New Intel 15th Gen Arrow Lake going to be any good?

[THE LORD]

In terms of power consumption, heat output and performance?

Or is it going to be trumped by AMD's 9800X3D coming next month?

Intel doesn't have a good track record the last few years.

[r.linder]

Depends on how well they do the architecture, because they don't have HyperThreading to pick up any slack in regards to multi-core performance anymore.

Power consumption for the K SKUs is still high, 250W being the default max boost power. Those CPUs will still run hot, that isn't going to change overnight.

I would expect the 9800X3D to continue AMD's dominion over the gaming space.

[_I_]

wait for reviews and performance tests

the next gen naming is another trainwreck
intel should just drop the 3/6/7 series naming, and just use numbers those are redundant anyways

its no longer going to be i series, but now ultra, and 3 digits, which was basically what it was before without extra 00 'zeros'

its only the lats 2 gen that were problematic to some users and will fail sooner than other cpus

but also wait for reliable reviews of the 9800x3d
the other 7800x3d was kind of a flop compared to 7700x

[Illusion of Progress]

Not to nitpick, but... it's not the 15th generation. Intel has retired the "Core i" nomenclature and restarted a new series. It's just Core (or Core Ultra).

Even if they were still doing the old naming, this would be the 16th generation instead of the 15th, as the predecessor exists but was skipped on the desktop (so it's like Broadwell/5th generation mostly was). That's why you'll be seeing 200 numbers instead of 100 on the new desktop CPUs, as the Meteor Lake 100 series was mobile only. So LGA 1851 might actually be another LGA 1150 and be a single generation only?

As for all the details... there's a lot of speculation and I imagine much of it will come to fruition, but as it's just speculation, wait for reviews.

Names are officially known. There's the Core Ultra 9 285K, Core Ultra 7 265K, and Core Ultra 5 245K. Later, the non-Ultras, like "Core 2xx" will likely come.

Hyper-threading is gone, so expect a multi-threaded performance regression. I expect p-core/e-core amounts per tier is otherwise staying the same but I'm not sure.

A single threaded uplift is speculated; apparently IPC is going up (sub-10%) but clock speeds are down so the actual uplift will be smaller than the IPC uplift (synthetics will probably peg it higher but real world results will differ as usual). Despite the single-threaded uplift, the improvements in gaming are speculated to be low. Intel themselves claim it won't even beat the 7800X3D (they said 5% slower, and the 9800X3D is speculated to be on the heels of the release), and the 285K may not even surpass the 14900K all the time. For productivity, they also compare it to the 7950X3D instead of the better (in productivity) 7950X (remember, X3D drops clock speed and cache doesn't help in most productivity stuff), which seems pretty telling to me that they know they don't have a real performance winner.

Power draw should be down, so it will be more efficient than Raptor Lake, but that's the lowest bar to clear. Still, improvements are improvements.

All in all, it seems speculated to be pretty muted, but... speculated is the key word. The Silver lining here is that Zen 5 was pretty muted too, thus the 9800X3D likely will be (but will still likely become the fastest in gaming), and X3D pricing is way up right now in general, so Intel has breathing room, especially if they price them well.

But wait for reviews which come late October, by which point you may as well want to consider waiting for the 9800X3D, since it's speculated to be announced/release as early as November. Unless you know price, performance, and intangibles, you're buying blind.

You might see some nice sales on older stuff too.

[A&A]

Intel has recommended customer prices
9 285K ≈$599.00
7 265K ≈$400.00
5 245K ≈$300.00

The F variants are about $10-$20 cheaper. Strange considering they are one of the chiplets.

[Electric Cupcake]

I'm more interested in seeing where RISC-V goes.

What shenanigans will the US government pull to try to hold China back? Will it get general purpose uses and eat into the old standard x86_64 market share and ARM market share?

[A&A]

Electric Cupcake eredeti hozzászólása:

I'm more interested in seeing where RISC-V goes.

What shenanigans will the US government pull to try to hold China back? Will it get general purpose uses and eat into the old standard x86_64 market share and ARM market share?

Made in China 2025?

idk but I'm sure the 13th fastest (was number 1) computer and the fastest Chinese computer runs on RISC V.

[Edu]

Illusion of Progress eredeti hozzászólása:

Not to nitpick, but... it's not the 15th generation. Intel has retired the "Core i" nomenclature and restarted a new series. It's just Core (or Core Ultra).

Even if they were still doing the old naming, this would be the 16th generation instead of the 15th, as the predecessor exists but was skipped on the desktop (so it's like Broadwell/5th generation mostly was). That's why you'll be seeing 200 numbers instead of 100 on the new desktop CPUs, as the Meteor Lake 100 series was mobile only. So LGA 1851 might actually be another LGA 1150 and be a single generation only?

As for all the details... there's a lot of speculation and I imagine much of it will come to fruition, but as it's just speculation, wait for reviews.

Names are officially known. There's the Core Ultra 9 285K, Core Ultra 7 265K, and Core Ultra 5 245K. Later, the non-Ultras, like "Core 2xx" will likely come.

Hyper-threading is gone, so expect a multi-threaded performance regression. I expect p-core/e-core amounts per tier is otherwise staying the same but I'm not sure.

A single threaded uplift is speculated; apparently IPC is going up (sub-10%) but clock speeds are down so the actual uplift will be smaller than the IPC uplift (synthetics will probably peg it higher but real world results will differ as usual). Despite the single-threaded uplift, the improvements in gaming are speculated to be low. Intel themselves claim it won't even beat the 7800X3D (they said 5% slower, and the 9800X3D is speculated to be on the heels of the release), and the 285K may not even surpass the 14900K all the time. For productivity, they also compare it to the 7950X3D instead of the better (in productivity) 7950X (remember, X3D drops clock speed and cache doesn't help in most productivity stuff), which seems pretty telling to me that they know they don't have a real performance winner.

