Question Anyone currently using an Intel Core Ultra 7 265K?

[coercitiv]

I don't personally want to fool around with either but if someone gave me say a system for free, I'd make it work.

A product that can lead to material loss and/or data corruption is a net loss even when offered for free. Don't confuse the hobby of making things work with the science of making things that work.

 

[lakedude]

A product that can lead to material loss and/or data corruption is a net loss even when offered for free.

Every phone and laptop with a lithium battery can lead to material loss and no storage device I know of is 100% reliable.

I mean I get your point. Componants should not catch fire or trash your data. The products in question are "bad" for an average person.

My point is that with enough skill and a big enough discount these bad products could be "good", especially the RL chips which are not dangerous to life or limb.

I'd stick one in a BOINC crunching machine (with no personal data on it) and run it 24/7 until it died and then RMA it!

 

[igor_kavinski]

My grandsons Intel machine boots in 5.

Specs?

DDR5 generation I don't think any mobo, AMD or Intel, cold boots in 5 seconds. DDR5 needs to be trained. The higher the speed, the longer it will take to train. Memory context restore may cut the time down to less than 10 seconds for the training but once that's over, the mobo has to go do the rest of its POST routines. So I'm skeptical that any current DDR5 mobo will do a 5 second cold boot.

 

[igor_kavinski]

Bartlett Lake is supposedly releasing in H1 2025. 12 P-cores and no lame E-cores. Hopefully by now Intel has learnt its lesson and finally releases a stable Intel 7 CPU? I'll most likely jump for a drop-in upgrade to my 12700K, if the price isn't outlandish.

 

[Thunder 57]

Yes, agreed, however...

If someone is handy they may be able to modify a product or modify the way they use a product.

If a PS has bad caps (or other parts), those can be replaced.

RL can be run at conservative clocks and given a low enough price (free?) might represent a good value.

I don't personally want to fool around with either but if someone gave me say a system for free, I'd make it work.

There are bad products. Even if you get them for free. Not going to derail this thread so I will not discuss further.

 

[DrMrLordX]

Bartlett Lake is supposedly releasing in H1 2025. 12 P-cores and no lame E-cores. Hopefully by now Intel has learnt its lesson and finally releases a stable Intel 7 CPU? I'll most likely jump for a drop-in upgrade to my 12700K, if the price isn't outlandish.

Is it still an LGA1700 CPU?

 

[igor_kavinski]

Is it still an LGA1700 CPU?

Yes, according to this:

https://twitter.com/x/status/1786318690528432348

 

[DAPUNISHER]

163 is the average frame rate and $2.42 is the CPU cost divided by the average frame rate. The problem with ARL in terms of gaming isn't that it performs poorly across the board, but rather it performs competitively in some games but not competitively in others. So, taking the average smooths over some really rough performances where ARL falls on its face. Whether those cases matter to you (and if a future game you may want to play performs poorly) will depend on the individual person, but you should definitely look deeper than the average and this is especially true for ARL.

Exactly. I am a 💩 head for using that cost per frame chart. I don't care much for them, geomeans, or 25-30 second test runs. But it was a quick and dirty way to convey that arrow is a bad bang for buck pick for gaming builds. In the end, there are better options for the same or less money, without the failings, is what I consider the salient point.

I do agree that ARL isn't a BD or P4 type product where it just gets blown out,

Correction: it did get blown out in gaming every bit as bad as Bulldozer. Someone provided the comparisons in the Zen 5 speculation thread a month or so back. There are games where it gets ROFLstomped by 50-60-even 75%

 

[DAPUNISHER]

Arrow to the knee indeed - https://wccftech.com/intel-newest-0...dly-fails-to-resolve-arrow-lake-s-cpu-issues/

 

[Hulk]

ARL performance was not "expected" as per Intel's internal testing. I'm really tired of this nonsense.

 

[igor_kavinski]

Arrow to the knee indeed - https://wccftech.com/intel-newest-0...dly-fails-to-resolve-arrow-lake-s-cpu-issues/

Use the Force, Intel!

 

[DAPUNISHER]

Use the Force, Intel!

I don't know what that is supposed to mean. Are you suggesting a Jedi mind trick -

*Hallock waves hand* "This performance will be fine"

Me -"No, it won't"

*Frustrated Hallock tries again waving hand slower* "THIS performance Will be fine"

Me - "No it won't. Mind tricks don't work on me, only money"

That they have yet another briefing about the performances issues scheduled is just sad trombone noises at this point.

 

[Hulk]

Time for Intel to finally be truthful. Here's an example of a spin that would be somewhat credible.

"With ARL we focused on areas of performance where the user is actually waiting on the CPU. These applications include rendering, media encoding, machine learning, and software development to name a few areas where ARL excels. We also made significant improvement when it comes to efficiency compared to our previous generation. As far as gaming, outside of competitive gamers the vast majority of people aren't going to notice the difference between 175fps and 208 fps. At the end of the day ARL addresses the real world performance improvements in applications where they can be most noticed."

