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

[Tyler-98-W68]

@Tyler-98-W68 I appreciate the testing you have done. I always love new graphs and new data points to ponder over. There is no single answer for every use case when it comes to the best CPU. Compromises have to be made one way or the other.

I have both Arrow Lake (245KF) and 9950X3D too. If I want to be very honest about how I feel about both of them, I think both are overpriced.

245KF often has trouble beating 12700K in benchmarks.

9950X3D (I paid almost $870 in 4 installments for it because of possibility of loser trump doing something idiotic and causing prices to go even higher) is bandwidth starved in multi-core scenarios and really needs quad channel RAM or EXPO DDR5-9000 (which we might get next year with better mobos maybe?).

I understand the majority going for the 9800X3D because generally, it does better in the average game(s) that people are most likely to play. Your gaming examples, while very enlightening, probably don't matter to the average user because they may not even want to play those games in their entire lifetime.

So I'm just thinking, why are you bothering to argue? Just present your data and let us enjoy the graphs

Appreciate the support, You are right, I'll just provide the data.......and let people digest that, I just get sick of hearing how when there is any Arrow Lake data that is positive, it's a 5 alarm fire and the AMD folks come rolling in like white on rice.

 

[igor_kavinski]

https://www.overclock.net/posts/29523597/

Microsoft seems to have no interest in helping Intel as 25H2 has crippled its performance badly. Just look at that gap between Ghost Spectre ISO and 25H2.

This guy managed D2D 40 with a really high voltage: https://www.overclock.net/posts/29523477/

I can't go above D2D 35 on my 245KF without touching voltage.

I hope Arrow Lake Refresh has at least 5% improvement in stock performance. Then I can abuse my current CPU with overvolting and if it dies early, I can pick up a Refresh CPU.

 

[LazyGamer]

I hope Arrow Lake Refresh has at least 5% improvement in stock performance. Then I can abuse my current CPU with overvolting and if it dies early, I can pick up a Refresh CPU.

I just hope that they manage to make it better than Raptor Lake in latency-sensitive stuff, and / or put out a version with triple the L3 cache.

I'd switch back to Intel if Intel had a credible X3D alternative (much to the detriment of @Tyler-98-W68's internal monologue).

 

[igor_kavinski]

or put out a version with triple the L3 cache.

Not happening for Arrow Lake Refresh. They only had Adamantine cache which they cancelled because it was either too expensive or failed to meet their expectations. Even with Nova Lake, they will make CPUs with large cache a very premium product. No cheaper options for peasants like 7600X3D or 7500X3D.

 

[LazyGamer]

Not happening for Arrow Lake Refresh. They only had Adamantine cache which they cancelled because it was either too expensive or failed to meet their expectations. Even with Nova Lake, they will make CPUs with large cache a very premium product. No cheaper options for peasants like 7600X3D or 7500X3D.

X3D has always been a premium product - the SKUs using reject dies typically come much later, Intel has done this plenty too.

Of course the only thing stopping Intel, is Intel.

 

[igor_kavinski]

Of course the only thing stopping Intel, is Intel.

That is SOOOO true. I haven't seen a company as big as Intel shooting itself in the foot so much, like every chance it gets and still stay alive. It's amazing, miraculous, mind boggling and extremely annoying all at the same time.

Had they taken performance seriously, Apple would just still be considered a mobile CPU maker with performance fit for mobiles only, not challenging desktop CPUs.

 

[DAPUNISHER]

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.

Ignoring the personal attack, let's break this down.

The reason Intel had great sales in the past without longevity, is there was no viable alternative. DIYers were conditioned to tick tock. As AM3+ hadn't had a performance boost since Vishera in 2012 and it was getting starched worse and worse by each new Intel generation. AM4 changed all that. Here is how important platform longevity is to DIYers now that they have experienced the advantages it can bring.


That's over 37K respondents. You can get accurate results that extrapolate to any entire group, within a margin of +/- 3% with only 1K responses. Meaning only 7% of DIYers share your opinion that platform longevity does not matter. Fun fact: while correlation doesn't equal causation, that tiny percentage with the margin to stretch it slightly, is basically Intel's current retail market share. Almost like people buy the hardware that matches their opinion on longevity.

Steve and Tim have a long conversation about the topic in their latest podcast. There is no escaping the advantages of platform longevity. It is now an expectation. Reviewers and gamers said there would be hell to pay if Zen 6 was not on AM5. Now the rumors say Zen 7 may be as well. The rumors also say Intel's next platform will support upto 4 gens. I am looking forward to how important platform longevity becomes to the remaining 7% of DIYers when it likely favors Intel in a couple of years.

 

[igor_kavinski]

I am looking forward to how important platform longevity becomes to the remaining 7% of DIYers when it likely favors Intel in a couple of years.

It's pretty easy to predict that they will start citing that as the number one reason to go Intel.

"Hey, if the refresh disappoints, we can just wait for the tock hitting in the 3rd year of the platform and get our money's worth of investing into this platform. Intel is awesome like that!"

 

[LazyGamer]

Steve and Tim have a long conversation about the topic in their latest podcast. There is no escaping the advantages of platform longevity. It is now an expectation. Reviewers and gamers said there would be hell to pay if Zen 6 was not on AM5. Now the rumors say Zen 7 may be as well. The rumors also say Intel's next platform will support upto 4 gens. I am looking forward to how important platform longevity becomes to the remaining 7% of DIYers when it likely favors Intel in a couple of years.

