NotebookForums.com › Forums › Notebook Manufacturers › Dell Forums › Dell Home (Inspiron, XPS, Studio) › Better Cooling for the 7800GTX
New Posts  All Forums:Forum Nav:

Better Cooling for the 7800GTX - Page 11

post #201 of 416
Thread Starter 
Quote:
Originally Posted by ZX81
No I get 7123 with background stuff off, 7021 with crap on

It only has a 1.73processor but I guess not having to carry around the xmas lights gets it close to your score
Benchmark on high performance settings, everyone else does , and that's worth ~200pts.

Wow, beautiful place you have there Doc . RTHDRIBL is a great program, but it's too old to bring these cards up to max temperature. FEAR, ATiTool, Far Cry w/ HDR, and Splinter Cell: Chaos Theory are the most taxing I've found.

I re-applied my AS5 to a VERY thin layer and it dropped temps 5-6C or so. So now I'm loading at 98-100C in I9KFanGUI. Still too hot in my opinion, but the card doesn't seem to have any problems. I think I will add the copper shim tomorrow. I think the problem with the heatsink is that it doesn't put enough pressure on the core to insure proper contact. I still think there is too much of a gap. Anyway, I'll see tomorrow .
post #202 of 416
Quote:
Originally Posted by Mr. K6
Benchmark on high performance settings, everyone else does , and that's worth ~200pts.
I always just test with the default settings. What exactly are the accepted "standard" test settings?

Quote:
Originally Posted by Mr. K6
Wow, beautiful place you have there Doc
Thanks. Leaving it behind in 7 days. Rio is a little too big for us, so we're moving to a slightly smaller neighborhood:



Saba. Less than a mile and a half in diameter, but they have DSL!

Quote:
Originally Posted by Mr. K6
I re-applied my AS5 to a VERY thin layer and it dropped temps 5-6C or so. So now I'm loading at 98-100C in I9KFanGUI. Still too hot in my opinion, but the card doesn't seem to have any problems. I think I will add the copper shim tomorrow. I think the problem with the heatsink is that it doesn't put enough pressure on the core to insure proper contact. I still think there is too much of a gap. Anyway, I'll see tomorrow .
I can't imagine that the heatsink doesn't screw down nice and tight against the GPU... I mean, that's what they'd want even with the stock thermal pad. I'm going to call nVidia Monday and really dig around for an answer on this.

-Doc
post #203 of 416
so Mr. K6, u said ur getting better temps with reapplyin the AS5, how thin is thin because when i did it i rubbed it in both it and the whole copper underside of heatsink and then applied a layer on core so that it was not transparent at all, prob a grain of rice size
So, ur applyin to the point where u can read nvidia, like a haze of that silveryness of AS5
I'm getting 50/94 idle/load i think after awhile gaming is that high for 445/1250? maybe i need to reapply so that its haze and i can sort of read the nvidia on the core?
post #204 of 416
This whole thing about applying a thin layer... I like the idea of doing it the other way the AS5 site directs. To rub it into the heatsink and then wipe it off, which leaves the microscopic cavities filled in, and then put a tiny bit of AS5 in the center of the GPU and put the heatsink on. This minimizes the chances of getting an air bubble in there. The pressure from the heatsink and the heat from the GPU will make sure that the stuff is spread thin and even, and most importantly, bubble free. That's how I'll do it when I dig into it again. (If I do, it's maxing out at 92C with a little OC action in there.)

-Doc
post #205 of 416
Hmmm, you can always replace the screws . Of course, then it's hard to know when to stop screwing.
post #206 of 416
wtf... can't edit post?!? What's this? Anyway, doc, you're supposed to only apply that grain of rice to the center and put the HS on? Huh.... I max out at 87 fangui (93) at 435/1200.
post #207 of 416
Quote:
Originally Posted by Master_Axe
Hmmm, you can always replace the screws . Of course, then it's hard to know when to stop screwing.
(snicker)
post #208 of 416

:pirate:

ok officially jealous now



may you find sunken treasure!


K6 I only test what I run, no one stop wonders, I like the quality settings.
post #209 of 416
http://www.arcticsilver.com/arctic_s...structions.htm

The bag method is for AS 1-3, not AS5.

-Doc
post #210 of 416
Looks like I had my info wrong as well. I guess you're supposed to spread it trowel-style with the edge of a razor or credit card.

