O/C'd CPU works fine... EXCEPT it will not reset/restart. Help?

[Arschloch]

Hello.

First, the pertinent system specs:
Abit KG7-RAID
Athlon XP 1600+ (AGOIA stepping)
256 MB Crucial PC2100 DDR (CAS 2.5)
Gainward GeForce4 Ti 4200
Win98 SE

When running at the default speed (1600+) and voltage, the computer is completely stable.

I have been able to push it up to 2000+ (FSB of 159), and it is stable if I bump up the memory voltage to 2.65V.

The problem is this: if I overclock it to 1800+ speeds or faster, even though I can get it to run completely stable in Windows, the computer will not completely restart. For example, if I select "Restart" from the shutdown menu, the computer WILL get out of Windows and begin to restart, but then the monitor light turns from green to amber, and the computer never makes it back. The same thing happens when pressing the reset button on the case -- the restart begins, then the monitor goes off and the computer doesn't make it into POST.

If I then completely power off and turn back on, the computer boots just fine -- no problems at all.

Has anyone seen anything similar before? Any suggestions? Help.

Thanks.
-Arschloch

 

[paralazarguer]

no video on boot is common with nvidia and VIA combos. If you are able to turn it on and it works, however, I'd say it's obviously a result of the overclock. If you hear a post beep then you are overclocking the AGP bus too much. If you do not hear a post beep then the CPU needs more voltage. When a PC boots or reboots all of the devices in it draw their maximum potention current. An AMD setup like this really isn't good for overclocking. You might want to look into an intel setup where you can just change the 100fsb to 133 and get a 533mhz overclock without brining your PCI or AGP busses our of spec. Going above the default voltage is always a bad thing.

 

[rancherlee]

OR get a Asus Nforce board (dang intel people j/k), the PCI/AGP bus stays at 33/66mhz no matter what the FSB/Ram speed is
My 1600xp+ is running 2134+ @ 1747mhz with 1.8V totally stable and COOL, never hits above 50*C on a HOT day (AGOIA)
Your Memory may be hang you up on the reboot, I'm running pc 2700 on mine and its at a 167fsb/167ram setting (334DDR)

 

[Arschloch]

Thanks for the responses.

I am not getting a beep at all on reboot. So based on what you said, 7757524, the problem is that I'm not getting enough juice to my CPU. I suppose I could bump the voltage up a bit. How much is too much? Is 1.9V too much? 2.0?

And yes, I also thought that the memory might be a problem as well, since after all it's only rated to 133 MHz. But it has no problems at all except for resetting, so I assumed the memory wasn't the problem here.

Outside of getting new parts, do you have any suggestions of things I can try with my current setup?

Thanks.
-Arschloch

 

[Barrei]

Thats a classic symptom of lack of cpu voltage. As far as anything over default voltage being a bad thing that is complete nonsense.

 

[paralazarguer]

If you are OC'ing your FSB then your ram is also overclocked. It could be the processor or the ram. You will likely have to raise the voltage on both. Try raising it a little on the ram or use less agressive settings and see what happens.

 

[paralazarguer]

Okay, Barrei, increased voltage makes it hard for the CPU to determine the difference between a 1 or a 0. As transistors get smaller, voltage must also or the signal changes will get confused (open of closed gates) It was fine to have 5 volts on a Pentium 1 for example because it used such large transistors. Using that kind of voltage on a .18 r .13 would not work. Raising the voltage on a process can shorten its life drastically. You're a fool if you think there's no consequence to raising your voltage. It can lead to stability issues (corrupt data,) extra heat, and lowered lifespan.

 

[paralazarguer]

For your education Barei, any newbie overclocking guide like this one has information about the dangers of raising your voltage.
"Too much voltage and you run the risk of permanently damaging the processor."

here too:

"However be VERY CAREFUL whenever you vary the voltage and only increase in small percentages as too high a voltage may possibly damage your processor."

another one

" Increasing voltage on any CPU is very dangerous, in fact, it's even more risky than overclocking. Proceed with extreme caution!"

"Too much heat and too much voltage are the most frequent methods of frying components.
-"OVERCLOCKING BASIC TRAINING" www.overclockers.com

 

[paralazarguer]

here's an entire article which discusses voltage tollerances in CPUs and how running above voltage spec can lead to the same stability issues as running fsb above spec etc.

 

[paralazarguer]

"However, additional voltage (power) causes the circuits comprising the processor core to produce additional heat. This heat must be dissipated in order to prevent damage to the processor core, as well as other components. Temperature concerns aside, applying high voltage across the processor core can lead to electromigration whereby electrons burn new paths through the processor core, bypassing those constructed by the manufacturer. This phenomenon alone can render a processor useless."