Part of this is, now that we have DDR5 and PCIe5 knocked out, there's very little room for platforms to grow - aside from Intel's on-again-off-again FIVR implementation, we're topped out on economical upgrades of consumer socket / chipset PCIe connectivity, memory channels, power delivery (really: heat dissipation), and so on.

Notably AM4 held AMD back a bit; thankfully AM5 has most of what's needed (see: lower PCIe connectivity than Intel, but still more than sufficient).

On the Intel side, I don't think that there's a good reason for them to have forced the move from LGA1700 to LGA1851 to Nova Lake's upcoming LGA1954. That's probably just my ignorance though; still, they all support DDR5 and PCIe5, as well as more PCIe lanes than we see on AM5.

"Hey, if the refresh disappoints, we can just wait for the tock hitting in the 3rd year of the platform and get our money's worth of investing into this platform. Intel is awesome like that!"

Overall worth the mockery, but Microcenter bundles can be great to get decent baseline Intel stuff from a generation or two at steep discount for... activities.

 

[igor_kavinski]

On the Intel side, I don't think that there's a good reason for them to have forced the move from LGA1700 to LGA1851 to Nova Lake's upcoming LGA1954.

My guess is that LGA1851 has extra pins for the NPU while LGA1954 maybe needs extra pins for supporting dual tile based CPUs?

 

[Denpepe]

That's over 37K respondents. You can get accurate results that extrapolate to any entire group, within a margin of +/- 3% with only 1K responses. Meaning only 7% of DIYers share your opinion that platform longevity does not matter. Fun fact: while correlation doesn't equal causation, that tiny percentage with the margin to stretch it slightly, is basically Intel's current retail market share. Almost like people buy the hardware that matches their opinion on longevity.

Steve and Tim have a long conversation about the topic in their latest podcast. There is no escaping the advantages of platform longevity. It is now an expectation. Reviewers and gamers said there would be hell to pay if Zen 6 was not on AM5. Now the rumors say Zen 7 may be as well. The rumors also say Intel's next platform will support upto 4 gens. I am looking forward to how important platform longevity becomes to the remaining 7% of DIYers when it likely favors Intel in a couple of years.

saying something is important does not mean anything more then just that. Early AM4 mobo's were barebones and could barely hold 1 gen worth of UEFI in their bios chip, how many people have bought a high end 5xxx series CPU for an entry level 3xx series board?

DIY is a small (but fun) part of the market but where the bulk of the sales are, they sell whole pc's and they come with the newest mobo's anyways and they rarely get upgraded.

I would like to see some actual data of how many people actually upgraded without changing mobo iso some hypotethical question.

 

[igor_kavinski]

I would like to see some actual data of how many people actually upgraded without changing mobo iso some hypotethical question.

Without doing extensive data mining (and not sure how legal that would be), here's one example: https://browser.geekbench.com/v6/cpu/14827623

I'm on the side of the little guy. I want computing to be affordable because any computer user, be it productivity focused or gaming focused, is ultimately doing that activity for the pursuit of happiness and you can't have happiness with your bank account being drained by brazenly capitalistic companies like Apple/Nvidia.

The above GB result's user could have upgraded his B450 mobo or he could have bought it new and cheaper than a B550. Either way, that is validation for AMD's socket longevity strategy because the outcome in the end is affordability and ultimately customer satisfaction. What could this user have gotten instead of a 5700X/B450 combo? Maybe 12400/B660 combo? I doubt that user could have gotten the Intel combo cheaper than AMD since Intel tends to command a higher price, especially in developing markets.

I like and prefer that AMD continue on their existing path of serving people, especially the struggling ones. Were it not for them, we would still be paying $200 for four core eight thread CPUs. AMD completely disrupted the market and Intel is in trouble because they failed to see the economic power in the hands of the little people. They voted with their wallets to ruin Intel.

 

[Brian_B]

That's over 37K respondents. You can get accurate results that extrapolate to any entire group, within a margin of +/- 3% with only 1K responses. Meaning only 7% of DIYers share your opinion that platform longevity does not matter. Fun fact: while correlation doesn't equal causation, that tiny percentage with the margin to stretch it slightly, is basically Intel's current retail market share. Almost like people buy the hardware that matches their opinion on longevity.

I'm only one data point, but...

If you ask me - hey, is longevity important to my PC parts selection - well yes, it is. (I am saying this in the context of the poll that was linked - so yeah, CPU slot upgradeability)

In practice - how often have I ever just upgraded a CPU on an existing platform? Maybe once... the vast majority of the time, if I'm buying a new CPU, it's because I'm doing an entire new build around it. And that means that slot compatibility is ... not a big factor really.

Now - that said - that's only one facet. There's another facet I'm more interested in though than just upgrading a CPU. That's one of hardware maturity. When a socket sticks around for a while - it gets a nice level of consistency and maturity. Sure, chipsets get upgraded, often with every new CPU generation. But we still get this effect where every new generation isn't also accompanied by a brand new socket with new heat sink layouts, or new pinouts on motherboards that need some time to work out issues with errata in peripheral equipment, etc.

So for the second part, that's why I really like the upgradability. Not because I want to upgrade, but because I think (I have no data to prove though) that it results in more stable hardware offerings. I just find if you are jumping into a platform on first revision with new socket, new chipset, new generation CPU - there always seems to be a period where there's a whole lot of revisions/patches/glitches, and it's not until the second or third hardware revision (not just BIOS upgrade) that things really start to stabilize out.