-Doc
post #211 of 416
using a razor edge or card edge??? i used a baggy when i did it

Damn even using a baggy at sometimes it felt like it wasnt going on enough, and with a card edge seems like that wouldnt spread but just take it off

Anyways when u apply it is the layer supposed to be so thin u can see the core and the nvidia etched on it? like more of a haze instead of a layer?
cause when i did it i applied enough to where it was compeletely covering core so i applied til the layer was no longer translucent.

When reapplying, we should still use baggy method for smearing and wiping off a coat for rubbing it into heatsink and core, and then a card for getting a layer on core itself, i guess i want to make the layer so i can still see nvidia on core and translucent.
Clean off all previous AS5 when reapplying, even stuff rubbed into heatsink??? or just clean off core and do a rub layer and then translucent layer

I just hesistate to put a layer that is translucent, just doenst seem like enough, but i guess if people are getting good results that way i will reapply
post #212 of 416
Thread Starter 
Quote:
Originally Posted by Doc.Caliban
Saba. Less than a mile and a half in diameter, but they have DSL!
LOL! Oh man, looks great, just dont take that last step off the back porch .

And yes! You answered the question before I could. That "grain of rice in the center" method is ONLY for CPUs WITH an IHS (heatspreader). All other, uncovered CPUs, GPUs and chips MUST have the entire surface of the chip (the glass, silicon part, not the green cermanic, that'd fry it ) be covered with AS5 to ensure complete contact with the heatsink. Otherwise you can heat stress fracture the silicon in the chip (over a period of time). The way I'm doing it now, without a shim, I apply a small dab in the middle of the core. Then I take a plastic microscope slide (+ point of being a biochem and molecular bio major ) and spread the AS5 over the entire core. I then scrape off any excess so that there is a relatively thin layer on the core. NEVER USE A RAZOR OR ANY METAL OBJECT. Depending on the metal used, it can scratch your core! Then, with the heatsink off the back side of the card, I put the front part on and press it against the core and rotate it slowly back and forth to ensure connection and that the AS5 is spread out, air bubbles are released, and that it is pushed into microscopic gaps. Then I put the back plate on and screw it all together. I'm still loading at 100C in FEAR (was 104C+ before) and 92C in Serious Sam 2 (HDR lighting). I dunno, I would like it to be lower, but the card is 100% stable. I'm going to try putting on the copper shim tomorrow for the heck of it. I'll see how much it lowers temps.
post #213 of 416
The thing I love is the "wand" that comes with the AS adhesive. It's about the only thing I got out of killing a Ti4600 of mine (while trying to epoxy nickel RAMsinks to it) when I thought "hey, it couldn't be TOO hard!"

Hellooooooo yellow bars on boot.

But that wand is goddamned genius.
post #214 of 416
Well, I was just going by what the official Arctic Silver 5 web site says. "A razor blade or the clean edge of a credit card can be used as the application tool." Common sense would dictate that you don't use the razor like an ice scraper. :-)

-Doc
post #215 of 416
Isn't the cooler screws springloaded? I tighten them until I meet resistance, and the a few more degrees to be sure will not loosen themselves. Trying to tighten them more will not do any good if they are springloaded. The 10th of a mm you gain in more pressure is eaten up by the spring.

If they are not springloaded, this is all BS.
post #216 of 416
Thread Starter 
Quote:
Originally Posted by destruya
The thing I love is the "wand" that comes with the AS adhesive. It's about the only thing I got out of killing a Ti4600 of mine (while trying to epoxy nickel RAMsinks to it) when I thought "hey, it couldn't be TOO hard!"

Hellooooooo yellow bars on boot.

But that wand is goddamned genius.
Yah, I have one of those wands too, but I like the bigger surface of the microshope slide. Still, it is great and can't be beat for applying their adhesives . Btw, welcome to the "I have killed a vid card club"

Quote:
Originally Posted by Doc.Caliban
Well, I was just going by what the official Arctic Silver 5 web site says. "A razor blade or the clean edge of a credit card can be used as the application tool." Common sense would dictate that you don't use the razor like an ice scraper. :-)

-Doc
I wish that arctic silver would do a better job of separating its directions for different applications. The razor can be used on the large IHS (heatspreaders) of desktop CPUs, as those are metal and will not be harmed by the razor. It CANNOT be used on an open CPU core, nor can you use it on a heatsink as it can scratch the copper/aluminum. It's always best to stick to plastic or some softer material, refer to Moh's hardness scale in case you're wondering what else to use .

