Review Intel Core Ultra 9 285K Review: Intel Throws a Lateral with Arrow Lake

[TheHerald]

Seems like clocking the cache does wonders for gaming on the 285k. From the alca.cz review

 

[bit_user]

I find it somewhat funny that everyone is crying about the flat or slight fall in gaming numbers. Intel's sells the VAST majority of its chips to OEM's, of which the VAST majority of those systems are for business purposes.

I'm curious about this, because a lot of business PCs are either laptops, little mini-PCs, or servers. I wouldn't expect the "VAST majority" of mainstream socketed machines are going into the corporate market, any more.

Right now, that is super good news for Intel since workload performance and power efficiency is tops on all those guys lists.

Okay, so let's look at performance:

​

The 9950X wins on outright performance and the 9700X is close enough that it could compete with the 245K on price, if it dropped a bit more.

Effeciency-wise, the 9950X is also just 1.7% worse, according to the above Handbrake test. However, if you look at Phoronix' review, the 9950X stomps the 285K on both performance and efficiency. It delivers a better Geomean by 17.3% and 19.6% better efficiency.

https://www.phoronix.com/review/intel-core-ultra-9-285k-linux
​

Now, maybe that's specific to Linux... I have yet to see a similar efficiency metric over a broad suite of MT apps on Windows.

 

[usertests]

You sure about that?

Yes, I looked at other reviews and not just memebench

 

[bit_user]

The PC Gaming Audience has a disproportional amount of sales vs marketing, & customer psychological influence.

You have to understand how important they are.

Our noise influences ALOT of purchasing decisions.

No, not our noise, but rather the popular reviewers and influencers on youtube, tik tok, and other social media.

 

[magbarn]

Decent productivity CPU, 50% increase in idle power usage. Must be a chiplet disadvantage, but AMD can't seem to fix their idle consumption which most computers are running at.

 

[TheHerald]

Now, maybe that's specific to Linux... I have yet to see a similar efficiency metric over a broad suite of MT apps on Windows.

It's the avx support. Imporant for the server workspace, useless for us mere mortals. That's why you see differences in phoronix and not in other review.

EG1. You can see it in ycruncher. Using the AVX version a 14900k needs double the wattage to match a 7950x. Even my 12900k barely edges ahead a 7700x at ~3 times the power.

Yes, I looked at other reviews and not just memebench

How about handbrake? It leads in that too.

It's fine, don't expect you to accept it. AMD is ahead, if that makes you happy, im happy.

 

[helper800]

It's great to see Intel getting some productivity wins and huge efficiency gains even if its at the cost of 7% or so gaming performance compared to last gen. I would make the assumption that with some BIOS updates and scheduler work Intel can claw back some gaming performance as well. Intel has long needed to catch ryzens top 16 core chips in performance and efficiency and with this generation it effectively is par with AMD now. I assume that a tweaked 285k vs a tweaked 9950x in the interest of efficiency over performance it will be tight for who wins on most MT applications. I would consider this to be Intel putting AMD on notice. Competition is great!

 

[TheHerald]

It's great to see Intel getting some productivity wins and huge efficiency gains even if its at the cost of 7% or so gaming performance compared to last gen. I would make the assumption that with some BIOS updates and scheduler work Intel can claw back some gaming performance as well. Intel has long needed to catch ryzens top 16 core chips in performance and efficiency and with this generation it effectively is par with AMD now. I assume that a tweaked 285k vs a tweaked 9950x in the interest of efficiency over performance it will be tight for who wins on most MT applications. I would consider this to be Intel putting AMD on notice. Competition is great!

Scheduler is an easy thing to test for, if reviewers care. Just run the entire game on pcores and see what's up.

I'm afraid it's not the scheduler though. A lot of games perform better with ecores on (at least until 14th gen) because they put non critical gaming tasks on those. But with 285k, since HT is gone, core heavy games might be offloading some primary threads on the ecores which basically kills the performance. Cyberpunk is a prime example since it seems like it's one of those games that performance has tanked. It's losing to my alderlake for gods sake, lol.

Overall the whole lineup is top notch bar gaming. Gaming is well, let's call it acceptable and nothing more.

 

[TheHerald]

Also this power draw graph is pretty crazy. Lowest gaming power draw amongst it's peers, by far.

