Intel Core Ultra 9 285K & Ultra 5 245K CPU Review

[Tyler-98-W68]

Real question though..........Do you personally use and test both Arrow Lake / Raptor Lake, and 7 and 9 series X3D chips in games and real world scenarios? Funny thing is I do....yet I'm still the paid shill

I scored 38 052 in Time Spy

AMD Ryzen 7 9800X3D, NVIDIA GeForce RTX 5090 x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

I scored 24 591 in Time Spy

AMD Ryzen 9 7945HX3D, NVIDIA GeForce RTX 5090 x 1, 98304 MB, 64-bit Windows 11}

www.3dmark.com

I scored 40 941 in Time Spy

Intel Core Ultra 9 Processor 285K, NVIDIA GeForce RTX 5090 x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

I scored 31 912 in Time Spy

Intel Core Ultra 5 Processor 225, NVIDIA GeForce RTX 5090 x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

Paid intel shill, yet still run AMD X3D systems as well

 

[David_Schroth]

There's a real lack of civility going on in this thread and I'm pretty disappointed with how the new guy is being treated. I'll be glad to drop the banhammer on shills, but based on PMs exchanged when he joined, you've got the wrong guy. Please address the post content in a civil manner instead of making it personal. There's also a report button available.

Please return when you've completed more research, thanks!

This is not a humorous or civil response.

Some of us have been around a long, long, time. And we have lived through the Nvidia and Intel guerilla marketing, gaslighting, and astroturfing + Anti consumer and Anti competitive practices. Never again. I nuke that crap from orbit in the forums I police. We insist on honest and open dialogue, free of advertising, misinformation, and disinformation. Along with a higher level of decorum than allowing lazy labels like
"fanboy". I've probably banned you at least once already.
The only delusional people, are the arrow lake acolytes trying to pimp a dead on arrival platform with CPUs that are worse for gaming( You know, the thing that drives retail sales/DIY) Than the generation that came before it.

Address the content as opposed to taking aim at the poster. If you think he's a shill, hit the report button and support it in your report. Otherwise, please address the post content as opposed to expanding to classifying people based on their post.

 

[DAPUNISHER]

This is not a humorous or civil response.

And these are?

I bet I've been around as long if not longer than to, building systems from the 486 days. But again this just proves my point even more. Good thing you cant nuke me, I'll see myself out as this place is no different from everywhere else. Delusional amd fanboys and people like you thinking they are all high and mighty "policing" things.

Also please tell me where im paid to post anything. What a bunch of horseshit, that's about as likely as you thinking you are someone important. I'm an independent person who simply buys hardware and tests it out on games I play. I have zero sponsors, receive zero money. Feel free to check out my extensive BeamNG testing on both amd and intel systems.....but again you can go away with accusing me of being paid.

Be consistent.

 

[Grimlakin]

Yeesh I didn't mean to kick odd forum whine 2025 edition...

Wow...

Apologies ops.

 

[David_Schroth]

And these are?

Fair enough. They are not either.

 

[LazyGamer]

Real question though..........Do you personally use and test both Arrow Lake / Raptor Lake, and 7 and 9 series X3D chips in games and real world scenarios? Funny thing is I do....yet I'm still the paid shill

So, to be clear - you're getting the reception you're getting because you're countermanding all established performance metrics, aside from those actually pushed by Intel shills.

And it has nothing to do with AMD vs. Intel. It's more cache vs. less cache, and the effect on frametime delivery. Intel would still be #1 for gaming if they made a top-end part that had eight of their current P-cores and replaced the die space consumed by the eight or sixteen E-cores with more cache. Intel is better at basically everything else; stronger cores, better platforms and so on.

But Intel has chosen not to address their cache differential with AMDs X3D parts. Note that this doesn't apply to AMDs non-X3D parts either; they're in the same boat as Raptor Lake.

Intel made Arrow Lake objectively worse for gaming. For reasons that likely have much more to do with getting products shipped than what they're actually capable of, but unfortunately, we consumers can only buy what Intel is willing to produce and sell.

 

[Tyler-98-W68]

Please return when you've completed more research, thanks!

Research eh? Funny because I actually did TESTING aka real world tests in the games that I play to come to the determination that Raptor lake was ideal for what my use was going to be.

Having bother a 7800X3D system and 14900K system, I play BeamNg alot, I also play Derail valley. So when AL was released I was intrigued with similar performance but less power consumption.