Just use the old playbook Intel. Play up your strenghts and play down your weaknesses. Hinting at a magic fix is only going to hurt you in the end. There are some areas where ARL is quite performant.

 

[igor_kavinski]

There are some areas where ARL is quite performant.

I blame their management trying to mimic AMD's chiplet approach. They were in no position to "experiment" and even then, they should've limited it to mobile chips. If ARL had been a monolithic architecture, it could've been an easy win with TSMC N3B. Why is it that average people on these forums were able to predict well in advance that there would be serious tile latencies involved in ARL and yet no one at Intel took that seriously? It's like they don't care about their bread and butter business. Maybe there are still stalwarts in their midst who remember that something as awful as Pentium 4 didn't destroy Intel so ARL wasn't gonna either. These narrow minded fools destroyed Intel's mindshare. The only remaining hope is Bartlett Lake (if it doesn't melt at 6 GHz) and a slim hope that they will miraculously fix any serious issues they overlooked in Pat's maniacal quest to ship before end of 2024 and surpass 6 GHz RPL gaming performance with just 5.7 GHz of Lion Cove sauce.

IF Bartlett Lake is released, I will become genuinely hopeful that we will see a P-core only monolithic Lion Cove refresh before end of 2025 or early 2026. That would be a sane strategy to ensure they don't rush Nova Lake and release it as its best possible version without major snags.

 

[mikeymikec]

With 7xxx you need to set in BIOS "Power Down Enable" to "Enabled" if MCR is also enabled. So if MCR and PDE are both enabled = no issues. If MCR is enabled but PDE is disabled = bluescreen.

With 9xxx, AFAIK, this bug is solved. So with MCR enabled and PDE disabled = no issues

Huh, I didn't realise it was processor related. I must try this on my 7800X3D because I've been running the memory at the system default speed because EXPO / DDR5-6000 caused the POST time to increase by about 30 seconds.

(I've had this post as a draft for a while) I've just enabled it on my setup and the good news is that it didn't insta-crash like it did when I only enabled MCR. POST times look pretty quick too, I haven't counted the seconds but it wasn't far off what I would call a normal POST time.

Thanks in advance!

 

[Hitman928]

Time for Intel to finally be truthful. Here's an example of a spin that would be somewhat credible.

"With ARL we focused on areas of performance where the user is actually waiting on the CPU. These applications include rendering, media encoding, machine learning, and software development to name a few areas where ARL excels. We also made significant improvement when it comes to efficiency compared to our previous generation. As far as gaming, outside of competitive gamers the vast majority of people aren't going to notice the difference between 175fps and 208 fps. At the end of the day ARL addresses the real world performance improvements in applications where they can be most noticed."

Just use the old playbook Intel. Play up your strenghts and play down your weaknesses. Hinting at a magic fix is only going to hurt you in the end. There are some areas where ARL is quite performant.

That doesn’t seem right, the tech media told me that Intel was truthful in their launch of ARL, that’s why they criticized AMD for Zen 5 but not Intel. You need to fix your thinking to be less consumer centric. /s

 

[Hulk]

I blame their management trying to mimic AMD's chiplet approach. They were in no position to "experiment" and even then, they should've limited it to mobile chips. If ARL had been a monolithic architecture, it could've been an easy win with TSMC N3B. Why is it that average people on these forums were able to predict well in advance that there would be serious tile latencies involved in ARL and yet no one at Intel took that seriously? It's like they don't care about their bread and butter business. Maybe there are still stalwarts in their midst who remember that something as awful as Pentium 4 didn't destroy Intel so ARL wasn't gonna either. These narrow minded fools destroyed Intel's mindshare. The only remaining hope is Bartlett Lake (if it doesn't melt at 6 GHz) and a slim hope that they will miraculously fix any serious issues they overlooked in Pat's maniacal quest to ship before end of 2024 and surpass 6 GHz RPL gaming performance with just 5.7 GHz of Lion Cove sauce.

IF Bartlett Lake is released, I will become genuinely hopeful that we will see a P-core only monolithic Lion Cove refresh before end of 2025 or early 2026. That would be a sane strategy to ensure they don't rush Nova Lake and release it as its best possible version without major snags.

Intel knows why Lion Cove isn't competitive in a variety of applications/workloads but they aren't telling us. I'm not sure it's latency. I think it's a more complicated mix of factors.

If we're doing armchair quarterbacking then Intel should have just "refreshed" Raptor Lake again and swapped out the Gracemont clusters for Skymont. ST would have remained as competetive as it is today and MT would have been much improved. Efficiency would have been better on the new node as well.

