Intel Core Ultra 9 285k 'Arrow Lake' Discussion/News ("15th gen") on LGA-1851

[Journey]

Yes. The truth is that even the most stalwart intel fan is struggling with this one.

But I still think that the gaming performance, or lack of, is exaggerated by the testing methods of the reviewers. Few enthusiasts play at 1080P, and, I don't know about you, but I buy a new PC often enough that I will probably never be CPU bound. I think the reviews are scaring people away from the intel CPU's when they really would be fine for many people.

You are soooooooo wrong it is unreal.

I just did an upgrade the other day for someone, from I7 11700k to a 7800X3D because they play at 1080p in CS2 and want to get the most they can in FPS terms. I have done many of these sorts of upgrades, 1080p is a huge market for gamers to play at in Fortnite/CS2 etc. Don't say the methodology is wrong because it doesn't fit your own very narrow use case. I know way more people that play at 1080p than 1440p/4k.

 

[Rroff]

I know way more people that play at 1080p than 1440p/4k.

Outside of those playing competitive FPS/e-Sports/"Pro" gamers though most of those people playing at 1080p aren't using anything like a 4090 or a higher end CPU.

 

[RSR]

These were all completed with "Load Optimized Defaults" set, rebooted and the XMP1 set.

MB: Asus Z890 Apex
CPU 285K
RAM: Team XTREEM 48GB CUDIMM DDR5 8800 - FXD548G8800HC42ADC-CU01
Windows Version: 24H2 - 26100.2605


UEFI 1101 - MC 0x113
ME 19.0.0.1753

UEFI 1202 - MC 0x113
ME 19.0.0.1827

UEFI 1203 - MC 0x114
ME 19.0.0.1854

UEFI 1302 - MC 0x114
ME 19.0.0.1854v2.2

So, in addition to my earlier comments, it does seem that MC 0x114 increases the latency, as there was a 1202 and 1203 update that introduced the 114 MC.

 

[RSR]

I need to see where the G2 memory profile ends up, as when running defaults, the D2D, NGU, and Cache are all set fairly low, which, when increased, brings the latency down quite a bit.

That said, I have mostly been using it as stock with just XMP and it's been motoring long fine.

 

[SpellowHouse]

You are soooooooo wrong it is unreal.

I just did an upgrade the other day for someone, from I7 11700k to a 7800X3D because they play at 1080p in CS2 and want to get the most they can in FPS terms. I have done many of these sorts of upgrades, 1080p is a huge market for gamers to play at in Fortnite/CS2 etc. Don't say the methodology is wrong because it doesn't fit your own very narrow use case. I know way more people that play at 1080p than 1440p/4k.

You consider 1440P and 4K "narrow use" ? I wouldn't exactly agree with you on that.

Well, we will have to agree to disagree. I can honestly say that I don't know a single enthusiast who plays in 1080P. But, in fairness, one persons experience doesn't really count for much....and that applies to you as well.

But I still think that the reviewers are answering the wrong question. They are answering "what is the best CPU?" when they should be answering "what is the best upgrade?". The two are not the same. The crazy thing is that all they need to do is run comprehensive tests in 1440P and 4K to give us a hint, yet, often, in their haste to get their results out, they don't bother.

There is more to life than just the fps highs. There are the lows. The temperatures. Power consumption. These are all things I would like to get a handle on before I buy a new CPU. I can't do that if they run unrealistic tests.

 

[Lord-Jaffa]

You consider 1440P and 4K "narrow use" ? I wouldn't exactly agree with you on that.

Well, we will have to agree to disagree. I can honestly say that I don't know a single enthusiast who plays in 1080P. But, in fairness, one persons experience doesn't really count for much....and that applies to you as well.

But I still think that the reviewers are answering the wrong question. They are answering "what is the best CPU?" when they should be answering "what is the best upgrade?". The two are not the same. The crazy thing is that all they need to do is run comprehensive tests in 1440P and 4K to give us a hint, yet, often, in their haste to get their results out, they don't bother.

There is more to life than just the fps highs. There are the lows. The temperatures. Power consumption. These are all things I would like to get a handle on before I buy a new CPU. I can't do that if they run unrealistic tests.

If you're talking about user statistics then Steam has this covered, 55% are using 1080p as a primary monitor.

Steam Hardware & Software Survey

As for Arrow Lake, what a disaster!

 

[RSR]

Here is another interesting one on how to decrease the latency further by changing the D2D, NPU, and CR while still using XMP1 timings

System Spec - All defaults other than XMP and options are changed below.

