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

Shaolin Chicken · 9 Apr 2025 at 15:56

Glanza said:
Had similar issues with a 8400 kit on the QVL list, even got another identical set in case it was faulty and same was happening with random BSOD when it would boot, swapped it to my known working 6000 sticks and it was fine for the remaining time I was on Intel before switching back to AMD.

Thanks been messing about with it all afternoon, can get it to boot at 8200, but will not post with the correct timings.

RSR · 9 Apr 2025 at 15:56

Shaolin Chicken said:
Now having problems with my RAM, booted fine on bios it came with at XMP profile 1 8200, now fails to boot on profile 1, RAM is on the QVL too, don't know what is going on.

What is the board and ram kit?

RSR · 9 Apr 2025 at 16:21

Looking back through the thread:

ASRock Z890I NOVA WiFi

https://pg.asrock.com/mb/Intel/Z890I%20Nova%20WiFi/index.asp#BIOS

I'd recommend using the latest BIOS update:

2.26

2025/3/28

1. Support Intel Maintenance Release 1 (MR1).
2. Update Intel Microcode 0x117.
3. Improve Memory Compatibility.

Is this still the same kit of RAM?

32GB (2x16GB) Corsair DDR5 Vengeance RGB Black, PC5-64000 (8000), Non-ECC Unbuffered, CAS 36, XMP 3.0, 1.5V

Shaolin Chicken · 9 Apr 2025 at 16:29

RSR said:
Looking back through the thread:

ASRock Z890I NOVA WiFi

https://pg.asrock.com/mb/Intel/Z890I%20Nova%20WiFi/index.asp#BIOS

I'd recommend using the latest BIOS update:

2.26 2025/3/28

1. Support Intel Maintenance Release 1 (MR1).
2. Update Intel Microcode 0x117.
3. Improve Memory Compatibility.

Is this still the same kit of RAM?

32GB (2x16GB) Corsair DDR5 Vengeance RGB Black, PC5-64000 (8000), Non-ECC Unbuffered, CAS 36, XMP 3.0, 1.5V

Hi RSR yes that is the board, the ram is G.SKILL Trident Z5 CK RGB 48GB (2x24GB) DDR5-8200 CL40, F5-8200C4052G24GX2-TZ5CRK, as I changed the ram, still have the older ram though.

RSR · 9 Apr 2025 at 16:32

Shaolin Chicken said:
Hi RSR yes that is the board, the ram is G.SKILL Trident Z5 CK RGB 48GB (2x24GB) DDR5-8200 CL40, F5-8200C4052G24GX2-TZ5CRK, as I changed the ram, still have the older ram though.

As above, I'd update the bios if you have not done so already. I'd also perform a full bios reset and then set XMP, but change the 8200 to 8000 and test.

Also make sure its setting the VDD and VDDQ voltages.

Shaolin Chicken · 9 Apr 2025 at 18:21

RSR said:
As above, I'd update the bios if you have not done so already. I'd also perform a full bios reset and then set XMP, but change the 8200 to 8000 and test.

Also make sure its setting the VDD and VDDQ voltages.

Thanks, I tried everything you said and no joy, sadly.
Can get it to boot at 8200 with all the main timings correct, but CAS is really high at 66 instead of the 40, which it is rated for.
Will probably try the other ram tomorrow, even though it is only 32GB not the 48GB of the G-Skill.

RSR · 9 Apr 2025 at 18:24

Shaolin Chicken said:
Thanks, I tried everything you said and no joy, sadly.
Can get it to boot at 8200 with all the main timings correct, but CAS is really high at 66 instead of the 40, which it is rated for.
Will probably try the other ram tomorrow, even though it is only 32GB not the 48GB of the G-Skill.

Are you using a contact frame or anything like that? As they can often affect memory.

The other thing would be to check and remount the CPU

Shaolin Chicken · 9 Apr 2025 at 18:30

RSR said:
Are you using a contact frame or anything like that? As they can often affect memory.

The other thing would be to check and remount the CPU

But it was working on the original shipped bios which is over a year old, works on XMP profile to 6400, no contact frame, don't really want to check and remount CPU it was a pain getting it installed the first time round, will try the other RAM first.

Shaolin Chicken · 10 Apr 2025 at 07:42

Well the Corsair 32GB DDR5 8000 worked first time.
Loaded BIOS defaults rebooted and then loaded profile 1 and it was still working, woohoo!

I'll install an operating system later today.

Shaolin Chicken · 10 Apr 2025 at 10:55

Can anyone tell me what are the drivers I should download from the manufacturers website please?

RSR · 10 Apr 2025 at 11:19

Shaolin Chicken said:
Can anyone tell me what are the drivers I should download from the manufacturers website please?

Asrock website should have the latest: https://pg.asrock.com/mb/Intel/Z890I Nova WiFi/index.asp#Download

However, you can double check them against the Asus Firmware/ Driver resource page: https://rog-forum.asus.com/t5/hardw...-drivers-firmware-software-threads/m-p/827232

jigger · 10 Apr 2025 at 11:23

Rroff said:
This is the problem with a lot of reviews not showing the full spectrum especially how Intel CPUs are still often significantly more power efficient than AMD in idle through to moderate utilisation while AMD currently hold the advantage under heavy utilisation.

