just walk away for a while.

02/27/08

When I shut the system off, the idea was just to get away and leave it alone for a while?.. Yah right! There are two major things I lack, that would be a lack of resolve and patience.

Lack of Resolve??.
Got home from work this morning and promptly (even before feeding the critters) fired up the Tech Station and adjusted the nView?s Bios.

HTT back to 295 and bumped the Vcore up to 1.45v (I know ud you said one bump ? 1.42v!) and threw on Prime. Then hit the sack.

Lack of Patience??
After the nap and doing a few other things I stopped the Prime run at 4.5 hours ? which was an hour longer than yesterdays 3.5 hour crash. Had to go out on some errands so turned off the system for 5 minutes to get my ?cold boot? and fired her up at 296 started Prime and off I went. When I got home I could here an audible wee-woo, wee-woo or however you want to describe that two-tone alarm that reminds me of the European style PD sirens.

Got inside and found Prime running merrily along, the screen wasn?t lock, and everything seemed fine except for the alarm. Did a screenie and rebooted the system with the alarm stopping a boot-up. Man am I glad I did the screen shot! When I pulled it up and was looking at it, I found what should have caused the alarm?..in Spdfan

Temp1: 19C
Temp 2: 112C
Temp3: 47C
HD0: 22C
Temp 1: 0C
Core: 20C



Now I need someone to tell which temp is what, but it is obvious something was ?over temp?! But what I find interesting is in Bios:

Temp Alarm is set to 80C
Over Temp Shut Down is set to 90C

In two to three days of playing never did I see any of the temps in Spdfan reach 60C much less 112C. Okay I have raised the Vcore from 1.4v to 1.45v, but come on that shouldn?t cause that much of a temp change!!

Also to make it more ?curiouser? I started it all over again with everything set the same and Core #2 failed after 5 minutes. Now trying stresscpu v2 to watch the temps as the heater is on and ambient is around 21C (assuming Temp 1 is = to case temp).

While I was up in the wife's "office" typing up the thread reply, the system lockup running S'CPU v2 at 296 / 1.45v. Unfortunately the screen saver was running when it locked up so I don't know when it happened.

THIS IS WHAT IS DRIVING ME TO DISTRACTION!

First run with these settings was chugging alone fine (except for the flaky high temp alarm) in Prime for over 1.5 hours. Second run has a Core #2 failure after 5 minutes of Prime. Third run has a system lockup in Stresscpu v2 in less than 15 minutes.

So it is back to turning off the system and contemplate my navel for a while. What I think is really going to happen is I?ll see if I can get the 295 to run Prime without crashing (if I can repeat the interrupted 4+ hour run) and be satisfied with what appears to be a 80 ? 90% OC. Worst-case situation would be somewhere between 290 and 295?? and be happy!! Then all I have to do is figure out how to keep everything cool enough while running in a shoebox.

Working on that with my W/C thread!!!

02/28/07

Went back to piddled some more but it appears I may be at the limits of the CPU & board without getting ridiculous with the Vcore. So I dropped it back to 1.425v @ 294 and started a Prime run around 14:00. Somewhere around 18:30 I ended up with the BSOD. Friday will give it another try but drop back to 293 @ 1.425 as my next challenge will be heat issues in the case.

Later!

Oh boy is this thing out of date!!

Anyway since this last entry I managed to smoke both the mobo and a brand new evga 8800GT SC. Is seems something was a little wonky with the GPU and when it blew it took the board with it. Anyway got a another mobo off of fleabay and the GPU was a great RMA experience at evga. So I'll skip all of the test between and begin here.

03/14/08

Two boxes came in from ebay.......

1. Zalman CNPS8700LED, turns out it is a referb from zalman so that's cool.

2. A dirty, smelling of ozone NF-M2 nView with several caps that wiggle To add insult to injury I paid 41.50 + shipping, the one on ebay Saturday went for 32.50

Busy trying to get BH rDx v2.0 put together so will wait ?til Saturday to see what I have.

03/15/08 Midmorning

Got the nView and everything hooked up (BTW the zalman 8700 is huge!!) hit the power button and it immediately went into alarm. Got the warning to reset bios because of new cpu......tried, set to default, got the same error message... for S & Gs did F1 and it loaded windows as the alarm kept doing ye old wee-woo thing. Windows loads up among the noise and I don't have a keyboard or mouse.

Turned it off, unplugged psu, reset the cmos jumper, waited 60 seconds, put everything back, start and the same thing.

I have popped the cmos battery out, unplugged everything, did the cmos jumper again and will leave it for an hour.

Let's hope that does it otherwise??????????????

03/15/08 Evening

Did I say the CNPS8700 was huge???? You decide



Found the reason for the alarm......

You may not remember, but back when I was having all the problems with alarms on the original setup I asked around what the SYS FanEQ Reference Temp, SYS FanEQ Target Temp, etc was. Never could figure it out until after the blow up as I started paying closer attention to the manual.

I?m sure you abit experts knew this?..it seems one can monitor the "System Fan 1" just like the CPU fan with all of the controls, etc. In a highlighted area of the manual I ignored, it explained if your hsf is 3-pin (as in the Tt SmartFan2 & the zalmans) instead of 4-pin you should plug it into the SysFan1 socket (3-pin) and you could set all of the auto temp/speed controls goodness for you 3-pin hsf. Forgot one very important setting to change, the fan fail alarm was still set to CPU and it needs to be SysFan.

So here it is plugged in and it is so neat to watch (as it is hard to hear) the fan ramp up as the temperature rises, using SpeedFan to monitor. Did make one additional revision in bios, set the target temp to 52C with 90% fan speed..



As far as oc?ing is concerned 292 bombed in 1-hour 30minutes and 291 hit the wall at 5 hours 30. I managed a 14 ? 15 hour run with the original board at 290 but it crashed here after 4 + hours. Let?s hope it was because AVG tried to do an update as I forgot to turn that off. Also the garage is a bit warmer than earlier today.

For S & Gs, I?ve bumped up the Vcore to 1.450v, Chipset Core to 1.24v, and the Hypertransport voltage to 1.22v. HTT is back to 292 and the NB & Mosfet coolers are now actually feeling warm.

So it seems even though the board smelt and looked funky it is holding its own. Let?s just hope it lasts a while!


Benchies:

Can tell it is late and I?m getting punch drunk! Why else would I decide to do some PCMarks ?05 and CrystalMarks bench test???? At weird HTT settings?????

HTT: 289 for 3.0GHz
Vcore: 1.450v
Vdimm: 2.1v
Chipset Vcore: 1.24v
Hypertransport V: 1.22v

PCMarks = 7216
CrystalMarks: ALU = 22102 / FPU = 24398 / MEM = 15405 with a TOTAL of 124289

HTT: 295 for 3.1GHz
Vcore: 1.500v
Vdimm: 2.1v
Chipset Vcore: 1.24v
Hypertransport V: 1.22v

PCMarks = 7414
CrystalMarks: Not Done

HTT: 299 for 3.14GHz
Vcore: 1.500v
Vdimm: 2.1v
Chipset Vcore: 1.24v
Hypertransport V: 1.22v

PCMarks = 7436
CrystalMarks: ALU = 23308 / FPU = 25523 / MEM = 16397 with a TOTAL of 129348

If I ?read? this correctly my overall increase probably in real life isn?t enough to be worth pushing much beyond 290HTT because of the need for more voltage and stress on the CPU. The system will not boot above 299 and most likely even when I go W/C I doubt that more v will help.

I going to combine this and my Water Cooling with the worklog so head on over to DreamCatcher v2.0's worklog for the rest of the story!