OCing an Opteron 175 socket 939

mk000000 · TweakNOOB Joined: 29 Dec 2004 Posts: 78 Location: Philly

I got my Opteron 175, I love it. So I'm trying to OC it to see what it can really do. I had it up to 2.7Ghz and unstable. But that was only my first testing. Instead I started over pushing the Vram up to the max 3.0v and the Vcore up to the max 1.5v (even though CPU-Z reads 1.535v).

The first OCing I did, I put Vcore and Vram up to 2nd or 3rd top, but not top, and then did the OCing. This time I'm doing it the other way.

I OCed at 230 ram & HT, 4x HT, 10.5 clock multiplier, and 2.5,3,3,8,2T. It's now at 2.415Ghz. But it's running hot. Even hotter when I started
Anyway, you guys think

I know the voltage is making everything hotter. And it's managable. But I am wondering what settings to try next. It's zero errors in 7 hours of Prime95 torture testing for both cores. I'm getting 115F on idle and 131-136F on torture test. Keep in mind, non-OC this runs a cool 98F idle and 105F processing.

Any advice on timings and voltage to try next?

Also, what's the real deal on burn in? Does a lengthy torture test on an OCed processor really burn it in? Just wondering.
_________________
AMD Opteron 175
ASUS A8N-SLI DELUXE
Corsair 2GB CMX1024-3200 CAS 2 LED
ASUS EN6600GT
ATI Theater 550 PRO
WD Raptor 74GB 10K RPM
3 x WD 320GB SATA150
Seagate 300GB 7200.8 SATA150
NEC DVDRW DL
Windows XP Professional x64
Dell 24" widescreen LCD

Google · Sponsor

[KoG]^weaZel · Posted: Tue Jan 24, 2006 10:57 am Post subject:

well the first thing you will want to do is drop the vcore back down. You don't want to run that up the very first thing. You want to find what it will do at stock voltage.
_________________
I tweaked and it tweaked back! So I Tweaked some more!

"Barney is like the Michael Jackson of PBS." - James Tybeerious

ToggleHead · TWEAKGURU Joined: 03 Mar 2004 Posts: 4360 Location: Jersey

agreed....only up the vcore when the OC you have in place becomes unstable...then up the vcore a bit and push some more.....and so on.....

you want as little power to the cpu as needed for a stable OC to keep the temps down
_________________

racing87stang · Posted: Tue Jan 24, 2006 3:55 pm Post subject:

not to mention, you dont want to just blow the chip away with the highest setting you have.
_________________
Remember When Sex was safer than racing???

mojo1340 · UberTweaker Joined: 19 Jun 2004 Posts: 1022

Loosen up your memory timings, put your voltages back to stock, leave the multiplier stock (11?) and see how far you can go. I go up 5 MHz at a time and run OCCT to check stability. When the CPU becomes unstable, drop back a few MHz and post back with your results.

Good luck!

mk000000 · TweakNOOB Joined: 29 Dec 2004 Posts: 78 Location: Philly

mojo, I thought dropping the multiplier was step 1 of OCing. I thought you can squeeze some MHz out of that, but dropping it was the only way to make substanial gains on the FSB & HT, no?
_________________
AMD Opteron 175
ASUS A8N-SLI DELUXE
Corsair 2GB CMX1024-3200 CAS 2 LED
ASUS EN6600GT
ATI Theater 550 PRO
WD Raptor 74GB 10K RPM
3 x WD 320GB SATA150
Seagate 300GB 7200.8 SATA150
NEC DVDRW DL
Windows XP Professional x64
Dell 24" widescreen LCD

mojo1340 · UberTweaker Joined: 19 Jun 2004 Posts: 1022

I always start with the stock multi. I have a 148 (2.2 ghz) at 2.8 and change with 260X11(stock multi)

mojo1340 · UberTweaker Joined: 19 Jun 2004 Posts: 1022

Also, do not use half-multipliers (10.5). This can cause funny things to happen to your RAM speeds.

mk000000 · TweakNOOB Joined: 29 Dec 2004 Posts: 78 Location: Philly

I lowered the voltages until stability was reduced and put it back to the one before. What I got was an OC of 235 x 10.5 = 2.467Ghz, 4xHT, 2.5,3,3,8 on Vcore = 1.475v and Vram = 2.7v. Not bad, and the lower voltages of course helped on heat. My torture test maxed at 136F.

