Which PC Build would you choose out of these two?

ronferri

ronferri

ronferri

Associate
Joined
31 Oct 2017
Posts
81
Which PC build would you choose from the below for heavy competitive online shooters for lowest latency, highest FPS and highest fps dips? Both are same exact same price.
1)
-Motherboard: MSI Z790 EDGE Wifi, 16+1+1 power stage at 90A
-CPU: i9-13900KF
-GPU: MSI Trio X 4090
-RAM: DDR5 Kingston Fury Beast 5600MHz 16GBx2 unknown CL
-EKWB AIO 360mm liquid cooling (older model), top mounted
-2 Front Case Fans Intake, and 1 Rear Case Fan Exhaust.-Case: H700 based on coolermaster design

OR

2)
-Motherboard: MSI MPG Z790 Carbon Wifi, 19+1+1 power stage at 105A
-CPU: i9-13900KF
-GPU: TUF 4080 OC
-RAM: DDR5 Kingston Fury Beast 6000MHz CL36, 36-38-38-80, 16GBx2
-EKWB EK-Nucleus AIO CR360 Lux (newer model), top mounted
-2 Front 140mm Case Fans Intake, 2 Bottom Noctua 140mm intake fans, and 1 Rear Case Fan Exhaust.
-Case: Fractal Meshify 2

Considering thermal performance and heavy gaming performance, which one would you choose? (both are Same exact price)
 
Last edited:
4090 is a more powerful card so it should easily have the best performance in the majority of games (and higher fps = lower latency), BUT it does depend on which games you're playing and what resolution/settings you plan to use, since the memory speed has a bigger impact when you're CPU bottlenecked and for 13th gen both PCs have slow RAM (slow if you're going to be heavily CPU-bottlenecked most of the time).

If you're competitive gaming (i.e. you're not just playing competitive games, you're actually competitive) then I'd suggest checking out some benchmarks so that you know where the bottleneck is likely to be and where best to spend your money.
 
What I am trying to point out are 4 things, the 4080 build has:
-A mobo that can deliver more power to the i9-13900KF,
-also has a better AiO cooler for CPU,
-2 extra 140mm case fans for increase airflow,
-DDR5 with slightly faster speed and tighter timings
 
Tetras said:
4090 is a more powerful card so it should easily have the best performance in the majority of games (and higher fps = lower latency), BUT it does depend on which games you're playing and what resolution/settings you plan to use, since the memory speed has a bigger impact when you're CPU bottlenecked and for 13th gen both PCs have slow RAM (slow if you're going to be heavily CPU-bottlenecked most of the time).

If you're competitive gaming (i.e. you're not just playing competitive games, you're actually competitive) then I'd suggest checking out some benchmarks so that you know where the bottleneck is likely to be and where best to spend your money.
Click to expand...
the z790 motherboards have compatibility lists for DDR5 kits and for those two MSI mobos these are the only two compatible DDR5 kits available in my market. Nothing over 6000MHz. Not sure what you mean by bottlnecks when I choosing the top of the line cpu and gpu available. Just because the DDR5 is 6000MHz and not 7200MHz does not make it a bottleneck in Battlefield 2042 or warzone. The 7200MHz+ are very hard to find and be compatible, mostly scalped, and very expensive.
 
ronferri said:
-A mobo that can deliver more power to the i9-13900KF,
Click to expand...

According to TPU's review of the 13900K, in games the average power consumption was only 117.9 watts and even if the CPU reaches 150-200, my judgement would be that it is very unlikely that either board will VRM throttle while you're gaming. If your PC was likely to do long-run heavy CPU workloads (I mean, like Blender), then I would care more about the VRM.

ronferri said:
-also has a better AiO cooler for CPU,
Click to expand...

I don't know the difference between these two models, but as with the board, I doubt your CPU will hit 100 degrees while gaming, with a 360mm AIO (unless it is crap).

ronferri said:
-2 extra 140mm case fans for increase airflow,
Click to expand...

According to my googlage, the HAF has 2x 200mm front fans pre-installed, so I'm not worried and if I were worried, fans cost something like £10-£20 and aren't usually that hard to install, while the 4090 is (an estimate) around £300 more expensive.

ronferri said:
-DDR5 with slightly faster speed and tighter timings
Click to expand...