 

[-Fran-]

It's great to see Intel getting some productivity wins and huge efficiency gains even if its at the cost of 7% or so gaming performance compared to last gen. I would make the assumption that with some BIOS updates and scheduler work Intel can claw back some gaming performance as well. Intel has long needed to catch ryzens top 16 core chips in performance and efficiency and with this generation it effectively is par with AMD now. I assume that a tweaked 285k vs a tweaked 9950x in the interest of efficiency over performance it will be tight for who wins on most MT applications. I would consider this to be Intel putting AMD on notice. Competition is great!

Keep in mind:

1- AMD is not using the new I/O die they're using with EPYC, so that is a big performance loss if you compare there.
2- Intel is using a better process for all the tiles AND a better packaging process than AMD and they're still barely reaching efficiency parity on some tasks. This implies, vis a vis, Intel SKUs are more expensive to manufacture, so margins may be razor thin for Intel.
3- CU-DIMM and regular U-DIMM high speed RAM is very expensive still and most "positive" numbers you see from ArrowLake are using those more expensive kits.
4- Arguable, but for me it's a minus: rumoured single gen platform.

I do not think this is positive for Intel, at all. The only saving grace is the X890 platform, which at least on paper, reads quite nice. The CPU that goes with it is "mid" at best.

Regards.

 

[thestryker]

Appreciate the review and specifically the different memory configurations. It seems like this is going to be very important for ARL as overall memory latency seems the likely gaming regression cause. It seems like both AMD and Intel have run into somewhat of a wall as they're redesigning for the future. I imagine that whatever comes next from each company will see more universal uplift.

@PaulAlcorn the memory kit latencies aren't included in the test setup graphs, and given the additional latency with ARL this seems doubly important.

 

[thestryker]

2- Intel is using a better process for all the tiles AND a better packaging process than AMD and they're still barely reaching efficiency parity on some tasks. This implies, vis a vis, Intel SKUs are more expensive to manufacture, so margins may be razor thin for Intel.

Intel's using N6 for IO/SoC, N5 for graphics and N3 for CPU. So they're using the same process for IO, better for graphics (this hardly matters) and better for CPU.

 

[helper800]

Keep in mind:

1- AMD is not using the new I/O die they're using with EPYC, so that is a big performance loss if you compare there.
2- Intel is using a better process for all the tiles AND a better packaging process than AMD and they're still barely reaching efficiency parity on some tasks. This implies, vis a vis, Intel SKUs are more expensive to manufacture, so margins may be razor thin for Intel.
3- CU-DIMM and regular U-DIMM high speed RAM is very expensive still and most "positive" numbers you see from ArrowLake are using those more expensive kits.
4- Arguable, but for me it's a minus: rumoured single gen platform.

I do not think this is positive for Intel, at all. The only saving grace is the X890 platform, which at least on paper, reads quite nice. The CPU that goes with it is "mid" at best.

Regards.

I disagree on a couple points and agree on others. Thanks for the info about the the underlying processes of both the new Intel CPUs and the current AMD. Personally, I see this as Intel coming up for air after treading water for 4 years with their top chip's power efficiency, and performance (gaming excluded).

Gen on Gen Intel has done good work here. I presume a lot of the expected let downs in performance as a comparison to what nodes they are using on specific dies to come down to its first major implementation of chiplets in combination with its bigLittle strategy. If Intel can withstand financial pressures and given the time to iron out some kinks they may be able to get a fair few gains from the next attempt and with updates.

Remember how rough AMDs first ryzen generation was at launch compared to a couple years down the line with scheduler optimizations and AGESA tweaks? I see this performance at launch as the staging ground for some considerable improvement.

 

[Elusive Ruse]

Also this power draw graph is pretty crazy. Lowest gaming power draw amongst it's peers, by far.

What do you mean by gaming performance among peers? As in flagship gaming CPUs?

 

[Geekaycee]

Changing a generation for Intel used to be tick-tock. It is now flip-flop.

 

[-Fran-]

I disagree on a couple points and agree on others. Thanks for the info about the the underlying processes of both the new Intel CPUs and the current AMD. Personally, I see this as Intel coming up for air after treading water for 4 years with their top chip's power efficiency, and performance (gaming excluded).

Gen on Gen Intel has done good work here. I presume a lot of the expected let downs in performance as a comparison to what nodes they are using on specific dies to come down to its first major implementation of chiplets in combination with its bigLittle strategy. If Intel can withstand financial pressures and given the time to iron out some kinks they may be able to get a fair few gains from the next attempt and with updates.

