• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Rare GPUs / Unreleased GPUs

Hello everyone,

I have a few more items to share in the hopes of motivating myself to get around to thermal testing the next round of coolers that I discussed previously.

First off I acquired an engineering sample HD3870 X2 to go with the non-qualification sample HD3870 that I posted earlier.
View attachment 195893View attachment 195894

I also picked up a 7800 GTX 512MB which was a very short lived flagship that I've been after for awhile. Boy does it need a good dusting.
View attachment 195895View attachment 195896

Lastly I acquired two coolers that I consider groundbreaking in the history of GPU cooling. I plan to run some tests with both, a stock 3870, and the vapor chamber 3870 that I posted earlier but man its going to be a lot of work especially with different fan configurations.
View attachment 195897 View attachment 195898
I have one of those Accelero S1s (rather heavily modified) cooling the APU in my HTPC :D Works great, runs passively the vast majority of the time, with the 140mm fan I stuck to it just turning on occasionally. I used to have two on my CF 4850s back in the day. Still have the second one lying around too, should try to find a use for that as well!
 
Hello everyone,

I have a few more items to share in the hopes of motivating myself to get around to thermal testing the next round of coolers that I discussed previously.

First off I acquired an engineering sample HD3870 X2 to go with the non-qualification sample HD3870 that I posted earlier.
View attachment 195893View attachment 195894

I also picked up a 7800 GTX 512MB which was a very short lived flagship that I've been after for awhile. Boy does it need a good dusting.
View attachment 195895View attachment 195896

Lastly I acquired two coolers that I consider groundbreaking in the history of GPU cooling. I plan to run some tests with both, a stock 3870, and the vapor chamber 3870 that I posted earlier but man its going to be a lot of work especially with different fan configurations.
View attachment 195897 View attachment 195898

Very late 3870X2 and with the Ruby sticker I'm going to guess it's a vendor sample of some kind. Probably not much about it that isn't final spec. Anything earlier doesn't have Ruby on it. Very cool still.
 
TITAN Z Engineering Sample
01.png
 

Oh this is one of the press/review samples. They're the only ones that got the shroud stickers. These pop up on eBay every now and then, but they're not exceptionally common.

TITAN Z Engineering Sample
View attachment 199661

Heeeey yeah an even earlier board! What does the BIOS say and what dates are there on the BIOS/ASIC/board? Definitely cool! I still regret selling mine all those years ago.



Speaking of engineering samples...

Here's a neat one I found while browsing eBay. Another early RV670!

IMG_20210507_195741~2.jpgIMG_20210507_195803~2.jpgIMG_20210507_195818.jpgIMG_20210507_200352.jpgIMG_20210507_201127~2.jpg

This card is warped, scraped, missing parts, and totally inoperable in its current state so no BIOS dump yet. This ASIC is actually from the same lot as my 2950 Pro, but dated a week earlier. That makes this the earliest dated RV670 ASIC yet to be found on a completed board! With a date of 29th week 2007 this ASIC was cut and fused a mere 9 weeks after the 2900 XT (R600) launched to market, and 18 weeks before RV670 itself would appear as the 3870 (ignoring the 2950 phase and rumors from September 2007, I already mentioned all that in my 2950 Pro post ages ago.) That paints a very interesting picture for the end of R600's developmental life; 55nm RV670 ASICs being nearly in ATi/AMD's labs at the same time the beleaguered R600 was still being prepared for its debut. This is mostly interesting as it really punctuates how long R600 languished in development hell starting all the way back in October 2006 with the first delivery of A11 revision chips to ATi. In summary: It took so long for R600 to get to customers that its die-shrink/tweaked replacement on an entirely new unproven node was functionally ready for assembly testing and internal qualification. Imagine the alternate timeline of ATi/AMD scrapping R600 entirely after 6 months of rumors and just shipping RV670 with the original [R600] promised clocks of 800/1000 on a dinky board with that single 6-pin. (Although I kinda think R600s incredible stumble is what led to RV670s success in a way. Even with the same architectural problems, it was shown to be a substantial improvement. Without R600 there to compare to I think RV670 would have been very poorly rated by reviewers.)

Add in ATi's internal turmoil after the AMD buyout and there's probably a hell of a story hiding in there somewhere.
 
Last edited:
I have an ICEQ HIS 4670 1gb AGP card that might be rare.

Still almost brand new, only used it for about a month with a northwood p4 with HT.
been sitting in a box for probably 12 years now.

I'm pretty sure this was the best consumer agp card made
can post some pics if anyone wants to see it

I still have that old 478 msi mobo and chip too, I could set it up again and run some modern benchies.
I remember that p4 being a pretty big bottleneck though
 
I have an ICEQ HIS 4670 1gb AGP card that might be rare.

Still almost brand new, only used it for about a month with a northwood p4 with HT.
been sitting in a box for probably 12 years now.

I'm pretty sure this was the best consumer agp card made
can post some pics if anyone wants to see it

I still have that old 478 msi mobo and chip too, I could set it up again and run some modern benchies.
I remember that p4 being a pretty big bottleneck though

They're definitely sought after, but the HD 3850 AGP is still the higher performance card. You just can't beat the full 16 ROPs even with the improvements of RV730. The 4670 has the benefits of slightly higher ILP and better AA resolve (since it actually has dedicated hardware for it, rather than resolving AA through the SPs.)
 
I have an ICEQ HIS 4670 1gb AGP card that might be rare.
Nice. That fits well over in the thread below. Make sure to post pictures!
 
Last edited:
Oh this is one of the press/review samples. They're the only ones that got the shroud stickers. These pop up on eBay every now and then, but they're not exceptionally common.



