So I have an asus strix oc and I do alot of benching. I am looking into getting another 2080 ti for for Nvlink I have read that it works with mismatched cards as in SLI in general. But does it work well with benching 2 different cards? I know one will downclock to the other but I am afraid of things not quite working right when spending this amount of money. So I was hoping someone in these forums has experience in this particular area? And if so with one card being able to overclock better than that the other its basically just going to match the speed of my Asus card, that Asus card is a very good Overclocker with the Asus unlimited bios. I can get 2190 on core and 8300 memory through firestrike extreme. Managed to get 45th on x1 GPU in firestrike extreme so as long as I can the kingpin to match this cards capabilities I will be happy for SLI. This will also allow me to try and better just my single GPU score as well.

so the main problem here will be in card width and I think ftw3 and kingpin a wider than rog strix.
kingpin is 139.3 mm wide when rog strix is 130.30 mm. and reference is 115.7 mm.
 
and also it is a problem find a 2nd gpu that can get up to 2190mhz. mine strix was able to run at 1950-1980 only while my ftw3 can go up to... 2130 max? 2085-2055 mean?

well, that is not only about money.
kingpin is wider and there is no flexible nv link bridges, right? so you just cant do it at all.
 
also big problem here to win silicon lottery 2nd time in a row.

wyattjneill
https://www.3dmark.com/fs/21646579

You can see the core and mem clocks on the 3dmark result.

In this post I said firestrike extreme which I checked was 2175 core but regular firestrike is 2190 I'm 45th on firestrike 1x so I miss stated that. My current strix is on water as well as shunt modded, and the lack of kingpin blocks worry me at the same time, for the sake of compatibility and cleanliness I'm leaning towards another strix card now.

Thank you your comments probably just going to pick up another strix this week.

Hummm.   Ran firestrike (once, ever) a while back, on my normal gaming OC settings, beat that GPU score.  Temps around 50C max on an XC+hybrid kit with push pull fans.
 
https://www.3dmark.com/fs/20383702

wyattjneill
Well as you see you are running a 9900k at 5ghz which the higher clocks do wonders for fps that's the reason your GPU score is better.

I did not know that, so the gpu score is partially impacted by the CPU performance?  How many threads does it prefer?  Thanks!

wyattjneill
Yes that's the reason the 9900k is still gaming king threads do matter, but not quite as much as the frequency I have some scores with a 9900k at 5.5ghz and this 2080 ti I'll try to find them for you and edit this.

https://www.3dmark.com/spy/6469824

No firestrike extreme ones but timespy for comparable reasons

Night raid
https://www.3dmark.com/nr/79738

Port Royal
https://www.3dmark.com/pr/59673

 
Cool thanks, here's my best timesp - normal gaming gpu OC, stock 9900K clocks and potato memory.
 
https://www.3dmark.com/spy/8734958
 

wyattjneill
Not bad though, that memory is haha here is my best TS

https://www.3dmark.com/spy/10296823

I've managed to at least get on half the x1 GPU leaderboards used to be top 10 in one.

Yeah thanks, have upgraded to 2x16gb of 4000CL19 that runs stable at XMP on my board.  Picked up about 8% in timespy cpu.

emmett
You could use ribbon riser cables and alternate mounting, it would not look great, but should work?

That is a perfect idea, I can add washers where the pcie sockets into the bottom of gpu and rise it accordingly, add longer screws and washers where the IO bracket mounts and I could absolutely make it work. This is a very good idea I have the cables I would just need the vertical mount for the 011D xl. Great idea thank you for your reply.