201mhz FSB?

[ZeroRift]

Okay, up front, I'm a complete n00b with overclocking, so I need some ideas on an issue I'm having. I dropped a line in the tech support forum, but no bytes after a week. So maybe you guys can help me.

I'm getting random crashes to desktop running games. most crashes occur after 10mins of gameplay. I noticed in the critical error messages that my CPU clock was listed as 2415mhz. However, my BIOS tells me that my FSB is 200mhz and my CPU multiplier is 12x. I ran everest home edition (just cause I had it layin around) and it said that my FSB was correctly 200 mhz but my memory clock was at 201mhz, thus my CPU was running at ~2412mhz. I was wondering if this clock issue could be causing my stability issues. I've already looked at drivers as a possible issue. I'm currently running a fresh install of win xp pro sp2, and the system has not been hooked up to the internet yet. Speedfan and my BIOS say that my tremps are well within tollerance during gaming: ~60C processor core and ~70C gpu at the highest.

So, you guys being experienced with clock settings, do I need to address this as a possible problem, or should I look for something else as the culprit?

Here's the build:
AMD64 newcastle 3800+ (fixed mult. 12x; 200FSB)
2x 512 Mushkin blue pc3200 (201mhz?)
GA-K8N Ultra-9
GeForce 6800U PCIE
FSP Bluestorm 500w ATX2.0
seagate 200gb SATA

Any help is appreciated. thanks.

 

[theMan]

have you tried cpu-z ? i dont think that could cause those issues. whats the voltage set at?

 

[ZeroRift]

I can't really download CPUz right now

CPU voltage = 1.50
although my Vcore A sensor says 1.55
Ram voltage = 2.60

 

[SNM]

I don't know what the tolerances are for a 6800, but from what you've said it sounds like the vid card's overheating.
Or else it's a software issue having nothing to do with your hardware.

In any case, I promise an "overclock" of 15 MHz on the cpu and 1 MHz on the ram won't do anything.

 

[ZeroRift]

well, according to my GPU drivers, anything up to ~125C is fine. And after that the GPU is supposed to automatically downclock and notify me that it is overheating to prevent damage. I only ask about the clock because I'm running out of options. I thought such a small up would not cause stability issues, but it bothers me that I can't set it back to be in sync with my FSB. I guess I'll keep looking at software.
thanks.

btw: Is it typical for ram to have odd stock timings these days?

 

[imported_michaelpatrick33]

125C? Are you kidding? 257 degrees? You must mean 125F. That board would melt quite quickly at 125C. It does sound like a video card overheat or lack of psu power (which I doubt judging by your sig psu)

Edit: Many motherboard makers overclock their "FSB" by one or two points to gain a lead in benchmarks from reviewers. That little of an overclock would definitely not cause ram or the cpu to cause problems unless you have ram timings that are too tight to begin with. I would make sure there is no overclocking occuring on your video card and take the side off the case and see if the problem persists. If it doesn't than you know it is heat/video card related.

 

[ZeroRift]

seriously, I mean 125C. I didn't believe it eaither, but the driver says Celsius and the scale goes to 125. eaither way, 67F is too cold to be true as all of my cooling is stock, and my card is know to be one of the hottest.

I'll try taking the case off when I get home anyways.

It bothers me that mobo manufacturers do this. It's unfair to the consumer to lie about actual FSB freqs just to get ahead.

I have run no software that could be overclocking my GPU (as far as I know) and my PCIE bus is set to default at 100mhz.

 

[imported_michaelpatrick33]

67C is much more reasonable under load. I think the driver screwed up the Celcius and Fahrenheit letters. Anything above about 70C and you are entering dangerous territory for the gpu.

67F is only 20C. Obviously that isn't your temperature under load.
67C is 152.6 degrees. That is approaching the top end temperature tolerance.

They screwed up the C and F symbols. Be careful.

Edit: They had to mess them up. There is no physical way a piece of silicon can run at over 250 degrees Fahrenheit. 80C I could see possibly, but 125C? No way.

 

[ZeroRift]

sorry I rounded

I'm going to do some reasearch with that, because it seems odd that such a basic and obvioius problem would go unnoticed after as many updates as nvidia does with their drivers. not to mention how much damage this would do to their marketing if word got out that nvidia tehcs don't know the difference between C and F.

 

[imported_michaelpatrick33]

Originally posted by: Zero RIFT
sorry I rounded

I'm going to do some reasearch with that, because it seems odd that such a basic and obvioius problem would go unnoticed after as many updates as nvidia does with their drivers. not to mention how much damage this would do to their marketing if word got out that nvidia tehcs don't know the difference between C and F.

