• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Intel's Core Ultra 9 285K Performance Claims Leaked, Doesn't Beat i9-14900K at Gaming

btarunr

Editor & Senior Moderator
Staff member
Joined
Oct 9, 2007
Messages
47,670 (7.43/day)
Location
Dublin, Ireland
System Name RBMK-1000
Processor AMD Ryzen 7 5700G
Motherboard Gigabyte B550 AORUS Elite V2
Cooling DeepCool Gammax L240 V2
Memory 2x 16GB DDR4-3200
Video Card(s) Galax RTX 4070 Ti EX
Storage Samsung 990 1TB
Display(s) BenQ 1440p 60 Hz 27-inch
Case Corsair Carbide 100R
Audio Device(s) ASUS SupremeFX S1220A
Power Supply Cooler Master MWE Gold 650W
Mouse ASUS ROG Strix Impact
Keyboard Gamdias Hermes E2
Software Windows 11 Pro
The Chinese tech press is abuzz with slides allegedly from Intel's pre-launch press-deck for the Core Ultra 2-series "Arrow Lake-S." The most sensational of these are Intel's first-party performance claims for the top Core Ultra 9 285K model. There's good news and bad news. Good news first—Intel claims to have made a big leap in energy efficiency with "Arrow Lake," and the 285K should offer gaming performance comparable to the current Core i9-14900K at around 80 W lower power draw for the processor. But then there in lies the bad news—despite claimed IPC gains for the "Lion Cove" P-core, and rumored clock speeds being on par with the "Raptor Cove" P-cores on the i9-14900K, the 285K is barely any faster than its predecessor in absolute terms.

In its first party testing, when averaged across 12 game tests, which we used Google optical translation to make out the titles of, Intel used performance numbers of the i9-14900K as the mean. The 285K beats the i9-14900K in only four games—Warhammer 40K: Space Marine 2, Age of Mythology Retold, Civilization VI: Gathering Storm, and F1 23. It's on-par with the i9-14900K in Red Dead Redemption 2, Total War: Pharaoh, Metro Exodus, Cyberpunk 2077, Black Myth: Wukong, Rainbow Six Siege. It's slower than the i9-14900K in Far Cry 6, FF XIV, F1 24, Red Dead Redemption 2. Averaged across this bench, the Core Ultra 9 285K ends up roughly on par with the Core i9-14900K in gaming. Intel also compared the 285K to AMD's Ryzen 9 9950X, and interestingly, even the Ryzen 9 7950X3D.



The Ryzen 9 7950X3D isn't AMD's fastest gaming processor (which is the 7800X3D), but Intel chose this so it could compare the 285K across both gaming and productivity workloads. The 285K is shown being significantly slower than the 7950X3D in Far Cry 6 and Cyberpunk 2077. It's on par in Assassin's Creed Shadows and CIV 6 Gathering Storm. It only gets ahead in Rainbow Six Siege. Then there's the all important comparison with the current AMD flagship, the Ryzen 9 9950X "Zen 5." The 9950X is shown being on-par or beating the 285K in 8 out of 12 game tests. And the 9950X is the regular version of "Zen 5," without the 3D V-cache.

All is not doom and gloom for the Core Ultra 9 285K, the significant IPC gains Intel made for the "Skymont" E-cores means that the 285K gets significantly ahead of the 7950X3D in multithreaded productivity workloads, as shown with Geekbench 4.3, Cinebench 2024, and POV-Ray.

View at TechPowerUp Main Site | Source
 
This is before or after the castration of the 14900k with baseline and patches? Imagine if they didn't remove the HT, Arrow Lake would have been on par with Alder Lake from 2021

On the next Q-report Pat will be like:

As I walk through the valley of the silicon and tech
I take a look at my company and realize there's nothin' alright
'Cause I've been crying and begging' for so long that
Even my momma thinks that my career is gone
 
There are also some IPC leaks:

Intel Core Ultra 200 "Arrow Lake-S" to offer 9% IPC gain on P-Cores and 32% on E-Cores - VideoCardz.com

It looks like much higher IPC E-cores is 1:1 replacing HT and slightly higher IPC P-cores is allowing Intel to drop the clocks to help with power consumption and runaway voltage issues. In the end, it looks like Zen 4, Zen 5, Raptor Lake Refresh and Arrow Lake-S are going to be within 10% of each other performance-wise on average across the board. Desktop gains are slowing down generation to generation as the industry focuses on data centers and AI.
 
Oooof. Man that is unfortunate. Oh well looks like 9800X3D in my future.
 
Not all software is executable on E-Cores. Comparing E-cores with decent processors with only P-cores is nonsense in the first place.

If you do, I assume that your software is badly optimised and only uses the capabilities of the bad E-Cores, not the good P-Cores. Which contradicts the purpose of a newer processor with newer instruction sets and capabilities.
 
They seem to be going for the same efficiency push as AMD did with Ryzen 9000 series. Same-ish performance at a noticeably lower power point. -80W in gaming test is significant.
Probably hoping that against Ryzen 9000 it looks better than would otherwise be expected.
 
