rjbarker
Still have this new system set up on my bench table....
3080Ti OC (not too extreme as still on air) +165 / +1100 for some runs, some runs 120 / 675...
No OC at all on the CPU, just XMP enabled.
 
Will be finishing up the test bench stuff over the next few days then install into my 900D with Loop,,,Velocity 2 and Vector Water Blocks..Rads RX 480, RX 360 and EX 240
 

That 3080 ti looks beautiful. Ugh. I wish I had queued for that instead.

ericbartman
Lol!!! Yes you are. I am sure you know that there is a difference between time spy extreme Stress test and time spy extreme right. Just wished to make sure.

Again...thanks for clarifying man....I'm a complete noob with all this stuff ;)

All good. LOL!! noob or not, gotta make sure that we are all on the same page. ;)

Well I almost hit 21,600 for my GPU score, made it to the top 1 %!!!!
 

 

Eddiewax6150
Well I almost hit 21,600 for my GPU score, made it to the top 1 %!!!!
 
 
 

Pushed the GPU Clocks a bit ,,,still on air though,,,,  +180 / +1175
 
https://www.3dmark.com/spy/25520312
 
https://www.3dmark.com/pr/1390578
 

rjbarker

Eddiewax6150
Well I almost hit 21,600 for my GPU score, made it to the top 1 %!!!!

Pushed the GPU Clocks a bit ,,,still on air though,,,,  +180 / +1175
https://www.3dmark.com/spy/25520312
https://www.3dmark.com/pr/1390578

I never really pushed mine on air. I waited to push seriously until the Optimus block came in.
If you're interested, here's a head-to-head of your Time Spy vs. mine.
Same comparison with Port Royal.
 
I'm quite fascinated how badly your 12900k trounced my 5900x in the CPU portions of Time Spy, given my 5900x is pushed as aggressively as I know how and my hand-tuned RAM is running within nanometers of as fast as my given CPU could possibly run even in theory. That kind of sums up where the CPU lines are at, at least for loads similar to what Time Spy does. Anyhow these comparisons at least show somewhat the headroom that can exist when you go from air to water cooling on the 3080ti.

sethleigh

rjbarker

Eddiewax6150
Well I almost hit 21,600 for my GPU score, made it to the top 1 %!!!!

Pushed the GPU Clocks a bit ,,,still on air though,,,,  +180 / +1175
https://www.3dmark.com/spy/25520312
https://www.3dmark.com/pr/1390578

I never really pushed mine on air. I waited to push seriously until the Optimus block came in.
If you're interested, here's a head-to-head of your Time Spy vs. mine.
Same comparison with Port Royal.
 
I'm quite fascinated how badly your 12900k trounced my 5900x in the CPU portions of Time Spy, given my 5900x is pushed as aggressively as I know how and my hand-tuned RAM is running within nanometers of as fast as my given CPU could possibly run even in theory. That kind of sums up where the CPU lines are at, at least for loads similar to what Time Spy does. Anyhow these comparisons at least show somewhat the headroom that can exist when you go from air to water cooling on the 3080ti.

Yep this 3080Ti Card under water with a Vector Block on my 9900K system did very well OC'ing, would expect the Avergae Boost speed to be somewhat faster.....Im just about ready to move everything into my 900D and Custom Loop.
This is an RMA 3080Ti  (refurb) so I havent tried it yet underwater.....my 3080Ti that $%&* the bed (pop n poof...Red Light District after 3 months) would reach +180 / +1200 Max, but the Average Boost was higher due to max temp of 43c, we'll see how this Card does under water.
 
The scores are damn good when comparing to most 3090"s (about the same or a tad less ) and only 1K less or 1.5K less than a premium $$$$ KP Card....all good!
 
