Bastiaan_NL

Nice, I need to get me one of those converters asap so I can give that theory from @Avacado a try Thank you @Fluxmaven! And if you are planning to do another run for fun, the biggest bottleneck for me was CPU clock

Stupid question, but was it showing as 'SLI enabled' in nvidia control panel? I had mine running pretty good on a 4770k with a bit of an overclock and it was loading both cards. If you need live help or whatever hop on discord, usually I'm there with @Avacado as much as we can so you can always ask questions there (and some other guys too). I hope you can get it working so you can battle it out with @Fluxmaven We need you @tictoc, you can check the spreadsheet in the OP for basic info and links to the benchmarks. I bet you know how it works And that's a 1000 points

Of all those cards, I still ike the Zotac cooler design the most (as weird as it looks)... But I would still put a waterblock on it I'm always afraid that they sacrifice some board quality when they try to make a card that small. I know that the passthrough design leaves a lot of the PCB space unoccupied but still I feel like there is more hardware on the PCB. For now the Zotac card is on top of my list, let's hope they deliver quality..

But did you see the hwbot results for an AIO cooled 7950x?

Don't worry Jan, being there for your buddy is way more important than running a 2080Ti for a benchmark. Ah that sucks man, I would do the same as @The Pook suggested. You can just post the link here like I fixed in the quote above and it'll work just fine No pressure, but it would be nice to have that result on the board too We are in 5th now, proud of y'all

I'd like to bench a 4090 but not sure about keeping it.

A waterblock fixes a lot of issues, like bad looking large chunks of cooler..

I like those chassis for simple tech bench sessions, I had one running as F@H system in my bedroom too. Cheap and the quality isn't bad.

That's exactly what I thought it looks like

Sadly only two of those boxes are the originals, the others are just there to put the motherboards on something solid. When looking at your stream I noticed all the motherboards, you win for sure

I remember saying to @Avacado that 2 GPU's was the max I would get for the competition.... @Fluxmaven when is that AA meeting again?

And I might have purchased some extra stuff: iGPU DDR3 890GX incoming And a Phenom 960T:

Some stuff showed up, now I can do 3 more benchmarks for the team competition

For what you'll have to spend for a new board, ram and chip I totally agree. Your system doesn't get slower when new stuff comes out. Those results at 210W is impressive, I wonder what I draw when I'm playing around but it's never something I look at as long as it's not overheating

I think it was 800 on the z690 and for the X299 I can't check it, pretty sure it was over 1k. Some airflow over this board is not a bad thing anyways though

I'm pretty sure this board warns you about heat with the higher switching frequencies, the max frequency is way higher than on the ASUS Z690 board for example. I don't know how hot stuff gets around the CPU but some airflow over the board is not a bad idea if you go for that max.

@ENTERPRISE increased the AVX offset to see if that was causing the instability. I think I ran no offset at all, cause yolo, but my voltage was higher (yolo again ). As far as memory goes I think it should be able to clock a decent bit higher. I don't know what chips they are, but there is always something to gain. For now I would focus on the CPU though, 3200mem is fine for testing around 4.9ghz. I've been reading about voltages some time ago when I was playing with mine and I found a lot of 'opinions' on the matter. The biggest issue is heat, 1.35 on this 10c20t will be pretty hot so finding a stable overclock at 1.3v is a nice goal and so is 1.2v VCCSA.

Here you have your VCCSA, AKA Voltage Common Collector System Agent I just love how every manufacturer decides to give something their own name, so you have to search for hours in a bios just to find out it was right under your nose.. All of the other settings seem to be correct, or at least how I would do it. I agree with this as far as AMD vs older stuff goes, I don't have the same trouble with Intel though besides some funny things which are different between boards. In the end it's still boosting the BCLK or multiplier and upping a few voltages.

This time I thought about taking a picture

Looks good @Fluxmaven! A leaking fitting, dripping on the GPU... That stuff is scary

I should get me some of that But I'm not sure if I want to put two 3090's above an expensive motherboard with just that metal strapping. At least I know I can get the 3090's to work on that board, maybe one day it'll be a bit colder with a certain type of cooling

Yeah, I have a pair of 4.0 risers but I have no proper way to mount those heavy cards above the board at the moment. I used to have the board one level lower but to modify all that will take me a lot of hours. I managed to score about 330 global points with 5 results (not taking into account the loss of points from other runs dropping out of the list) so I'm pretty happy with that. Btw, the Superposition score was way lower with two cards than one as expected. Also, most of the hwbot benchmarks don't scale at all with NVLink. Only six benchmarks I did are worth it, five of them with good global points.

It might look like another one of my messy benchmark systems but this one was the most powerfull (and power drawing(and expensive)) system ever on my bench! 3090 FTW3 in NVLink with the 12900KF maxing out at 5.6GHZ in 8P with HT for some benchmarks. With the 1000W bios on both cards I saw up to 1500w powerdraw. It kinda sucks that the watercooled card doesn't fit in the first slot, I bet the score could be higher with a higher clocking card as 'display' card..

I totally agree, I'd love to have a kit but I don't want to waste a lot of money on it before knowing what I can do with it. With all the hardware coming out soon I need to watch my wallet a bit, don't want to spend it on something that turns out to be useless in combination with next gen AMD for example..