XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

[momaka]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

Wow, I had to look it up, did not believe it could already have been eight years!

Ha! Me too. I thought EP1 came with the Orange Box for some reason. Man, and I just played it last year. I'm so behind with tech, lol.

Guess the flood gates have opened now

*Ahem*
Even that can be used a reference to some HL/HL2 level, so be careful

[ChaosLegionnaire]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

Yeah, don't get me started on that. I just played EP 1 and 2 last fall. After EP2 ended, I was like, "WTF?! Is this how they are really going to leave the series to end?"

lol they have to stop writing the endings of video games to be like that of tv series. always a cliffhanger ending for the season finale. really smartass thing to do. the entertainment industry is so competitive. what if u get cancelled for the next season? that will piss off a lot of fans of the series. anyone here in the entertainment line who can shed some light on why they like doing this cliffhanger ending thing on tv?

Also, when I get some work done on the dead fan from this 6800, I will post the results here. Just to make it look like we were on-topic the whole time .

errr yea me too. in the interest of looking on-topic, when u get the card fully up and running, we can compare benchmark scores between the 6600 gt, which has a similar core config to the 6800 xt, as well as the 6800 ultra to see if the 6800 xt is indeed half a 6800 ultra.

the fact that its a comparison between 2 agp cards and a pci-express one shouldnt skew the results or comparison much anyway since there isnt much difference going up from agp 4x anyway. there! that should be enuff yakking to look on topic... hehehehe

[ChaosLegionnaire]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

hmm i got a radeon 9600 vanilla/non-pro second hand last week and an x1050 previously from my bro's comp. the x1050 is the pci-express version of the rv350 core and can also be called the pci-express version of the radeon 9600 se.

funny thing about both of them is they are both passively cooled AND they have both failed! u'd think cool running mid-low end cards would last a long time... think again... being cool running only made manufacturers smartasses and they smacked on passive coolers for these cool cards making them NOT cool!

im really extremely unimpressed by manufacturers' planned obsolescence stunts... heat also seems to be the nemesis of bga chips irregardless of the use of leaded or unleaded solder. even low tdp normally cool running bga chips like the rv350 can fail if u run them passive and they overheat.

good thing i decided to bolt a 50mm fan to my conventional pci mpeg4 accelerator card, the radeon hd 5450. it only overheated once to 80+°C so keeping my fingers crossed that it will last.

really disappointed that my purchase of a second hand radeon 9600 ended up with a lemon right out of the shop. straight away, upon plugging it in, i already noticed that the font in the bios setup screen looked weird and sort of blurry. i noticed dotted vertical lines all over bios dialogue boxes. also noticed the same albeit very slightly and hard to see on the winxp boot logo. the improper shutdown chkdsk screen was chock full screen of these dotted vertical lines even before driver installation. tried the omegadrivers and they wont install/boot. upon the driver being loaded, windows blue screens with ati2mtag stop errors.

next, i tried the official amd catalyst 10.2 drivers which are the final drivers to support r300-r500 gpus and those install and boot fine. but upon running games and benchmarks, the screen is chock full of artifacts in every one of em. the artifact patterns range from the "standard" vertical dotted lines, dotted lines around the outline of objects, tiled triangular dotted lines(never seen this type of artifact before. first time) and spike artifacts on sprite textures like grass.

funny thing is, when i tried a linux ubuntu 12.04 live session, it seems to work fine on the gallium 0.4 drivers. i tried glxgears and glmark2 to test the card and dont notice any artifacts... really strange. those are the only two tests i could find that work on 32-bit ubuntu. they dont seem to stresss the gpu enuff to expose any defects, i guess. the gputest program from geeks3d requires 64-bit so couldnt run it on my p4 northwood.

ah well... guess i'll take it back for a refund. *cry*

[momaka]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

funny thing is, when i tried a linux ubuntu 12.04 live session, it seems to work fine on the gallium 0.4 drivers. ... they dont seem to stresss the gpu enuff to expose any defects

Yup. I've had that happen with my Radeon 9700s as well. Would work fine under XP desktop but crash in 3D mode. Reflow and fixed away.

ah well... guess i'll take it back for a refund. *cry*

Or reflow it?
Those older Radeons have a high success rate after reflow... unlike the GeForce 6, 7, and 8 series. If your GeForce video card has artifacts and you reflow it and it works, then definitely go play the lottery too, because it's probably your lucky day .

