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

[AORUS]

Can anyone help! New Build

I want to order two packs of the same Vengeance DDR5 48gb (2x48) DDR5 CUDIMM 8000T/s I know I won't get 8000T/s only 6400 - 7000T/s with XMP when using 4 stick of ram 96gb. on Asus Extreme Z890 MB with Intel 285k Note I'm not using any mix ram here.

On ASUS memory QVL MB webpage it shows SS (4x24) CUDIMM 7000T/s but there isn't any DDR5 SS CUDIMM (4x24) on market only sold in 48gb (2x24) packs that is single module. I don't really want UDIMMs if I can use CUDIMM! And don't want to use two sticks of ram

 

[Legion]

I don't really want UDIMMs if I can use CUDIMM! And don't want to use two sticks of ram

You don't need CUDIMM for what you are trying to achieve. CUDIMM won't realistically get you any further with 96GB than a UDIMM kit would. Your CPU's IMC and motherboard choice will decide the limits, not because you bought CUDIMM over UDIMM at the frequency range. Due to silicon lottery every CPU is a little bit different on how far they can push mem or tolerate 4x sticks at higher frequencies.
Both Intel and AMD don't bin Memory Controllers, you get what you get.

You'd be better off (and have better results) with a 6400 2x 48GB UDIMM kit.
A 2x 48GB UDIMM kit is easy to run these days on either Intel or AMD.
There is no advantage to running 4x 24GB CUDIMM sticks over a 2x 48GB UDIMM kit at 6400-7000.
CUDIMM isn't a magic bullet, it's only really useful on 32GB and 48GB kits at 8600+ where novices would struggle to stabilise a UDIMM kit on hardware that can actually push those frequencies. At 6400-7000 you'd just be throwing away money in CUDIMM kits that give you no advantage over a UDIMM kit.
A 2x 48GB 6400 UDIMM kit will run at XMP on just about any Z890 or B860 motherboard, even dirt cheap 6 layer boards. Most kits using Hynix M 24gb IC's will even move up a few frequency bins without even touching volts or timings.
Every 6400 2x 48GB kit out there with Hynix M 24gb IC's is listed with 32-39-39 primary timings. You'd want to avoid any kits that don't have that.

As for the Z890 Extreme, that's just beyond sensible. That board doesn't do anything a significantly cheaper board will do, you're just throwing money away there for no good reason.

From what I can see in your motherboard post, you've decided 96GB of system ram is the right size for what you want to do with it. That's fair enough. But after watching some trash vid from a no name Youtuber. You've got it stuck in your head you somehow need 4x 24GB CUDIMM over a 2x 48GB UDIMM kit.
4x single rank dimms have no advantage on Intel systems over 2x dual rank dimms, where have you got this idea from?
It's always preferable to run 2 sticks over 4 on current desktop hardware in just about every scenario.

 

[AORUS]

You don't need CUDIMM for what you are trying to achieve. CUDIMM won't realistically get you any further with 96GB than a UDIMM kit would. Your CPU's IMC and motherboard choice will decide the limits, not because you bought CUDIMM over UDIMM at the frequency range. Due to silicon lottery every CPU is a little bit different on how far they can push mem or tolerate 4x sticks at higher frequencies.
Both Intel and AMD don't bin Memory Controllers, you get what you get.

You'd be better off (and have better results) with a 6400 2x 48GB UDIMM kit.
A 2x 48GB UDIMM kit is easy to run these days on either Intel or AMD.
There is no advantage to running 4x 24GB CUDIMM sticks over a 2x 48GB UDIMM kit at 6400-7000.
CUDIMM isn't a magic bullet, it's only really useful on 32GB and 48GB kits at 8600+ where novices would struggle to stabilise a UDIMM kit on hardware that can actually push those frequencies. At 6400-7000 you'd just be throwing away money in CUDIMM kits that give you no advantage over a UDIMM kit.
A 2x 48GB 6400 UDIMM kit will run at XMP on just about any Z890 or B860 motherboard, even dirt cheap 6 layer boards. Most kits using Hynix M 24gb IC's will even move up a few frequency bins without even touching volts or timings.
Every 6400 2x 48GB kit out there with Hynix M 24gb IC's is listed with 32-39-39-39 primary timings. You'd want to avoid any kits that don't have that.

