From 10900K to 12900KS, how much more FPS in 4K?

[Peter]

I'm gonna retire my dear i9 10900K that has served me so flawlessly for 4 happy years, together with the ddr4 3200.

I'm now going to upgrade to the mighty 12900KS {which I got for a good price} and 6000 M/T ddr5.

In both cases I'll be using the same gpu, my beloved and reliable Aorus 3090 xtreme. {Don't worry I've got a Aorus 4090 master and a 13900K, but there is still so much life left in the 3090}

My question is, how much more FPS can I expect, in 4K resolution, from such an upgrade? In theory, since the 3090 is already gpu limited, and fully saturated, in 4K resolution with the 10900K, am I right in expecting next to no uplift in FPS with the 12900KS and ddr5 6000 M/T

[󠀡󠀡]

+Broken

[Trout]

keep the cpu. upgrading to 13th or 14th gen intel is playing expensive russian roulette

[Karumati]

you are looking at 5-10fps increase on average if you get 12900ks
upgrade only makes sense if you get 4090 after

[Tonepoet]

I just wouldn't. By the time you're spending $375 on a 12900ks you only have to spend $50 more to get a 7800x3D, which usually beats the 14900k in gaming applications, and you'll have possible upgrade path open to you at least until 2027. It's hard to ignore the fact that LGA 1700 is an effectively dead socket.

Moreover techspot did a rereview on the past two generations of C.P.U.[www.techspot.com], and what we see in the 13 game average is that the 7700x beats the 12900k. 151 vs 150 average but more importantly 104 vs 109 in the 1% lows. As marginal of an advantage as that is, a 12900ks is just a binned. 12900k, so we're only expecting marginal gains in the first place.

Moreover, we're looking at a Windows Update that enhances the performance of Ryzen chips, which may change the performance comparison metrics in A.M.D's. favor even further.

Even Zen 3 performance said to be affected, which leaves me in a position where I'd have to reevaluate my previous stance that LGA 1700 still obsoletes AM4. I hate to say that 'cause I gave advice predicated on the old numbers. >_< It is something like a 10% uplift on average[www.techspot.com]

We're simply looking at the 7700x being a significantly better C.P.U. than the 12900k under the most recent version of Windows, unless there's a corresponding update to bolster LGA 1700 perf. too. I see no reason to invest onto LGA 1700 if you're going to be spending about the same price on the processor as you would a 7700x. This would've been enough so the case with their previous performance parity, but now that the 7700x is looking like it's gaining a not so insignificant edge I find an upgrade

This is especially so considering Intel's also looking rather shady these days with the raptor lake fiasco. Also,let us suppose you have a moral duty to use Intel chips for a moment here, because Andy Grove saved your life when A.M.D. went all Twisted Metal on you and ran over your dog or something. we are expecting Arrow Lake chips in Octobor anyway[www.tomshardware.com].

I'd be flabbergasted if Arrow Lake launch performance was as disappointing as Zen 5 launch performance, and if Intel pulls a Raptor lake twice in a row, it shall effectively be an act of suicide by the company.

But let's say you're anticipating a price drop come october or black friday and the 12900ks becomes more.

But anyway, to answer the question these chips are basically just golden samples of the 12900k and the 10900k, so I don't really anticipate the performance uplift being proportionately different between the 12900k and the 10900k or the 12900ks and the 10900ks, so we can just compare those numbers and look at the percentage difference.


We P.C. compared the 10900k, the 11900k and the 12900k with the following results[www.wepc.com]:

---

New World:

12900k
1440p: 125
2160p: 126

10900k:
1440p: 108
2160p: 106

---

Gas Station Simulator

12900k
1440p: 125
2160p: 75

10900k
1440p: 112
2160p: 79

---

Far Cry 6

[Nullified entry: No 10900k results to compare]

---

Days Gone

12900k

1440p: 142
2160p: 84

---

10900k
1440p: 149
2160p: 91

---

Cyberpunk 2077

12900k

1440p: 99
2160p: 69

10900k

1440p: 103
2160p: 70

---

Counter Strike: Global Offensive

12900k

1440p: 355
2160p: 263

10900k

1440p: 258
2160p: 320

---

Control

12900k
1440p: 129
2160p: 69

10900k

1440p: 131
2160p: 69

---

Six Game Averages

12900k

1440p: 162.5
2160p: 114.33

10900k

1440p: 143.333333333
2160p: 122.5

---

From these figures, we expect the 12900k to be 13.37% stronger at 1440p?

Did I get my data entry right? If so I'm surprised to say that the 12900k is weaker than the 10900k at U.H.D. resolution for some reason. It is worth noting that the 10900k doesn't use little-big architecture, so it has 10 equal cores, instead of eight performance cores and eight efficiency cores. If my numbers are accurate, the 12900k is 7.14% weaker than the 10900k at U.H.D.

Perhaps Counterstrike is just poorly optimized for the 12900k and skewing the data with its high F.P.S. count. Nobody really cares about Counterstrike performance since the game is so easy to run, so let's see if that brings the averages more in line with expectations. That brings our avergae 4k result down to 84 for the 12900k and for the 10900k it's 83 F.P.S. A mere 1.2% increase in performance at 2160p, likely because of a G.P.U. bottleneck based on what we see in Control and Cyberpunk.