People like to say BNG is poorly optimized and not a good game, mainly because it runs better on intel systems because it uses as many cores/threads as you can throw at it. The general rule of thumb is 1 thread per vehicle spawned. If you want to have lots of traffic and/or props in maps then you want multithreaded performance. Of course everyone's natural response is "get a threadripper" the problem is that in addition to multithead performance, BNG also wants fast single thread performance, and right now, AL is the fastest single threaded performance CPU (excluding the apple M3).

So last year I built 14900K and 7800X3D systems with a 4090 to test out which CPU would be better for BNG.

Without going in to a huge amount of detail, but I guess I will because you wanted research. Quite simply Beam wants more cores and fast ones.

So in my testing of a 7800X3D and 14900K, I found that the 7800X3D chocked HARD when more than 20 vehicles/props were spawned.

At lower car counts the 7800X3D had slightly higher frame rates, but as the complexity was added performance started to tank badly.

14900K + 4090 0 traffic

About 160fps

7800X3D + 4090 0 traffic

About 160fps

14900K + 4090 10 cars

about 140fps

7800X3D + 4090 10 cars

About 140fps

14900K + 4090 20 cars

About 110fps

7800X3D + 4090 20 cars

https://www.youtube.com/watch?v=-LX6tAcgMNg

About 40fps

14900K + 4090 30 cars

https://www.youtube.com/watch?v=yED8RVRCh6U

About 70fps

7800X3D + 4090 30 cars

https://www.youtube.com/watch?v=DE0UIj-fW0c

about 20fps

14900K + 4090 40 cars

https://www.youtube.com/watch?v=q-zmx1BrQwI

About 65fps

7800X3D + 4090 40 cars

https://www.youtube.com/watch?v=msaT2AhxNTM

about 15fps

But then people would argue that's not a fair comparison because the intel chip has so many more cores, but on the other side of the equation X3D chips are the best for gaming bar none. But i'm getting slightly off topic so how does this related to arrow lake?

Well for fun I found a package deal of a Core Ultra 5 225 MB/Ram for a very fair price and thought I would build a test system out to see what it would do. Paired it with my 5090 and 5070Ti so I could take the GPU out of the equation in certain tests (although when spawning large amounts of cars the game gets CPU limited and not GPU limited even at 4K so for that type of testing the gpu isn't that important.

The shocking part when I started testing (using direct 3D mode.......was how 10 cores was nearly as fast as my 14900K with 24C/32 Threads)

225 + 5090 + 40 Cars

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

www.youtube.com

14900K + 3090 + 40 cars

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

www.youtube.com

So based on that information to grab a 285K and a Minisforum BD790iX3D (7945HX3D basically a lower power 7950x3d) and start doing a ton of testing to see which systems gave me the best performance for lots of traffic in Beam.

I enjoy playing with a ton of traffic so I want lots of cores, and I saw a generational increase from a 14900K to a 285K

14900K + 4090 + 80 cars

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

www.youtube.com

285K + 5090 + 80 cars

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

www.youtube.com

Then I slightly OC'd the E-Cores and found even more performance

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

www.youtube.com

But I"m getting off topic, why did I choose arrow lake for BNG because it simply likes the fast single threaded performance and the E-Cores that AL provides. Absolutely X3D chips work great in other games, I don't think I denied that anywhere but you wanted a reason why.

And just to show i'm not a paid intel shill.

Here is a perfect example of how the 9800X3D is much faster in the derail valley game i mentioned before. Actually I'll just post all the systems to show.

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

www.youtube.com

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

www.youtube.com

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

www.youtube.com

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

www.youtube.com

You'll notice all systems except the 9800X3D have similar framerates and then drop big time once a lot of explosions start happening, the 9800X3D is on a level of it's own in this game, then again you also won't find any testing of Derail Valley done by another else either.

I have 100's of hours of testing games that I play, doesn't mean the cpu's I use are the best for everything but I can at least identify how they performance in real world games, and whether it's intel or AMD is based on the game.

But like I said before blanket statements like "AL is bad for gaming" is incorrect.

 

[Tyler-98-W68]

So, to be clear - you're getting the reception you're getting because you're countermanding all established performance metrics, aside from those actually pushed by Intel shills.