As for the Z890 Extreme, that's just beyond sensible. That board doesn't do anything a significantly cheaper board will do, you're just throwing money away there for no good reason.

From what I can see in your motherboard post, you've decided 96GB of system ram is the right size for what you want to do with it. That's fair enough. But after watching some trash vid from a no name Youtuber. You've got it stuck in your head you somehow need 4x 24GB CUDIMM over a 2x 48GB UDIMM kit.
4x single rank dimms have no advantage on Intel systems over 2x dual rank dimms, where have you got this idea from?
It's always preferable to run 2 sticks over 4 on current desktop hardware in just about every scenario.

I only came here to ask a question on RAM not to get into a why question on what to buy, I waited more then 8 years to build an High End PC within 7K. There must be reason for High end vs B version motherboards its the same for the RTX 50 series GPU cards and nearly everything we buy today, I understand your meaning on running 2x48 UDIMM kit the only advantage is (CKD) CUDIMM with more stable performance so running 4x24 is more to do with aesthetics for me than speed of cause I will get 6400T/s XMP with Intel 285K build on Asus & gigabyte QVL MB's Max 7000T/s if I want to.

Of cause there are many write ups on 2x24 CUDIMM built for speed I get that but limited in Ram 48gb maybe that will change over time! It's was the same when 16 and 32 came out also M.2 drives 3.0 vs 4.0 and today New tect 4.0 vs 5.0 drives it's seems to me we're on a merry go round when it comes to new technology and the disagreement are endless a nightmare for New PC building.

Everything is an overkill unless it a lower value budget PC!

 

[Vimes]

@Rroff I think that I am right to suggest that you have built a Intel 265k system..? If so have you noted what the highest E Core speed you are seeing in use..? According to Intel it is supposed to have a max turbo of 4.6Ghz. And yet, according to HWiNFO I am seeing some cores go up to 5.2Ghz.

Thanks

 

[Rroff]

@Rroff I think that I am right to suggest that you have built a Intel 265k system..? If so have you noted what the highest E Core speed you are seeing in use..? According to Intel it is supposed to have a max turbo of 4.6Ghz. And yet, according to HWiNFO I am seeing some cores go up to 5.2Ghz.

Thanks

I've not built a 265K system. Do you have any motherboard enhancements enabled in the BIOS?

 

[t2na]

@Rroff I think that I am right to suggest that you have built a Intel 265k system..? If so have you noted what the highest E Core speed you are seeing in use..? According to Intel it is supposed to have a max turbo of 4.6Ghz. And yet, according to HWiNFO I am seeing some cores go up to 5.2Ghz.

Thanks

I’ve got a 265k system so can take a look at this today. Pre-enabling 200s boost I had manually adjusted my E cores to 4.8Ghz and hadn’t checked since - so can have a look to see.

 

[Vimes]

I've not built a 265K system. Do you have any motherboard enhancements enabled in the BIOS?

Ah, fair enough.
I asked the creator of HWinfo, to make sure that it was reporting correctly.
He replied.....

It's probably due to Turbo Boost Max 3.0 Technology which can override the maximum per-core Turbo frequency limit or Thermal Velocity Boost.

I do not think that the board has any particular feature other than defaults applied. That sits not mean nothing though..!

Changing from a 14700k to the 265k it did surprise me how little voltage the CPU uses, even after applying a - 0.06v undervolt for the RL CPU.
For the cost the 265K is a worthy consideration.

 

[Vimes]

I’ve got a 265k system so can take a look at this today. Pre-enabling 200s boost I had manually adjusted my E cores to 4.8Ghz and hadn’t checked since - so can have a look to see.

Many thanks. At this point I have not enabled that feature. I just wanted to see how it was at stock, as such. I was surprised to see what a couple of the E cores are boosting to.

It was in a game that I noted the E Core speeds.

BTW are you happy enough with the 265k system you have built..?

 

[t2na]

Many thanks. At this point I have not enabled that feature. I just wanted to see how it was at stock, as such. I was surprised to see what a couple of the E cores are boosting to.