Heeeey yeah an even earlier board! What does the BIOS say and what dates are there on the BIOS/ASIC/board? Definitely cool! I still regret selling mine all those years ago.



Speaking of engineering samples...

Here's a neat one I found while browsing eBay. Another early RV670!

View attachment 199900View attachment 199901View attachment 199902View attachment 199903View attachment 199904

This card is warped, scraped, missing parts, and totally inoperable in its current state so no BIOS dump yet. This ASIC is actually from the same lot as my 2950 Pro, but dated a week earlier. That makes this the earliest dated RV670 ASIC yet to be found on a completed board! With a date of 29th week 2007 this ASIC was cut and fused a mere 9 weeks after the 2900 XT (R600) launched to market, and 18 weeks before RV670 itself would appear as the 3870 (ignoring the 2950 phase and rumors from September 2007, I already mentioned all that in my 2950 Pro post ages ago.) That paints a very interesting picture for the end of R600's developmental life; 55nm RV670 ASICs being nearly in ATi/AMD's labs at the same time the beleaguered R600 was still being prepared for its debut. This is mostly interesting as it really punctuates how long R600 languished in development hell starting all the way back in October 2006 with the first delivery of A11 revision chips to ATi. In summary: It took so long for R600 to get to customers that its die-shrink/tweaked replacement on an entirely new unproven node was functionally ready for assembly testing and internal qualification. Imagine the alternate timeline of ATi/AMD scrapping R600 entirely after 6 months of rumors and just shipping RV670 with the original [R600] promised clocks of 800/1000 on a dinky board with that single 6-pin. (Although I kinda think R600s incredible stumble is what led to RV670s success in a way. Even with the same architectural problems, it was shown to be a substantial improvement. Without R600 there to compare to I think RV670 would have been very poorly rated by reviewers.)

Add in ATi's internal turmoil after the AMD buyout and there's probably a hell of a story hiding in there somewhere.
This TITAN Z engineeing sample has more power limit(default 250W,max 265W each core)and more frequency(default 876Mhz,Boost 928Mhz)
and date of parts(GPU: 2013 week 45,PCB: 2013 week 47 ,BIOS Bulid Date: 2013-11-26)

02.png01.png

Thank you share this cool RV670 engineering sample and good story about this card. I used to heard a specific version of RV670 engineering sample got 900Mhz default core clock but i dont have clear evidence to prove it 100% ture. R600 delay too much in that days and shown poor performance,this reason maybe promote a secret plan of this card.
t
 
Anyone have any idea what these dip switches do? Its a 5700 non-xt engineering sample card. Got it off ebay "dead" but it just had a Vega 10 bios on it :P

Works fine aside from not running OpenCL at all. Miners flat out crash and GB5 gets about 30 second in the OCL bench. Looks like the voltage isnt boosting at all when running OpenCL while clocks remain high and occasionally mem voltage drops to 1.2v. I suspect the dip switches are all for the voltage controller, and some of them were on when I got the card but I turned them all off when I was trying to fix it. Runs games completely fine however and gets 25k in Firestrike in this condition.
 

Attachments

  • image0.jpg
    image0.jpg
    1.3 MB · Views: 155
  • download.jpg
    download.jpg
    4.7 MB · Views: 168
Anyone have any idea what these dip switches do? Its a 5700 non-xt engineering sample card. Got it off ebay "dead" but it just had a Vega 10 bios on it :p

Works fine aside from not running OpenCL at all. Miners flat out crash and GB5 gets about 30 second in the OCL bench. Looks like the voltage isnt boosting at all when running OpenCL while clocks remain high and occasionally mem voltage drops to 1.2v. I suspect the dip switches are all for the voltage controller, and some of them were on when I got the card but I turned them all off when I was trying to fix it. Runs games completely fine however and gets 25k in Firestrike in this condition.
But have you pushed the red button? :laugh:
With my luck it would self destruct.
 
Watch it go for $2500.
it's not my preferred cooler for the NV30, won't be bidding i don't think. just love the NV30 Story, and supposedly the OTES cooler was the inspiration for Flow FX? don't know for sure

Translucent Green Flow FX (ftw)
 
It's currently up to $1026. WTH? It's a rare card but damn! Not THAT rare.. Credit where it's due, Abit scored a win with that card.
Some are questioning this particular card’s authenticity. They say it’s a qaudro fx 2000 pcb mashed with another siluro fx 5600 ultra heat sink and cooler components
 
Some are questioning this particular card’s authenticity. They say it’s a qaudro fx 2000 pcb mashed with another siluro fx 5600 ultra heat sink and cooler components
Who is questioning? Regardless, if you look at the pictures in the ebay page, there are photo's of the VGA BIOS boot screen. Additionally, I've seen that card in person. Sold and installed a bunch of them at the PC shop I worked at BITD. Looks perfect to me. I'm not questioning the legitimacy of the card, I'm questioning that it has $1000 value.
 
Who is questioning? Regardless, if you look at the pictures in the ebay page, there are photo's of the VGA BIOS boot screen. Additionally, I've seen that card in person. Sold and installed a bunch of them at the PC shop I worked at BITD. Looks perfect to me. I'm not questioning the legitimacy of the card, I'm questioning that it has $1000 value.
Heard from fellow collectors on a discord and Facebook group this morning about the questioning, I’ve never personally seen one of these abit siluros in real life so unqualified to speak to it myself
 
Having my doubts about this card after looking can’t find that there ever officially was a FX5800 Ultra From Abit Siluro
Don't know what to tell you. Abit didn't make many video cards so data on them seems to be limited. I know I installed a bunch of them, 5800 Ultra's included.
 
Back
Top