Yes the 12900K is a beast!!!
Also this 12900K is completely stock (no OC at all), I will start playing with OC'ing it once I have it in my Loop....thanks for the comparison...great feature ;)
 

rjbarker
Also this 12900K is completely stock (no OC at all), I will start playing with OC'ing it once I have it in my Loop....thanks for the comparison...great feature ;)

The boldest part scares me. Metaphorically speaking of course, since I'm perfectly happy with my machine. Your stock 12900k did that to my 5900x that was pushed as hard as I know how to push it, with RAM running close to theoretical best possible. That is beastly.
 
I did some comparison of my score (that score is currently #1 for the 5900x/3080ti combo) with 5900x/3090 machines and they're not only beating my GPU scores by a k or so because they're 3090s but most of the top several dozen scores also seem to be running manual CPU overclocks with locked all-core speeds of like 4.8GHz or so, which is higher than the all-core speeds I'll see running my CPU pushed with Precision Boost Overdrive 2 and the Curve Optimizer. Those are AMD tools that allow you to "OC" the Ryzens by coaxing the CPU to boost higher and for longer than they would in stock configuration. It's an amazing utility but you don't end up with as high of all-core speeds as you can do manually with a forced locked-speed manual OC. The tradeoff is that under non-all-core situations my CPU will boost up over 5GHz while theirs won't exceed whatever they're locked at, in this case 4.8GHz. I also allow my CPU to boost up and down, with the attendant drops and rises in voltage, as the loads change. Since my machine isn't going balls to the wall most of the time I'm happy to let it rest in between games, renders, or benchmarks. Keeps my room a little cooler too. :-)
 
I wondered why I managed to capture #1 for the 5900x/3080ti class and then realized that all the "big dogs" who are playing the benchmark game seriously go all out just for these benchmarks with their locked manual overclocks and whatnot. I guess 3080ti isn't the highest of the high end so doesn't attract the same attention. The GPU OC I used for those Time Spy and Fire Strike Extreme runs is not one I use in any game, nor would I want to - it was balls to the wall. Nothing I play will notice the difference between that "benchmark stable" Time Spy OC and the realistic OC I actually use daily. My CPU and RAM OC, on the other hand, are my daily driver profiles.
 
I'd be interested in seeing how high you end up getting with your 12900k once you start pushing it. Assuming you have the cooling and PSU and whatnot to support that, which I'd assume you do.
 
Oh, if you're waiting on an Optimus block for the 3080ti FTW3, that thing is a complete beast. Trying to come up with a colorful way to describe it, all I came up with was a Swiss bank vault door mated with a tank and gave birth to a 3080ti FTW3 with an Optimus water block on it.

sethleigh

rjbarker
Also this 12900K is completely stock (no OC at all), I will start playing with OC'ing it once I have it in my Loop....thanks for the comparison...great feature ;)

The boldest part scares me. Metaphorically speaking of course, since I'm perfectly happy with my machine. Your stock 12900k did that to my 5900x that was pushed as hard as I know how to push it, with RAM running close to theoretical best possible. That is beastly.
 
I did some comparison of my score (that score is currently #1 for the 5900x/3080ti combo) with 5900x/3090 machines and they're not only beating my GPU scores by a k or so because they're 3090s but most of the top several dozen scores also seem to be running manual CPU overclocks with locked all-core speeds of like 4.8GHz or so, which is higher than the all-core speeds I'll see running my CPU pushed with Precision Boost Overdrive 2 and the Curve Optimizer. Those are AMD tools that allow you to "OC" the Ryzens by coaxing the CPU to boost higher and for longer than they would in stock configuration. It's an amazing utility but you don't end up with as high of all-core speeds as you can do manually with a forced locked-speed manual OC. The tradeoff is that under non-all-core situations my CPU will boost up over 5GHz while theirs won't exceed whatever they're locked at, in this case 4.8GHz. I also allow my CPU to boost up and down, with the attendant drops and rises in voltage, as the loads change. Since my machine isn't going balls to the wall most of the time I'm happy to let it rest in between games, renders, or benchmarks. Keeps my room a little cooler too. :-)
 