It was in a game that I noted the E Core speeds.

BTW are you happy enough with the 265k system you have built..?

So with 200s boost enabled and running a few games (AC Shadows, The Finals, Space Marine 2, Doom) the max E-Core frequency was 4.6Ghz. P-Core was 5.2Ghz - both reported from HWInfo.

If I run a benchmark/stress test from Intel XTU it reported a peak of 5.34 for P-Core and 4.70 for E-Core - but neither of those values ever appeared in HWInfo so I'm not trusting them all that much.

I then disabled 200s boost and swapped to my settings before (bumping E-cores to 4.8 Ghz, and raising the ring clock to 4000) and it was as you'd expect. E-Core frequency was then 4.8Ghz and P-Core still at 5.2Ghz - so it does seem to follow my settings. Is yours all at stock? No adjustments at all?

Yeah - super happy with the 265k system! It runs really nice and cool at idle with typical PC usage - and then even at load playing games it doesn't ever get too hot or loud. Overall the system feels 'snappy' and really responsive (which I appreciate is a bit of an odd measurement) regardless of what I'm doing or how much I've got going on at once.

If you need a PC for more than just gaming right now, the current pricing and value of the 265k makes it a bit of a no brainer. It's also not bad for gaming either (despite what people seem to keep repeating). It matches or beats my 5800x3D in all of the games that I currently play so it made a lot of sense for me.

 

[Vimes]

So with 200s boost enabled and running a few games (AC Shadows, The Finals, Space Marine 2, Doom) the max E-Core frequency was 4.6Ghz. P-Core was 5.2Ghz - both reported from HWInfo.

If I run a benchmark/stress test from Intel XTU it reported a peak of 5.34 for P-Core and 4.70 for E-Core - but neither of those values ever appeared in HWInfo so I'm not trusting them all that much.

I then disabled 200s boost and swapped to my settings before (bumping E-cores to 4.8 Ghz, and raising the ring clock to 4000) and it was as you'd expect. E-Core frequency was then 4.8Ghz and P-Core still at 5.2Ghz - so it does seem to follow my settings. Is yours all at stock? No adjustments at all?

Yeah - super happy with the 265k system! It runs really nice and cool at idle with typical PC usage - and then even at load playing games it doesn't ever get too hot or loud. Overall the system feels 'snappy' and really responsive (which I appreciate is a bit of an odd measurement) regardless of what I'm doing or how much I've got going on at once.

If you need a PC for more than just gaming right now, the current pricing and value of the 265k makes it a bit of a no brainer. It's also not bad for gaming either (despite what people seem to keep repeating). It matches or beats my 5800x3D in all of the games that I currently play so it made a lot of sense for me.

Thanks for that. This is simply a snapshot of a single game, a city builder. Nothing taxing for GPU or CPU................

you can see the E core boost that I have talked about.

It is proving to be a most capable setup and has pleasantly surprised me at its overall performance.

I am using a Z890 Gigabyte Elite Aorus board, which are you using..?

 

[t2na]

Thanks for that. This is simply a snapshot of a single game, a city builder. Nothing taxing for GPU or CPU................

you can see the E core boost that I have talked about.

It is proving to be a most capable setup and has pleasantly surprised me at its overall performance.

I am using a Z890 Gigabyte Elite Aorus board, which are you using..?

Ah yeah! I'm sure someone a lot cleverer than I am will be able to explain it - I can run a few more checks throughout today to see if I can trigger the same thing.

Yeah, definitely agree. I think now Intel have adjusted it's pricing it really is a great choice.

I'm using an Asrock Z890 Taichi Lite board, has been fantastic so far with a really easy to use BIOS. First time using an Asrock board as well (typically have stuck with Gigabyte or Asus for all of my previous builds).

 

[Vimes]

Ah yeah! I'm sure someone a lot cleverer than I am will be able to explain it - I can run a few more checks throughout today to see if I can trigger the same thing.

Yeah, definitely agree. I think now Intel have adjusted it's pricing it really is a great choice.

I'm using an Asrock Z890 Taichi Lite board, has been fantastic so far with a really easy to use BIOS. First time using an Asrock board as well (typically have stuck with Gigabyte or Asus for all of my previous builds).