Mojo, what you said about ram settings getting messed up with .5 multipliers struck something that I was questioning already. I notice that while I manually set the ram to 235, in CPU-Z, the HT reads FSB 235 and the ram reads 224.9. Is this the kind of thing you mean? I couldnt understand why the HT and ram speed werent the same. I left the ram at DDR400, so it should be a 1:1 divider.

So you think I should see what a multiplier of 11 can do with say 220, and keep pumping that up 5 until a problem? How about 10?
_________________
AMD Opteron 175
ASUS A8N-SLI DELUXE
Corsair 2GB CMX1024-3200 CAS 2 LED
ASUS EN6600GT
ATI Theater 550 PRO
WD Raptor 74GB 10K RPM
3 x WD 320GB SATA150
Seagate 300GB 7200.8 SATA150
NEC DVDRW DL
Windows XP Professional x64
Dell 24" widescreen LCD

mojo1340 · UberTweaker Joined: 19 Jun 2004 Posts: 1022

Yes, that's exactly what I mean.
Set the voltages back to default. Don't lower them below that for now.

Let's be clear about something here. Overclocking is an endevor that really should not be hurried. It sometimes takes me weeks to get a maximum stable overclock that I'm happy with. I then back off a few MHz and enjoy it.
If you have a target, and it's realistic, then by all means make haste toward it, but don't expect to "set and forget". The very same components can yeild vastly different results on the same board, so don't depend on someone else's results.
I always go up by 5. If you want to do 10, be my guest. The closer you get to the maximum oc, however, you'll need to reduce your jumps.
Also, keep a notebook, or journal of what you did, what you changed and the results during testing. This will save a lot of time and headaches when trying to troubleshoot instability.

mk000000 · TweakNOOB Joined: 29 Dec 2004 Posts: 78 Location: Philly

When I ask how about 10, I mean for the clock multiplier, not the increase in FSB. Do you think I should try OCing at 10x?

And I am shooting for a target, 2.5 - 2.6 Ghz. That should not be unreasonable considering my parts. I just dont want heat to rise too much more. Last I checked, I'm 25 Mhz away from the 2.5 lower end of that goal. But that's on 10.5x. I wonder what the chip will do at 11x.

Any suggested timings / settings is apprciated, and I do understand each rig requires fine tuning, but look at where Ive been and am now, and you can make suggestions.
_________________
AMD Opteron 175
ASUS A8N-SLI DELUXE
Corsair 2GB CMX1024-3200 CAS 2 LED
ASUS EN6600GT
ATI Theater 550 PRO
WD Raptor 74GB 10K RPM
3 x WD 320GB SATA150
Seagate 300GB 7200.8 SATA150
NEC DVDRW DL
Windows XP Professional x64
Dell 24" widescreen LCD

racing87stang · Posted: Wed Jan 25, 2006 3:07 pm Post subject:

Are you using a stock HS/Fan. If you are having temp issues already with such a LOW overclock. I would have to check out the cooling solution. I myself will have aOptie 175 running in my new system (hopefully in the next week or so.) and with that I know I will be water cooling that at least.
_________________
Remember When Sex was safer than racing???

mk000000 · TweakNOOB Joined: 29 Dec 2004 Posts: 78 Location: Philly

I'm not holding the countless good reviews and user opinions to translate to my results, but obviously they are good indicators of chip batches. Anyhow, the current 2.475GHz @ 136F is not bad... we are again talking 2x Prime 95 torture test temps, and idle temps are @ 110F. I would obviously be happier with lower temps and the possibility of getting a higer OC. I still have some Vcore fiddling to do as well as with the clock multiplier. I havent tried anything but 10.5x yet.