 

[Barrei]

I don't think the issue is that you shouldn't go over default voltage but how MUCH you should overvolt your cpu , many people that I know overvolt there cpu's and have been doing so for a LONG time and there computers still run fine , if you've read the posts in this forum most of the overclockers run well above the default cpuv with no problems as long as you don't go above 1.75volts on the p4 it won't be a problem , until I see alot of the overclockers processors dying from running above default voltage all the articles you qoute from are meaningless.

 

[paralazarguer]

It's not nonsense as you say. Obviously. I hope that you stop making false blanket statements. The voltage is already so high on the higher end athlon XPs because they're pushing them so hard I wouldn't go very high above it. They're pretty close to tollerance. Might lose your processor in less than a year. It happens.

 

[paralazarguer]

They're, there, and their

 

[Arschloch]

I haven't been one in the past to subscribe to the "burning in" theory, but...

This CPU is definitely OC'ing better with each passing hour. Right now, I'm running at 1909+ (10.5x153) speed, and I actually bumped DOWN the CPU voltage to 1.725. (Isn't the default voltage 1.75V for XP's?) I've tried rebooting a few times, and it has worked at this speed.

Idle CPU temp (as measured by the diode right by the CPU on this motherboard) is in the upper 40's C, and under full Prime95 load it's about 57 C. I'm aware that different mobos read differently -- some report higher temps than others. Based on the info that I have an Abit KG7-RAID, are these temps acceptable?

Another question -- my case is an Enlight 7237. I have one case fan (in the front of the case), and the PSU supply, of course. I'd like to get another case fan to mount in the back. However, I have no idea where I'd be able to mount a fan in this case. Suggestions? Seen a fan-mounting guide?

Thanks.
-Arschloch

 

[Duvie]

I do agree any degree of overvolting starts shortening the life of the processor...shortening the life as described by the manufacturer that is...For many this may be too much time!!!Like me who upgrades at least 4-6 months....

However I fully accept and understand running my cpu at 1.66v will not run possibly to the stated lifetime from Intel. I accept that and take my chances...

As for Arschloch,

You got yourself a fine chip there...Many of that particulr stepping is doing quite fine...If you can undervolt it to 1.725v and not have any adverse affects to stability by all means do it. Please take voltage set in the bios with a grain of salt as often mobos will add or take a little from it.

The temps look ok...BUt, with socket a temperature readins it is hard to know the true accuracy and comparing to other motherboards is down right futile....A handful of the newer boards read the internal diode and give true core temps. Who knows where the temp is bening read on yours so the accuracy can always be called into question. Just be watchful for instability often related to heat as the summer progresses. If you are stable then heat isn't much of a concern...

I recommend you run and test it as you will use it...If you play games for 2-3 hours or encode divx movies then run a program like prime95 for 3 hours plus to test stability...More games throw a looping in of 3dmark2k1...sable is not merely starting and shutting down and letting it sit idle with some repeated sandra benches...

As for cpu burn in!!! Don't belive in it...Either chip does it or not. resarced this a bit and the more technical sites and reports also disagree with the science of it.

Personally I don't like sandra burn in for a stability test as well...I have seen it not detect errors when prime95 or a divx encoding will find it....I don't think highly of this test...

 

[paralazarguer]

The CPU was burned in at the factory. It was tested HARD. As for you questioin :isn't the athlon XP voltage 1.75?" No...on the new .13u 1700+ it's only 1.5 on the 2200+ it gets up to 1.8 I believe. AMD has to do that. Basically, the processor comes overclocked from the factory.

 

[speed01]

According to AMD, The Palomino chips run at 1.75v while the Tbreds run at 1.5 to 1.65v. None go to 1.8v.

Go to AMD's Self Service Tech Support page and type in "default processor voltage" if you would like to see the breakdown....

 

[paralazarguer]

I find it unnerving that AMD has to now raise the voltage from one speed bump to the next. Traditionally Tbirds have had 1 voltage as well as classic athlons, the p3, and the current P4s....

 

[Arschloch]

Thank you for all the helpful comments.

I'm still running well at 1909+ (1606 MHz), but now I bumped down the CPU voltage to 1.675V. I hope I can work out the kinks in my system so I can see how far this CPU can REALLY go!

Actually, I made another improvement: I reversed the front case fan. Whereas before I had it blowing out, now it's sucking air in (the PSU fan is blowing out the back). Before, it was at 56-57 C at 100% load using Prime95); now, it's down to 51 C. I can live with that.

All signs are positive with power/voltage: Motherboard Monitor reports the +5.00 V circuit at 5.15 V.

Next improvement: the front case fan is not blowing air on to the CPU. The CPU is on the clear opposite corner of the motherboard (next to the PSU). So, I'm going to build some kind of a duct (perhaps made out of cardboard) that will direct some of that air to the CPU. Has anyone done anything similar to this?

Thanks again, guys.
-Arschloch