MB: Asus Z890 Apex
CPU 285K
RAM: G.Skill Trident Z CK RGB - 8800
Windows Version: 24H2 - 26100.2605
BIOS: UEFI 1302 - MC 0x114
CSME 19.0.0.1854v2.2

XMP 1

XMP 1 + D2D set to 38 - Reduction around ~4ns

XMP 1 + D2D set to 38 + NPU set to 35 - Reduction around ~3ns

XMP 1 + D2D set to 38 + NPU set to 35 + CR set to 42 with downbin disabled. - Reduction around ~2ns

Manually setting up the memory will also see a further reduction in latency

 

[Journey]

You consider 1440P and 4K "narrow use" ? I wouldn't exactly agree with you on that.

For your interpretation yes, because that is how you'll take it anyhow.

I can honestly say that I don't know a single enthusiast who plays in 1080P. But, in fairness, one persons experience doesn't really count for much....and that applies to you as well.

No it doesn't apply to me since I deal with actual real people who actually need machines, not just someone who sits and watches YouTube videos all day then complains that the review is done badly.

But I still think that the reviewers are answering the wrong question. They are answering "what is the best CPU?" when they should be answering "what is the best upgrade?".

No they are not reviewing upgrades., they are reviewing a new CPU, and in some cases the related platform.

Intel CPU's are a mess right now, changing the way the review in done to spin them in a more positive light is only hurting the consumer as they'll buy the part that isn't as good overall. So you are saying data should be obfuscated and allow the product to look like it is better than it is, well you need to maybe go work for Intel marketing?

 

[DarrenM343]

If you're talking about user statistics then Steam has this covered, 55% are using 1080p as a primary monitor.

Steam Hardware & Software Survey

As for Arrow Lake, what a disaster!

Note the CPU speeds. I don't think many are paying £300, £400, £500+ for a new CPU to game at 1080P.
'Best gaming CPU' claims at 1080P should always be taken as a best case scenario I think for that CPU. It's a valid test but not necessarily relevant in the likely use cases for the CPU's. At higher resolutions the results will be closer
However, a good 1080P gaming CPU will likely perform better at higher resolutions as GPU's become more bottlenecked at those higher resolutions, but I think by that time many would have upgraded again anyway

 

[Game]

You consider 1440P and 4K "narrow use" ? I wouldn't exactly agree with you on that.

Well, we will have to agree to disagree. I can honestly say that I don't know a single enthusiast who plays in 1080P. But, in fairness, one persons experience doesn't really count for much....and that applies to you as well.

But I still think that the reviewers are answering the wrong question. They are answering "what is the best CPU?" when they should be answering "what is the best upgrade?". The two are not the same. The crazy thing is that all they need to do is run comprehensive tests in 1440P and 4K to give us a hint, yet, often, in their haste to get their results out, they don't bother.

There is more to life than just the fps highs. There are the lows. The temperatures. Power consumption. These are all things I would like to get a handle on before I buy a new CPU. I can't do that if they run unrealistic tests.

there will be differences at 1440p but the main issue is only using a 4090 which pretty much makes any test pointless for most people

would rather see a good spread of graphics cards tested with each processor at different res see how they behave with each other

 

[RSR]

CPU-Z

CPU Seems to have had a small increase in performance since the release of bios and today.

UEFI 0806 + XMP

UEFI 1302 + XMP

Two things to note, when comparing them both where stock CPU setting between those two tests its running two different RAM kits and the Windows OS version is the non-patched 24H2.

 

[Bidley]

there will be differences at 1440p but the main issue is only using a 4090 which pretty much makes any test pointless for most people

would rather see a good spread of graphics cards tested with each processor at different res see how they behave with each other

Ok so lets say 10 games tested as a bare minimum (and that set will vary from person to person, so good luck choosing ones that nobody will moan about), at 4 different resolutions (1080p, 1440p, 3440x1440 [because I play at that resolution and apparently need benchmarks at the exact resolution I play at to make a decision] and 4k), with 3 current gen Nvidia cards, 3 current gen AMD cards, Intel's GPUs, and perhaps 3 from each from last gen because comparatively few upgrade every gen (I say 3 because that would in theory cover high-mid-low end).

Then you have all the related thermal testing and power consumption. I suppose we'll be wanting those numbers from the above tests as well.

