Intel Core Ultra 9 285K + Z890 Motherboard for Gaming?

[Azza ☠]

Curious to know if anyone has the latest next generation Arrow Lake CPU + Motherboard from Intel? How well does it perform for gaming purposes and compare to your previous build?

How would it compare to a Intel i9-14900K CPU + Z690 Motherboard?

I'm just interested in the amount of power usage reduced and no more hyperthreading vs performance still keeping up or better than previous generations.

Seems to be doing quite well for itself?

Ideal for laptops and AI creation, etc... but what's the desktop gaming like?

[r.linder]

Turns out that the 285K is realistically 8% BELOW the 14900K.

https://www.youtube.com/watch?v=XXLY8kEdR1c

If Intel can explain why the performance is so much worse, they need to do it now and fix it if they can, because it's looking like this generation is going to be completely pointless. Raptor Lake users have zero reason to upgrade based solely on performance, and the 285K is still inefficient.

[Rod]

Gamers Nexus review just confirmed what we knew. Its a whole generation about efficency. This is exactly what they needed two of them for free and getting rid of Hyperthreading and adding cache. I said if Intel could do all of this they will catch AMD and likely with the help of CHIPS Act surpass AMD.


The big problem is these two node shrinks are not free AMD are firing out the 9800X3D where really Intel just need time to freeze. How do you get the two node shrinks? You cant lol.

[Illusion of Progress]

Seems my guess as to one of the reasons for the lower gaming performance, the non-monolithic design adding latency, was pretty spot on.

https://www.tomshardware.com/pc-components/cpus/intel-core-ultra-9-285k-cpu-review

"Intel’s decision to split the memory controller and PHY into their own tile (I/O tile) was to improve yields, but this creates memory latency issues that contribute to the lower gaming performance we see on the next page."

"As you can see in the AIDA tests, on a like-for-like basis with DDR5-5600, the remote memory controller and PHY add 15 ns of memory latency in our test. Intel says we can expect a 15 to 20 ns increase in memory latency for Arrow Lake. That's a pretty significant change, and not in a good way. It will definitely impact certain workloads, gaming in particular."

Compared to the 14900K, the 285K It has lower latency at some points (especially when still within cache, which has increased in size), but once it needs to go to memory, it is higher than the prior generation. So despite "measured single core performance" increasing (which does lend to tasks that aren't latency sensitive seeing uplifts), this hurts its gaming performance.

I was thinking this might happen when it became known that they would move to a non-monolithic design, but I was hoping they'd have some magic approach that would counteract it (like keeping the memory controller local to the cores, or brute forcing it with enough IPC/clock speed increase to compensate the higher latency, but we know why that wasn't possible here).

[Mr White]

The new Intel Processors maybe like the first Ryzens. Plus driver updates will help

[r.linder]

They essentially rushed a new design, it needs more improvements before it'll actually start winning, it took AMD until the third iteration to be actually good, and then the fourth blew it all away.

[A&A]

Messaggio originale di Illusion of Progress:

Seems my guess as to one of the reasons for the lower gaming performance, the non-monolithic design adding latency, was pretty spot on...

Yes, there is added latency, and it's an insanely high increase. Ryzens also have an i/o tile, but they managed it to perform as a monolithic.

i9 13900K
https://www.techpowerup.com/review/intel-core-i9-13900k/5.html

Ultra 7 265K
https://www.techpowerup.com/review/intel-core-ultra-7-265k/7.html

Ultra 9 285K
https://www.techpowerup.com/review/intel-core-ultra-9-285k/7.html

Ryzem 9 7900X
https://www.techpowerup.com/review/amd-ryzen-9-7900x/6.html

[Illusion of Progress]

That's interesting. I thought Ryzen generally had higher memory latency than Intel's CPUs of the same times.

This really means Intel should benefit that much more from stacked L3 cache, and now that they've moved onto a non-monolithic design, I wonder if we'll see that from them. With that, they would probably take the gaming performance spot back from AMD unless AMD makes some major IPC or clock speed increases before then.

[r.linder]

The difference is that AMD has been working with a dedicated I/O die since Zen2 and they've refined their architecture and design. This is more or less brand new to Intel and it could take them several years to properly refine it.

[A&A]

They really should bring back the L4 cache until they refine their memory controller ngl.

[_I_]

yea amd has been working with chiplets for much longer
it will take intel some time to catch up with amds 'gluded' design

not that it was bad either way, but intel needs to learn how to do it better to catch up with amd again

[Bad 💀 Motha]

Especially when an all new Intel Chipset + latest high end CPU to compete with 14th Gen unlocked i9 is only netting approx 8% gains; that make moving to the newer platform pointless. It's only good perhaps if someone is wanting to upgrade or build/build a new machine who is maybe coming from 8th/9th Gen stuff or perhaps from 10th/11th Gen... if you're in a 12th Gen i7 or i9 you're fine really.

[r.linder]

There are no gains that are relevant to most people here, because in gaming it's slower than both the 14900K and 7800X3D by a laughable margin.

Arrow Lake is nothing more than a stepping stone generation like original Zen.

[_I_]

except that the first zen was miles ahead of what fx could ever do

[r.linder]

Messaggio originale di _I_:

except that the first zen was miles ahead of what fx could ever do

Minus the Ryzen 3 1200 which was able to be outpaced by some 8 core FX chips in some situations and configurations

But that CPU wasn't really great to begin with, and lower end Zen was barely any better than Sandy Bridge

[Bad 💀 Motha]

Yea I had kept my FX8350 and 4790K PCs until the Ryzen 2600 came out. Then I sold that MB & CPU once B550 came out and used a 3700X for quite a while. Ryzen 1st Gen was a waste of time considering how good the 4790K was. But ultimately it paved the way forward for AMD.