Intel is at a fork in the road in my opinion. The next major release is going to make or break them. ARL is "good enough" for Dell, HP, and all the rest but those manufacturers know what is going on. If Intel stumbles again I think many people (like myself) and the big box guys are going to start having second thoughts about Intel. AMD only needs to be able to handle the volume that might be coming their way to complete the role reversal of the Intel/AMD positions from 10 years ago.

 

[Hitman928]

Intel knows why Lion Cove isn't competitive in a variety of applications/workloads but they aren't telling us. I'm not sure it's latency. I think it's a more complicated mix of factors.

If we're doing armchair quarterbacking then Intel should have just "refreshed" Raptor Lake again and swapped out the Gracemont clusters for Skymont. ST would have remained as competetive as it is today and MT would have been much improved. Efficiency would have been better on the new node as well.

Intel is at a fork in the road in my opinion. The next major release is going to make or break them. ARL is "good enough" for Dell, HP, and all the rest but those manufacturers know what is going on. If Intel stumbles again I think many people (like myself) and the big box guys are going to start having second thoughts about Intel. AMD only needs to be able to handle the volume that might be coming their way to complete the role reversal of the Intel/AMD positions from 10 years ago.

I still have yet to see any evidence that Intel has made the 285k in any real volume. You can at least buy it now but it's only from 3rd party sellers or in boutique pre-builts, nothing from the big OEMs. Looking at Dell, they don't even offer lower end ARL, seems all the OEMs are all still shipping with RPL. Granted, desktops are small business for them now, but I would have thought at least Intel's bestie Dell would have bought into ARL.

 

[jpiniero]

but I would have thought at least Intel's bestie Dell would have bought into ARL.

You'll have to wait until the locked parts are available presumably. Given that Raptor is cheaper and faster in games, it almost doesn't make any sense to adopt Arrow really.

 

[Markfw]

You'll have to wait until the locked parts are available presumably. Given that Raptor is cheaper and faster in games, it almost doesn't make any sense to adopt Arrow really.

Since AMD is the champ for games, no Intel makes any sense for games. And ARL is not better than the 9950x for anything. The whole lineup is worthless IMO.

 

[jpiniero]

Since AMD is the champ for games, no Intel makes any sense for games. And ARL is not better than the 9950x for anything. The whole lineup is worthless IMO.

AMD probably doesn't have the volume to satisfy OEMs with Zen 4/5d. Raptor is competitive enough with regular Zen 5 in games, even if it does draw way more.

 

[Markfw]

AMD probably doesn't have the volume to satisfy OEMs with Zen 4/5d. Raptor is competitive enough with regular Zen 5 in games, even if it does draw way more.

But Raptor lase is the one that blows itself up ! Unless you set it below 5 ghz, then Zen wins.

 

[Hulk]

I still have yet to see any evidence that Intel has made the 285k in any real volume. You can at least buy it now but it's only from 3rd party sellers or in boutique pre-builts, nothing from the big OEMs. Looking at Dell, they don't even offer lower end ARL, seems all the OEMs are all still shipping with RPL. Granted, desktops are small business for them now, but I would have thought at least Intel's bestie Dell would have bought into ARL.

This is a good point regarding 285K volume. It has been OOS at MC literally since the day it was released. Why? TMSC yield issues? Package integration issues? It's a nice little mystery. If it was on an Intel process we'd know immediately yield/binning is the issue. But with TMSC we're used to good yields. But still no parts?

So many things "not quite right" with Intel these days. I'm worried for them. Only thing that has gone reasonable okay lately is Lunar Lake and Battlemage. But Lunar Lake isn't available in a lot of systems and the 258V part is as rare as hens teeth it seems. Battlemage has buyers lined up but no parts to put in their carts. What the heck is going on with Intel? It's almost like they're trying to fail.

Meanwhile AMD is just rolling along. Selling tons of Zen 5, 4, and 3. Zen 5 X3D is a big hit and they are seemingly selling good numbers with the big dog version on the way. Everything on track and A okay.

 

[Hulk]

But Raptor lase is the one that blows itself up ! Unless you set it below 5 ghz, then Zen wins.

While overall the 9950X is the superior CPU, there are a number of applications where the 285K is more performant. It's not a total wash out.

Intel Core Ultra 9 285K Review

Finally! Intel's new Arrow Lake architecture is launched. The new CPUs are full of design changes, like removal of Hyper-Threading, new Lion Cove P-Cores, an improved Thread Director and more. In our review we got surprising results that were both impressive and disappointing.

www.techpowerup.com

 

[igor_kavinski]

While overall the 9950X is the superior CPU, there are a number of applications where the 285K is more performant. It's not a total wash out.

Page 7 - RapydMark CPU benchmark

Page 7 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

Someone with a proper CS background and deep knowledge about implementation of those algorithms can shed light on what could be causing Arrow Lake to get beaten in so many of those tests.

However, I think HT would've made the fight a little more fair.