WoW. I think ive managed to burn out my Tbird!

[lhgpoobaa]

well strange things were ahoy with my new setup.
i recently upgraded to spiffy corsair PC3200 ram and the epox 8k3a+.

think iis ive experienced a number of bad BSOD's.
and doing some tinkering its become more apparent that my somewhat OVER enthusiastic CPU overclocking attempts have fried my Tbird Core.

not fried as in "take the HS off while running" fried. too smart to do that
but it crashes with and significant overclock, despite the cpu running fine at such speeds over the past number of months.

here is what my old tbird 1200C AXIA can and cant do currently. Temps are kept within spec thanx to my MCX-462.
(remembering that at 166fsb the pci/agp is within spec with the 1/5 divider)

<font color=purple>the cpu has never been particularly OC friendly, despite its AXIA code... behaves more like an AVIA.</font color=purple>

1500(9x166) @ 2.05v or above - stable in idle, divx crashes it, prime95 immidetialy dies with errors and superPI gets calculation errors.

1466(11x133) @ 2.00v or above - no stability. BSOD before windows finished booting. strange considering its slower than 1500@166. multiplier problem maybe?

1420(8.5x166) @ 1.95v - random BSOD every couple of hours with divx encoding.

1400(10.5x133) @ 1.95v - was extreemly stable untill recently. now dies regularly within an hour doing any CPU intensive app.

1333(10x133) @ 1.85v - only overclock state that seems stable anymore. will try 1330(8x166) soon.

so there u have it.
anyone else seen such behavior from a tbird?
is it on its way out?
comments?



My Next Performance System!!! - P4 Celleron, 128k cache, SDRAM, Integrated graphics, 5400rpm HDD!

 

[phsstpok]

I heard that Epox boards draw heavily from the +5 volt line. If you have Motherboard Monitor 5 set it up in log mode and it will record your voltages at regular intervals. (I think 10 seconds is the shortest interval). See how the +5 rail fairs under your tests.

Now, I don't have anything as nice as an 8K3A+ but here is what I have seen with my 8KTA3PRO and Tbird 1.0, AYHJAR.

At 1000 mhz (133*7.5), 1.75 volts, 5 volt rail is at 4.91 volts.

At 1333 mhz (133*10), 1.87 volts, 5 volts rail drops to 4.78 volts.

At 1500 mhz (150*10), 1.87 volts, 5 volt rail is at 4.66 volts.

At 1500 mhz (150*10), <b>1.97</b> volts, 5 volt rail is at 4.58 volts.

Note that at 1500 mhz running something intensive like Toast, as CPU temps go up the 5 volt reading goes down. 4.52 volts is the lowest reading I have seen before my system crashes (usually an auto restart).

By the way, I have an Enermax 330 watt power supply.

<b>I have so many cookies I now have a FAT problem!</b><P ID="edit"><FONT SIZE=-1><EM>Edited by phsstpok on 05/22/02 09:31 PM.</EM></FONT></P>

 

[lhgpoobaa]

i have watched MBM considerably, mostly for temps but i also kept an eye on other factors.

the 5v line behaves itself, mostly due to the GRUNTY 550W Enermax PSU

for low to medium overclocks its around 4.95v, dipping to 4.89 worst case scenario @ 2+ volts on the core + excessive overclock.


My Next Performance System!!! - P4 Celleron, 128k cache, SDRAM, Integrated graphics, 5400rpm HDD!

 

[phsstpok]

Well, I'm out of real guesses.

I've never heard of processor losing some of it's ability to overclock not without dying completely. I would tend to think it has to be something in the new configuration. Have you tried the CPU in the old motherboard just to double-check?

<b>I have so many cookies I now have a FAT problem!</b>

 

[lhgpoobaa]

uh no
the old mobo is completely dead.

anyway.
gonna run for a few days at 1333. then again at the same speed but with 166fsb

must admit it IS strange
but the overclocks i was getting were never fantastic. so ill probably end up using it until i can pick up a mid level tbred CPU.


My Next Performance System!!! - P4 Celleron, 128k cache, SDRAM, Integrated graphics, 5400rpm HDD!

 

[phsstpok]

It doesn't take much of a change to affect overclocking.

My own CPU used to be stable right up to 1500 mhz (and a little higher) but I got tired of the noise from the 60mm Delta fan. I swapped the HSF out for a 70 mm unit the fan of which pushes the same amount of air (supposedly) as the Delta. However, the new HSF raised my CPU temps by about 2 degrees, 2 measly degrees. Now 1333 mhz is about as high as I can overclock.

Since this is more than enough to drive my Geforce2 GTS-V it is not a big deal.

If and when I get a better graphics card it might become an issue (but having only an SDRAM platform will be a bigger issue).

<b>I have so many cookies I now have a FAT problem!</b>

 

[lhgpoobaa]

and yes, going from a sdram mobo (KT133A) to a efficient DDR one (KT333) does give a significant performance boost regardless of cpu Mhz. 17.6 to 20.2 fps as a rough and dirty guide with divx encoding.


My Next Performance System!!! - P4 Celleron, 128k cache, SDRAM, Integrated graphics, 5400rpm HDD!

 

[lhgpoobaa]

nods
well im using the same cooler. same speed fan.
only diff could be the amount of thermal paste, but i always use an extreemly small amount.

if i can be bothered i may reapply the paste... just to be sure

My Next Performance System!!! - P4 Celleron, 128k cache, SDRAM, Integrated graphics, 5400rpm HDD!

