Intel Alder Lake Chips for Desktops Faster Than M1 Max in Benchmarks, But Use Much More Power

[Zdigital2015]

It's impossible to judge how well Intel's 12th Gen H-Series is going to do until they are out and in various form factors. Given Intel's innovative naming structure, we're getting a 12700H, 12800H, 12850H and a 12980HK. The first two will make up the majority of gaming and high performance laptops with a few takers for the high end models. I suspect a Sager clocking in at 2" think will make the 12980HK hum along pretty well and for those who want that sort of HOSS laptop, I say good luck. Unfortunately, 2 hours and reduced performance while on battery does not a beast laptop make.

Ironically, had Apple stayed with Intel, they would have the E-Cores figured out long before MS and their developers. I also believe that Linux will have it figured out before MS does, provided that Intel didn't botch the whole implementation altogether at the hardware level. It's quite satisfying watching Intel trying to learn how to walk and chew gum at the same time and still trip on their shoelaces.

 

[grkm3]

Depends. There are significant differences in the PLs, and if you run a 700H at PL4, it probably outperforms an 800H at PL1.

The key question will be what any of these CPUs will do in something that resembles a laptop.

You didn't understand my point. The 700h was 1300 single and a way higher multi core score than the 800h multi core with a 1600 single score.

These are es samples and not running final specs. How can the 700h beat the 800h in multi score?

 

[chucker23n1]

You didn't understand my point. The 700h was 1300 single and a way higher multi core score than the 800h multi core with a 1600 single score.

These are es samples and not running final specs. How can the 700h beat the 800h in multi score?

Read my post again. The 700H can potentially beat the 800H by being set to a higher PL (automatically, because there’s more thermal headroom) or cTDP (manually by the vendor).

 

[grkm3]

Read my post again. The 700H can potentially beat the 800H by being set to a higher PL (automatically, because there’s more thermal headroom) or cTDP (manually by the vendor).

The 800 had a higher single score man. Do you think laptops allow pll changes and shutting cores off? This is clearly an es sample being shown on sites as click bait to say its under preforming. This is not an unlocked k version that allows you to nake these changes and you can clearly see in the geekbench run saying it's in balanced performance mode..... not high performance!

 

[chucker23n1]

This is clearly an es sample being shown

Yes, could be. We’ll see.

and you can clearly see in the geekbench run saying it's in balanced performance mode..... not high performance!

Geekbench is notoriously unreliable at such details for unreleased CPUs.

 

[throAU]

So, can an Alder Lake-P CPU beat an M1 Max on performance? My guess is the CPU can, the GPU cannot, and either way, your laptop will run significantly hotter. Something like a Dell XPS or Microsoft Surface will probably be well below M1 Max performance numbers.

That's somewhat fair but I'd go further:

The CPU may in more power, the GPU can't and the chip doesn't even have dedicated coprocessors for video, AI/ML, etc. that can offload work from the CPU/GPU.

 

[chucker23n1]

That's somewhat fair but I'd go further:

The CPU may in more power, the GPU can't and the chip doesn't even have dedicated coprocessors for video, AI/ML, etc. that can offload work from the CPU/GPU.

This is true, but those are for specialized workloads, and the app needs to take advantage of them. Lots of people who buy an MBP do CPU-bound workloads, such as compiling.

 

[throAU]

This is true, but those are for specialized workloads, and the app needs to take advantage of them. Lots of people who buy an MBP do CPU-bound workloads, such as compiling.

Sure, but the point is measuring processing performance and claiming victory on CPU only tests ignores the fact that for a lot of workloads that aren't compiling, the CPUs are basically idle while doing that and you can continue to use the machine for other things (e.g., like compiling) while its say, rendering out a bunch of video. With basically full performance in both tasks (assuming you have enough RAM).

 

[hot-gril]

No gamer cares?? That's why gamers are putting freezers into their PC builds? LOL.

Haha. They pretend to care, but it doesn't actually matter unless they're severely overclocking. And at the end of the day, they're not doing anything useful on their machines.

 

[hot-gril]

I was talking about the end user. All this talk about Apple’s chips being so power efficient and cool in comparison. That’s great but no gamer cares.

Gamers don't even care about CPU performance that much since most of it falls on the GPU. But when they do, they care about single-core performance or 4-core performance at most, which the M1 does very well in.

Also, low TDP on these earlier chips suggests Apple has a lot of room to scale the M1 into something more powerful.

 

[hot-gril]

That's somewhat fair but I'd go further:

The CPU may in more power, the GPU can't and the chip doesn't even have dedicated coprocessors for video, AI/ML, etc. that can offload work from the CPU/GPU.

The Intel CPUs have coprocessors for efficient H.264/5 encoding/decoding and other things like that, which for regular people is important for Zoom and playing videos. Not ML, though.