Opteron 170 OC questions

Astral Abyss

Astral Abyss

2[H]4U
Joined
Jun 15, 2004
Messages
3,066
So I got this Opteron 170 from Newegg a week ago with a supposedly good stepping of CCBBE 0610DPMW, and I was hoping for the best. It replaced an Opteron 146 running @ 275x10 for about a year.

The new 170 will go to 250x10 no problem. Doesn't even need extra voltage. I started to get excited... :cool: Try to go to 260MHz and the system will just power down at random times in WindowsXP Pro. :( Never seen that before where it just shuts down. No BSOD, no freeze, just turns off completely. Bumped voltage incrementally up to 1.425v and still does it. :mad: I guess I'll try 1.45v next but I'm wondering what the deal is. I've been overclocking for a long time and never seen a mobo just power down like that when the OC fails.

The real problem is I'm watching temps in CoreTemp hit 60-61C before it shuts down at 1.425v. The TCaseMax for this 170 is 57C (which seems really low) and has a 75W (I think it is, I'm pulling this from memory) power rating.

Question #1: TCaseMax varies between different CPUs and steppings. Is this a theoretical max or just a max temp to maintain the rated power consumption? I know a lot of people have to be over their CPUs TCaseMax if they're only going by BIOS/mobo software to get CPU temps.


Question #2: Should I be worried if I'm hitting 60+C on core0 and 56-57C on core1 in CoreTemp? Nvidia monitor shows temps at 50-51C like it did for my 146, which is obviously not as accurate as CoreTemp. The temps probably going to go even higher if I bump the voltage to 1.45v or 1.475v.


While I guess I should be happy with 2.5GHz out of a 2.0GHz CPU, it's actually a bit of a downgrade considering my 146 was running at 2.75GHz.

I guess I'm just bummed I got up to a totally stable 2.5GHz at stock voltage to complete, utter failure at 2.6GHz at even high voltage with equally high temps.


Any help or suggestions? RAM? Running at 250MHz, 3-4-4-8. When I bumped it to 260 I lowered the divider to 166MHz, which obviously didn't help. Tried 3x and 4x for HyperTransport. See sig for system specs...
 
I'm going to assume those 60c temps were at load? Even still, that's REALLY hot for your chip. I'm running an Opty 170 at 280 HTT, and my cores are about 35c idle, 45c load (and I'd like to lower it further).
 
I run my Opteron 175 at 250 FSB to net 2.75 GHz, 24/7 with 1.45 V, on air. So, 1.45 is not that bad.
 
Are you using anything that measures the different voltage rails while in Windoze?

I like using OCCT ( ftp://ftp2.ocbase.com/ocbase3/OpenBeta/OCCTv0.91.exe ) to stress my system when I *suspect* a non-stable overclock.

OCCT will log the voltages during the stress test, and allows you to determine if you have a power issue. I noticed a serious ripple and dip in my 3.3v rail during stress testing of my Venice 3200+, and couldn't get a stable OC over 2.35GHz. I replaced the Antec SL-350 I was using with an OCZ powerstream and tried again... ripple was reduced across all voltage rails and the 3.3v line had no more drops. I was able to get that 3200+ to 2.5GHz with 1:1 on my ram running 24/7.

System shutdowns without warnings (error messages, BSOD, temp warnings, etc...) are usually indicative of power supply issues. Monitor your voltages.

Good Luck.
 
If you can monitor the vcore with a dmm. I found that my DFI lan party is actually not as stable as I thought w/the vcore. CPUZ, speedfan and all other programs report 1.520 and rock stable but with a dmm I'm showing switching between 1.59 to 1.61. I was kind of surpised by this and don't really trust software monitoring programs anymore.
 
NordicRX8 said:
Are you using anything that measures the different voltage rails while in Windoze?

I like using OCCT ( ftp://ftp2.ocbase.com/ocbase3/OpenBeta/OCCTv0.91.exe ) to stress my system when I *suspect* a non-stable overclock.

OCCT will log the voltages during the stress test, and allows you to determine if you have a power issue. I noticed a serious ripple and dip in my 3.3v rail during stress testing of my Venice 3200+, and couldn't get a stable OC over 2.35GHz. I replaced the Antec SL-350 I was using with an OCZ powerstream and tried again... ripple was reduced across all voltage rails and the 3.3v line had no more drops. I was able to get that 3200+ to 2.5GHz with 1:1 on my ram running 24/7.

System shutdowns without warnings (error messages, BSOD, temp warnings, etc...) are usually indicative of power supply issues. Monitor your voltages.

Good Luck.
Click to expand...
It has the makings of a power issue, I agree. I'll give that program a try.
 
My brother agreed to buy my 550W Antec, so I'm applying that towards a 750W Silencer Quad. I figured I'd take power out of the equation completely. I'll report back when it arrives and I install it.
 
I wouldn't worry about temp difference between cores. My 165 does it too. 2-3C difference. I do agree that this has all of the makings of a power issue though.
 
I just thought I'd update this problem and solution I found...

I got the Silencer 750W Quad and I still had the same problem. The computer would just shutdown minutes or seconds after booting into Windows when overclocked to 2.6GHz or higher.

More CPU and RAM voltage didn't help. Lowering CPU temps didn't help.

I gave up and sold the 170, only to find out from the buyer he got it to 2.95 on water... doh! :eek:

So I bought another 170 from the Egg...

Same stepping: :) CCBBE 0610DPMW. Slightly different TDP: 85W, 59C. (new) vs 75W, 57C (old)


Alright I thought, this is going to be better. Got it to 2.4, no problem. 2.5 locks up in Orthos, so I bump volts to 1.375. Stable.... 2.6 system powers down. Nooooooooo. :( I figured it had to be something else. Turns out it was. The RAM. Even though this same RAM was stable at 240MHz FSB with my Opteron 146 it was failing utterly at lower speeds with the 170 to the point the entire system would power down, and this is at CAS 3-4-4-8 so there's not much I can do. Well, I relaxed a couple of the other RAM timings by only 1 tick and bam... rock solid. No shutdowns, no lockups. I'll just say at this point that it's definately worth your time to check those other RAM settings no one really mentions if you're having system instability.

I wanna say it was Bank Cycle Time (Trc) from 8 (default) to 9 and Row to Column (Trcd) from 10 (default) to 11 that I changed.

I'm now up to 2.7GHz @ 1.40v and Orthos testing for stability. We'll see just how high we can go after my Tuniq Tower gets here next week. :p
 
You must log in or register to reply here.
Back
Top