Look, efficiency and all is cool but realistically nobody except 3.5 enthusiasts and a few die-hard Intel fans with long overdue upgrades from something like 9th gen is going to spend around $800 on Z890mb+cpu combo to be slower than both past gen and AMD. This is a complete mess if true, there's no good spin here. Damn.
 
Last edited:
Look, efficiency and all is cool but realistically nobody except 3.5 enthusiasts and people with long overdue upgrades from something like 9th gen is going to spend around $800 on Z890mb+cpu combo to be slower than both past gen and AMD. This a complete mess if true, there's no good spin here. Damn.
You meant to say 1200$+
 
The funny thing is, my 14900K & KS will end up having more warranty than them too : )

But let's wait for official benchmarks before we get carried away yeah.
 
They seem to be going for the same efficiency push as AMD did with Ryzen 9000 series. Same-ish performance at a noticeably lower power point. -80W in gaming test is significant.
Probably hoping that against Ryzen 9000 it looks better than would otherwise be expected.

Still, 400W on avg for "gaming" is kind of absurd.

X3D will need not even 1/3rd of that to even produce more or higher FPS.
 
The funny thing is, my 14900K & KS will end up having more warranty than them too : )

But let's wait for official benchmarks before we get carried away yeah.
Arrow Lake was never expected to beat Raptor Lake by anything approaching a significant margin. Intel is recorrecting right now as both Skylake and Raptor Lake were dead-ends.
 
I'd say no one in their right mind would expect Arrow lake beating Superposition Raptor lake in everything given the W differences. I'm not saying it's impossible, but it's highly unlikely.

No HT™ on top of that, although same amount of Economy-cores™. Better IPC so I guess they're Economy Plus?
 
Last edited:
Arrow Lake was never expected to beat Raptor Lake by anything approaching a significant margin. Intel is recorrecting right now as both Skylake and Raptor Lake were dead-ends.

I think you're right. In my own, way-off the mark thread here, this is what we generally thought:
1728384214325.png


The scary part is: Intel's "First-Party Benchmarks" are usually fantastically optimistic.... and a large chunk of people did think that there would be SOME improvement. So IMO this looks like it's going to generally miss expectations in the broader market.

I legitimately thought the APO team, faster ram etc. etc. would net some gains in games (since tuning ram and turning of HT on a 14900K can get you +10% over stock, reduce power draw, without any IPC increase or other improvements).
 
Last edited:
They seem to be going for the same efficiency push as AMD did with Ryzen 9000 series. Same-ish performance at a noticeably lower power point. -80W in gaming test is significant.
Probably hoping that against Ryzen 9000 it looks better than would otherwise be expected.
For the 9000 series that's just mostly from lower power limits. GN tested both arch on 45W and the 7000 series actually did better. At higher power draw >65w 9000 won so the lower power limits for 9000 series are actually more of a disadvantage.

Still, 400W on avg for "gaming" is kind of absurd.

X3D will need not even 1/3rd of that to even produce more or higher FPS.
C'mon, it's obviously total system power...
 
Last edited:
Or, maybe they saw what happened with AMD's 9000 launch and got a bit humble about it this time around.
It's possible but I'm not holding my breath on that one. For example, here's the 14th Gen (13th Gen) slide deck:
1728385364455.png

Steve Harvey Lol GIF by Digg
 
It's possible but I'm not holding my breath on that one.
I think all the "Up to" numbers should be ignored, while the rest should be taken with a grain of salt as usual.

It's a more interesting launch than the last time, that's for sure. Not because of performance, but because it's different, and less insane..

I'm curious if this is the final CPU for LGA 1851 tho.
Lots of people won't upgrade the CPU the following year, I know lol
 
One benchmark showing 4% faster while being down 165W?!? Here is to hoping Intel finally fixed their insane power consumption. At the very least new architecture is interesting.
 
This is before or after the castration of the 14900k with baseline and patches? Imagine if they didn't remove the HT, Arrow Lake would have been on par with Alder Lake from 2021

On the next Q-report Pat will be like:

As I walk through the valley of the silicon and tech
I take a look at my company and realize there's nothin' alright
'Cause I've been crying and begging' for so long that
Even my momma thinks that my career is gone
I'm not sure where do you see castration. Yes setting a 253W power limit is a bit restrictive but it only matters for all core loads like Cinebench and not in gaming. Also my 14900K is a bit faster with the latest Intel microcode from a few days back. It's scoring 39.9K at Cinebench at 253W, so I'm happy.
 
I think all the "Up to" numbers should be ignored, while the rest should be taken with a grain of salt as usual.

It's a more interesting launch than the last time, that's for sure. Not because of performance, but because it's different, and less insane..

I'm curious if this is the final CPU for LGA 1851 tho.
Lots of people won't upgrade the CPU the following year, I know lol
Yes, Arrow Lake last proc. for 1851. Nova Lake will have 1900+ pins
 
Yes, Arrow Lake last proc. for 1851. Nova Lake will have 1900+ pins
Well it had a good run, I guess.

Funny thing is that the only game where it beats the 7950X3D is Rainbow Six Siege that's NINE years old, and it has an asterisk lol.
1728387040258.png
 
Last edited:
Isn't there a latency advantage by removing HT?

That's important for gamers, and sound engineers.
 
Back
Top