As for cooling... I am using the stock cooler right now. It got such excellent reviews and it is well put together. It has an 80mm fan and 4 copper pipes going through it. My after market HSF is a gigabyte rocket, and while it was supurb on my old chip (3500+), I feel the stock Optie HSF is way better.

One concern for me was that the stock HSF uses the thermal pad, whatever that stuff is called, the kind that is on there waiting to just stick to your cpu. Now in the past I was always using Artic Silver 5. I even have a tube of it still available. I was considering cleaning off the crap they stock the fan with and exchange it for some AS5. That could easily be another 5 degrees F off the temps under torture.

Comments/Suggestions welcomed.
_________________
AMD Opteron 175
ASUS A8N-SLI DELUXE
Corsair 2GB CMX1024-3200 CAS 2 LED
ASUS EN6600GT
ATI Theater 550 PRO
WD Raptor 74GB 10K RPM
3 x WD 320GB SATA150
Seagate 300GB 7200.8 SATA150
NEC DVDRW DL
Windows XP Professional x64
Dell 24" widescreen LCD

racing87stang · Posted: Wed Jan 25, 2006 4:16 pm Post subject:

it will most definetely be better than the pad.
_________________
Remember When Sex was safer than racing???

PhRoZeN_X_ · SirTweaksabit Joined: 19 Jun 2004 Posts: 410

^^^ yep ditch the pad right away. AS5 would be a big diff.

mk000000 · TweakNOOB Joined: 29 Dec 2004 Posts: 78 Location: Philly

I'm not sure 'pad' was the right term...

look at the link, it's a review of the new opties HSF.

http://www.madshrimps.be/?action=getarticle&articID=399

Anyway, the copper block has a square of paste, but its preapplied, u know. That's a pad, right?

Also, I tried 11x multiplier and different ram speeds. Keeping at 2.5 CAS i got 2475GHz no prob, 138F, and 2530Ghz with one error in a 25 min 2xPrime95 torture test. Vram at 2.75v and Vcore at 1.475v.

Nothing crashed but obviously that error points to instability. So whats next, one more Vcore bump, or CAS bump?
_________________
AMD Opteron 175
ASUS A8N-SLI DELUXE
Corsair 2GB CMX1024-3200 CAS 2 LED
ASUS EN6600GT
ATI Theater 550 PRO
WD Raptor 74GB 10K RPM
3 x WD 320GB SATA150
Seagate 300GB 7200.8 SATA150
NEC DVDRW DL
Windows XP Professional x64
Dell 24" widescreen LCD

racing87stang · Posted: Thu Jan 26, 2006 5:23 am Post subject:

TRY THE MEMORY TIMINGS FIRST. THE increase in FSB will surely take out any effects of slowing RAM timings.
_________________
Remember When Sex was safer than racing???

mojo1340 · UberTweaker Joined: 19 Jun 2004 Posts: 1022

As I mentioned in my first post on this subject, LOOSEN UP YOUR RAM TIMINGS! We'll worry about them later. Right now, we want to see what the chip will do. As 'stang said above, you'll see much better performance with more MHz than you will from tight timings.

[KoG]^weaZel · Posted: Thu Jan 26, 2006 11:21 am Post subject:

Only overclock one part at a time, that way you can find its maximum operating range.

My advice is OC your memory first and check it using Memtest86+. Then back it down and OC the cpu and check that using Prime95. Then once you have the max one both then you can find the middle ground and best performance.

and stop using the 1/2 multipliers, you are only handicapping yourself by using them.
_________________
I tweaked and it tweaked back! So I Tweaked some more!

"Barney is like the Michael Jackson of PBS." - James Tybeerious

mojo1340 · UberTweaker Joined: 19 Jun 2004 Posts: 1022

^^^^ Good advice! I do the CPU first, but it's dealers choice.

[KoG]^weaZel · Posted: Thu Jan 26, 2006 5:14 pm Post subject:

I find the max memory before the OS is even installed.
_________________
I tweaked and it tweaked back! So I Tweaked some more!

"Barney is like the Michael Jackson of PBS." - James Tybeerious

mojo1340 · UberTweaker Joined: 19 Jun 2004 Posts: 1022

Interesting idea. Never thought of that.