ICE ICE baby :) (low temperature topic)

[Shodan486]

Re: ICE ICE baby (low temperature topic)

Still - having two pieces of CPU chips on a single motherboard is sexier than more cores on a sole die. :P

[paul_h]

Re: ICE ICE baby (low temperature topic)

Yeah, like I said, the dual p3 with some nice HSFs on it looked cool.

[Scenic]

Re: ICE ICE baby (low temperature topic)

indeed.. got a recapped Gigabyte 6VXD7 here. (Specs)
maxed out as far as i could
- 2x 1GHz PIII Coppermine
- 4x 512MB PC133 Kingston
- Geforce 6800GT (slightly overkill )

has 2 Arctic Cooling Copper Silent II TC on the CPUs, and the 6800GT has the Arctic Cooling nVSilencer 5 on it. combined with a 350W Delta w. 120mm fan -> VERY quiet..

edit:



edit2: forgot to mention... this board also had the crappy Choyo 25V 330uF that Gigabyte liked so much back then EVERYWHERE except CPU VRM (Sanyo) >.<!

[paul_h]

Re: ICE ICE baby (low temperature topic)

Yeah the GB 6vxd7 was the mobo I had. Was OK for the most part, but a little funny.
One of the caps near the RAM was swollen, one of the caps between the PCI slots was too.
It was me first attempt at a recap, but ever since I went near it with a soldering iron, it hasn't posted or worked properly so I killed it more or less.
Just before I killed it I bought 1.5GB SDRAM and 1Ghz CPUs (only had 866mhz CPUs and 756MB RAM before), what a waste.

[Shodan486]

Re: ICE ICE baby (low temperature topic)

Hi guys!

After a while of rapin' my system, I come back to you with my latest findings (and strange behavior). Again, I need to highlight the fact I live in Slovakia, which was supposed to be in a medium weather conditions, meaning no serious temperatures or humidity...but the climax changed and now we have 30 degrees Celsius under shade and 40 right on the direct sunlight. That's just the environmental presumption...

When we browse this thread (and more from me concerning the VP6), we can see I was able to reach a stable level of operation with my system...I was able to play Fallout 3 on it for 4-5 hours without any glitch, fall, BSOD...under these settings:

VCore - 1.85v (I'm firmly sure that's OK)
VCC3.3 - 3.4v
Vtt - 1.48v

MPS: 1.4
SDRAM timing: 2-2-2-5 CL2
Chipset buffer delay : 0.5ns
AGP Driving Value : Auto (improves stability when set to the highest values, from 00 to FF, but it depends on the VGA used)

So these are the main settings which affect the overall stability and performance, mainly the VCC3.3 and AGP driving value. So let's get to the point:

I'm playing fallout3 for about max 45. min and I get one of these : just a regular windows error, after which I can run the game again and play it for just as long as after the boot up OR just a simple hard reset...Occasionally I get a pure freeze, the game holds still till the reset button pushed - that I can tell for sure is a temperature problem (had this phenomena for a while, suspected some temperature issues, solved by replacing the chipset heatsink). By these experiences, I have the following questions for you, mainly theoretical:

1. Could I relate this behavior to the rise of ambient temperature? (2 months ago everything was fine due lower degrees) Can I suspect the MX-2 thermal paste having some air bubbles created within it from these temperatures? Or can any chip be overheated from the bottom side (where soldered to PCB)?

2. Mainly for the best experienced people of you: Can I connect this issue to the caps? Why? Of course due to the sudden temperature/humidity change, though it's relatively dry in Slovakia. OR have I ''raped'' the caps by frequent changes of VCC3.3v/VCore values?

[Scenic]

Re: ICE ICE baby (low temperature topic)

i'd say the system was running very close to it's (thermal) limits..
the change in ambient temp may have pushed it over the border.. one of the CPUs doing wrong calculations, leading to crashes.. (might even be the chipset or RAM..)

i've had that happen before with an Athlon XP Palomino 1700+ that ran at the thermal limit (90°C and slightly over) for some time (clogged heatsink/fan)..
even after cleaning everything, i get VERY strange windows errors or drivers randomly not working..
i've tried to run Prime95 with that CPU.. sometimes it worked fine for almost 14 hours.. sometimes i get calculation errors after as little as 30mins..
that CPU is more or less toast. i still have it though.. using it for testing unknown boards as it does work, but isn't quite calculating right anymore due to the overheating.. (Pentium1 FDIV bug anyone? )

[Shodan486]

Re: ICE ICE baby (low temperature topic)

Thanks Scenic, you reminded me I forgot to put som temperature readings

CPUs : 35-45 celsius
Chipset : lower than 30 celsius
GPU : 45-55 celsius
HDD : 40 celsius max

And I have this stuff inside the HAF932, there are 10 active coolers installed ...I guess the flow of warm breeze won't help either.

[Shodan486]

Re: ICE ICE baby (low temperature topic)

Cause identified - loosened heatspreader on the chipset ...running fine at 172,5MHz all the time...but those 173MHz aren't gonna happen again I guess.

[RJARRRPCGP]

Re: ICE ICE baby (low temperature topic)

So I've tested this system so far and it seems all OK. By knowing that if aggressive timing settings & high clock frequency is bestowed upon the memory modules, I can say that if there is no single error from the start of the computer, no BSOD, no random reboot - there is nothing to worry

---------------------------------------------
Start
Successful OS start (WIN XP Professional SP2)
Successful Battlefield 2 game start
Successful Loading map
10 Minutes playing the map
End
---------------------------------------------

You may be speaking too soon, watch out when you run Prime95! You're likely gonna get an error or a PC crash! Usually, NOTHING beats Prime95!

