OC Gods, Help! - Am I getting a good OC right now?

[alwayslate]

So the dust is now settling on my upgraded setup:

AMD T-bred 1800+ (just got one of the OEM cheapies from newegg)
OCZ PC3500 512MB DDR (1 stick)
Abit KX7-333R (newegg refurb, grey caps) w/vcore mod
D-tek TC4 water block w/eheim 150gph pump and Chevette heater core (3/8" setup)

I've "settled" on a rock-solid 1900MHz (189*10) at 2.0vcore. I'm assuming the extra 10MHz is coming from the "enhance for benchmark" setting in the BIOS or something. CPU integrity is a big deal for me and this seems to be the highest speed I can get and still run Prime95 reliably for days at a time.

Don't get me wrong, I think this is a great speed to be running a $69 chip at. But I feel like I have to scratch my head over a couple of things and need the OC gods to tell me if my situation is to be expected or if I should be spending even more time trying to make the rig run better/faster/cooler:

1. TEMP - My system runs hot-- 50C idle and 58C under load. I'm running Shin Etsu G751 thermal compound (1st day so I expect temps to drop slightly). When I run at stock speed & voltage (1533MHz / 1.5v) I get 32C CPU idle with mobo at 30C so it seems like the water rig and my application of the block/grease is working. My pre-RMA attempt with a 1700+ chip showed similar temp behavior. It seems that as soon as I get above 1.75vcore my temp goes wayyyy up. IS THIS RIGHT? am I missing something here? Now that I'm stable should I even care about running 50-58C? I know that many people would say "I like to keep it under 40C" or "stay under 50C"... but if I can run Prime95 stable, then is it just shorter CPU life that I risk?

2. VCORE - My vmod seems to take me up to 2.02 volts before kicking in overvolt protection and refusing to POST. I just used the pin 7 vmod. This seems low based on what I've read. Given my temp issues in #1 above I guess I shouldn't think too much about going higher... but is it normal for me to be limited at 2.0? Would I need to use the pin 10 vmod to get past this?

3. TEMP MEASUREMENT - I'm using the built-in winbond temp sensors and MBM to report temps. Is this a good way to be measuring?

4. SPEED - Other than temp and vcore limitations everything else seems stable. It feels like I could easily squeeze a 200FSB and an even 2GHz if I could squeeze a bit more vcore (see #2 above) to say 2.1v and keep temps in check (see #1). Clearly this is the whole challenge of a good OC, but given that this is my 2nd T-bred acting the same w/respect to vcore and temp I'm inclined to believe this is to be expected with these chips, although maybe I'm just getting chips from less than optimal batches. I believe my 1700 stepping was RIWGA and my 1800 is AIUGA but i'm not sure on the 1800 (I looked at the CPU and it seemed to be the same stepping as the other threads were reporting on the recent 1800 purchases from Newegg so I didn't bother writing it down). Is 2GHz just that elusive for these chips?

4. SUGGESTIONS? - Other than splitting hairs over which components could be "just a little bit better" such as Corsair vs. OCZ or "use one of Cathar's blocks instead of the TC4"... what else could I be using/doing differently? I have some doubts over my thermal grease technique, but I wonder if it is really that easy to screw it up and cause high temps (seems to work great at stock speeds).

thanks in advance!

[JNav89GT]

personally I think you should be getting over 2ghz with that cpu. From what I hear peeps are getting 2.0-2.2ghz with these. At least that's what I expect from mine that I just ordered.