I wondered why I managed to capture #1 for the 5900x/3080ti class and then realized that all the "big dogs" who are playing the benchmark game seriously go all out just for these benchmarks with their locked manual overclocks and whatnot. I guess 3080ti isn't the highest of the high end so doesn't attract the same attention. The GPU OC I used for those Time Spy and Fire Strike Extreme runs is not one I use in any game, nor would I want to - it was balls to the wall. Nothing I play will notice the difference between that "benchmark stable" Time Spy OC and the realistic OC I actually use daily. My CPU and RAM OC, on the other hand, are my daily driver profiles.
 
I'd be interested in seeing how high you end up getting with your 12900k once you start pushing it. Assuming you have the cooling and PSU and whatnot to support that, which I'd assume you do.
 
Oh, if you're waiting on an Optimus block for the 3080ti FTW3, that thing is a complete beast. Trying to come up with a colorful way to describe it, all I came up with was a Swiss bank vault door mated with a tank and gave birth to a 3080ti FTW3 with an Optimus water block on it.

Technically the "System" is OC'd with XMP enabled (XMP I 5600 / Cl36), but the CPU is completely stock since putting this together 3 days ago.
 
This 12900k Z690 is replacing a 9900K / Z390 Maximus XI-E, but its going into the same case with the same Custom Loop.(see my Mods Rig, the system will pretty much look the same)...as in 480*360*240 Rads (fairly large loop that takes about 1.5L of coolant to fill).
 
Once I start playing around with OC'ing I'll post back. I was able to run my 9900K with this same Loop all cores @ 5.1 Ghz (24/7) and 5.2 - 5.3 Ghz for bench runs.
This 12900K Im going to take a different approach to OC'ing, as in "per Core" vs "All Cores".....should be able to get 5.3 Ghz to 5.4 Ghz I would think (for say 6 Cores), then 5.2 - 5.3 Ghz for 8 cores.
We shall see ;)
Yes the 12900K not only exceeds your 5900X but in most cases also exceeds the 5950X, I had been looking at upgrading my Z390 / 9900K to possibly a 5950X, but waited a few months as I knew Alder Lake was debuting in Oct and the benchmarks I saw amazed me.
Heat is an issue with these chips, as in "no to air", imho minimum 240 aio (its what I have with it currently on my bench) although I would recommend 360 aio or custom loop.
 
btw all my System Specs are in my Signature below....PSU is a Corsair AX1600i ;)
 
One thing after building custom pc's since 2012.....there is always something faster n better around the corner ;)
Will be interesting to see AMD Zen 4...I'll be really curious to see what Intel Meteor Lake looks like in 2023! Raptor Lake I dont expect much more than what Intel has now in Alder Lake.

Got the hybrid cooler and installed on my 3080 ti ftw3.. it still power limits at anything over 420 watts, but that's okay. It's dead silent at 50% fan speed and Noctua NFA25s. VRAM temps did increase by about 8C as I expected, but still staying under 82C so far. Previously they always hovered in the 70Cs. I placed 2.5mm thermal pads on the back to give extra VRAM cooling. Honestly not sure if it helped. Love this GPU and $1000 less than going on Reddit and paying $2700ish for 3090 FTW3 that's maybe 2-3% faster.

I also managed to dethrone all of my old 3090 FTW3 Hybrid scores.. somehow? LOL

Port Royale 14623
Time Spy 19718 - GPU Score: 21921
Time Spy Extreme 10334 - GPU Score: 11072
 
The AIO is definitely worth it.. Just sucks that I can no longer use my ITX cases. I collected quite a few over the years and I think the only one that could accommodate it is the FormD T1, but then my poor 5900x suffers with a super small CPU cooler. I saw a Noctua 92MM fan mod someone did on their FTW3 that could be an option. I could also go full custom liquid cooling, but the reason I rarely do this is cost and the fact that I just love to swap out PC parts so often.