[Shodan486]

Re: ICE ICE baby (low temperature topic)

You may be speaking too soon, watch out when you run Prime95! You're likely gonna get an error or a PC crash! Usually, NOTHING beats Prime95!

You're right definitely, but this particular issue involved loosened chipset heatsink, which resulted in immediate hang-ups and so, not even 2 minutes and the system would fail. Now I attached the heatspreder a little more carefully - the results were immediate - no worries, played fallout yesterday for 2 hours - no problems...Even managed to play it at 173,5MHz - finally found the best BIOS combo setting.

And now, folks, it's time for some Prime95 - now I feel the machine is ready for this torture .

[Per Hansson]

Re: ICE ICE baby (low temperature topic)

Prime95 is childs play, try passing the Intel Linpack tests, Linx is a user friendly way to run it;
http://www.xtremesystems.org/forums/...d.php?t=201670

[RJARRRPCGP]

Re: ICE ICE baby (low temperature topic)

Prime95 is childs play, try passing the Intel Linpack tests, Linx is a user friendly way to run it;
http://www.xtremesystems.org/forums/...d.php?t=201670

I seen Prime95 fail at 5 hours or about that, after a short Linpack run. So 30 minutes in LinX don't always equal 24 hours of Prime95!

[Per Hansson]

Re: ICE ICE baby (low temperature topic)

Well, I've had a computer run 48h Prime95 but fail after 15mins on Linpack
But you need to use heavy settings in Linpack if it's really going to test the hardware (this is where Linx comes in, makes it user friendlier)

[Shodan486]

Re: ICE ICE baby (low temperature topic)

SO!

Ran some P95 test - it fails within 2 mintues , once even hard-reset. Solution was quite easy - jumping from 1.85v to 1.9v for VCore. That resulted only in one failure (probably caps needed to familiarize with the higher value of voltage), tried once again and held stable for 45min. for the first test...will try more when time available.

BTW, been on a LAN party, which provided a great environment for testing (both for logic & hardware) - Playin FEAR combat quite stressed all HW in there and we even had some air conditioning, which I do not have at my place - NO crashes for 1 & half hour when the Vcore was 1.9v, on the other hand too many crashes when 0.05v lower.

[Phaihn]

Re: ICE ICE baby (low temperature topic)

I put my PC outside, to my balcony, where it could be like -2 Celsius and the computer acted more stable. So it looks it likes it ...still I don't like the AGP problem - bus frequency raised to 86MHz and its voltage is 3.6v, though cards I'm testing (HD2600XT & N7600GS) only require 0.8v, but anyways the Rialto & HSI chips may go hot very soon, mainly the ATI converter gets pretty hot and eventually hangs up sooner than the GeForce HSI. I only need an 6800 class VGA to prove my self wrong...I'm definitely leaving this problem and solution to mother nature

ive done something simular to that i just used bottles to rout air from out side my window to the intake fan on the front of my computer.

[Shodan486]

Re: ICE ICE baby (low temperature topic)

Okay, that's enough!

Just finished my testing. To summarize my findings briefly, I'll just provide very simple info extracted from my thorough testing procedures:

-solidly usable under heavy stress at 172,5Mhz FSB

-unusable under any stress at 173,5MHz FSB, although normal usage is OK.

-ROCK SOLID stable at 166MHz FSB

How I determined rock solidness ? Playing games solely isn't sometimes much of a stress, but running Prime95 + playing is quite reliable when speaking of legit results - the best combo (and playable ) seems to be P95 and Flatout2. Flatout isn't a stresser at all, can play it FullHD with highest details at 45FPS, but adding some pressure to CPU or to the whole motherboard showed BSODs, hang-ups, WIN errors etc.

So right now I'm using the rig at the 172,5MHz normally, playing games, running videos, youtube....all ideal unless some HEAVY math stress tests with SuperPI / P95.

So what's wrong? Well definitely Sanyos are quite strong caps and sustain a lot of raping , but I think I've crossed the line - seems the electrolyte is not as good as it seemingly used to be, 10 months ago it was easily possible to run anything at 173,5MHz, now I cannot fully rely on 172,5MHz, but it's working so far, but I guess not for long.

Solution?

R.E.C.A.P

[Shodan486]

Re: ICE ICE baby (low temperature topic)

Well Need to say I've failed...aaaand this is a nice one to admit.

The research done upon all the failures causing various behavioral occurrences and / or the BSOD / cold reboots was a bit preliminary as expected. Simply put I jumped into the complex stuff before doing some basic checks. What I mean is that I've found the core problem - the machine was behaving differently at distinct places (my flat, workplaces, the balcony) and also erratic behavior has been observed rather in summer than winter. So this lead myself to conclude the fact that it still should be the temperature issue, not the caps condition, ripple or ESR...it wasn't it, it was the temperature. And it still has been the temperature - I'll try to make a shot o my RAM modules and point directly at the responsible part for all of my problems. It's some sort of a chip down on the RAM stick and it fires up when games hungry for memory are run. I was just browsing the mobo with my index finger and trying all the chips on it - found none. Then tried the memory when I suddenly got burned by this bug, so I tried putting a huge 90mm fan on the 2 of my sticks just for the hell of it.

Not only I've managed to acquire TOTAL stability (173,5MHz) at my workplace (we don't heat up), but experienced 40 minutes of playing fallout NV at FSB 175MHz, then somebody cranked up our radiators. Anyways I got IT! I bought some AC heatspreaders, so they should eliminate the need of the fan usage.

Guess that's all, this is my Christmas present this year, just wanted to share a little of progress I've made. Happy New Year everyone and best of luck!

Me

[Shodan486]

Re: ICE ICE baby (low temperature topic)

Solution?

R.E.C.A.P

Please ignore