Maybe the put up to 125C just as an example to allow overclockers a higher maximum setting before throttling occurs to keep damage from occuring to the card.

For example, many motherboard makers set the maximum "FSB" to 450 knowing that it won't ever reach that level but it looks good in marketing.

 

[ZeroRift]

hmmmm. it seems as if most people are running their 6800s at peak temps of 64C. My case has plenty of passive venting (2 vents on the side and one huge one on the back), but I haven't installed any additional fans. Currently the only fans in my system are: The CPU fan, The GPU fan, and the huge freakin ps fan that moves a lot of air. I guess I could also check my ambient temps, although, like you said, the best way to test is to take the cover off and get some more air flowing.

Regardless of any marketing reasons, if the GPU will melt at 120C (which is the correct driver max) it's lying to make a scale that goes that high and say that the GPU will only die near 120C.

I suddenly have a great distaste for the computer business. The last time I bought a new system was when the gf 3 ti series was brand new. I guess some things have changed since then, besides the new technology.

Thanks for all your help, I'll post again tomarrow if overheat is not the issue.

 

[imported_michaelpatrick33]

Originally posted by: Zero RIFT
hmmmm. it seems as if most people are running their 6800s at peak temps of 64C. My case has plenty of passive venting (2 vents on the side and one huge one on the back), but I haven't installed any additional fans. Currently the only fans in my system are: The CPU fan, The GPU fan, and the huge freakin ps fan that moves a lot of air. I guess I could also check my ambient temps, although, like you said, the best way to test is to take the cover off and get some more air flowing.

Regardless of any marketing reasons, if the GPU will melt at 120C (which is the correct driver max) it's lying to make a scale that goes that high and say that the GPU will only die near 120C.

I suddenly have a great distaste for the computer business. The last time I bought a new system was when the gf 3 ti series was brand new. I guess some things have changed since then, besides the new technology.

Thanks for all your help, I'll post again tomarrow if overheat is not the issue.

It isn't real different from other industries' marketing FUD. Look at my S10 Chevy pickup truck. It's odometer goes to 140 m.p.h. yet it has a cutoff at 98 or so (found out once on a desert highway lol). Why the high odometer. It looks good in the dash.

 

[ZeroRift]

yes, but your odometer is not telling you that your are going at 15mph when you are really going at 20mph just to make the ride more thrilling.

 

[imported_michaelpatrick33]

Originally posted by: Zero RIFT
yes, but your odometer is not telling you that your are going at 15mph when you are really going at 20mph just to make the ride more thrilling.

You'd be surprised, LOL. Just kidding. The temperature isn't higher than what they are claiming or their driver isn't working right. NVIDIA wouldn't lie about temperatures in the graphics card or they would face a public relations (and business relations) disaster. Under load I would imagine your card hits the mid 60C level. I still think the 125C mark is just an odometer that goes to 200 with a speed throttle at 100. Looks good, but is laughable in execution.

 

[ZeroRift]

Originally posted by: michaelpatrick33

The temperature isn't higher than what they are claiming or their driver isn't working right. NVIDIA wouldn't lie about temperatures in the graphics card or they would face a public relations (and business relations) disaster. Under load I would imagine your card hits the mid 60C level. I still think the 125C mark is just an odometer that goes to 200 with a speed throttle at 100. Looks good, but is laughable in execution.

I was referring to my FSB. I'll see if I can get my card to run in the mid-low 60s tonight.
Thanks again.

 

[Continuity27]

Originally posted by: michaelpatrick33
67C is much more reasonable under load. I think the driver screwed up the Celcius and Fahrenheit letters. Anything above about 70C and you are entering dangerous territory for the gpu.

67F is only 20C. Obviously that isn't your temperature under load.
67C is 152.6 degrees. That is approaching the top end temperature tolerance.

They screwed up the C and F symbols. Be careful.

Edit: They had to mess them up. There is no physical way a piece of silicon can run at over 250 degrees Fahrenheit. 80C I could see possibly, but 125C? No way.

You are very wrong sir.

125 - 135C is not good territory for 6800s, but it is indeed very doable. These are the set THROTTLING temperatures. I have personally had my video cards at over 100C with no stability problems whatsoever. Not that you'd want it that hot, but they can definately take it.

 

[ZeroRift]

actually, now that I think about it, you're right, continuity. I have an old amd 1.4ghz t-bird and it runs at ~145F under load. 150F would not be a far reach for a 6800U under load, however I would be worried about my card at 100C, especially with stock cooling. I'll still try the open case thing, who knows, maybe my drivers are reporting bad info. But if I'm not overheating, wtf?