And it has nothing to do with AMD vs. Intel. It's more cache vs. less cache, and the effect on frametime delivery. Intel would still be #1 for gaming if they made a top-end part that had eight of their current P-cores and replaced the die space consumed by the eight or sixteen E-cores with more cache. Intel is better at basically everything else; stronger cores, better platforms and so on.

But Intel has chosen not to address their cache differential with AMDs X3D parts. Note that this doesn't apply to AMDs non-X3D parts either; they're in the same boat as Raptor Lake.

Intel made Arrow Lake objectively worse for gaming. For reasons that likely have much more to do with getting products shipped than what they're actually capable of, but unfortunately, we consumers can only buy what Intel is willing to produce and sell.

Speaking of Cache.

I also played around with different Core configurations in BNG to see how the game used cores and found some interesting results

What's not covered in those graphs and I discovered was when there was rumor of a Core Ultra 3 205 (4P 4E Core AL CPU) I made my Core Ultra 5 225 into a 205 and did some tests

So 40 cars spawned with full core count on a 285K

40 Cars with a simulated 4P 4E CPU

Big difference like it lost 50% of it's performance.

But what happens when we make the 285K into a 4P 4E Core CPU?

I was shocked that it only lost about 10fps vs the full 24 Core configuration.

I came to this conclusion but not verified by anyone

For the 285K, it's possible that 4P 4E configuration is leaving half the P-core stops active, and then leaving one E-core in each cluster. If that's the case, each remaining E-core suddenly has 4x the L2 cache to itself, but In reality the 3D cache really isn't overly beneficial in BNG, but could explain why I didn't lose hardly any performance getting rid of 16 cores on the 285K.

So, again I've done a wee bit of research and testing.


As i mentioned before and no one seems to want to answer, based on pricing in Canada.

You can have a 265K and 5070Ti vs a 9800X3D with a 5070 for the same price.

The pricing gap is also similar going 265K with 5080, vs a 9800X3D with a 5070Ti,

So basically you would go with a better gaming CPU with a weaker GPU instead of an arrow lake cpu with a more powerful GPU?

 

[Grimlakin]

I dont mean this to sound offensive... I really dont.

But you balk at price yet build out systems left and right to test for fun? Ok.. sure.

 

[Tyler-98-W68]

I dont mean this to sound offensive... I really dont.

But you balk at price yet build out systems left and right to test for fun? Ok.. sure.

Where was i complaining about pricing. I was highlighting pricing as an example. And yes I do build systems for fun? Apparently that's a problem these days. I have a hobby, I spend money on this (and other hobbies) however my situation likely doesn't represent most "enthusiasts" so basically when someone has lots of high end hardware they are a paid shill? My YT channel started as a place to document technical information on vehicles and tuning, then I started benchmarking games most major review sites dont do. I don't whore my YT channel out, I don't reach out and ask for money, I simply buy what I want and test it, yet people like you highlight that and somehow try to weaponize the fact I have money to spend so I must be paid by someone.

Harsh reality I worked for *** off for many years to get to a point where I can enjoy the hobbies I want, but that is now seen as a liability because someone is jealous.

 

[DAPUNISHER]

It's the same sales pitch for Arrow Lake on every message board I visit. They need a better one. Because AMD pitches a retail shutout week after week here in the U.S. No one cares about testing done by internet randoms. Zen 3 is over 5 years old and some SKUs are still consistently outselling everything Intel.



It isn't just CPU swaps either. The boards are top sellers This has been the trend since last year.


It is a little hard to understand why AM4 is still selling so well. There has to be a knock on effect from Raptor Lake degradation hurting Intel sales. That knock on effect combined with 1851 being effectively DOA and failing to beat Raptor in gaming. Otherwise they would not have to do fire sales to crack the top 10 and outsell 5 year old CPUs for a single week.

Rumor is, Intel's next desktop platform will be long lived. That's good news if true. Because retail sales the last few years have made it clear that platform longevity is part of the value proposition for the vast majority of shoppers.

 

[Tyler-98-W68]

Wasn't trying a sales pitch, then again its the same BS dribble you post online everywhere else whenever AL is mentioned. You are an internet no one as well. So you have tested and played with arrow lake? Again when was having to upgrade motherboards a dead end platform? That never stopped intel from having great sales years ago.