Yeah, but it is a very small difference, I think you'd have to be playing at a super low resolution, with very low settings, for this to make a meaningful impact on your FPS and in many games you won't have this kind of bottleneck, so I'd never dump a 4090 for it.

ronferri said:
Not sure what you mean by bottlnecks when I choosing the top of the line cpu and gpu available. Just because the DDR5 is 6000MHz and not 7200MHz does not make it a bottleneck in Battlefield 2042 or warzone. The 7200MHz+ are very hard to find and be compatible, mostly scalped, and very expensive.
Click to expand...

The reason I say this is because competitive gamers often have VERY specific requirements.

For example: competitive gamers often play at very low resolution and/or settings and may play old games (not to mention that esports are generally not that demanding anyway) so they can be hitting a CPU bottleneck in scenarios that 95% of gamers would not be. This means that they balance their system differently and might, for example, buy extremely fast memory that is nearly irrelevant to the other 95% of gamers.

BF 2042 and Warzone are fairly recent games, pretty demanding and I wouldn't really class either game as "esports", but I guess that's a subjective definition.
 
The second option has a CARBON mobo which has 19+1+1 DIRECT phase stages at 105A, whereas the first option has the cheaper EDGE WIFI mobo which has 16+1+1 MIRRORED phase stages which means using phase controllers to try and replicate the effect of 16 direct phase stages. This affects how clean and stable the voltage is reaching the CPU which is very important even for non-overclockers. This is why I expected more engagement on a platform such as this. Throttling due to fluctuating voltage can hurt a competitive player in a CPU-bound online game. Hence I do not find it an easy choice between the two builds.
 
Last edited:
ronferri said:
What I am trying to point out are 4 things, the 4080 build has:
-A mobo that can deliver more power to the i9-13900KF,
-also has a better AiO cooler for CPU,
-2 extra 140mm case fans for increase airflow,
-DDR5 with slightly faster speed and tighter timings
Click to expand...
None of that is going to make a dent in closing the gap between a RTX 4080 and a RTX 4090. It's like buying a regular VW Golf and adding a sports air filter thinking it might make it faster then a Golf GTI which is clearly won't.
 
ronferri said:
The second option has a CARBON mobo which has 19+1+1 DIRECT phase stages at 105A, whereas the first option has the cheaper EDGE WIFI mobo which has 16+1+1 MIRRORED phase stages which means using phase controllers to try and replicate the effect of 16 direct phase stages. This affects how clean and stable the voltage is reaching the CPU which is very important even for non-overclockers. This is why I expected more engagement on a platform such as this. Throttling due to fluctuating voltage can hurt a competitive player in a CPU-bound online game. Hence I do not find it an easy choice between the two builds.
Click to expand...

I'm happy to be proved wrong, but I find it very unlikely that this would have such a big impact on the frametimes, most (all?) of the high-end boards are built way overspec for gaming load.

You're saying that the motherboard's VRM has a bigger impact on performance when paired with a 4080, than the performance benefit from a 4090 and there is no mainstream source (on YouTube or magazines) that says anything like this. The general consensus is that if the motherboard has a VRM powerful enough not to throttle, then the performance difference between motherboards is zero.

If you have such specialised knowledge, then you know more than most of us.
 
I wish but 7800X3D is not available unless I order online and assemble myself. Definitely better value and efficency.

The bottom line: Yes I will definitely get Higher Max FPS and higher Avg FPS on the 4090 build. But it is the Dips and microstutters that I am worried about when a player randomly jumps in front of my screen with sudden explosions and war effects. There is a chance I might get lower dips on the 4090 build than on the 4080 build, and based on over 20 years of experience I believe these dips can make or break for a competitive player. My FPS is most likely to always be at 144Hz or 240Hz which is my screen refresh rate no matter what build I choose, but the FPS dips will vary between the two, in my opinion. You need two things out of a quality motherboard: 1. VRMs that deliver stable and clean voltage to CPU under prolonged heavy load, and 2. Good thermal design that keeps the VRMs cool under heavy load so that they would not throttle the voltage delivered to the CPU. Having enough power delivered by the chipset to the CPU is not enough, it should be maintained cool to ensure stable high voltage and boosting clocks. That's why higher end mobos use different material to keep their VRMs cool as opposed to the mobo used in the 4090 build above.

Here is an insightful graph that shows the impact of properly cooled VRMs in a high-end mobo on the CPUs performance, even if the CPU's temperature on itself is kept low.

