Ratdude's Main rig V3.5

[ratdude747]

Re: Ratdude's Main rig V3.5

my main concern with a burn in period is not being able to use the system for 200 Hours... is that the case?

[rhomanski]

Re: Ratdude's Main rig V3.5

You have to use the system, the heat is what burns it in. Just no heavy lifting for 200 hours.

[ratdude747]

Re: Ratdude's Main rig V3.5

You have to use the system, the heat is what burns it in. Just no heavy lifting for 200 hours.

What if simply tuning it on is heavy use? Or so it seems to be...

[mariushm]

Re: Ratdude's Main rig V3.5

You're exagerrating the importance of burn-in... it's really not rocket science... the paste needs to heat up just like it would heat up normally...
Just don't run occt or cpu benchmarks for 24 hours in a row or something stupid like that.

In the normal state, it's a paste and as it heats up it goes liquid and enters those microscopic imperfections in the metal, trapping whatever air atoms (or whatever) are still in the middle. So basically, you no longer have air between both metal surfaces.

You should know but just in case someone stumbles on this post in the future...

Just apply the thermal paste the proper way... clean both surfaces with isopropyl alcohol (or regular sanitary alcohol), dry the surfaces, put less than pea sized drop of paste in the center (or a thin line if the die is very big) and press the heatsink on top of that drop.
As you press, the drop flattens and pushes the air between cpu and heatsink out.

Don't put paste and spread it over the whole surface and then apply the heatsink.. if you do that you just trap a bit of air between paste and heatsink.

It doesn't have to cover the whole chip die and overflow, as the paste gets hot and more "liquidy" it spreads by itself under the heatsink.

Basically what this guy says : http://www.youtube.com/watch?v=-hNgF...ature=youtu.be

[ratdude747]

Re: Ratdude's Main rig V3.5

^ I take it you meant thermal paste not solder paste... I'll edit in a fix if you want.

Noted on not needing to cover. Never knew that... Duh. Noted.

Orderign some AS5 then.

edit: getting it in bulk: http://www.newegg.com/Product/Produc...82E16835100008

Since I'll be redoing a lot over time (when laptops are due for cleaning, V2's CPU's could use a redo as well.). I figure it doesn't expire that quickly so it's probably good to have around.

[Sparkey55]

Re: Ratdude's Main rig V3.5

You're exagerrating the importance of burn-in... it's really not rocket science... the paste needs to heat up just like it would heat up normally...
Just don't run occt or cpu benchmarks for 24 hours in a row or something stupid like that.

In the normal state, it's a paste and as it heats up it goes liquid and enters those microscopic imperfections in the metal, trapping whatever air atoms (or whatever) are still in the middle. So basically, you no longer have air between both metal surfaces.

You should know but just in case someone stumbles on this post in the future...

Just apply the solder paste the proper way... clean both surfaces with isopropyl alcohol (or regular sanitary alcohol), dry the surfaces, put less than pea sized drop of paste in the center (or a thin line if the die is very big) and press the heatsink on top of that drop.
As you press, the drop flattens and pushes the air between cpu and heatsink out.

Don't put paste and spread it over the whole surface and then apply the heatsink.. if you do that you just trap a bit of air between paste and heatsink.

It doesn't have to cover the whole chip die and overflow, as the paste gets hot and more "liquidy" it spreads by itself under the heatsink.

Basically what this guy says : http://www.youtube.com/watch?v=-hNgF...ature=youtu.be

Yeah, Do not spread the paste out in thin layer, very bad. Always apply the paste in globule form at center of core and let the heatsink spread it. A lot of people are misinformed about this.

[Per Hansson]

Re: Ratdude's Main rig V3.5

You're exagerrating the importance of burn-in... it's really not rocket science... the paste needs to heat up just like it would heat up normally...
Just don't run occt or cpu benchmarks for 24 hours in a row or something stupid like that.

Don't worry about running OCCT or any other benchmarking programs.
We are talking about a temperature difference of 30°C if using a "normal" heatsink.
Not a problem in any way, actually just good because it allows the paste to creep out, the hotter it gets...
What they mean with the "burn in" period is that when that time has passed the paste will have settled and offer maybe 1 or 2°C lower temps than before it had settled completely.

[rhomanski]

Re: Ratdude's Main rig V3.5

When I got the Thermaltake water cooler a few months ago it already had paste applied. Spread out all over the bottom. When I do output transistors, I always cover the whole backside of the transistor. I don't think it matters whether it's spread or put in a blob. Just so it covers the whole contact point when mechanically fastened.

[NUKLEAR-SLUG]

Re: Ratdude's Main rig V3.5

Aye, the only way you're going to get air bubbles between a heatsink and cpu shield is if there's a mismatch between the surfaces anyway in which case your paste isn't going to be doing a lot to help either way. So long as you put a thin layer on it should be fine, just not smear it on like toothpaste.

[shovenose]

Re: Ratdude's Main rig V3.5

Burn-in periods are over-rated.
Temps will go down a couple degress - your CPU won't overheat before then just run that much hotter which is really just a little!!

[ratdude747]

Re: Ratdude's Main rig V3.5

Just a minor update:

a month and a half ago I redid the GPU and CPU heatsink compounds... with AS5. Having gotten my 200 PoH, I furmarked it last night... one GPU got to 95C, but furmark's benchmark ran w/o breaking 100.