Now, about your temps. I think you have a kink in your line somewhere. No way your temps should be that high unless there is something wrong with your setup. I mean on air my cpu's don't get that high really unless pushed further. Maybe the A rev has something to do with that though. I still feel that the H20 setup should allow more mhz at that vcore and lower temps

[KnightElite]

1. Your temperatures are fine. Remember that power used (heat output) varies with voltage squared, so a small voltage increase results in a high CPU temperature increase.
2.Mine cuts out around 2.05V on my AT7, but after that changing the voltage in the BIOS to 1.85V gives me about 2.15V.
3. Not the best way, since you aren't actually measuring the on-die temperature, but the socket temperature, but it's definitely the most hassle free method.
4. 2GHz could probably be achieved with better cooling, possibly more VCore. I'm stuck at 1900Mhz with my 1600+ Palomino, and it's really annoying, since it does go into windows at 2GHz... ;).
5. About the only thing I could suggest would be to get a TEC and TEC capable waterblock. That's the best way for you to get more speed at this point.

[alwayslate]

Thanks for the replies.

No kinks in the line. I have great flow and the chevette heater core seems plenty big for this application. I forgot to mention that I have a custom fan shroud with a thermaltake 92mm smartfan attached at low speed. Good flow of fairly warm air getting pulled out of there.

It's frustrating to fall short of the 2GHz mark. Not that it makes all that much difference since the primary reason I upgraded from my P3 750MHz is for MPEG2 encoding-- 100MHz barely even seems to buy me 1FPS encoding difference. I think it's just psychological-- if others can achieve it on similar setups it always seems like I should be able to as well. Oh well, I guess.

I guess I should have posted some pics to help...
(yeah I know the cables and wires are still a mess)

http://66.218.57.24/Dsc02555.jpg

http://66.218.57.24/Dsc02556.jpg

http://66.218.57.24/Dsc02557.jpg

BTW another question related to my lame thermal grease skills-- using the 4-screw post and springs setup, how hard should I be torquing the thumbscrews? The block came with no instructions.

[kchip]

No reason you shouldnt hit 2.1 IMO

[alwayslate]

Argh! My thoughts exactly.

Which is why I seek the help of the OC gods... :rolleyes:

[lefthandedgoat]

its all luck. I have a 2.4b pentium 4 that wont go much over 2.7ghz with any reliability. I just got it like 2 weeks ago too. Most have been getting 2.9 really easily, and if they are lucky at all 3.1

[sharpbw]

One of the first things I would play with, is to put a 120mm fan on that heater core, and see if it does anything to your temps.

[Havokator]

I had a similar problem with my water cooling setup when I first started using it. I got really high temps. THe problem was that the block was not tightented enough with the springs. Once I torqued it all the way the temps drop significantly. May be worth a try for you.

[JNav89GT]

yeah I think your temps should be about 5-10C better. And then of course your oc will go quite higher.

The fan on the heatercore may not be enough to shed that heat, and what the other poster said about seating the waterblock. My maze3 I tighten the crap out of.

[ryanpgroovy]

58C on a water cooled setup seems mighty high to me. At first I though you were on air and had a decent OC going on.

[ABIT Marketing]

Where is Opp and Randi when you need them!!?? Grrrr

[alwayslate]

Ok I think I'm zeroed in on the "problem" here.

Cranking full speed on the radiator's TT smartfan (5400RPM) nets me about 1.5 degrees C. Not really worth the added noise, IMO.

BUT... Just opening the case up and cranking down on my existing block mounts-- 4 degrees lower for both idle and load. :D

Thanks for the replies. I'll be pulling the block off this weekend to re-mount it again with the G751 grease and torquing the hell out of it (well... hopefully not enough to break anything!) and checking the final results.

My guess is probably low 40's or better when the dust settles again.

If it goes well maybe a pin 10 vmod to get me up to 2.1v and maybe 2GHz is within reach after all...

[Cathar]

It might sound silly, but get an 80mm case fan and sit it atop your video card and point it at the CPU socket area.

Usually good for around a 5C improvement in temps with water-cooling if you have poor case ventilation.

The problem with WC is that it takes away the airflow around the socket area and this really does contribute quite greatly to the CPU's temperature as quite a surprising amount of heat is conducted through the CPU pins and through the traces to/from the surrounding components. The motherboard is a heatsink and it does need to be cooled. Crazy, but true.