Here is an easy and informative short read on VRMs and how their power rating, build quality, and temperature can tremendously impact the CPU's performance:
A Short Guide To Motherboard VRMs [The Secret To CPU Overclocking
 
Last edited:
Freddie1980 said:
None of that is going to make a dent in closing the gap between a RTX 4080 and a RTX 4090. It's like buying a regular VW Golf and adding a sports air filter thinking it might make it faster then a Golf GTI which is clearly won't.
Click to expand...
totally this.
option 1 is clearly the superior choice
none of the fancy schmancy of option 2 will change the fact that the 4090 wipes the floor with the 4080
 
ronferri said:
it is the Dips and microstutters that I am worried about when a player randomly jumps in front of my screen with sudden explosions and war effects. There is a chance I might get lower dips on the 4090 build than on the 4080 build
Click to expand...

ronferri said:
Here is an easy and informative short read on VRMs and how their power rating, build quality, and temperature can tremendously impact the CU's performance:
A Short Guide To Motherboard VRMs [The Secret To CPU Overclocking
Click to expand...

The part with a graph in the article where you can see the VRM throttling and having a big impact on performance is from hardware unboxed, who have been doing VRM thermal testing in their motherboard reviews for a long while.

The numbers look bad, but this is after 1 hour of running the multi-threaded test in Cinebench R32 (which isn't equivalent to typical gaming load) and it's fairly well known at this point that low-end Asus motherboards don't perform well in these tests. I'd never buy an entry-level B board to run an i9 in long-run multi-threaded loads.

I assume you have other sources that you're using to confirm the impact on frametimes and I can comment on those if you want, but I've never seen anything on hardware unboxed that suggests very high-end motherboards have less dips/stutter than high-end motherboards when gaming.

In heavily CPU bottlenecked scenarios there are two things that definitely do impact the lows: 1. CPU clock speed, 2. fast RAM. In the examples you mentioned, of sudden explosions and war effects, it's hard to know exactly where the bottleneck is, since it depends on the engine and how the game is optimised and in some cases it's nearly impossible to avoid happening, because the bottleneck isn't always the hardware, but the game engine itself.

Mainstream hardware channels/magazines don't really look that deeply at the frametimes, so if it's not caught by the 1% and 0.1% lows in their benches, then it may be missed.
 
ronferri said:
see new post above you two
Click to expand...
ronferri said:
Here is an easy and informative short read on VRMs and how their power rating, build quality, and temperature can tremendously impact the CPU's performance:
Click to expand...
thanks for teaching us to suck eggs :P
as in your graph above, you've just pointed out how b660 boards can handle a 12900k, what makes you think that a mid-range z790 board won't be able to handle a 13900k?

don't let us stop you from wasting your money though ;)
 
Tetras said:
In heavily CPU bottlenecked scenarios there are two things that definitely do impact the lows: 1. CPU clock speed
Click to expand...
And to ensure stable and boosted CPU clock speeds when you need them, you need high quality and properly cooled VRMs which are more available in Build#2 above than in Build #1. Assuming all other variables such as RAM and CPU cooling are normalized. I rest my case.
 
Last edited:
Tetras said:
The numbers look bad, but this is after 1 hour of running the multi-threaded test in Cinebench R32 (which isn't equivalent to typical gaming load) and it's fairly well known at this point that low-end Asus motherboards don't perform well in these tests. I'd never buy an entry-level B board to run an i9 in long-run multi-threaded loads.
Click to expand...
tbh it's very disingenuous to even use a graph from the b660 boards when there's z690+12900k graphs to use, which will make OP's point totally nonsensical.
even the most entry level z690 board will peg at 75c with a 250w load

 
ronferri said:
And to ensure stable and boosted CPU clock speeds when you need them, you need high quality and properly cooled VRMs which are more avilable in Build#2 above than in Build #1.
Click to expand...

Yeah, you do, but typical gaming load is not equivalent to long-run multi-threaded load and I've never seen any benchmarks that show a motherboard VRM throttling a CPU while under gaming load *. The tests hardware unboxed do are deliberately exaggerated to show you the difference between them (the more recent tests use blender, if I recall correctly) and are only really applicable for workstation use. The average user will very rarely encounter a VRM throttle, even on a $100 entry-level board.

* I should use a disclaimer there, I've never seen a high-end motherboard have any impact on the frametimes (if you used a H610 motherboard with a 13900K then yeah, I wouldn't be surprised about that throttling the CPU!).

If you can show me some benchmarks where the motherboard VRM of a high-end board is having a meaningful measurable impact (e.g. on the 1%/0.1% lows) I'm keen to see them, because it would immediately change the suggestions I make for "money no object" gaming builds.
 
You must log in or register to reply here.
Back
Top Bottom