Which system should i keep for gaming? (older i9 or newer xeon)

[masterdeejay]

Thanks for your service.
I don't care to deal with Intel's first-gen hardware or its early drivers but I'm pleased people are buying them, that's what will bring competition back to the stagnant value proposition of the AMD+Nvidia duopoly.

I really like this card. I dont have any issues. Only the oldest games are slower than my old tesla m40 (titanx) but still very playable. I can use dxvk to make older titles faster.
The newer games (dx11-dx12) is way faster.
High quality, almost silent but little coil whine. I think it is a better deal than 3060 or 6600xt if it is cheaper. Raytracing performance is much better than 6600xt.

 

[Mussels]

All-core frequency is not that important, for gaming what can be sustained on 2-4 cores is what matters.

I'd say what they sustain on 4-6 cores, going by how modern games are performing. That gap is starting to show in things like the 3+3 ryzen setups where 4 is becoming a minimum


Also keep the cache sizes in mind, as the ryzen x3D chips have shown higher cache sizes help minimum FPS quite a lot - if you get chips that are more less equal with their boost clocks but one has triple the cache, you'd go with that one.


Looking at the specs:

Gold 6138

silver 4215R

7940x

Looks like the 7940x has the highest boost by 400Mhz/10%, and a lot of cache
Only negative is the higher TDP, but i'm sure like all CPU's its possible to control that in the BIOS if need be for cooling purposes


I dont trust these automated websites much, but this one claims the CPU is equal to a 9900K (at least at stock)
i9-9900K vs i9-7940X - Fortnite Battle Royale with RTX 2080 Ti Benchmarks 1080p, 1440p, Ultrawide, 4K Comparison - GPUCheck United States / USA

 

[Chrispy_]

I really like this card. I dont have any issues. Only the oldest games are slower than my old tesla m40 (titanx) but still very playable. I can use dxvk to make older titles faster.
The newer games (dx11-dx12) is way faster.
High quality, almost silent but little coil whine. I think it is a better deal than 3060 or 6600xt if it is cheaper. Raytracing performance is much better than 6600xt.
View attachment 284537View attachment 284538View attachment 284539View attachment 284540View attachment 284541View attachment 284542View attachment 284543

Yeah, from what I'm seeing around the web, the drivers are probably now at the point where they should have been at launch - in that most of the worst problems are fixed.
Unfortunately the drivers were absolutely shocking at launch five months ago, and first impressions count.

I have no need for GPUs right now but they look good enough that I'd seriously consider one as an alternative to a 3060 or 6600. I'm still not sure I can recommend them to friends who want a hassle-free GPU upgrade though, they're probably still reserved to the class of PC gamer who is okay with replacing DLLs, messing with .ini files to get older games working properly and launching games with additional command-line switches.

 

[P4-630]

Im running the dimms at 3200mhz 19-19-19-43-CR1 1.4v and the ram temps are fine 45-55C on memtest. It is passed for half hour.
I hope i can oc to 3600.

3dmark timespy (cpu clocks at default non oc, but mesh and ram is oc-d to 3200)
Video card Intel A770 LE 16Gb at stock.
View attachment 284309
View attachment 284316

If you went with a i7 12/13700K you would have a more powerful gaming PC.

 

[masterdeejay]

If you went with a i7 12/13700K you would have a more powerful gaming PC.

And that is cost 2x much more than this... for what 20-30% extra fps? I have a fullhd 60hz monitor, and my card just an intel a770 that is not botlenecked by the cpu for 60fps.
I got the cpu for 270 usd, and the mobo for 160 usd (1y wr) and the dimms was free because i have a lot of ddr4 ecc reg from my older servers.
And i need a second at least 8x pcie slot for my raid card (few sas drives that was also free from older servers)
So i think the motherboard with same features like this is not cheap for a new platform.

 

[Chrispy_]

And that is cost 2x much more than this... for what 20-30% extra fps? I have a fullhd 60hz monitor, and my card just an intel a770 that is not botlenecked by the cpu for 60fps.
I got the cpu for 270 usd, and the mobo for 160 usd (1y wr) and the dimms was free because i have a lot of ddr4 ecc reg from my older servers.
And i need a second at least 8x pcie slot for my raid card (few sas drives that was also free from older servers)
So i think the motherboard with same features like this is not cheap for a new platform.

not even 20-30% extra FPS.

The 12900K is only going to be faster if you give it a 4090/4080. AMD and Intel GPUs don't have the same CPU overhead as Nvidia, and less powerful GPUs mean that the GPU is rarely waiting on the CPU for anything unless you do what CPU reviewers do which is test lower resolutions to artificially increase the framerate beyond normal GPU bottleneck levels.