[ChaosLegionnaire]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

Yup. I've had that happen with my Radeon 9700s as well. Would work fine under XP desktop but crash in 3D mode. Reflow and fixed away.

same. i had that happen with my bro's x1050. it was perfect and flawless in 2d mode. cannot find any artifacts in 2d. but put it in 3d mode and immediately the screen blanks out and blue screens with a video driver stop msg.

Or reflow it?
Those older Radeons have a high success rate after reflow...

well i paid good money for a working unit. if its faulty, i should get at least a partial refund. and if i could get my hands on some bga flux, i would try reflowing it too since its a cheap mid-low end card, there's less to lose.

however, i cant find any flux suitable for bga reflow in electronics shops around my country. most of the flux being sold is for hobby and wiring use and i found them to be of acidic nature and unsuitable for use on pcbs. guess they dont wanna sell bga flux as its more for industrial use and i aint buying any of those chinese e-commerce bs.

[momaka]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

well i paid good money for a working unit. if its faulty, i should get at least a partial refund.

If that is the case, you should get a full or nearly-full refund.

and if i could get my hands on some bga flux, i would try reflowing it too since its a cheap mid-low end card, there's less to lose.
...i aint buying any of those chinese e-commerce bs.

As much as I try to avoid cheap Chinese stuff too, I have to say that Kingbo RMA-218 flux is pretty decent for BGA. It's a cheap generic flux (NOT made in Japan, as the sticker says, obviously), but it works well nonetheless. Also, a standard 100 g tub will last you for quite a while. It's been 3 years since I've had mine, and I still have over 4/5 left in the tub. Just avoid the fumes or air your workplace afterwards. It doesn't smell too bad, but I doubt it is good for you.

[ChaosLegionnaire]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

ok, this is starting to really piss me off now. yesterday, i went back to the shop and explained the faulty video card. however, they refused a refund because it was past the 7 day deadline/warranty. i took too long to diagnose the problem and come back to them. however, as a goodwill gesture they allowed an exchange.

after digging through their stock of video cards inside out, i managed to find what looked like an oem radeon 9600 green pcb. i recognised it because it had the same heatsink shape as the previous card except a different colour, the pcb layout is similar to the faulty card as well, its an agp 8x card and i recognised the part number printed on it (109-A03500-00) that is similar to the part number of the radeon 9800 pros i had (109-A07500-00). so i accepted the exchange.

then comes time to test it. boots up fine. no weirdness on the bios setup font and dialogue boxes like the previous card. proceeded to boot windows and install the omega drivers and bam. bsod again when rebooting with a driver error stop msg. im dreading the card is faulty also as i've found the omega drivers are very good at detecting faulty cards. next, i installed the official catalyst 10.2 drivers instead. those installed fine but when testing in oblivion and FEAR, i noticed strange rectangular tile artifacts on textures. other games i tried dont seem to expose any artifacts. i also tried quake 4 and neverwinter nights. those games use opengl for the graphics rendering and they both run flawless. im suspecting opengl does ecc on the hardware and masks any errors. that also explains why the previous faulty card ran fine in linux which uses opengl.

after my nightly tv fix, i proceeded to test the card again. this time when idling at the desktop, i noticed rectangular artifacts started appearing on the desktop even at idle! i tried a lot of tricks to make the artifacts disappear, including changing resolutions, downclocking, running a fullscreen 3d program briefly and then exiting to refresh the texture mem but to no avail. then i tried loading up FEAR again to see how bad things look in games. to my surprise, after running around for a short while the artifacts started to reduce albeit a very tiny bit. so i proceeded to try the fast and quick method of diagnosing a cracked solder joint: bending/flexing the video card. unfortunately, that didnt change or fix the artifact pattern like it did with the 9800 pros i had.