FWIW, I did the same compound swap on V2.5 as well... bit since doing that, I haven't even powered it on... remind me to find a use for it...

[ratdude747]

Re: Ratdude's Main rig V3.5

Looks like a few minor updates never made it here:

First, I found a BD-ROM Lightscribe drive for it at a goodwill for $20 a while back... but the crappy HP firmware kicked my ass at first. The saga that fixed said drive (firmware hack) is here:

https://www.badcaps.net/forum/showthread.php?t=36626

Also, before that upgrade, I added a pair of Dell 1703FPt's to the setup... they're the same size and resolution as the LCD's used before (1280x1024 @ 17"), but these have DVI and USB hubs... and telescoping rotating bases. Really nice picture quality for the vintage...

Last, I managed to SLIC mod the BIOS... mainly to free up the upgrade copy of 7 I had been using. Using a .bat I wrote, I was able to swap the certificates and keys w/o reinstalling 7.

[ratdude747]

Re: Ratdude's Main rig V3.5

Once again this rig is getting an upgrade... mainly I'm upgrading to a 9550sx-4lp controller (scored for cheap, I needed another BBU'd card for my other EATX build)... and around Xmas I may try getting a pair of geforce 9600's for it, as while the 8600 GTS's aren't bad, there are better fish in the pond.

[ratdude747]

Re: Ratdude's Main rig V3.5

Didn't like the raid card, it booted all right but my 2nd GPU began not working... first both gave a code 43 (nvidia reported a problem), re-installed the latest (compatible) drivers, then got the primary working and the second reporting code 12 (not enough resources)... tried again, now the secondary doesn't show up at all.

When I'm home next week I'll try a live linux disc to see if I can get everything to chime in and if so, I'll look to a reinstall of 7 (it could use it)... if not, then I'll reseat all the cards and whatnot... (insert u-235 making some wiseass comment about my stuff never working, until the RAID card addition it had been solid for many months buddy)

[RJARRRPCGP]

Re: Ratdude's Main rig V3.5

Didn't like the raid card, it booted all right but my 2nd GPU began not working... first both gave a code 43 (nvidia reported a problem), re-installed the latest (compatible) drivers, then got the primary working and the second reporting code 12 (not enough resources)... tried again, now the secondary doesn't show up at all.

When I'm home next week I'll try a live linux disc to see if I can get everything to chime in and if so, I'll look to a reinstall of 7 (it could use it)... if not, then I'll reseat all the cards and whatnot... (insert u-235 making some wiseass comment about my stuff never working, until the RAID card addition it had been solid for many months buddy)

Those codes are likely the BIOS being gay!!

[ratdude747]

Re: Ratdude's Main rig V3.5

^ I'll reset CMOS and see

In the mean time I've found that 9600's have come way down in price to the point where for $45 I can score a pair... If one saw the V4.1 thread, the pair of 8600 GTS's aren't cutting it (a single underclocked 9800 GT left them in the dust) and looking at the benchmarks and reviews, the 9600 GT is not far below the 9800 (what I'm looking at has only 512MB but similar benchmark scores) and nearly double the processing power of the 8600. Sure, the current setup *was* working, but IMHO that's not a lot for a substantial upgrade. Should run reasonable cool as well.

edit- Wasn't able to work out said deal... the best I can seem to do is $22 a card plus shipping... $55-$60 is a little much.

[ratdude747]

Re: Ratdude's Main rig V3.5

WTF?

I removed everything but the GPUs and the RAID card, no dice.

I tried moving the RAID card down to the 133 slot (where it should have been) with no change.

But when I put back in the 9500 (and reinstalled windows as it didn't re-reconize the array), the issue goes away.

How? Why? I know for a fact Topcat's former K8WE ran with a nicer 9550 AND twin Geforce 7600's... Maybe my 8600's are being bitchy. I have a pair of junk Quadro fx 1400's with SLI connectors that I can try...

To re-iterate, the lower of my two cards (no monitors) is in device manager reporting code 12, not enough available resources.

[ratdude747]

Re: Ratdude's Main rig V3.5

I've tried:

-updating the card's firmware. No dice
-swapping the GPUs for the pair of clay pidgion Quadros. Worked.

My only conclusion is for whatever reason the 8600's and the 9550SX fight for the same IRQ... and for another arbitrary reason the quadro's and the 9500S both go for different ones.

Here's my question, which may not have an known answer: would a pair of say, 9600's or a different (newer) pair of SLIable GPUs also work?

[brethin]

Re: Ratdude's Main rig V3.5

Remind us again what motherboard this is?

Try just 1 GPU and the raid card and see what it does. If that works then add the second GPU and check it. Then last Enable SLI and see what happens.

[ratdude747]

Re: Ratdude's Main rig V3.5

It's a Tyan S2895 Thunder K8WE, with the Nvidia Nforce Professional 2000 chipset.

Here's what happens when:

-Two 8600 GTS with the 9550sx: The first card (which both monitors are connected) is installed and fully operational. The second shows in device manager as code 12'd. SLI cannot be enabled.

Swapping for either the 9500s or a pair of Quadro fx 1400's results in a working system where SLI can be enabled.

I did not try a single 8600, as since one was working and the other wasn't (slot dependent), pulling a card didn't seem like a logical move. I will try it when I get home but I expect the one card to work, just like it has been (the issue is when a second card is added).