Again no one wants to answer what I asked before. Why buy 2 CPUs on AM5 when you could buy a 265k or 285k on 1851 and not need to upgrade for a few years because you got the performance you need, not having an upgrade path isn't that big of a deal, you make it seems like everyone upgrades their cpus multiple times without changing their MBs that's about as common as everyone owning super high end GPU's, despite the steam hardware survey showing otherwise.

Quit trying to portray that the majority upgrades all the time when in fact they don't.

Tell me what performance am I missing out on with a 1851 platform? For rendering, editing and gaming?

What AMD platform (excluding hedt) can give me 3 Gen 5.0 M2 slots, and 4 Gen 4.0 slots on the same board?

When im ready to upgrade (likely next year because I upgrade every year [doesn't everyone?? Because being able to swap cpus on the same MB is so important] ) I don't mind the fact I may have to grab a new motherboard.

But actually being realistic, the system i have now is likely going to be solid for 3-4 years. So again when im ready to upgrade in 3-4 years the cost of a new motherboard isn't going to make or break me, but i guess im in the minority when it comes to how I think

 

[Grimlakin]

@tylyou know what. Youre right. For you. Congratulatiins?

for me i use one high end large nvme.

On my platform I know i have two or more generations of cpus I can drop into it. If amd releases a 24 core 48 thread cpu for it I can update my bios and swap to it.

Yes Intel has more nvme slots. And pcie lanes are.important. if I was running a home nvme NAS I would be even more interested.

For what I do my 16 core 32 thread x3d cpu is top dog in the vast majority of scenarios. But I also have ran Intel before as well. So I dont want them failing. I want them trading blows and keeping each other on the cutting edge.

You'll note im also sporting a AMD video card. Its a solid 4k card. I am looking at 5090s but also waiting to see if AMD does another high tier card as well.

Congratulations for you. Maybe it's your career but your posts are very professional. Very thought out... it's as if you work in the industry or just use Ai to write them. Or are a forum pro. Or maybe you like to argue about it. Still feels gorilla to me. So hats off. You're so well represented it feels paid for.

 

[LazyGamer]

Tell me what performance am I missing out on with a 1851 platform? For rendering, editing and gaming?

None, for you. That seems to be the problem here - you believe that your testing is representative of everyone else's experience, and so therefore you're right and everyone else is wrong. And you believe this so much that you're throwing around vitriol and general insults at random internet strangers that have different experiences from you.

I have 100's of hours of testing games that I play, doesn't mean the cpu's I use are the best for everything but I can at least identify how they performance in real world games, and whether it's intel or AMD is based on the game.

Perhaps you should put this up front, with a dose of humility? Your game selection is, shall we say, relatively discerning compared to the masses. You can be absolutely correct about your testing, and then be absolutely incorrect in the application of your findings to situations outside of your testing. Tests good, conclusions wrong, etc.

But like I said before blanket statements like "AL is bad for gaming" is incorrect.

Well in general, and essentially I'd guess for 98% of folks that might be choosing between the two, it's correct. One has to get into very niche games to find the exceptions like you have.

I'll also say that 'bad' is poor wording. It's not 'bad', it's just far less optimal in terms of frame delivery. And I say this as someone that plays on a range of hardware, including Skylake-era IGPs. Sometimes just being able to play is all that matters, right?

Tell me what performance am I missing out on with a 1851 platform? For rendering, editing and gaming?

I said this above but it bears repeating: I feel that Intel has the better platform all around for everything except gaming. You'll find my posts here in the past complaining about the lack of NVMe slots on AM4 boards; only when ASRock released their X870E Nova did I find a board that was 'acceptable', and now I get to see posts about dead CPUs on ASRock boards every time I open Reddit. I hadn't used an AMD CPU since the Athlon64 x2 days, or put another way, I'd been using Intel CPUs primarily since the Core 2 days.

But I'll also say this: for rendering and editing, either one's performance needs aren't great (unpaid work / not time critical), or they're using a different platform. Rendering folks that do it for a living are on Threadripper, and editing folks that do it for a living are on a Mac. My general advice for folks that find a desktop CPU limiting for editing is to get a Mac regardless as even the cheap ones fly, and it's a much more economical solution relative to trying to build a computer that 'does everything' these days.

What AMD platform (excluding hedt) can give me 3 Gen 5.0 M2 slots, and 4 Gen 4.0 slots on the same board?

So, here's a fun one: what do you think you need those slots for? Unless you're racing CrystalDiskMark like you're racing 3DMark, they serve very little utility and provide almost no (or, effectively no) boost to the user experience for desktop applications, including games.