then i tried the second way: letting it heat up. i unplugged the 140mm fan that was blowing air over the pci slot area. within a minute, the artifacts cleared up and were almost gone. next, i tried reloading the level. the screen appeared flawless after that. the artifacts at certain angles and on specific textures and objects were no longer there. also tested it in oblivion and no artifacts. but put the fan back and leave it idling at the desktop, the artifacts reappear again. so i think i have properly diagnosed with 99% certainty a cracked gpu solder joint. i have used the same method to diagnose faulty solder joints on my 9800 pros.

so im focking pissed off that the idiot techs at the second hand shop do not properly test their cards before selling. i ended up with TWO lemon cards! do u think i should file a complaint with the local consumer's association against the shop? they have no goddamn clue what exactly they are selling and fail to run it through the proper gauntlet of tests and i end up having to be their technician and beta tester instead! i think this shop sure needs an official complaint lodged against it so others can be wary of that shop and avoid getting taken for a ride like me!

Just avoid the fumes or air your workplace afterwards. It doesn't smell too bad, but I doubt it is good for you.

dont worry. i got a 3m p100 mask with a multi-gas filter cartridge for mining, construction and industrial use. the multi-gas filter should protect against soldering fumes as well. i bought the mask as protection against smoke haze/smog blown in from nearby indonesia on a regular basis. those indonesians sure love burning down the forest to farm their palm oil trees. they're such treehuggers... NOT!

[momaka]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

i ended up with TWO lemon cards! do u think i should file a complaint with the local consumer's association against the shop? they have no goddamn clue what exactly they are selling and fail to run it through the proper gauntlet of tests and i end up having to be their technician and beta tester instead!

I don't know how much (if anything) your local consumer's association will do about that. If they received many different complaints from that place, then your feedback will probably matter. But if not, all the effort will go unnoticed.
I would say just keep getting exchanges until you get something working. If they have online reviews, then definitely post something there to warn others.

i bought the mask as protection against smoke haze/smog blown in from nearby indonesia on a regular basis. those indonesians sure love burning down the forest to farm their palm oil trees.

I'm not surprised about that. Cooking oils are used in a lot of everyday products, so there is a lot of money in that business. It's too bad their lax environmental laws are allowing them to produce cheap palm oil at the cost of the environment and their health (and all of ours too, actually... because we all live on the same Earth, after all).

Also, I hate how hard it is these days to find chocolate foods that don't use palm oil. Had to quit eating Nutella because of this, along with a few other foods that now exclusively use palm oil in their products. Oh well, their loss.

[ChaosLegionnaire]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

so regarding the ON KE 1 2 switch, after some extensive and exhaustive digging around the web, i came across a review of a radeon 9600 pro by legit reviews. in the review, they talk about a dip switch found on the 9600 pro which they found to be a switch to change the output from ntsc to pal for the tv output.

i also found an image around the web of a gigabyte radeon 9600se. that pcb has writing on it explaining which position to flip the switch to get the pal or ntsc output that you want. i have attached the image in my post. for convenience, archival and reference reasons, i have also attached a printer friendly version of the radeon 9600 pro review in its entirety in web archive mht format in case the site ever goes down or they del the page cuz its obsolete info. hope that helps anyone trying to figure out what the switch on their radeon 9000 series video cards do.

[momaka]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

^ Great info!
Moreover, I think this not only applies to the Radeon 9000 series, but possibly newer Radeons as well. I have a HD 3870 that also has a place for a similar switch on its PCB. I don't remember if I saw it on my low-profile HD 2400s. Will have to look.

[momaka]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

A bit of an update on the thread... (anyone still care? Lol.)

In post #2, I said that my XFX 6800 XT ran around 60°C under load, which is completely wrong. The video card actually runs close to 70°C under load with the original heatsink. However, I didn't find this out until I started using the "Charts" tab in SpeedFan to monitor the temperatures. The "Charts" tab basically plots various temperatures of your PC over time (i.e. CPU, GPU, HDD, etc. – whatever you choose). This is much more accurate than Alt-tabbing and looking at the instantaneous GPU temperature, which can fall by as much as 5-10°C as soon as the card is unloaded (depending on the GPU heatsink.)