So the actual power consumption over a whole day can be a very different story depending on your usage.

Third graph down. 47 aggregate test results.

AMD Ryzen 7 9800X3D Review - The Best Gaming Processor

The Ryzen 7 9800X3D establishes AMD as the leader in gaming performance. This Zen 5-based X3D chip is not only fast, it also comes with full support for overclocking. Besides gaming, application performance is considerably improved over the 7800X3D, but that comes at a price.

www.techpowerup.com

Rroff · 10 Apr 2025 at 11:35

jigger said:
Third graph down. 47 aggregate test results.

AMD Ryzen 7 9800X3D Review - The Best Gaming Processor

The Ryzen 7 9800X3D establishes AMD as the leader in gaming performance. This Zen 5-based X3D chip is not only fast, it also comes with full support for overclocking. Besides gaming, application performance is considerably improved over the 7800X3D, but that comes at a price.

www.techpowerup.com

Which is a graph of application benchmarks i.e. high CPU load not a reflection of the CPU utilisation profile of those applications over a longer time period outside of things like file compression and video compression which has a fairly binary profile - but then you'd have to look at the difference in power consumption over idle periods vs loaded periods for different usage patterns.

jigger · 10 Apr 2025 at 11:48

Rroff said:
Which is a graph of application benchmarks i.e. high CPU load not a reflection of the CPU utilisation profile of those applications over a longer time period outside of things like file compression and video compression which has a fairly binary profile - but then you'd have to look at the difference in power consumption over idle periods vs loaded periods for different usage patterns.

You said full spectrum which is a range of varying workloads from minimal to full load and what that graph represents in aggregate.

The idle usage would be the 5th graph down.

Rroff · 10 Apr 2025 at 12:09

jigger said:
You said full spectrum which is a range of varying workloads from minimal to full load and what that graph represents in aggregate.

The idle usage would be the 5th graph down.

That graph doesn't represent varying workloads at varying utilisation levels - it is synthetic benchmarks of those applications under load conditions, idle usage alone doesn't tell you much when you don't know the pattern of idle vs load usage over a full day.

The Unreal Engine benchmark for example does a short compile of a simple test level which takes 1-2 minutes to run and represents how the CPUs stack up under one of the more demanding parts of using the application - it doesn't represent the power consumption of someone working in Unreal Engine for several hours.

jigger · 10 Apr 2025 at 12:16

Rroff said:
That graph doesn't represent varying workloads at varying utilisation levels - it is synthetic.

It’s a representation of aggregate power use at various loads, so yeah, it represents various workloads. I think what you’re looking for are bias results.

Rroff · 10 Apr 2025 at 12:38

jigger said:
It’s a representation of aggregate power use at various loads, so yeah, it represents various workloads. I think what you’re looking for are bias results.

I dunno why I have to spell it out - the bulk of those tests are contained, synthetic, portions of high CPU load in those applications like compressing a file, running a post-processing filter, sorting a document, etc. they mostly don't represent long term use in different load conditions, the Office tests do have a bit more varied conditions but it is still a relatively short run that isn't representative of someone doing their normal work day use.

What I'm looking for is the representation of actual usage, I don't blame reviewers for avoiding the complications of having to try and faithfully reproduce long term runs over many different usage profiles to give a fuller picture but I do blame people for not being able to extrapolate those performance points into the real world.

jigger · 10 Apr 2025 at 12:43

Rroff said:
I dunno why I have to spell it out - the bulk of those tests are contained, synthetic, portions of high CPU load in those applications like compressing a file, running a post-processing filter, sorting a document, etc. they mostly don't represent long term use in different load conditions, the Office tests do have a bit more varied conditions but it is still a relatively short run that isn't representative of someone doing their normal work day use.

What I'm looking for is the representation of actual usage, I don't blame reviewers for avoiding the complications of having to try and faithfully reproduce long term runs over many different usage profiles to give a fuller picture but I do blame people for not being able to extrapolate those performance points into the real world.

You can find plenty of reviews that lack objectivity and show non real world results. YouTube has tons.

Rroff · 10 Apr 2025 at 12:45

jigger said:
You can find plenty of reviews that lack objectivity and show non real world results. YouTube has tons.

And we descend to the childishness because you don't like what I'm saying... and you know deep down it is true.

EDIT: PS I don't like ARL much at all as can be seen in this thread - if I was being bias it would be against it :s

jigger · 10 Apr 2025 at 12:56

Rroff said:
And we descend to the childishness because you don't like what I'm saying... and you know deep down it is true.

EDIT: PS I don't like ARL much at all as can be seen in this thread - if I was being bias it would be against it :s

What you’re looking is out there. Where you’re going wrong is trying to attach terms like ‘full spectrum’ and ‘real world’ to what are fringe results and not representative of typical use cases.