P.S.: The melting temp of silicon is ~1000C (1600K) the bricks in my house would melt beofre the silicon in my GPU would. I would be concerned about the rest of the board though. I don't know what boards are made of, but the board would likely melt before the silicon in the chip would.

 

[bozicek]

Does this happens only when playing games? I belive that temperature of videocard is the problem.
Me have 6600GT; normaly 48-52C; then I OC-it. And here is the problem. I played Tony Hawk U2 and it crashed. Take a look at the temperature and it was 72C. When I reached 70C on GPU of a card it crashed.
Try: Put 2 vents (say 80mm) on bottom and upper side of a videocard for same fresh air to flow on videocard (have the case open). Play games. If they dont cras, temperature of a core is your problem

 

[Continuity27]

Originally posted by: bozicek
Does this happens only when playing games? I belive that temperature of videocard is the problem.
Me have 6600GT; normaly 48-52C; then I OC-it. And here is the problem. I played Tony Hawk U2 and it crashed. Take a look at the temperature and it was 72C. When I reached 70C on GPU of a card it crashed.
Try: Put 2 vents (say 80mm) on bottom and upper side of a videocard for same fresh air to flow on videocard (have the case open). Play games. If they dont cras, temperature of a core is your problem

Keep in mind, it's not always the video card that's responsible for the cras hitself, at least not directly. When the video card gets hotter, other components like the northbridge may be affected by the excess heat, and THAT may be what's failing. I just can't see how my video cards (2 of them even) were able to handle 100C while yours cannot. More likely, I think the issue is another important component that can't handle the heat like the video card can.

 

[imported_michaelpatrick33]

Originally posted by: Continuity27

Originally posted by: michaelpatrick33
67C is much more reasonable under load. I think the driver screwed up the Celcius and Fahrenheit letters. Anything above about 70C and you are entering dangerous territory for the gpu.

67F is only 20C. Obviously that isn't your temperature under load.
67C is 152.6 degrees. That is approaching the top end temperature tolerance.

They screwed up the C and F symbols. Be careful.

Edit: They had to mess them up. There is no physical way a piece of silicon can run at over 250 degrees Fahrenheit. 80C I could see possibly, but 125C? No way.

You are very wrong sir.

125 - 135C is not good territory for 6800s, but it is indeed very doable. These are the set THROTTLING temperatures. I have personally had my video cards at over 100C with no stability problems whatsoever. Not that you'd want it that hot, but they can definately take it.

You are right and I was wrong about the specification. Good correction of my earlier post. I haven't seen anyone who has the 6800 series except you get reliable function out of the video cards over about 70C on a consistent basis. Over 100C seems way to high for a graphics card to run reliably for any stretch of time.)

Maybe you can help explain a disparity to me? temperature differences in review between Nvidia gpu probe and in use tempature of 38C?

The review shows the GPU probe reading 100C but the in game full usage temperature reading of 62C (seems more normal in reading). Are they measuring two different parts of the core or are they averaging the temperature of the whole card? It doesn't make sense to me.

 

[sangyup81]

many BIOSes don't even bother with 201 fsb and just skip to 202

 

[ZeroRift]

Okay, It's been a while, but I had to be at least 90% sure of my solution before I posted it.

Here's what I did:
I took the case off my system and ran it. My fail time changed from 5~10mins to ~25. So I put a house fan in fron of it, just to see. My fail time went up to ~2hrs. Obviosly an overheat issue. To see if it was my GPU, I enabled 8xAA and 8xansio. Still no overheat issues. So I guess it must be my CPU. I ran speedfan, and my peak temperatature was ~60C beofre I'd crash. Suddenly, I realized my error. I had re-set my BIOS to fail-safe defaults not to long ago, and I forgot to set my CPU stop temp to something other than 60C. Doh! I set it up to 70C, (since I'd rather my CPU not run hotter that ~180F anyhow) and suddenly everything's stable. YAY! I also bought a 120mm fan to put in my case, and now my GPU maxes at ~60-63C at the hottest, but typically runs (under load) at about 57C. My CPU maxes now at ~155F.

So far, aside from a few minor bugs in the system, I've run KOTOR1 for ~7hrs, and played all the way through Halo without issue. UT2k4 has some issues, (varying "critical errors") but I believe they are configuration-related.

Thanks for all the help and input, and special thanks to michaelpatrick33 for suggesting the case-off solution.