That said, here is the temperature curve I got for the 6800 XT with the stock heatsink (with the GeForce 7900 Delta squirrel cage fan) and 21.1° C (70F) room temperature:


First, I would like to point out that the "Ambient" temp in that curve (in red) is not actually the ambient temperature in the computer case. The case for my test PC does not have sides, so it runs open. Thus the case temperature is pretty much the same as the room. According to GPU-Z, that "Ambient" temp above should actually be of the card's PCB.

The green graph shows the GPU core temperature. You can see that the video card idles at 56°C (which IMO is quite high already) and jumps as high as 68°C under load! :\ The unofficial threshold of the bumpgate problem is around 60-62°C, so that's not good. Moreover, my load test is only a few minutes long. After all, note that the entire time interval of the temperature curve is 13 minutes. Actually, my typical load test consists of launching Portal or Half-Life 2 Episode 2, loading a level, and then playing for a few minutes. This is why the load graph has two "peaks" in it: one after the game loads the 3D menu, and another when the game loads the level and I play for a bit. While the level is loading, the GPU is not doing anything - hence the dip between the peaks. I like this particular test, because it shows me how quickly the temperature of the GPU core will change between idle and full load (i.e. GPU delta T). A badly-designed or inadequate heatsink will have a large delta T. You can see from my temperature curve that the GPU core jumps 9°C almost instantly (from 56 to 65°C). And after the dip between the peaks, the delta T is 8°C.

These temperatures didn't look very good to me, so I decided to play around with the card. First, I tried a different fan – the stock one from my XFX GeForce 8600 GT, which is exactly the same fan (or very similar) to what probably came with this 6800 XT. And here is how it looked:
https://www.badcaps.net/forum/attach...1&d=1463872565
Guess what, though? – The temperature curves hardly changed at all. But at least the ARX fan from the 8600 GT was much quieter (though still audible).

Of course, I still wasn't happy with those temperatures, so more needed to be done. This is when I took the modified Xbox 360 rev. 2 CPU heatsink from my 8600 GT and put it on the 6800 XT. Now, remember the drastic change in running temperatures I had with the Xbox 360 heatsink on the 8600 GT? Well, it was even better on this card. Performing the same set of load tests on the same test PC with the same 120 mm fan blowing air from far away on the Xbox 360 heatsink of the video card (i.e. same test setup as with the 8600 GT), this is what I got for the temperature curves:



You can clearly see on the first graph that both my idle and load temperatures have dropped by at least 10°C. The second graph is after about 1 hour of playing Half-Life 2 Episode 2 – the max temperature didn't go past 56°C at all. Moreover, this was with a warmer ambient room temperature of 25°C (77-78F). And if you look closely at the spikes and dips of the temperature curve, you can also see that the idle-load delta T is about 5°C – almost half of what it was before!

So with all of that being said, I think it is clear that many video card stock heatsinks are inadequate for the job. Perhaps I shouldn't judge them as harshly, since they are much more compact after all. But it's hard not to (especially nowadays) when you have mid and high-end GPUs dissipating almost as much power as the CPU in the PC.

Anyways, next up I will be trying an Xbox 360 rev. 1 CPU heatsink on this card, just to see how that performs.

Also, I believe I mentioned somewhere that the GeForce 6 series and the Radeon 9500/9700/9800 use different package for the RAM chips, but I was wrong again. Turns out both the Radeon 9500/9700/9800 and the GeForce 6600/6800 use the same 128-ball (144-ball with cooling pads) square BGA package. And speaking of the RAM, this XFX GeForce 6800 XT has eight Samsung K4J55323QF-VC20 chips. According to the datasheet, the VC20 indicates these are the slowest of the K4J55323QF, running at 500 MHz (1 GHz effective). The fastest is the VC14, capable of 700 MHz (1.4 GHz effective). Since the RAM is made by Samsung, I wonder how far I could overclock it. I already briefly tried overclocking the GPU core from 350 MHz to 400 MHz, but I didn't see any difference in gaming performance in the Source engine, so I returned everything back to stock speeds. I guess those hardware-disabled pixel and vertex shaders really do matter more than anything else (and they are locked on this card, so I can't re-enable them with Riva Tuner). At least the overclock temperatures didn't change much with the modified Xbox 360 heatsink. I obviously didn't try this with the stock heatsink as I'm pretty sure the temperatures wouldn't have been any good .

Finally, here is what really surprised me about this video card – it has H.264 video acceleration for up to 1080p!
http://www.bit-tech.net/news/hardwar..._decode_h264/1
That is awesome! I only briefly tested this feature as well, but I was actually able to have my overclocked Athlon 64 3200+ handle a 1080p MP4/MKV video file that it could never play before on its own (nor with my Radeon HD2400, which is supposed to have H.264 acceleration ). I know the GeForce 8 series and higher as well as the mid-range Radeon HD 2x00 series support H.264 well. But the GeForce 6800? Never expected that from such an old card. I think I might actually end up using this card in one of my PC builds now .

Anyways, that is all I wanted to add now.
Oh and look
<-----
I got 6000 posts now. I guess this is to celebrate the GeForce 6x00 series

[goodpsusearch]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

I wish 6000 series supported Youtube Hardware Acceleration

[Sparkey55]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

I have been collecting the AGP versions of the XFX 6800xt and recapping them and reapplying fresh TIM to the GPU. I get them for about 15 to 30 dollars on Ebay, have 4 so far. Nice LOUD fans but cooling seems good.

[momaka]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

I have been collecting the AGP versions of the XFX 6800xt and recapping them and reapplying fresh TIM to the GPU. I get them for about 15 to 30 dollars on Ebay, have 4 so far. Nice LOUD fans but cooling seems good.

Nice! I guess there are quite a few of us here that "collect" old PC hardware.

Does the XFX GeForce 6800 XT AGP you are talking about look like this?
http://www.newegg.com/Product/Produc...82E16814150157

If so, I am not surprised about the cooling being better. The heatsink on that particular card has the fins sitting right above the GPU core - and that's how it should be if you want good cooling. On the original heatsink of my 6800 XT (PCI-E), only the fan sits above the GPU core. So the core is relying on the thin aluminum of the heatsink to conduct away the heat, which is a poor design IMO. Hence the temperatures I was getting. Also, your fan looks more powerful than the stock one of my 6800 XT. I would guess it's similar to the replacement Delta fan I used, which is indeed a "screamer" .

By the way, which GPU core do those AGP XFX 6800 XT cards use? I wonder if the disabled pixel and vertex shaders are unlockable. A 12 pixel and 5 vertex config will make it more or less on par with the 7600 GT.

I wish 6000 series supported Youtube Hardware Acceleration

Or the 7 series for that matter .
...
...
Anyways, stay tuned as I have pictures of the Xbox 360 rev. 1 CPU heatsink coming soon. All I will tell you right now is that it fit my 6800 XT like a glove and doesn't look ghetto due to using proper bolts to mount the heatsink.

[Sparkey55]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

Nice! I guess there are quite a few of us here that "collect" old PC hardware.

Does the XFX GeForce 6800 XT AGP you are talking about look like this?
http://www.newegg.com/Product/Produc...82E16814150157

If so, I am not surprised about the cooling being better. The heatsink on that particular card has the fins sitting right above the GPU core - and that's how it should be if you want good cooling. On the original heatsink of my 6800 XT (PCI-E), only the fan sits above the GPU core. So the core is relying on the thin aluminum of the heatsink to conduct away the heat, which is a poor design IMO. Hence the temperatures I was getting. Also, your fan looks more powerful than the stock one of my 6800 XT. I would guess it's similar to the replacement Delta fan I used, which is indeed a "screamer" .

By the way, which GPU core do those AGP XFX 6800 XT cards use? I wonder if the disabled pixel and vertex shaders are unlockable. A 12 pixel and 5 vertex config will make it more or less on par with the 7600 GT.


Or the 7 series for that matter .
...
...
Anyways, stay tuned as I have pictures of the Xbox 360 rev. 1 CPU heatsink coming soon. All I will tell you right now is that it fit my 6800 XT like a glove and doesn't look ghetto due to using proper bolts to mount the heatsink.

yes that is the model that I have. If I remember correctly they have the NV41 cores with a pci-e to agp bridge chip. When I overhaul them I remove the old tim and replace the waxy cheese cloth with a silicon pad and some Artic Silver on both sides for better heat conduction. They run quite well now.

[momaka]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

Alright, here are the pictures of the XFX 6800 XTreme with the Xbox 360 revision 1 CPU heatsink:


Looks massive, doesn't it? But at least there is no more ghetto-ness with this one, because the heatsink is held down properly with two bolts instead of screws like the rev. 2 Xbox 360 heatsink. Of course, since I still haven't implemented anything to serve as a backplate, the bolts can warp the PCB if overtightened. Having a fine M3 thread pitch, though, I have them just tight enough to hold the heatsink firmly onto the GPU chip without warping the PCB. So I think I might actually forgo making a backplate.

Moving along to the temperatures:
Much like the Xbox 360 rev. 2 CPU heatsink, the temperatures with this (rev. 1) heatsink were quite good too. At the time of testing, my room was around 21.5° to 22°C (71-72°F). Using the same test PC setup as before, here are the GPU core temperature curves I got:


Basically, after a brief 10 minute session of playing various 3D games, the GPU core temperature peaked at 54°C under load. Leaving the GPU idle made the core temp settle to 43°C in 6-7 minutes. The change in temperature between loaded and idle GPU states was 5-6°C per unit of SpeedFan time (which is roughly 3-5 seconds)... so pretty much identical to what I got with the rev. 2 heatsink. Thus, I am quite satisfied with this performance.

Last week, however, I decided to try the 6800 XT (with this heatsink) in my secondary PC (which I often use for playing older games). The temperature results I got in it with this card were not good, though. And here is the reason why:
https://www.badcaps.net/forum/attach...1&d=1465183745

As you can see, there is very little space behind and under my modified GPU heatsink to exhaust the hot air. As a result, the video card ended running quite hot after an hour of gaming (Test Drive Unlimited) with a semi-closed PC case in a fairly-hot 26.6 to 27.2°C (80-81°F) room temperature. The idle GPU temps were, at best, 51°C. Meanwhile, the load temps reached as high as 68°C. This is with the same 120 mm fan blowing on the GPU heatsink from (more or less) the same distance. But I learned something from this experience: even a big heatsink does not guarantee low temperatures if there is no place for the hot air from the heatsink to be exhausted. In this case, the hot air from the GPU was pushed towards the motherboard and the system HDD (yeah, I know having the HDD like that is ghetto , but it gets better cooling like that when I have a fan blowing near it). Made the HDD run hotter than normal.

That leaves me with one more thing to try: rotate the heatsink 90° so that the hot air can move parallel to the motherboard versus perpendicular (and right into it). Thus, there will probably be one more post from me to update the results with that when I get to rotating the heatsink. I'm thinking the temperatures will be much better, but we will see .

[Wester547]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

That 360 heatsink on that 6800 card does look very .

Basically, after a brief 10 minute session of playing various 3D games, the GPU core temperature peaked at 54°C under load. Leaving the GPU idle made the core temp settle to 43°C in 6-7 minutes. The change in temperature between loaded and idle GPU states was 5-6°C per unit of SpeedFan time (which is roughly 3-5 seconds)... so pretty much identical to what I got with the rev. 2 heatsink. Thus, I am quite satisfied with this performance.

As you can see, there is very little space behind and under my modified GPU heatsink to exhaust the hot air. As a result, the video card ended running quite hot after an hour of gaming (Test Drive Unlimited) with a semi-closed PC case in a fairly-hot 26.6 to 27.2°C (80-81°F) room temperature. The idle GPU temps were, at best, 51°C. Meanwhile, the load temps reached as high as 68°C. This is with the same 120 mm fan blowing on the GPU heatsink from (more or less) the same distance.

Those temperatures don't sound too shabby to me, although considering the fact that these are cards with the bumpgate defect (temperatures of 60*C and higher causes the underfill to soften), it might be less than desirable.

But I learned something from this experience: even a big heatsink does not guarantee low temperatures if there is no place for the hot air from the heatsink to be exhausted.

Off-topic (sorry but that's my specialty ), but perhaps this also means that those chunky secondary heatsinks in those old Dell 250W-300W Hipro and Newton PSUs aren't such a good thing after all. They are a tad overabundant in the amount of space they occupy, which means less room for airflow in the PSU chassis (although it also means that the secondary heatsink gets more airflow).

In this case, the hot air from the GPU was pushed towards the motherboard and the system HDD (yeah, I know having the HDD like that is ghetto , but it gets better cooling like that when I have a fan blowing near it). Made the HDD run hotter than normal.

Not a surprise... that heatsink appears to be almost touching the HDD's PCB, so it's probably heating up the microcontroller quite a bit more than usual. On that note, I find it quite flabbergasting how quickly the HDD microcontroller (particularly the read/write channel) heats up during I/O operations without forced airflow over the PCB (much faster than the hermetic block itself). And that HDD doesn't look like it's secured to the case very well if at all.

[Sparkey55]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

Alright, here are the pictures of the XFX 6800 XTreme with the Xbox 360 revision 1 CPU heatsink:


Looks massive, doesn't it? But at least there is no more ghetto-ness with this one, because the heatsink is held down properly with two bolts instead of screws like the rev. 2 Xbox 360 heatsink. Of course, since I still haven't implemented anything to serve as a backplate, the bolts can warp the PCB if overtightened. Having a fine M3 thread pitch, though, I have them just tight enough to hold the heatsink firmly onto the GPU chip without warping the PCB. So I think I might actually forgo making a backplate.

Moving along to the temperatures:
Much like the Xbox 360 rev. 2 CPU heatsink, the temperatures with this (rev. 1) heatsink were quite good too. At the time of testing, my room was around 21.5° to 22°C (71-72°F). Using the same test PC setup as before, here are the GPU core temperature curves I got:


Basically, after a brief 10 minute session of playing various 3D games, the GPU core temperature peaked at 54°C under load. Leaving the GPU idle made the core temp settle to 43°C in 6-7 minutes. The change in temperature between loaded and idle GPU states was 5-6°C per unit of SpeedFan time (which is roughly 3-5 seconds)... so pretty much identical to what I got with the rev. 2 heatsink. Thus, I am quite satisfied with this performance.

Last week, however, I decided to try the 6800 XT (with this heatsink) in my secondary PC (which I often use for playing older games). The temperature results I got in it with this card were not good, though. And here is the reason why:
https://www.badcaps.net/forum/attach...1&d=1465183745

As you can see, there is very little space behind and under my modified GPU heatsink to exhaust the hot air. As a result, the video card ended running quite hot after an hour of gaming (Test Drive Unlimited) with a semi-closed PC case in a fairly-hot 26.6 to 27.2°C (80-81°F) room temperature. The idle GPU temps were, at best, 51°C. Meanwhile, the load temps reached as high as 68°C. This is with the same 120 mm fan blowing on the GPU heatsink from (more or less) the same distance. But I learned something from this experience: even a big heatsink does not guarantee low temperatures if there is no place for the hot air from the heatsink to be exhausted. In this case, the hot air from the GPU was pushed towards the motherboard and the system HDD (yeah, I know having the HDD like that is ghetto , but it gets better cooling like that when I have a fan blowing near it). Made the HDD run hotter than normal.

That leaves me with one more thing to try: rotate the heatsink 90° so that the hot air can move parallel to the motherboard versus perpendicular (and right into it). Thus, there will probably be one more post from me to update the results with that when I get to rotating the heatsink. I'm thinking the temperatures will be much better, but we will see .

It would cool alot better if you cut off about half the aluminum heatsink and turn it 90 degrees then flip the HDD over and leave about 1/4 inch of air space between it and the chassis. Then build a cardboard duct around all of that then remove the slot covers from the back and put an exhaust fan on the back slots to pull cool air over the heatsink and HDD.

[diif]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

Nice mod, factory

I also think that something could be done with the space that's got the two blanks in and a fan or two.

[momaka]

Re: XFX GeForce 6800 Xtreme – an eBay special (with LONG story)

That 360 heatsink on that 6800 card does look very .

See, that's (again) what I don't understand: why is everyone OK with having a big heatsink on the CPU but not on the GPU? IMO, this is what the stock heatsink on any high-power video card should be (and have a fan too, of course). I just played with a Radeon HD4850 yesterday... and I've got to say, whoever thought that a 110 Watt TDP card can be cooled with a single-slot cooler properly is/was absolutely mad. But that's for another topic I have coming sometime later this summer .

Those temperatures don't sound too shabby to me, although considering the fact that these are cards with the bumpgate defect (temperatures of 60*C and higher causes the underfill to soften), it might be less than desirable.

Yes, that is exactly why I want to keep this card under 60°C with a room temperature of close to 30°C. I know this is not a rare video card or anything like that, but I just want to see if it can be done.

I still haven't flipped the heatsink by 90° yet, but it's in the plans to do it sometime this or next week. Instead, I performed another test on my test PC but with a slightly higher room temperature of 27.7-28.3°C (82-83°F). I still got the same temperature curves as above, but this time the maximum temperature peaked only at 62-63°C, since my test PC has a full-size ATX motherboard and there is much more space under the heatsink... in addition to the case not having any sides.

Off-topic (sorry but that's my specialty ), but perhaps this also means that those chunky secondary heatsinks in those old Dell 250W-300W Hipro and Newton PSUs aren't such a good thing after all. They are a tad overabundant in the amount of space they occupy, which means less room for airflow in the PSU chassis (although it also means that the secondary heatsink gets more airflow).

It's not the big heatsinks in these units that are the problem. It is the passive PFC choke on the 250 Watt models that blocks almost a quarter of the fan and much of the primary heatsink. But they still run quite cool inside compared to an LCD power supply, for example. If they didn't, the one would expect to see burned and/or darked spots on the PCB. But I've never seen any on a HiPro PSU, save for a few models that had a hot dummy load resistor on the 12V rail (IIRC).

Not a surprise... that heatsink appears to be almost touching the HDD's PCB, so it's probably heating up the microcontroller quite a bit more than usual.

Yes, only 4 or 5 mm between that heatsink and the HDD PCB below it. I was using that front audio cable you see there to keep the heatsink from pulling down too much on the video card.

On that note, I find it quite flabbergasting how quickly the HDD microcontroller (particularly the read/write channel) heats up during I/O operations without forced airflow over the PCB (much faster than the hermetic block itself).

Indeed.
If it wasn't for the PCB on HDDs running hot, I'd never bother to cool them at all. The Hitachi Travelstar in my Dell Latitude C600 is regularly running 45-55C now in the summer.

And that HDD doesn't look like it's secured to the case very well if at all.

Nope.

It was a faulty HDD out of some NAS backup device. I couldn't even install Windows on it at first. So that's why I only "temporarily" placed the like that - no screws or anything to hold it in place. After many tries, I finally managed to install Windows XP... and then I just left everything as is. This "temporary" setup has stayed like that since the spring of 2013.

It would cool alot better if you cut off about half the aluminum heatsink and turn it 90 degrees...

I agree with flipping the heatsink 90 degrees. But cutting the fins in half would defeat the whole purpose of having a cool-running GPU that is nearly passively cooled.

Not to mention that our garage is going to be full of aluminum dust if I do that.

Then build a cardboard duct around all of that then remove the slot covers from the back and put an exhaust fan on the back slots to pull cool air over the heatsink and HDD.

Yeah, I was thinking of something similar as well.

Nice mod, factory

Lol, thank you.

I also think that something could be done with the space that's got the two blanks in and a fan or two.

A single 60 mm fan fits these heatsinks fairly well (only a tiny bit smaller than the heatsink) - I just checked today.

I think with the proper fan, I should be able to cool as much as 100 Watt-rated TDP chips with that heatsink.