I don't have the maths skills to calculate the quantity of tests - but it's far more than you need when you can get all that info from 1080p testing. The GPUs and resolutions don't matter because it's a CPU and won't give you more frames if you're GPU-limited.

Just buy the Intel CPU and stop looking for validation from Youtubers.

 

[Game]

Ok so lets say 10 games tested as a bare minimum (and that set will vary from person to person, so good luck choosing ones that nobody will moan about), at 4 different resolutions (1080p, 1440p, 3440x1440 [because I play at that resolution and apparently need benchmarks at the exact resolution I play at to make a decision] and 4k), with 3 current gen Nvidia cards, 3 current gen AMD cards, Intel's GPUs, and perhaps 3 from each from last gen because comparatively few upgrade every gen (I say 3 because that would in theory cover high-mid-low end).

Then you have all the related thermal testing and power consumption. I suppose we'll be wanting those numbers from the above tests as well.

I don't have the maths skills to calculate the quantity of tests - but it's far more than you need when you can get all that info from 1080p testing. The GPUs and resolutions don't matter because it's a CPU and won't give you more frames if you're GPU-limited.

Just buy the Intel CPU and stop looking for validation from Youtubers.

in no hurry atm more watching this out of interest due to it being a new platform sorted out my own issue and might just buy a gpu once we see the next gen

then upgrade the rest before xmas

 

[Cyber Ghost]

Here is another interesting one on how to decrease the latency further by changing the D2D, NPU, and CR while still using XMP1 timings

System Spec - All defaults other than XMP and options are changed below.

MB: Asus Z890 Apex
CPU 285K
RAM: G.Skill Trident Z CK RGB - 8800
Windows Version: 24H2 - 26100.2605
BIOS: UEFI 1302 - MC 0x114
CSME 19.0.0.1854v2.2

XMP 1

XMP 1 + D2D set to 38 - Reduction around ~4ns

XMP 1 + D2D set to 38 + NPU set to 35 - Reduction around ~3ns

XMP 1 + D2D set to 38 + NPU set to 35 + CR set to 42 with downbin disabled. - Reduction around ~2ns

Manually setting up the memory will also see a further reduction in latency

CPU-Z

CPU Seems to have had a small increase in performance since the release of bios and today.

UEFI 0806 + XMP

UEFI 1302 + XMP

Two things to note, when comparing them both where stock CPU setting between those two tests its running two different RAM kits and the Windows OS version is the non-patched 24H2.

Thanks for all the testing and info above

 

[humbug]

Yes. The truth is that even the most stalwart intel fan is struggling with this one.

But I still think that the gaming performance, or lack of, is exaggerated by the testing methods of the reviewers. Few enthusiasts play at 1080P, and, I don't know about you, but I buy a new PC often enough that I will probably never be CPU bound. I think the reviews are scaring people away from the intel CPU's when they really would be fine for many people.

The problem is the one that most would buy for gaming at £380 its trying to pass its self off as something it isn't, its only Intel and their loyalists who complain about 1080P testing, the reason being is it shows they are overpriced for gaming, of course they don't like that, they want you to spend near £400 on this thing thinking its going to be great for gaming, if you can see that much cheaper CPU's are just as good or better well then that's a problem...

Even with expensive 8200MT/s RAM, they are giving it the best possible chance they can.... its still garbage for its price. Think what its like with RAM that doesn't cost more than the CPU.

 

[throwaway4372]

That chart shows it doesn't matter what you buy if you have a 4090 at 1080p - which nobody does. So if anything it'll make people buy cheaper options.

 

[Journey]

That chart shows it doesn't matter what you buy if you have a 4090 at 1080p - which nobody does. So if anything it'll make people buy cheaper options.

No it shows if you want the best 1% lows and performance, you buy an Zen4 X3D CPU

 

[throwaway4372]

Only if you have infinite budget.

 

[humbug]

That chart shows it doesn't matter what you buy if you have a 4090 at 1080p - which nobody does. So if anything it'll make people buy cheaper options.

There are cheaper options.

14600K at £230
7600X at £190

No one would buy the 265K for gaming looking at that as you can see the CPU costing literally half as much is at least as good and its performance is what you will get on a sub £100 C30 6000MT/s 32GB Kit.

You see the thing is if you test them in such a way that they all look the same you're never going to think to buy the £190 CPU over the £380 CPU for gaming as most people are going to assume more expensive = better, that's how people think, they know it and that's why they are upset that this isn't being done.

 

[jigger]

Only if you have infinite budget.

And/or an IQ>90