 

[Tyler-98-W68]

What Do I need 18GB of Local NVME storage for? 6GB of which is gen 5. Video editing, I deal with large video files and having quick access to them is important before they go off into network storage. I don't chase 3Dmark and CDM tests, I simply have them to make sure things are working right, but I also do more than CDM as passmark has more thorough testing.

Well in general, and essentially I'd guess for 98% of folks that might be choosing between the two, it's correct. One has to get into very niche games to find the exceptions like you have.

Niche exceptions, always fun to hear that when something X3D doesn't look great. You and many many others always seem to portray the same thing it's "bad performance"

Example, if AR loses by 2fps to an AMD CPU it's "bad bad terrible gaming experience" but if an AL CPU "wins" by 2 fps to an AMD CPU it's irrelevant or within the margin of error. The amount of goal post moving is comical these days.

I used AMD CPU's in the 939 era, but the 939 Opteron's (144 and 165) because they overclocked very well.

I then had to transition to laptops to be mobile and when Conroe and Merom came out and stuck with laptops for a while until I didn't need to be mobile anymore then went back to AMD for the Phenom CPU's, but I've gone through so many systems I can't even keep track anymore.

If you gamed on Skylake IGP's did you ever play with Broadwell? The Iris Pro 6200 in the 5775c was quite impressive for it's time (although hard to get) and even today puts up a decent fight vs newer IGPs, I could also get in to how I tested Broadwell with Blackwell GPU's, but then I'd probably be called an Nvidia Shill.

 

[Brian_B]

I can't speak for the new AT folks, who all seem decent and I'm very glad they are here.

But for the older FPS guys in the crosshairs here - I can absolutely vouch they are not fanboys of anything, other than chasing performance. And they will let you know in no uncertain terms which color label has the performance today, and if that changes tomorrow, so will they.

 

[LazyGamer]

Niche exceptions, always fun to hear that when something X3D doesn't look great. You and many many others always seem to portray the same thing it's "bad performance"

It happens so very rarely - meaning that it's the general case, whereas poor performance (or just slightly lower relative to non-X3D AMD SKUs and Intel SKUs) is the exception - so yeah, that's a 'niche exception'. It's not an insult, it's an honest characterization.

Example, if AR loses by 2fps to an AMD CPU it's "bad bad terrible gaming experience" but if an AL CPU "wins" by 2 fps to an AMD CPU it's irrelevant or within the margin of error. The amount of goal post moving is comical these days.

The goal post is overall gaming performance. Confining that to very specific, very niche games is indeed 'goal post moving'.

If you gamed on Skylake IGP's did you ever play with Broadwell? The Iris Pro 6200 in the 5775c was quite impressive for it's time (although hard to get) and even today puts up a decent fight vs newer IGPs.

I didn't, no, but would like to have, and would like to see Intel reprise the concept.

I could also get in to how I tested Broadwell with Blackwell GPU's, but then I'd probably be called an Nvidia Shill.

If you choose to walk like a duck and quack like a duck, should you be surprised when people point at your posts and say 'there's a duck!'?

I'm not challenging your testing and conclusions for your usecases - I'm challenging your extrapolation of those to the broader market. And it isn't personal, any more than you make it to be.

 

[DAPUNISHER]

If you choose to walk like a duck and quack like a duck, should you be surprised when people point at your posts and say 'there's a duck!'?

It is just that simple. And it is never too late to dial it back and have better dialogues with others.

Intel itself admits Arrow Lake is a dud - "Despite receiving numerous updates and enhancements since its rough launch in October 2024, Intel's Arrow Lake desktop CPUs have not been especially popular, and for multiple reasons. At a recent technology conference, the chip company openly admitted that it had "kind of fumbled the football … particularly [the] high performance desktop side"

"Zinsner was a lot more frank about Arrow Lake, its multi-tiled processor for desktop PCs. "As you know, we kind of fumbled the football on the desktop side, particularly high performance desktop side. So we're as you kind of look at share on a dollar basis versus a unit basis, we don't perform as well and it's mostly because of this high end desktop business that we didn't have a good offering this year."

Article

Even Intel's big clients have been ordering Raptor Lake instead. Arrow Lake will be the baseline for demonstrating the remarkable improvements Nova Lake brings in benchmark suites. After that it will be immediately forgotten, especially by Intel.