News Intel hasn't sold a single Arrow Lake CPU at Germany's largest retailer — Core Ultra 200S sales stagnate after just one week

[Loadedaxe]

Ah the pitchfork crowd.

You guys do realize that AMD has had way more issues in the last 8 years with Ryzen yes? Way more!!

But I digress, its easier to jump on the "They Suck" wagon than actually do some research. You all act like Intel reads this vomit...grow up.

News flash they dont!

 

[Gaidax]

They need to discount their top end CPUs a good 20%, then they will be more competitive and likely that is what will happen.

 

[SyCoREAPER]

Ah the pitchfork crowd.

You guys do realize that AMD has had way more issues in the last 8 years with Ryzen yes? Way more!!

But I digress, its easier to jump on the "They Suck" wagon than actually do some research. You all act like Intel reads this vomit...grow up.

News flash they dont!

News flash, we don't all live in a communist mindset like you and are free to speak out minds. Don't like it, don't participate.

 

[Ferlucio]

They're just following the latest trend in tech: OuO- Overpromise, underdeliver, Overprice.

 

[bit_user]

here in Romania, I cannot find a 285K to buy. Either preorders with no delivery time, or out of stock everywhere.

I wonder if this suggests Intel's pivot to TSMC for Arrow Lake's compute tile (i.e. instead of Intel 20A) came fairly late and without enough lead for them to build significant inventory.

I do software development, lots of compiling, lots of small files processing, occasional video editing and the standard media consumption. And in all these areas, in all the benchmarks I've seen, 285k is king.

In these compilation benchmarks the 285K has a slight lead, but the difference is small enough that I think it's not decisive. On content creation, the R9 9950X still maintains an overall lead. If you look at the overall geomean, the 9950X remains the clear champion.

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

 

[bit_user]

Even at a 800mv/1800MHz volt curve my 3080 runs over 200w. My monitor also takes more power than the CPU.

Thanks for the info. Yeah, I got a VRR monitor and it definitely uses more power than my old LCD of the same size/resolution.

I am not concerned over the cost. For me it is more an acoustic issue and a vanity issue.

I'm concerned with heat. During the warm months, my air conditioner cools the entire home, not just the room where my computer is. So, that makes power consumption disproportionately more expensive for me, during that part of the year.

 

[bit_user]

Ah the pitchfork crowd.

You guys do realize that AMD has had way more issues in the last 8 years with Ryzen yes? Way more!!

But they were in the "rising from the ashes" phase, where as these are things Intel doesn't normally mess up. That's the difference. AMD is generally improving year-by-year, while Intel has been faltering.

You all act like Intel reads this vomit...grow up.

News flash they dont!

I think most of us don't. A handful posters on here seem to, but I haven't paid attention to whether they're in this thread.

 

[bit_user]

They need to discount their top end CPUs a good 20%, then they will be more competitive and likely that is what will happen.

Like I said before, Arrow Lake is almost completely made in TSMC's fabs. That means Intel likely doesn't have as much room to cut prices as they usually do.

I expect TSMC's wafer pricing on N3B will decrease, over time. So, price decreases might be possible, but they might not be as much as you're used to, and don't look for them soon.

 

[helper800]

I'm concerned with heat. During the warm months, my air conditioner cools the entire home, not just the room where my computer is. So, that makes power consumption disproportionately more expensive for me, during that part of the year.

I am in the same boat. When its 90-100+ degrees how can I reasonably dump another 400-600 watts of heat into the room with my desk without turning the AC on or having a heatstroke. That plus 0.43 cents per kWH is rough.

 

[EvenTide2]

I doubt the veracity of these sale numbers.

Intel might be moving fewer Ultra 2 chips than expected but asserting that they have moved none at all does not make sense to me. Checking mindfactory myself I can see that they have "at least 5 units sold" text under a few ultra 2 chips. At the very least tech reviewers and other hardware enthusiasts would pick them up and there might just be some users that have niche cases for the Ultra 2 series that make them choose it over AMD2.

I would strongly suggest to reach out to mindfactory for comment or verification.

 

[Loadedaxe]

News flash, we don't all live in a communist mindset like you and are free to speak out minds. Don't like it, don't participate.