Quote:
Originally Posted by hifiking
Isn't the cooler screws springloaded? I tighten them until I meet resistance, and the a few more degrees to be sure will not loosen themselves. Trying to tighten them more will not do any good if they are springloaded. The 10th of a mm you gain in more pressure is eaten up by the spring.

If they are not springloaded, this is all BS.
Yup, they're spring loaded. That's why they have lock washers on the bottom of the screws so that they meet the PCB and cannot screw in any further, so that the rest of the compression is due to the springs. What's the problem with this? The springs don't apply enough pressure. For example, desktop CPU's have a compression weight of I think 60-70lbs (I'll have to look it up exactly), and that's one reason why they run so cool: the mating of the surfaces is basically flawless. Do you think those dinky springs on the CPU heatsink or the GPU heatsink provide that kind of pressure? That's why you add the copper shim, so that it adds more pressure on the core
post #217 of 416
omgz omgz, ive got a running supply of those tiny screws, really long ones while were at it, i could replace all the screws . but then you don't know how tight to screw it, and my guess is dell has a reason for the springs etc.: maybe the cards are really fragile? Im not gonna test it, the two exchanged screws i got are tighter than the rest tho.
post #218 of 416
see i get that it needs to be spread its just the thickness of the layer i am unsure.

Should i wipe it to the point of making of the color of the core that dark silvery color? or enough on their that its just enough to NOT see the color of core???

Also would me not rubbing the heatsink to the core back and forth before tightening make it a lil hotter? cause i just applied the stuff and put it back on

Also for rubbing AS5 in heatsink, just rub it in and wipe with clean section of lint-free cloth (coffee filter in my case) or wipe it in and leave it?

Also should i rub it in the core and wipe it as well before actually applying it and everything?
post #219 of 416
Thread Starter 
ALL HAIL THE SHIM .

I just installed it after filing it down a tiny bit (it's probably 0.020" now or so), and although I managed to file off a finger nail or two in the process, it worked . Load temps in FEAR are now 90-91C and it idles at 46-47C instead of 49-50C. Installing the shim is a pain in the ass though. First, I put a thin layer on the core (VERY thin, almost transparent) and pushed the shim on the core with a bag over my finger, and then rotated the shim while applying pressure to it. I then rubbed AS5 into the heatsink using my plastic bag covered finger, and wiped away any excess with that bag covered finger. Finally, I smoothed out a thicker layer of AS5 on top of the shim. I then took the back plate off and pushed the heatsink on to the shim. I then rotated the heatsink EVER so slightly in both clockwise and counter clockwise circles in order to make good contact but not move the position of the shim. Finally, I tightened the screws until they locked, and then loosened them about 1/2 to 3/4 of a turn so that the springs had ever so slight give (barely detectable). Anyway, it works like a million bucks .

To review the temp changes:
Stock - God knows what, probably would have melted the plastic
Thick layer of AS5 - 104-105C in FEAR
Thin layer of AS5 - 100C, but note that the first attempt left a part of the core with no contact to the heatsink, although it was covered with a layer of AS5.
Thin layers of AS5 + shim - 90-91C load, gold

Master Axe - believe it or not, cards, mobos, chips, and what have you can take a beating as long as it isn't jarring or sudden. Compression doesn't really hurt the chip until you tighten a screw over it and crack it (which is very difficult to do with spring loaded screws).

strat- don't worry about the thickness that much man, that's what the rotation is for. If it's too thick, it'll push excess out to the sides . Strive to put on a very thin but opaque (for the most part) layer of AS5. The rubbing action is really important because it ensures good contact and fills all the microscopic valleys. If you want to rub AS5 into the heatsink, cover your finger with a plastic bag and use that to rub it in. Using any kind of cloth will still get debris in the AS5. The core, is, essentially, the texture of glass, so there are no valleys to fill in, so dont bother rubbing AS5 into it before applying.
post #220 of 416
How is Farcry with HDR in comparison to fear for load testing? With HDR enabled and running at 1920x1200, I peak at around 93C.

-Doc
New Posts  All Forums:Forum Nav:
  Return Home
NotebookForums.com › Forums › Notebook Manufacturers › Dell Forums › Dell Home (Inspiron, XPS, Studio) › Better Cooling for the 7800GTX