Assuming the data from my source is good and I figured all of the numbers properly. You're probably playing at 4k if you're only considering binned skews, so it's not even really an upgrade.

[A&A]

Tonepoet a écrit :

We P.C. compared the 10900k, the 11900k and the 12900k with the following results...

*sheets*

These benchmark numbers are so cursed
I don't know what they smoke, but what are these 0.1% lows? 11FPS in a game which has no problem reaching 1000fps. Also their CSGO graph have a mistake. There is no way 4K will perform better than 1440P. We also don't know how they tested this game because it doesn't have a benchmark. When you throw a smoke, the FPS drops a lot, because the GPU becomes a part of the problem.
13FPS in Cyberpunk. 0.1% lows.
WTF is this? Running on H motherboard or something?


And yes, forget about a CPU upgrade. In very niche games you can find use from a better one at 4K.

[De Hollandse Ezel]

the 10900k already gives no bottleneck to an 3090 running in 4k resolution so your gain is basicly zero.

but those numbers are not bases on an extreme version..if I factor that you have a model thats 5% faster than your average 3090... than maybe you gain 1 or 2 frames in fps..
(like your current average of 73ish fps gets pushed to 75fps.

[BlueBangkok]

If you can find a totally CPU bound game/scenario, where you absolutely need more frames, and assuming perfect scaling, no other bottlenecks, you will see about 30-35% more performance. But that's basically edge case scenario, although contrary to popular belief there ARE games that are CPU limited even at 4K with 3090.
But, you would have to be playing something like Stellaris or Factorio 24/7 for the CPU upgrade to make sense and even then, Ryzen 7800X3D is hands down better for gaming.

Fun fact, you would actually see much more performance uplift if you bought the 4090, too/instead of the 12900KS (50-70% at 4K and double that with frame generation).

[Tonepoet]

A&A a écrit :

Tonepoet a écrit :

We P.C. compared the 10900k, the 11900k and the 12900k with the following results...

*sheets*

These benchmark numbers are so cursed
I don't know what they smoke, but what are these 0.1% lows? 11FPS in a game which has no problem reaching 1000fps. Also their CSGO graph have a mistake. There is no way 4K will perform better than 1440P. We also don't know how they tested this game because it doesn't have a benchmark. When you throw a smoke, the FPS drops a lot, because the GPU becomes a part of the problem.
13FPS in Cyberpunk. 0.1% lows.
WTF is this? Running on H motherboard or something?


And yes, forget about a CPU upgrade. In very niche games you can find use from a better one at 4K.

I doubt it's inconsistent play. The 4k results are higher than 1440p across all three C.P.Us., which would mean the same mistake would have had to have been made in three separate play sessions. What seems most likely is that the 1440p and 4k results were accidentally reversed. I didn't notice that though so let's try this again. I didn't really notice how borked up these results were so let's try again:

12900k vs 10900k 1080p (tested with 3080)

Cyberpunk:

138
145

Forza Horizon 5:

157
168

Battlefield Ⅴ

243
241

Pub G
220
236

Hitman 3


197
216

Assassin's Creed Valhala

128 for both.

Microsoft Flight Simulator

84
93

Red Dead Redemption 2

169
162

Mafia: Definitive Edition

155
161

The Witcher 3

200
219

Average 169.2
176.9

So we're expecting 4.46% improvement based on those results. 1080p would be the least G.P.U. bound scenario, so I doubt a 3090 running at 4k is going to see the 12900k doing much better.

Also worth noting is that Hardware Unboxed has an update on the Windows Update. Turns out that the new Windows isn't magically making the Zen 5 processors better. It's just that Windows has slowly been making them worse and now that's fixed somehow.

What that means for how A.M,D. compares to Intel, I'm not sure. We'll need to wait for Win. 11 H24. to see formal release.

[smokerob79]

Peter a écrit :

smokerob79 a écrit :

you ducked NOTHING and your chip will still have problems....the way it was made is the real issues that can NOT be fixed....micro coding will only extend your time not having issues....but will have issues....

I take what you are saying as true only in the general sense that all cpus degrade over time. My 13900k is running in intel spec, with no mce enabled and 6000 ddr5 only. You should remind yourself how light the cpu load is, together with temperature, in 4K gaming, especially for a i9 processor. In fact everyone knows a i9 is overkill for gaming, but I can't help myself!

I'd love to own a 14900KS if the price is right. I think it's a historic cpu...6.2ghz bro!!!!!

NO.....yours has a defect that will shorten the life span no matter what they do.....its that simple.....

[smokerob79]

Bad 💀 Motha a écrit :

All those 13th and 14th Gen i7/i9 should be considered damaged goods. They are ticking time bombs. Avoid them. If you have one, RMA it; regardless of you having issues with it or not. Because at some point it will get damaged (by design) and that could be while you're in the middle of doing something important or that would then have that PC go down for days or weeks.

did you see the fact it now extends to even i5's??? its not just the 7's and 9's

[_I_]

i would even include the i3 and pentium g lineup from 12-13th gen