You completely missed the point gg

 

[YSCCC]

Meanwhile, here in Romania, I cannot find a 285K to buy. Either preorders with no delivery time, or out of stock everywhere.
And regarding all the reviews and opinions about the 285K, it may not be the best for gaming, I get that, but for all the rest of us, I can't wait to update my computer. And no, I don't play on my computer. But I do software development, lots of compiling, lots of small files processing, occasional video editing and the standard media consumption. And in all these areas, in all the benchmarks I've seen, 285k is king.
And luckily for Intel, about 70% of the PC (laptops excluded) industry is not gaming, but rather office work. So yeah. Intel is for me. AMD for the gamers. No problem. I'm cool with that.

On the production side the 285k isn't decisively faster than the 9950X either, and the X3D is yet to be seen how will it fare.

AndTBH, back in the day "no one gets fired buying Intel" is mostly because they although occasionally dethroned in performance, but except recent years Intel is basically the synom of "rock stable", which, in office or even rendering work, means a lot, imagine say AMD renders 20% faster than Intel, but in 1 out of 5 times it could get unstable and fails the render, it will offset the performance lead, worse still if those are massive rendering like taking a few hours to render, and once you try put it overnight to finish it, click t he button and leave office, just to find out the next day you get a welcoming BSOD.

That was why intel was always dominating the markett, even in days of Pentium 4 vs Athlon 64.

But this time around Intel have first the degradation of RPL, which is causing some developers a lot of trouble and thus the initial outrage on youtube channels earlier this year, then comes the ARL having scheduling and BSOD issues, while it is expected (and I truely believe) those are release teething issues and fix will arrive eventually, I am keeping some grain of salt on the timeframe they can fix those (e,g, core parking issue for Ryzen 7900 and 7950 X3D), which in the meantime it isn't a good choice for serious work to be done until those initial issues are fixed

 

[thestryker]

If you look at the overall geomean, the 9950X remains the clear champion.

You should always add in the asterisk of AVX512 when citing Phoronix, because without AVX512 those numbers absolutely do not look that way. I think it was 3dcenter who did a review analysis of a dozen or so reviews and Phoronix is the only place with such a broad gap. This is not to say the 9950X isn't still faster just that the overall geomean isn't accurate for the vast majority of people's usage.

I expect TSMC's wafer pricing on N3B will decrease, over time. So, price decreases might be possible, but they might not be as much as you're used to, and don't look for them soon.

All word has been that they're raising prices on N3 which makes me wonder if the rumors about Apple not buying up all of their N2 capacity are true. Of course it could also be as simple as a company profiting from market forces (aka AI). Either way prices shouldn't be dropping for Intel to take advantage on CPUs.

 

[SyCoREAPER]

To further add to this, they do run warm still, like previous gens.

DerBau3r showed a significant improvement in thermals deliding but will require more tools, like a heater he will be selling. There is a high risk of damaging it if not done right. He also stated TG will not be making a block for it despite all that.

So him being a business man with a recognizable brand doesn't see it worth making, it implies he doesn't forsee market share of these new CPUs.

 

[Amdlova]

@bit_user I have here now a 4060 8gb.
Still using the uhd770 as main display with the 4060 as slave card... feels like a 3dfx voodoo 2 time.
Because now I have one the system an u.3 enterprise ssd. My system has a little more idle power compsumation. But stay at 45w idling on windows with the the normal power profile. The power saver cuts 3w max.
playing the Icarus with a friend at quad hd (dlss) the machine uses 195w at the wall (something like 168w)
Still at 1.6w per Fps, maybe some day got the 1/1 modern games

 

[TheHerald]

You should always add in the asterisk of AVX512 when citing Phoronix, because without AVX512 those numbers absolutely do not look that way. I think it was 3dcenter who did a review analysis of a dozen or so reviews and Phoronix is the only place with such a broad gap. This is not to say the 9950X isn't still faster just that the overall geomean isn't accurate for the vast majority of people's usage.

If I remember correctly the 3d center numbers they are within 3-5% in both gaming and productivity, but 285k draws less power on average (although that's not tracked by 3d center, we've seen it in reviews).

It's down to what your workloads are rather than one being flat out better than the other, but it's an Intel cpu so people have to overly negative about it.

 

[DataMeister]

Probably what Intel should have done is kick everything down a notch in the product lineup and release the Ultra 9 in January or February. In other words once they saw the test numbers the Ultra 9 285K should have been labeled Ultra 7 265K and on down the line with the rest. Then go back and figure out how to build a significant upgrade to the Ultra 9.