Remember how rough AMDs first ryzen generation was at launch compared to a couple years down the line with scheduler optimizations and AGESA tweaks? I see this performance at launch as the staging ground for some considerable improvement.

It's fine to disagree, since interpretation and shoes not fitting all feet, etc.

The "gen on gen". Depends. Again, they're using a better process (with some asterisks; thanks thestryker) and packaging tech when AMD did it on the same nodes Zen1+ to Zen2 and still got a big generational improvement.

Sorry, but this is not "good work" to me. It barely qualifies for mediocre. The only thing they improve is power consumption, which is good, but also introduce regression in a lot of areas even when sacrificing Hyperthreading (and core count).

Too much of a mixed bag and way worse than any gen jump from AMD in recent years.

This is the second coming of Pentium 4 for them.

Regards.

 

[Thunder64]

It's fine to disagree, since interpretation and shoes not fitting all feet, etc.

The "gen on gen". Depends. Again, they're using a better process (with some asterisks; thanks thestryker) and packaging tech when AMD did it on the same nodes Zen1+ to Zen2 and still got a big generational improvement.

Sorry, but this is not "good work" to me. It barely qualifies for mediocre. The only thing they improve is power consumption, which is good, but also introduce regression in a lot of areas even when sacrificing Hyperthreading (and core count).

Too much of a mixed bag and way worse than any gen jump from AMD in recent years.

This is the second coming of Pentium 4 for them.

Regards.

Zen(+) was 14nm(12nm). Zen 2 was 7nm. You are probably thinking of Zen 2 to Zen 3, both on 7nm. The point is quite valid though.

 

[-Fran-]

Zen(+) was 14nm(12nm). Zen 2 was 7nm. You are probably thinking of Zen 2 to Zen 3, both on 7nm. The point is quite valid though.

Ah, you're right; I was mixing Zen2 and 3.

They went chiplets with Zen2 though, so my point is less valid on that specific tidbit.

Regards.

 

[Thunder64]

Ah, you're right; I was mixing Zen2 and 3.

They went chiplets with Zen2 though, so my point is less valid on that specific tidbit.

Regards.

Yup, I must've missed the "packaging tech part". Makes more sense now.

 

[TheHerald]

What do you mean by gaming performance among peers? As in flagship gaming CPUs?

No, I mean big multicored chips that are inherently inefficient in gaming. 9950x, 7950x, 14900k etc. It's a lot lower than all of them in power draw while matching the 9950x in performance. At least based on tomhardwares review.

 

[TheHerald]

I don't get the doom and gloom. According to both tomshardware.com and TPU review, the 285k is on par in both MT performance and efficiency with the 9950x (wins some, loses some), it beats it by 35% in st efficiency, it draws a lot less power in idle / semi idle / low power workloads (Autocad as tested by igors), offers identical gaming performance at way lower power draw.

I cant wrap my head around how is it any worse than the 9950x. It's better on every metric including the igpu.

The i5 and the i7 are obviously even more dominant but that has always been the case anyways since amds offerings are lacking core counts in that segment.

 

[oofdragon]

LMAO the 7800X3D owning 3 gens of Intel processors now. And Im beting it's still going to own Intel next gen 🤣🤣🤣

 

[dwd999]

As a non-gamer, I am VERY interested in this line...

It will also be interesting to see how the new ILM contact frame affects this chip's temps on the new motherboards. If this results in less bending and thus lower temps it will be a welcome change. Hopefully we'll see some tests of cpu and motherboard combos focused on this aspect in the coming weeks.

 

[USAFRet]

It will also be interesting to see how the new ILM contact frame affects this chip's temps on the new motherboards. If this results in less bending and thus lower temps it will be a welcome change. Hopefully we'll see some tests of cpu and motherboard combos focused on this aspect in the coming weeks.

Starting here:
https://www.tomshardware.com/pc-com...rrow-lake-cool-in-msis-mpg-velox-100r-pc-case

 

[bit_user]

It's the avx support. Imporant for the server workspace, useless for us mere mortals. That's why you see differences in phoronix and not in other review.

R9 9950X delivers better efficiency in Cinebench:

Source: https://www.techpowerup.com/review/intel-core-ultra-9-285k/24.html
​