2800 barton issues

[Sarian121]

Hey, first post here so might as well make it count

I'm having some issues with overclocking my Athlon XP 2800+.... For some reason, I cannot overclock past 12.5x178 and when i try to get it higher than that, windows wont load. I have tested stability at 12.5x178 with prime95 and it ran for 10 mins without any issues, but when i load a game i get random crashes. Im wondering if this is a power supply issue or what and maybe updating the bios might help? Any help would be appreciated.

Specs:
Athlon XP 2800+ barton
NF7 v2.0 mobo
2x512 PC3200 Corisar XMS
GeForce FX5600
80gb 7200 rpm WD HD
Allied 400w PSU

temps:
stock speed: 37C idle, 46C load
12.5x178: 38C idle, 46C load

 

[lobadobadingdong]

why are you running such an odd memory frequency? up it to 200, and lower your multiplier to 11, and test, then up it to 11.5 and test. this will make sure your memory is running @ spec.

 

[Sylvanas]

the barton 2800 has a locked multiplier. so all you can do is up the voltages and fsb. i got my 2800 to 185 fsb on 1.85v. make sure memory is at 1:1

 

[Sarian121]

Ive tried that lobadobadingdong, but i have to reset the bios to even get it to start again. And is 1.85v safe?

 

[Ketchup]

I would say 1.85 is safe, but I am sure there are people out there who would disagree with me. It is whatever you feel comfortable at, and how good your cooling is. You are only overvolting by .2 volts, and most people with AMD and Intel chips are very comfortable with that.

On the other hand, you shouldn't need to add any voltage at all for the speed you are at. If you do, something is wrong. Could be the power supply, as mentioned. Make sure the 4-pin PSU connector is plugged into the motherboard. If you still can't get it, may I recommend and Antec True Power power supply. It is a very good power supply and it won't break the bank.

 

[Sarian121]

Yeah, i have heard of people OCing a 2800+ to 12.5x200 at 1.75v, so 1.85 sounds high to me. I dont know what the problem is. If i try to go above 182fsb, windows has an error while starting and the system restarts. The 4 pin connector is in the motherboard. The only reason i think that the PSU is the problem is that it runs stable at 12.5x178 with prime95.. but as soon as i start running a game that requires a lot from the CPU, RAM and Video card, it crashes. I got it to start at 12.5x180, but as soon as i tried to run AIM, it restarts.

My PSU is only a $40 Allied 400watt PSU so thats why i was thinking that was the problem.. Cooling shouldn't be too big of a problem. I have a Zalman 7000 and running about 37C idle, 46C load with stock settings. Would a bios update make a difference with overclocking? i havent flashed the bios since i got the board.

 

[Sarian121]

I just tried 12.5x185 with 1.7v-1.85v, everytime i get the same problem. Goes to windows loading screen, flashes a blue screen and then reboots.

 

[Concillian]

It's not odd at all to end up with a desktop processor that can't push much more than 2200.

My more recent desktop 2500+ can't even get to 2200.
My older desktop 2500+ can push into the 2300s
My mobile can get to 2500, but only at very high voltages, I normally run it at 2400.

You sound like you got a ho-hum chip is all. I don't think you can expect wonders from any recent desktop AXP these days.

 

[maluckey]

Concillian

I disagree that it's normal for a Barton to need maga-juice for such low overclocks as you suggest.

I have three Bartons, all running at or beyond 2.4 Ghz, and the higheest vCore is 1.8 volts actual, and that one is at 205 x 12.5. My 2600+ Barton is at 211 x 11.5 with only 1.70 volts actual, and my 2400+ Barton is at 216 x 11.5 with only 1.82 volts actual. All systems are air cooled.


Sarian121

The first thing that I would do is to remove the heatsink, carefully clean the CPU, re-apply the HS compound and heatsink. It sounds like either you have a very poor heatsink, or a poorly seated HS/fan. If it is not seated correctly, or you have a very wimpy cooler, the exact symptoms that you have will appear. How do I know? I have had poorly seated HS from time to time. It's not always easy to gauge spring tension on a PAL8045 (still my favorite HS). The cure was always the same. Remove, clean, re-apply.

Good Luck!