Power draw should be down, so it will be more efficient than Raptor Lake, but that's the lowest bar to clear. Still, improvements are improvements.

All in all, it seems speculated to be pretty muted, but... speculated is the key word. The Silver lining here is that Zen 5 was pretty muted too, thus the 9800X3D likely will be (but will still likely become the fastest in gaming), and X3D pricing is way up right now in general, so Intel has breathing room, especially if they price them well.

But wait for reviews which come late October, by which point you may as well want to consider waiting for the 9800X3D, since it's speculated to be announced/release as early as November. Unless you know price, performance, and intangibles, you're buying blind.

You might see some nice sales on older stuff too.

source about 7800x3D announced by intel to be better than 15th genl?

[smokerob79]

wont matter to me...never buying intels garbage again.....let me know when they support a single motherboard platform for 7 year and release the best gaming chips ever made at the tail end as a loving send off VS getting told to buy a new motherboard every gen.....

[r.linder]

smokerob79 eredeti hozzászólása:

wont matter to me...never buying intels garbage again.....let me know when they support a single motherboard platform for 7 year and release the best gaming chips ever made at the tail end as a loving send off VS getting told to buy a new motherboard every gen.....

Intel has supported at least two generations for a long time, not just one. LGA1700 also supported 3, which is fine.

The alternative to having less generations supported means much greater work on the motherboard vendors and stretching out support which inevitably leads to issues with firmware. Partial firmware support for PCI-e 4.0 on older boards had to be cut from later BIOS revisions because they couldn't make room for it and some boards still don't properly work with or even support 5000 series due to issues.

I would rather have ~3 generations and good BIOS than 4+ generations and crap BIOS.

[Mr White]

The Big question is over the Integrated Graphics will Intel match AMD at least in low settings.

[A&A]

r.linder eredeti hozzászólása:

Intel has supported at least two generations for a long time, not just one. LGA1700 also supported 3, which is fine.

The problem with these three generations is that they run on the same architecture and you can't compare the platform to AM4, even to AM5.

[Çapgun]

I have looked at the first products. So as far as i can say new intel products much like amd x3d stuff. Just look to the 3 level cache size. Nothing new Just increasing and improving l3 caches imo. But better compatibility.

[Çapgun]

Intel developing new isa called avx10 so real new cpus will come for 2026

[r.linder]

A&A eredeti hozzászólása:

r.linder eredeti hozzászólása:

Intel has supported at least two generations for a long time, not just one. LGA1700 also supported 3, which is fine.

The problem with these three generations is that they run on the same architecture and you can't compare the platform to AM4, even to AM5.

It's no different, they require BIOS updates either way... There's Alder, Raptor, and Raptor Refresh, the differences are pretty much the same between each revision of Zen.

[Illusion of Progress]

Edu eredeti hozzászólása:

source about 7800x3D announced by intel to be better than 15th genl?

No 15th generation will ever exist to compare it to.

If you mean the generation after the 14th, that already exists, and the fastest one in it, the Core Ultra 185H, is already slower than the Core i9 14900H, let alone the 7800X3D.

If you're talking about the fastest upcoming desktop part, the 285K, then the statement comes from an Intel slide and statement by Robert Hallock.

https://hardwaretimes.com/intel-core-ultra-9-285k-5-slower-than-amd-ryzen-7-7800x3d-in-gaming-ryzen-7-9800x3d-launch-later-in-oct/

According to Robert Hallock, the VP and GM for Intel’s Client AI and Technical Marketing, we should see “about a 5% deficit” compared to the Ryzen 7 7800X3D, the current fastest gaming CPU. The 5% figure is a first-party estimate, so don’t be surprised if third-party reviews come up with a larger delta of up to 10% or more.

Here's the direct quote. The way he says "that part" makes it sound like he was questioned about the 7800X3D, hence the above may be an interpretation.

"We showed some data on the 7950X3D. Based on my understanding of the performance, that part is within a couple of percents so I think we will be about 5 percent back versus X3D which we feel really really good about considering that we have just the cache that’s built within the CPU and the great IPC of the product so you’ll see about a 5% deficit, I want to be clear about that."

But as I said myself, wait for formal reviews and benchmarks.

Çapgun eredeti hozzászólása:

Intel developing new isa called avx10 so real new cpus will come for 2026

This already exists as AVX512 to my understanding? Intel already supported it for like part of a generation, but then had to drop it for some technical reason related to e-cores I think?

This is also one of the major changes with Zen 5.

Stuff that uses it (modern emulation is one of the few that does) apparently gets a nice speedup from it.

r.linder eredeti hozzászólása:

Intel has supported at least two generations for a long time, not just one. LGA1700 also supported 3, which is fine.

I would rather have ~3 generations and good BIOS than 4+ generations and crap BIOS.

I'm not sure I can say anything bad about AM4 when it saw CPUs at the start that an overclocked Intel 4th generation/stock 6th generation would match, and ended up with CPUs outperforming the Intel 12th generation on average (albeit only in gaming), and sometimes the very fastest 14th generation in edge cases. That's one heck of an increase on one platform, and it displaced the legendary LGA 775 in that regard. Of course, that was lightning in a bottle and likely won't be repeated until a real breakthrough with CPU performance is found.

Intel can formally name things what they want, but a lot of its generations stretch the definition (not that there is a formal one), and the 14th generation was the worst example of that. It's Raptor Lake just like the 13th generation was.

In any case, it'll be interesting to see what happens with LGA 1851.