With an A770 and 1080p 60Hz monitor, I doubt the 13700K could add any benefit whatsoever.

 

[masterdeejay]

I started building a custom low cost water cooling for this rig.

Alphacool nexxxos 1080 radiator used only 30 usd (i got it)
I have a few Arctic P12 fan, that is the cheapest (5 usd new, i have 6 pcs)
EK-Quantum Velocity D-RGB Intel (never used) 75usd (ordered)
XSPC D5 Dual Bay Reservoir (used) 61 usd (not ordered)
ekwb d5 pump reservoir 60 usd used ordered
fittings 2-3 usd, i think i need 6-8 pcs.
tubing some cheap 10/13

So approx the total custom water loop system cost 230~250 usd. It is a little higher than a good aio but much better performance and quality.

 

[Mussels]

faster CPU's (single threaded faster) only help for high FPS situations - either fast Vsync above refresh rate, or high refresh monitors

Chrispy has it right there, for 1080p 60hz you'll only be changing/raising minimum FPS values and 0.1% lows by changing things around - it's not a demanding monitor setup



Having those extra cores idle and dedicating all the cache to the game, and an FPS cap keeping spare wattage/TDP for CPU and GPU to spare is going to give you a good low latency gaming experience - I dont know if intel have a fast Vsync equivalent yet, so a 58FPS cap at 60Hz might be your best setup

 

[masterdeejay]

i built a custom loop, and ocd the cpu to 4.5ghz 1.2v. But something is not right, i need to check it.
In games its working at 40C average, but on benchmarks it is overheating so i got lower points than expected.

 

[repsol23]

I have been following this simply because I have a 10900x on an x299 Asus Prime board and I was curious as to what your results would be. Looking at that last post I feel I was in the exact same temp scenario once I got into 1.2 and beyond on volts. I run my cpu at 4.6 on all cores with 1.266 V and in cinebench I have seen as high as 99C. To run 4.7 for me requires 1.350 and in minutes I easily see 105C so that was a no go for me. When gaming (tend to play escape from Tarkov only) the cpu package will show at max 70C and avg around 50's. Only thing I did to bring down temps slightly was change out my paste on CPU to corsair xtm70 and actually shaved off around 6 - 12 degrees compared to the AS5 I had used (old tube laying around). I think some of that drop was also related to a better job of getting the right amount pressure as too much was causing ram detection issues and I think I was too little pressure prior.

I do have a ton of radiator space, 2x320 and 2x220, with cpu and gpu (3080ti) in same loop. All corsair ml120 pros (x12) with fixed rpm at 1000 and ml140 pros (x4) at 850 rpm.

 

[ThrashZone]

Hi,
I can sum that r-23 run up as "Terrible contact" so yeah failed delid 78min 102max = 24c difference.
If you were at or within 10c differance you did well contact wise.

 

[repsol23]

Hi,
I can sum that r-23 run up as "Terrible contact" so yeah failed delid 78min 102max = 24c difference.
If you were at or within 10c differance you did well contact wise.

Yeah, forgot to even mention that. I had a difference of around 28C on my hottest/coolest cores. Once I adjusted the pressure correctly I am now seeing around 8C difference across my cores in benches and 5C in games. Noticed this improvement right away!!

 

[ThrashZone]

Take the oc off and do a bare bones stock run and see how bad the temp spread is than
Here's mine at stock I'm at 4c difference.

 

[masterdeejay]

I tightened the block screws, and turned up the pump to the max (it was on the lowest setting). It pumped out excess air from the loop.
Much better but still not good. (point not relevant, a few things run in the background)
it is difficult to access the screws so it is not fully tightened. I will ask someone with a smaller finger to tighten it.
With prime95 it maxed out at 70-80C (older version whitout AVX).

With the new dimms, still bad timings but a good start:

 

[ThrashZone]

Hi,
Your priorities are all messed up

 

[masterdeejay]

Hi,
Your priorities are all messed up

Totally.
It is now a fun project.

 

[Mussels]

ghtened the block screws, and turned up the pump to the max (it was on the lowest setting). It pumped out excess air from the loop.

Can you open er up and fill with distilled/clear coolant and top the juice up? Did this to my EK AIO, and it really helped reduce the startup gurgles

Can't really see the whole setup, but i assume you could find a spot to bleed it out


Edit: Oh right, the printer. The printer is your bleed/fill port, got it.