 

[Sarian121]

Well, about 3 days ago I received my Zalman 7000A-CU from newegg and installed it. Since i have a NF7 v2.0 board, which is listed as incompatible, i had to modify it so it would fit..which basically was removing one of the outer screwholes and switching the Silver and Blue mounting pieces around. The blue piece was wider than the other one and wouldnt fit because of some capacitors in the way which means it might not be aligned correctly. My CPU temps are reporting cooler than when i had the stock fan though... by about 13C. Ill try reseating the heatsink when i get a chance.

Thanks for the advice guys.

 

[Concillian]

Originally posted by: maluckey
Concillian

I disagree that it's normal for a Barton to need maga-juice for such low overclocks as you suggest.

I have three Bartons, all running at or beyond 2.4 Ghz, and the higheest vCore is 1.8 volts actual, and that one is at 205 x 12.5. My 2600+ Barton is at 211 x 11.5 with only 1.70 volts actual, and my 2400+ Barton is at 216 x 11.5 with only 1.82 volts actual. All systems are air cooled.

How many of those were obtained recently? How many are DESKTOP AXPs and not MOBILE AXPs?
Sure, your MOBILE 2400+ is getting good overclock speeds, not only is it a mobile, it is the best stepping of mobile.

I've seen the Desktop Bartons getting worse overclocks since the mobiles have become so popular. AMD is putting it's good chips into the 3200+ and the mobiles. The desktops are getting the rejects, and are not overclocking well at all. Used to be a Desktop 2500+ was a guaranteed 2200MHz, now you're LUCKY if you get that.

 

[Bypolar]

Yes the 2800 are not known for there great Overclocks. just find your fastest stable speed and be Happy.
I have never seen a 2800 go Past 2.3G

 

[maluckey]

Concillian

I agree that my 2400+ isn't a very good representation. It is the best of the best, and easily hits 2.65 Ghz on air cooling, but that's not the point. It is a Barton, and architechturally the same as the desktop CPU that it was made from.

A quick check of Xtremesystems.org shows people getting similar OC's from the desktop versions of the Bartons as the Mobile, though they aren't scaling with the base rating, IE: the 2800 doesn't tend to OC any higher than the 2500. The DFI based nF2 boards are really able to hit the highs with the locked Barton, but they generally spend the extra 5 dollars on a Mobile instead.

The reason that most 2800 cores are not reaching the high scores of the mobiles is that they ae locked. Many people either won't, or cannot OC their boards high enough to take advantage of the CPU capabilities. Those that can would likely spend the extra for the Mobile. This slews results to the lower side, when the average Joe uses the desktop, and the serious OC'erc uses the Mobile.....

 

[Concillian]

Originally posted by: maluckey
The reason that most 2800 cores are not reaching the high scores of the mobiles is that they ae locked. Many people either won't, or cannot OC their boards high enough to take advantage of the CPU capabilities. Those that can would likely spend the extra for the Mobile. This slews results to the lower side, when the average Joe uses the desktop, and the serious OC'erc uses the Mobile.....

FSB below 200 isn't going to have an effect on the ultimate speed of the processor. The OP's overclock is not limited by his motherboard or his FSB, it's limited by his processor. Desktop processors are simply not as likely to hit 2200 as they pretty much once were. Before the mobiles came out I personally know 4 people (including myself) who got 2500+ processors and run them at 2200 MHz.

What you cannot see in that database is the changes AMD has made to their speed binning over time. 2500+ - 3200+ used to get the good cores. Now those cores are going to the mobiles. Not just any AXP can run high speeds at 1.35v, so AMD bins the good cores to the mobiles. The leftovers go to the desktops, and these are simply not the same quality of cores that you used to be able to get in the past.

I think this persons experience is NORMAL for someone who has bought a DESKTOP barton core processor RECENTLY.

You have to ask yourself WHY does the serious OC'er buy the mobile? If the 2500+ processors tended to overclock better, you KNOW the OC'ers would be buying those. They don't... They buy the mobiles, BECAUSE they overclock better. It's partly a self fulfilling prophecy, but at the same time, I have seen a number of reports of poor overclocking from RECENT desktop cores, not just the OP of this thread. I have personally been affected as well, as my recent Desktop 2500+ isn't stable at 2200.