When they release their non-K chips I guess we'll see how sales really perform. I'm looking forward to seeing how efficient the Core Ultra 5 235 actually is compared to the 14500.

 

[YSCCC]

Probably what Intel should have done is kick everything down a notch in the product lineup and release the Ultra 9 in January or February. In other words once they saw the test numbers the Ultra 9 285K should have been labeled Ultra 7 265K and on down the line with the rest. Then go back and figure out how to build a significant upgrade to the Ultra 9.

from MLID in an pre release episode I recall that he said one intel employee in the engineering side wanted to scramble a U9 with more E cores and maybe clock higher to make it something like a power draw monster but keep the halo product over the 9950X, much like how the 13/14900k was to 7950X, inarguably better in production apps and on par/slightly better in games, while let the rest of the line be efficiency showoff, but somehow the top management banned the idea (of higher power draw I recall, but I could remember incorrectly). Maybe that is fake news all along but maybe also true that Intel top management have had enough of the RPL disaster they just prefer giving up the Halo product class to AMD and not hit the degradation wall again

 

[abufrejoval]

Not sure anyone will find this question at the end of this thread, but I couldn't find the info anywhere yet...

I wonder if CUDIMM support in Arrow Lake relieves the dual DIMM per channel constraints which have been hampering DDR5?

If CUDIMMs allowed Arror Lake CPUs to run with 128, 192 or even more RAM without compromising on bandwidth (or stability), that could at least be a niche for Arrow Lake productivity usage...

Of course CUDIMMs with real ECC and meaningful timings at an affordable price would become the next big challenge...

 

[bit_user]

I wonder if CUDIMM support in Arrow Lake relieves the dual DIMM per channel constraints which have been hampering DDR5?

If CUDIMMs allowed Arror Lake CPUs to run with 128, 192 or even more RAM without compromising on bandwidth (or stability),

I did a quick search and didn't find any info from Intel. So, then I thought I'd check the documentation of the 4-slot motherboards listing the fastest DRAM speeds. I checked two brands: Gigabyte and ASUS.

The Gigabyte boards listing the fastest speeds were the Z890 AORUS MASTER AI TOP and Z890 AORUS XTREME AI TOP, both claiming up to up to DDR5-9500 (OC). On the Gigabyte Memory QVL, the fastest DIMMs of either 32 GB or 48 GB sizes were only DDR5-7200 (48 GB), and presumably not CU-DIMMs. The Gigabyte manual and quick install guide didn't even mention CU-DIMMs, nor anything about how DIMM combinations affected speeds. I guess, because it's all OC, that they aren't going to make any statements that could be inferred as a guarantee that it would work.

In the case of ASUS, the ROG MAXIMUS Z890 EXTREME and ROG MAXIMUS Z890 HERO were the two 4-slot boards that supported the highest memory speed of DDR5-9200. When I checked their Memory QVL, they provided an option to show just CU-DIMMs. None of them were dual-rank, with the highest capacity being just 24 GB. As with Gigabyte, their documentation provided no information about achievable speeds with different DIMM configurations.

So, my guess is that your best bet is to find a reviewer who's actually tried it.

Of course CUDIMMs with real ECC and meaningful timings at an affordable price would become the next big challenge...

Yes, the fastest ECC UDIMMs I've yet seen are only DDR5-5600 (though I haven't checked in a couple months, since that's when I bought mine). I expect we'll next see DDR5-6400, which is the next step on the JEDEC speed ladder. I forget whether that requires a CKD chip, but I'm pretty sure it does.

 

[thestryker]

I wonder if CUDIMM support in Arrow Lake relieves the dual DIMM per channel constraints which have been hampering DDR5?

So far I haven't seen any tests, but we do have the official specifications to go by. It's not as low as ADL/RPL, but it's still showing pretty big drops:

MSI shows 2DPC ratings on their boards and show "4800+" for both 1R and 2R (on boards which are >9000 1R/>7000 2R on 1DPC). If I had to guess based on both of these I'd say it's not going to be much better.

Of course CUDIMMs with real ECC and meaningful timings at an affordable price would become the next big challenge...

Since ECC memory sticks to JEDEC timings these are probably going to be a ways out since it won't be until 6400+ and given the way enterprise pricing tends to work even longer to be affordable.