• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Alternate GF2 MX cooling <interesting i think>

Toysrme

Member
Ok. If you look back, just last night I posted on putting a TNT2u HSF on my gf2 MX. Now, wanting to keep my TNT2u around in a comp still.

(me and roommate have sevrial laying around doing stuff, yes i WILL join rc5 just as soona s i get around to it heh)
So i took the heatsink off the older TNT and placed it on the TNT2u might as well. That card is still quite adiquate for what that comp dose and dose really need to o/c much if any. So after taking the long way around =) I'll start the real thread heh.

Basicaly I ran out of thermal compound and wanted to get the MX with a hsf quickly(it runs VERY hot i think without a heatsink, not o/c) so i rand down to <GHASP> radio shack to pick up some <AHHH> generic thermal compound! (ya, to lazy to order some artic silver and didnm't want to wait)

To make a long post longer, While digging for the compound, I came accross a small stash of so called, &quot;Thermoelectric Cooling Module's&quot;. Hmmm to much amazement, they had baby siezed peilter eliments just hidden away. They had the price stickers on the rack for sized 15*15mm through 40*40mm, but all they had was 15*15mm that Me and the clerk could find. Welp, it cost about $20 with tax and the 6.5 of thermal compound, so i thought why not. Looks like a neat little toy.

ahhh now the story and question has a meeaning and we all see where it's going <g> So. I get home, take an old HSF off a p90 that really dosee do nothing but sit and collect dust. Cut it up to fit this 15*15mm peilter and here we are!

I have NO idea how this will turn out at all. the thermal compound on the HSF/peilter/GPU is brittle(typical radio shack) So far after 5 hours of benching in UT and playing no condinsation(thin silicon around the gpu and very thin layer along bottom of PCB around gpu)

I also have no idea on how long this little peilter will last,right now it's running at 2v @ 3 amps when it should run a maximum of 1.9V @ 3.3... close enough for me, but I'm hoping the peilters are pretty tolorable (any answer to that?)

Well just thought I'd post and see what people think, how long till it may fry/break or condinsation may start. (no problems yet like I said)

Not to bad tho, RadioShack or not... &quot;Thermoelectric Cooling Module 20x20
mm&quot; would be an absilute perfect fit for Nvidia chips.

So, what do yall think about this? Spent about $20 on it and running fine. I have no way of telling tempatures so no, i can't say. I will say the silicon on bottom is cold, and the air being blown on the HSF is hot...

btw no problems with case temp... the case is off and for the last 2 years exactuly this January(that's now) there has been an 18&quot; oscilating fan blowning into it set to &quot;warp factor low&quot; this is a noisy comp and no u honestly can't hear the big fan... (few large blades turning sow are much less noisy)

I guess most importantly it's a CL GF2MX
(it was cheap sorry guys, I don't like the DDR ram... not like it costs me 2 fps tho over the SDR MX versions, also has TV out that iw asn't expecting. It's an early version I can tell, much wasted PCB space)

Mem at 200 core is fine and stable at 265 (using RivaTuner)

will Bench it soon and see what I find out. (I doubt much... core aint limiting factor =( but everything counts LoL!)

p.s.
Some1 with some ballz and an Ultra try this! LoL!

hmmm think about this tho... if you could troubleshoot all the downsides peilters have... you could just cover an entire card with 2 40*40mm peilters like this for about $50...

Would like to see THAT done by someone with an ultra and some ballz!

found a link to the Little things on the RadioShack site.
HERE
 
Interesting ... I'd be curious to see the benchmarks for the core running at 260 vs. the default. Also, where is the condensation appearing?
 
After leaving it on running a UT demo I made (more intencive than Thunder, no bots so CPU dosn't come into play asmuch) I left the comp on all night running. It had crashed sometime last night. So I dropped the clock speed down to 250 just to be sure.

Honestly? It dosn't help very much. We all already know that all of the GF family are mostly memory bandwith limited.

Will run a timedemo sometime tonight and post it.

Condensation Isn't being a large problem right now. I noticed none this morning when I had checked on it, but like I said, alittle bit of moisture last night (on the silicon that's running around the little peilter).

Honestly I don't think I'm gaining much, it dosn't feel much fater (just someething that's kinda neat I think) I'm thinking next time I get $30 and somee time to do it right, I'll try again on someething that can maby actually benifit on this. Like a TNT2u which is still core limited, or I might try a diffrent size pad on a PII or something heh =)

Will keep posting if something comes up. btw working on getting a tempature reading. Will a digital thermomiter work? like the kind you put in your mouth?
 
That's true about the effect of overclocking the core's speed on overall framerate, but overclocking the core should affect polygon rates. If you have 3DMark2000, check the High Polygon Count test. I'd imagine a core overclock of close to 100Mhz or 60% ought to have a noticeable impact on that test.
 
Toysrme, how are you powering that peltier with 2v? I mean there is no 2v supply so what are you using, a custom made circuit or what?
 
Pelitiers will run great up until you start melting the solder on the hot side of the device. Which is usually at 100 C plus. So it takes alot. If you get it that hot chances are you CPU will be running hot as hell also. Figure at best 40 deg C less than the hot side temp on the cold side.

Perry
 
Thanks much PerryG

It's well under 100C i can assure you =)

3Dmark score of 1974 overall(old slow comp)

The high poly test did run better. When I bench it alone i get a N/A for score so... I take it that's cause it's a demo LoL!
 
No, the 3DMarks score is only comprised of the six Helicopter and Adventure tests - the others are for eye candy or additional info, depending on what you think of 3DMark 2000.
 
Back
Top