I don't think you'll find a single account of a mobile not stable at the speeds the OP is talking about, yet in this thread there are accounts of two desktop processors that are not... Seriously, have you heard a single account of a mobile not able to reach 2250? 'Cause I haven't.

 

[maluckey]

Sarian121 mentioned a 12.5 x 178 as his limit. Thats a fairly low OC for that CPU, even if it was purchased yesterday, and was a bad stepping. I would seriously look to other reasons for failure of the OC. first thing would be to check the HS/Fan. Even on the Anandtech forums, there are plenty seeing 2.4 Ghz with the 2800. I found these after a very quick WWW search.

Heres a few

This one has someone at 12.5 x 200

Here's a 208 x 12.5

I found others but the trend is that the people with the most experience get the higher OC, and the same people normally go for the Mobile so they can set the multiplier. The Mobile is no magic bullet, just able to idle at a lower vCore. As I said, all Bartons share the same architecture as the desktop. The only difference being that the Moniles idle at a lower vCore. The tend to need the same operating voltages as the desktop versions.

 

[Ketchup]

Originally posted by: Bypolar
Yes the 2800 are not known for there great Overclocks. just find your fastest stable speed and be Happy.
I have never seen a 2800 go Past 2.3G

Wow, that is weird since my desktop 2500 will hit 2.4 GHz. I have to agree with maluckey on this one, I think Sarian's chip can do better. Sarian, even though you shouldn't need to, have you tried raising the chipset voltage?

 

[Sarian121]

Yeah, i have heard of many people with 2800's being able to OC a lot higher than what im getting... I havent been around much to play around with it, so i havent had a chance to reseat the heatsink yet. I just tested it at 12.5x180@1.80v and was able to boot and run it for a while, but then after about 5 mins of testing it, the computer reset.

What i find funny is that i can run Prime95 for 10+ minutes at 178x12.5@1.65v without any problems whatsoever, but as soon as i load a game...it either crashes or the computer restarts.

I also dont understand why poor seating would affect OCing this much. Ive been monitoring the temps often and hasnt reported anything above 47C, even under full load at 12.5x178....so why would that matter. As i said earlier, the Zalman 7000A-CU i got wasnt completely compatible with my motherboard, so i had to modify it a way such that the horizonal alignment might be off by a little bit.

But the thing is, even before i installed the Zalman 7000A-CU and still had the stock heatsink on, it still would have the same problems as i do now.

12.5x178@1.65v = Solid with Prime95, but would choke as soon as i started gaming
12.5x180@1.80v = Same thing as above
12.5x185@1.80v = Windows will not boot. Gets to the loading screen then has a blue error and restarts.
12.5x200@1.80v = I have to reset the bios.

 

[maluckey]

Poor seating will allow the CPU core to be an uneven temperature. These hot and cold spots cause instabilty, even thought he overall temperature (which is measured by thermal diode in the airspace beneath the CPU) remains within specifications.

I have seated my PAL8045 wrong a time or two, and the OC stunk. With a poorly seated HS and my 35 watt 2400+ Barton (capable of 2.7 Ghz on air cooling), I was unstable at 200 x 12. Once the HS was reseated, it can now run cool and 24/7 stable at 216 x 11.5. I feel that speaks volumes about the importance of HS contact and design. Something so small as improperly applied HS compound can affect the OC of a CPU as well. The point is that it doesn't take much to ruin a good OC.

Also, just because your system passes Prime95, does not guarantee that it is stable. Prime95 does not test the video systems at all. For that, I use UT2004 and Gun Metal. UT2004 in particular can bring a "stable" system to it's knees when you turn up all the details to maximum, and enable FSAA/AF.

The voltage that you need to crank it over 195 Mhz FSB (likely 1.9-1.93v in BIOS) with the HS that you have may be too much for it to handle. Most all people with high overclocks on air and the NF7-S tend to use much more efficient (and likely larger and louder) HS. If you must go the quiet route, understand that quiet and air-cooled overclocking are not always campatible goals. Try to keep the temps under 50-53 C. if you can, and you should have enough working room to at least stabilize your setup at 185 - 190 Mhz FSB. NEVER trust the diode measured Core temperatures............That's why I mentioned low 50's C. as a good stopping point, at until you are a little more familiar with the way your board reads temps.

Mark.