 

[phsstpok]

You do Divx encoding? I would think you would be dying to get an Athlon XP and take advantage of SSE and hardware prefetch. I think Aceshardware.com found an XP had a 38% advantage over a similarly clocked Tbird for Mpeg-4 encoding which is basically the same as Divx isn't it?

The most intensive thing I do is play games and I'm not even a real big gamer. Other than this I surf and do a little word processing. I'm also doing a little MP3 encoding but not much of this.

I would be curious to see how an Athlon XP 2000+ (or so) combined with a Geforce4 Ti4200 does on a DDR system vs SDRAM system. Unfortunately, no one is benchmarking old platforms with new graphics cards.

<b>I have so many cookies I now have a FAT problem!</b>

 

[lhgpoobaa]

lol
i would try an XP in my old kt133a board... but i had to hit it with a hammer to get the MCX-462 heatsink standoffs out

generally what i use my system for is

A. cure for cancer 99% of the time & cpu load.
an upgrade to an XP may help a little with hardware prefetch (HP) but it doesnt use SSE.

B. browsing the net, downloading prOn, chatting. bah... could use a p1-100 for these

C. games. all my current games are quite happily satisfied with the cpu, generally the GPU is the limiting factor

D. audio stuff. relies more on memory size & sheer chrunching power.

E. divx. i use xmpeg, its decent, but the idt compression codecs i use are the high quality ones, and dont use SSE. yet. so i still proly wont be using the sse capabilites of the XP.


My Next Performance System!!! - P4 Celleron, 128k cache, SDRAM, Integrated graphics, 5400rpm HDD!

 

[rcj187]

had exactly the same with my tbird 1.2
absolutely stable at 1.4 temps within spec using the stock cooler. but 1.466 killed it. no physical damage but wont even post now... seems that might be pretty close to the architectural limit of the tbird

I'm out of my mind, but feel free to leave a message.

 

[phsstpok]

i would try an XP in my old kt133a board... but i had to hit it with a hammer to get the MCX-462 heatsink standoffs out

Hammer on a KT133A board...that's funny. LMAO!

XMPEG? Never heard of it but I don't follow video stuff at all. How long does it take to process 2 hours of video on your rig? What's needed to play DivX on a home theater system (not PC)?

I had more computing power than I needed when I had my Duron 600@1000+. Upgrading to an Athlon @1.5ghz would have been just a waste but I needed a system for my nephew so he got the Duron.

He and I could both use a new video card. Been waiting for the Ti4200 but now I'm eyeing a couple of Radeon 8500 LE's which have dropped below $100 USD.


<b>I have so many cookies I now have a FAT problem!</b>

 

[lhgpoobaa]

yeah it was pretty funny.

see i use the MVC-462, which has metal standoffs that go thru the 4 holes aorund the socket for mounting
and the kept twisting when i put the HS on, so i used some araldyte (epoxy resin) to keep em in place.
problem is the araldyte was a tad hard to remove!!!! :lol:

acetone loosened it enuf to get the hex nuts off the back of the standoffs,
but the standoffs themselves were firmly glued to the mobo... thus the HAMMER

the mobo was dead anyway... so it was good fun.
recovered the standoffs, soaked them in acetone, scratched off the araldyte and poped them on my new mobo.

XMPEG is an offshoot of flaskmpeg... but more user friendly. with my particular quality settings i get around 18 to 25fps for a movie that runs at 25fps.
so encoding a vid takes more than the run time of the movie.

course i use dual pass settings for better quality, but that takes approx twice as long.
the first pass is faster than the second, so for a 2 hour movie im proly looking at a 4.5 to 5 hour encoding run.
a fair amount of time, but often i just leave my comp running, or surf the net/chat and it churns away in the background... or it did until my cpu died.

turns out i cant even use 1333 @ 1.85v now... a very moderate overclock, so something is definately sick inside my box.

and all my video needs are currently handled by a leadtek geforce 2 pro. maybe ill upgrade to a GF 4 when the 5's come out. but no hurry on the video card.



:lol: Whats better, <font color=blue>Intel</font color=blue> or <font color=green>AMD?</font color=green> :lol:

 

[Bud]

I did the same to one of my Thunderbirds. 1200c axia. kept
pushing it 1400 & +. But it was never 100% stable at 1400 or more. It was 100% stable at 1370 & 1333. But like a tweaker, I would occasionally try for 1400+ again and again. Eventually this burned it out. Not a complete burnout though, just ever more frequent locks ups. At first I thought it was the PSU. Got a new 420w. same lockup problems. Then new RAM. Then new video card. Finally,
I tried UNDER clocking the CPU to 1064....viola....no more lockups. And now that 1200c only runs stable @1024. So i gave it to my dad as a 1024 CPU and it runs fine.

"Get the facts first. You can distort them later." - Mark Twain

 

[lhgpoobaa]

hmmmm
very interesting

your the second person to confirm whats happening.
thanx

so far it still goes acceptably well at 1200, so its not a complete loss.
proly hang onto it untill the tbred, then pass the CPU onto someone worthy who DOESNT overclock


:lol: Whats better, <font color=blue>Intel</font color=blue> or <font color=green>AMD?</font color=green> :lol: