Good luck with the 295. A friend shipped me a 9800GX2 years ago for free because it was broke. I thought maybe I could get it to work... nope. Wound up in the dumpster unfortunately.
Those tiny sheets of metal they call 'backplate' is actually the memory heatsink...
So many pads... I will reuse them, no chance I´ll get a replacement for them. This would cost 2x the current price of this card.
The discoloration on the shroud around the communication chipsets hints at a lot of heat, and those grey pads seem solid like a brick. One of those didn´t even have proper contact, this may have been the primary issue. And that paste, eww. It crackled as I loosened the screws.
Let´s see if this will run like it should afterwards. And with the different driver version that @Mr.Scott suggested.
May take me a day or two.
It doesn´t look bad from the outside, but since I see a cat around it there could be a lot of hairs inside. Not the nvidia hairworks you would want.
Maybe plug it in, check if it works and if temperatures seem reasonable. Does the fan spin, what noise does it make etc..
I have great memories of my GTX260, had to rma the first one I got due to cooler not making good contact. It was the first upgrade for my first pc!
And I need a GTS250, I would like to have the whole 200 series family. Got a 260 and 295 now.
@Muaadib I would like to get my hands on a 7900GX2, and a 9800GX2 for that matter, someday in the future. We will see what I can grab.
Again I have some mixed news for everyone. So I post the eyecandy first ok?
I think I might be developing feelings for this card that I should not have. Not much bare silicon but those two IHS facing each other. Kind of like that UT-map 'facing worlds'.
A more SFW shot:
And some silicon for good measure:
Now for what is troubling me. I wanted to reuse those pads, so instead of going brute force on the plate they stick on I took out my tweezers and carefully removed the larger lumps of dust around it, without disturbing that fine grain. I don´t want any of that close to the pad surface. I could not remove these pads in one piece, so I left them sticking. The white ones are still very soft but brittle.
The problem are the grey ones on those SLI chipsets that spread the 16x over Y to 8x/8x. These ones are hard as rock and crumble away into dust as soon as something hard touches them.
Poked the side of one with my tweezers, and they simply break apart. If I remount them under pressure I´m pretty certain they break and on top I think they lost their performance anyway. So forget what I said earlier, I´m buying new ones for those. Quality new ones. I just have to do this one right.
Now for something completly different! If you ever want to store itty bitty parts and screws, like the one from the cable-cover of the fan cable, get micro-mount boxes for mineral collectors. They are perfect to hold small batches of things, you can easily spot what´s inside, they stack and you can get 100 for less then 10$ or so.
And since the result of this one will now be delayed again, we will have a short intermission:
Managed to grab an EK-waterblock for ~10$. It may be dirty but it fits on GTX 260s and 280s. I will definitly use that at some point. Now let´s see if I can get this clean again. Maybe some toothpaste?
@Robert B 's Old hardware Emporium is always nice, but this thread is very cool too. Excellent pictures quality btw, i only wish there was a clickable version that would show them in all their HD glory.
I took a short detour. For 6€ I couldn´t really argue. BTW @Robert B ´s thread and posts are very impressive, I hope to get some soldering work in here too, if I get a case that needs it and I figure it all out. And his cleaning jobs are next level.
Here we got a partner in crime for the XFX card and the first stock DHE-cooler:
An EVGA 9800 GTX+.
Not much to see on the outside, looking decent already. Please note that screw right at the sticker. I will get back to it later.
Let´s jump into its current idle and load temps + power consumption figures.
We have 23°C ambient (measured at the front case intake fan btw) and get:
Idle 35°C // 98 W power-in
Load 78°C // 255W power-in
That is quite impressive to me. These blower style coolers can get pretty hot and I think these numbers are still ok. I don´t think it would reach the 78°C in any gaming scenario.
Now let´s take a look inside...
Oh. The previous owner sold me the card and some free noise insulation material. Let´s go with that.
This is one furry card, looking at the current full moon I start to wonder. Oh well those pads look fresh. Like really fresh. And that paste was still soft and not dry.
Now coming back to that screw next to the sticker, the spring on it was still UNDER the sticker and it was sticking to the glue of it. This and the paste tells me that this card was never actually opened before. We look at 9 years and about 6 months lifetime and it still performed like it did.
VERY nice EVGA. If my PSU from you guys has the same kind of lifespan design behind it, thumps up.
However, I don´t completly agree with the placement of this thermal pad:
How did I estimate the age of this card? Well the 'Magic'-Fan had a crystal-ball-bearing-... production date on it (20.11.2008):
Now that thing was loud. I know blower fans make noise, but this one had an unhealthy almost rattle-like tune. I decided to try an remove the sticker and on top let some oil on the bearing by building an oil-bridge:
They may be sealed, I´m not sure, but if there is a way in this oil will find it.
I use this stuff, intended for open yo-yo bearings. Not many would have this around, but if you do it´s great and you will never empty that bottle with your yo-yos no matter how much you play.
Now let´s get some close-ups while cleaning and then back into Furmark.
And it´s ready to rock:
This time we have an ambient of 21.3 °C
Idle: 33°C
Load: 73°C (actually some improvement)
And the biggest benefit, the fan did stop the annoying noise. It´s still loud but for one doesn´t spin up as much and the rattle like sound is gone.
I honestly did not think the temps would improve at all after seeing how good the paste and pads were looking. But looking at load temp, with 5°C drop and lower fan speed I think I did make a difference. Yes ambient is 1.7 °C lower too, and the lower the temp is, the smaller delta T gets because of density and transfer rate but I still think it improved beyond that effect.
To finish it off I have these pictures and some bonus material packed in the spoiler. WARNING tho NSFW-Screenshots from gameplay on this card with blood and gore.
I did test UT3 on it, because I had it installed. Well this 2007 title was absolutly no match even in HD. 60fps and bored with it. So I threw a more recent but not very demanding title at it: PoE
It struggled to hit 30fps and the drops made my hardcore char really anxious so I decided to not do an intensive test. I did notice that the Dx11 mode ran a lot faster on this old card then the Dx9 setting.
Now for @blobster21 ´s wish / suggestion, and thank you for these kind words on my pictures. I thought about the picture resolution and my problem is this: I take them all with my DSLR in RAW. This means big filesize and I need to convert them before uploading. Now I take every picture and crop it, if necessary and then convert it into a smaller jpg. I skip any post processing to speed things up. This file is in the original resolution of 4928 x 3264 and still at 5mb filesize for each picture. I want to use bigger pictures and not just the thumbnails in my posts, so you don´t have to open every specific one I write about manually. This means resizing them into a managable size, with the benefit of this page here loading even with slower internet connection and me not having to upload 3GB of photo material. My current ad-free pic hoster would not allow that anyway and I think TPU servers wouldn´t want to be piled with my 200GB photo history after a year or so.
Only Option I see would be, to upload either the full size JPGs, which would take forever with my 'quality' german internet, or do double the work for every photo and resize them twice. One version for this thread and one version in FullHD, the max. that my pic hoster allows if it´s below 1mb.
If anyone has a suggestion to easily upload high-res without the downsides, that would be great.
Ah, the 9800GTX+. You made me remember my poor 9800GT which I lost in a horrible accident involving overvolting a quad core processor, a cheap motherboard, and unfortunately a critical amount of negligence
It was the first time I had a system that could respectably tear through anything I threw at it. Then I swapped the whatever chip I had in there for a Phenom 9500 (I think it might have been my old 5200+) but it was, as OG phenoms were, slow, and I was desperate to push the clocks... I'm sure we've all had those "oops" moments...
Yes I use a Nikon D5100 with a 50mm 1:1,8 lens. Have this one for a long time and I´m afraid the mirror shutter might die soon as it´s nearing the 100.000 shot mark fast.
For my extreme close-ups I use close-up spacer rings between the camera and the lens. Works great, but really demands a lot of light. I don´t do this professionally and I do not own any studio lighting so I make use of my desk lamp as well as I can. It can be a PITA to hold the cam steady enough in your hand for a close-up shot at 1/5th sec. exposure time...
I´d say when it comes to PC-Hardware I was trained by my granddad to be very careful on what to do and what not to do. He tought me this stuff like overclocking, installing os, getting games to run, troubleshooting, building a pc. I was about 6 or 7 at that time and loved it, most of all the gaming stuff. Ahh good old times. Now its more then 15 years later and the only 'oops'-moment I can recall would be how I treated my first PC.
At one point it got so bloated with software (good old Win XP-times) that I had random bluescreens and couldn´t trace them down to a source. They only ever happened every 1-2 hours and that was really annoying. That constant feeling of 'it IS going to happen'. One day I was already a bit angered by something and after the 3rd BSOD that day I started to kick my pc with my foot. Until it shut down. Then instantly regret hit me. It booted back up, but I broke the front fan and the MB has some scars left. Audio ports no longer work and 2 RAM-slots are broken. The other two only boot stable if I put in max voltage.
Still have that board, with my original CPU and harddrive in a case and sometimes turn it on just to see it run again. Kept this one as a reminder to never use violence as a vent for anger. So far it worked. If I rage now like ingame, I only get vocal and if it get´s to much I just leave.
You can also use whatever is the latest date between the PCB and ASIC, since they are both dated. The ASIC on that particular card is 0841, or 41st week of 2008. The PCB is 0847, 47th week of 2008. So all three of your production dates indicate it was built in late November or early December, with the ASIC rolling off the line first and waiting quite awhile for a board to sit on.
You can also use whatever is the latest date between the PCB and ASIC, since they are both dated. The ASIC on that particular card is 0841, or 41st week of 2008. The PCB is 0847, 47th week of 2008. So all three of your production dates indicate it was built in late November or early December, with the ASIC rolling off the line first and waiting quite awhile for a board to sit on.
Yep. ASICs aren't always dated, but boards generally are. If you can't find a direct date on a card you can check the BIOS revision info for a build date with GPU-Z, or find a date on any supporting chip a card may have like a PLX bridge, or sometimes a VRM controller. It's not a guarantee those parts will match up well with the cards actual production date, but it'll give you a nice indication of, 'no older than X' timeline.
Well thank you for joining @Liquid Cool ! What a perfect name for my next post, because soon it´s time for some liquid cooling too.
Update on my EK-FC 280 waterblock.
Insides look a bit corroded, backside is ok.
My tip for getting the rubber gaskets out, something pointy but dull. You could use your average knife or a needle, but that just calls for an accident to happen. This plastic toothpick served me well:
I gave the copper two baths in citric acid. Not extremly high concetration and for about 30 minutes each time. Then a quick brush with toothpaste. For the acrylic make sure to NOT use any alcohol or aggressive chemicals! I had to remove the spots of corrosion from it and did so with a pencil eraser. You can use warm water with soap and a towel to wipe it off too. IF something happens with the acrylic and you see it get white/milky don´t use it on your block any longer. It get´s brittle and can crack very easy in that state, especially if under pressure from the screws and during heat cycles.
Results:
Btw I did not grease the gasket. Typically it´s not necessary for waterblocks I think and on top different types of grease are for different applications. In the worst case the grease could ruin the rubber = leakage.
Now I need a card for this. I´m already preparing a loop just for testing watercooling stuff, more on it when it´s done. With these different cooling solutions I wonder if I should add another chart with overclocking results on these cards. Do you want to see them run benchmarks?
Another update on my GTX 295 and the SLI issue: I have ordered some thermal pads, I´ve gone for several sheets in varying thickness to have some supply ready for future cards. This is kinda new for me and I was not thinking about everything I would need. I hope the 295 is not mad at me for letting it sit here another couple of days.
I did however make progress on the SLI-Issue. Huge thanks to @Mr.Scott for hinting at the driver version. I had the same BSOD escapade with my GTX 9800+ cards too when trying to boot on both. So I went for driver Version 332.21 and it booted no problem. However not in SLI. This was the moment I finally noticed that the EVGA is @55nm while the XFX is 65nm! Different device IDs. Both cards recognized by nvidia control panel, but only physX mode possible. Grrr...
I googled myself a whole encyclopedia of troubleshooting SLI configs and if it´s possible to use those two with the standard nvidia drivers. Everyone on the Internet back in the day reported it should not be an issue and that this case with the 9800 GTX cards is the only exception from the rule that only matching device IDs work.
Well it no longer works. I tried many different drivers, on about 50% of them I get a BSOD with pci.sys as soon as those two cards initialize on the nvidia driver. (They have no issue to boot up with the windows driver tho.)
I´m starting to believe that the chipset drivers for ryzen on windows 7 are a bit confused about what the hell I´m trying to do too. So I currently look into options on a system around LGA 775.
Today is the day we get back to that beautiful GTX 295. The pads arrived and I got to cover it with these squishy blue squares. Time to Go Down.
My first cut outs are a bit large, but I got better the more I tried.
EDIT: I initially went for 0.5mm thickness on the memory and VRM, but that does NOT make proper contact! Use 1mm all around.
I noticed the grey they had on there before had been a bit thicker then the white ones. It was very sketchy because I had no clue what thickness I had to go with but thinking the less the better, as long as they connect with the plate.
The tricky part is, that this plate on top is being held in place by screwing it together with the two seperate backplates, which each cover another set of memory on the back. So I had to cut and place all pads over every module and hope for the best. Really didn´t want to redo the whole thing. This card with it´s massive IHS surface and all the memory and chipsets and VRM needs about 1/3 of it´s total pcb surface, front and back, covered in TIM...
Well next up is placing the fan back on, taking a quick snap of it´s sticker:
Then I just had to re-paste it, attach the fan, put the little fan cable protection thing back in place and it should be done. Hopefully. As easy as one...
...two...
...and three.
You may have already noticed from the first pictures of it, that there is quite a gap between the plastic shell and the card. This is because some of the clips that hold it down are broken, and I did break another one on removal as I noticed, so now it´s only two left that hold onto the outer edges... It gives it a really beaten up and wonky look.
Maybe I can find a broken one with a decent cover for cheap. But it is time for the moment I was a bit anxious about. Will it run? Are temps ok? Did these damn pads make contact or will the VRM fail on me in furmark?
Boot went as normal, no beeps, LED on the back of card working. First sigh of relief. Driver install went fine, it boots back up and tells me SLI-compatible system recognized. Well it said that before too and crashed but atleast I didn´t break anything further!
New idle temps @ 20°C ambient this time (-2°C from initial test).
That is a small 2°C improvement compared to before, I´m ok with that.
But does it survive Furmark now with the new driver?
Yes it does!
It´s alive and feeling as good or even better then new!
68°C max while the system is inhaling 400W. I was really anxious during the first minutes of the stress test, probably hurt me more then the card. Afterwards I googled, because even tho my ambient is low, these numbers seem very low for this card. And indeed, these cards could get really hot, mostly due to lacking airflow. We should be thankful for modern airflow oriented cases with several 140mm fans. I remember my case back in the day had a single 120 as intake and a single 120 as exhaust...
I could have seen something like this: (NOTE not my card ofc., got this one from someone asking if this is a normal temp for the 295...)
This beast of a card was something I looked at back when I bought my GTX260, thinking to myself that I would never ever even get to see one of those. And now it´s sitting in my pc and I can play Sacred 2 on it with physX enabled, something that my 260 couldn´t handle. Oh I will enjoy this weekend
And I already have more cards lined up, moving slowly backwards through time.
You may have already noticed from the first pictures of it, that there is quite a gap between the plastic shell and the card. This is because some of the clips that hold it down are broken, and I did break another one on removal as I noticed, so now it´s only two left that hold onto the outer edges... It gives it a really beaten up and wonky look.
That happens on pretty much every card with those stupid clips. The plastic gets brittle and the clips snap. The GTX 295 doesn't suffer alone though, AMD used a nearly identical clip mount for the Radeon HD 6000 series reference coolers, and I'm always sweating bullets if I have to tear one of those down for cleaning. Worst design choice ever, screws work just fine.
That happens on pretty much every card with those stupid clips. The plastic gets brittle and the clips snap. The GTX 295 doesn't suffer alone though, AMD used a nearly identical clip mount for the Radeon HD 6000 series reference coolers, and I'm always sweating bullets if I have to tear one of those down for cleaning. Worst design choice ever, screws work just fine.
Now that´s cool. Engineering samples always have that special snowflake (in a good way) feeling to them. Love it!
And I agree on these clips, I´d be ok with anything else that doesn´t involve brittle plastic that can break. Make me do a raindance while yelling the lyrics of golden brown but pls no more of those clips. Even if you have a 3D-Printer ready it would be tough to replace...
I did have a dual PCB GTX 295 but sold it ages ago 
Oh how I wish I kept more stuff 