I can't remember using an Asrock board. I have never got on with an AMI based Asus BIOS.
Overall I am more familiar with a Gigabyte Award type.
The board is good enough and the price for the board and CPU was great value.

 

[t2na]

I can't remember using an Asrock board. I have never got on with an AMI based Asus BIOS.
Overall I am more familiar with a Gigabyte Award type.
The board is good enough and the price for the board and CPU was great value.

Yeah, that was largely one of the reasons I went with this board too. Was at a good price and had a cashback deal on it too - so was really happy overall.

I hadn't had a Z series board since a z390i back with my 8700k so it's been really fun to play around with everything again.

 

[t2na]

I have a query related to using the iGPU on the 265k and whether it'll impact overall CPU performance - but hoping someone can offer advice or at least guide me in the right direction.

If I'm running one monitor off my dGPU and another off my iGPU is there any impact to the CPU performance when I'm playing games? The monitor using the iGPU is only ever used for web browsing, Discord, working etc. and from my rudimentary tests I can't see a real performance drop - but wasn't totally sure if I was missing something.

 

[Vimes]

The only consideration that I have read about is potentially one of heat when using the iGPU, adding to the heat of the CPU. Even though I have two monitors I am using my 4080 to output to both of them.
I had not considered using the iGPU, as the secondary monitor is only ever used for light tasks.

Perhaps someone else can comment who has had greater experience.

 

[Tetras]

I have a query related to using the iGPU on the 265k and whether it'll impact overall CPU performance - but hoping someone can offer advice or at least guide me in the right direction.

If I'm running one monitor off my dGPU and another off my iGPU is there any impact to the CPU performance when I'm playing games? The monitor using the iGPU is only ever used for web browsing, Discord, working etc. and from my rudimentary tests I can't see a real performance drop - but wasn't totally sure if I was missing something.

Shouldn't be an issue.

 

[Vimes]

It seems that ArL uses a different core order for their CPU's....

No longer are the P cores grouped together, followed by the E cores.
Hence the problems that HWinfo had, being designated wrongly.

 

[hlennie]

I have a query related to using the iGPU on the 265k and whether it'll impact overall CPU performance - but hoping someone can offer advice or at least guide me in the right direction.

If I'm running one monitor off my dGPU and another off my iGPU is there any impact to the CPU performance when I'm playing games? The monitor using the iGPU is only ever used for web browsing, Discord, working etc. and from my rudimentary tests I can't see a real performance drop - but wasn't totally sure if I was missing something.

I run another small monitor off the Igpu webpage or hexus display page it runs at 10% so super light has no effect while im gaming on the 5090. Due to the tile desighn Im thinking it does not hog resources like on the previous intel chips

 

[t2na]

The only consideration that I have read about is potentially one of heat when using the iGPU, adding to the heat of the CPU. Even though I have two monitors I am using my 4080 to output to both of them.
I had not considered using the iGPU, as the secondary monitor is only ever used for light tasks.

Shouldn't be an issue.

I run another small monitor off the Igpu webpage or hexus display page it runs at 10% so super light has no effect while im gaming on the 5090. Due to the tile desighn Im thinking it does not hog resources like on the previous intel chips

Cheers all, much like everyone mentioned it really didn't make any difference at all.

I played The Finals for around 2 hours last night - one hour with the second monitor plugged into the iGPU and one hour with only one monitor plugged into the dGPU - with logging in HWInfo. The difference was within the margin of error (1-2% at most, fluctuating either way) with no difference in temps either.

 

[Vimes]

Cheers all, much like everyone mentioned it really didn't make any difference at all.

I played The Finals for around 2 hours last night - one hour with the second monitor plugged into the iGPU and one hour with only one monitor plugged into the dGPU - with logging in HWInfo. The difference was within the margin of error (1-2% at most, fluctuating either way) with no difference in temps either.

Good result for you, well done.

Did you make any changes to try and undervolt the CPU.?

Did you also noted your max power draw when running R23.?
I had thought the 265K was limited to a mac of 250w.? But I see a max of 201w when running that.
Thermally it is not restricted, but it did tell me, HWiNFO, that the P and E cores are power limited.