The problem will always be that Intel has to use 5-10x the power just to match the efficiency of Apple Silicon. No matter what node they're using. It's the design philosophy that is the issue.
Thread director and efficiency cores are basically the same as why Apple silicon is efficient. When M1 came all the cores in Intel CPUs were high performance and the same. Apple as the advantage of tighter control of power management and the kernel that addresses their hardware than Intel.
If you just mean x86 decode that is a small overhead in fraction of a die at this point. That was a huge overhead for a 486 but there are orders of magnitude more transistors now.
I don't think Intel will win on power vs Apple because of the software aspect but the gap should close to maybe 10% or something.
Apple? Thats an ARM processor. Not the same thing at all. The problem with Intel lately is that they use 1.5-2x the power just to match the efficiency of AMD Silicon
I'm very curious about Meteor Lake right now. If it delivers 50% better efficiency than raptor lake, it should bring the efficiency fight right back to AMD in slim laptops. Just doesn't scale up to desktop class yet, but laptops sell way more honestly.
And then only spending one gen on Intel 4 before already moving onto 20a. Next few years will be very interesting for Intel.
Any chance of asking Intel the question as to what node Lunar Lake is on. Its weird they mentioned 20A for Arrow Lake and 18A for Panther Lake but nothing about Lunar Lake. Its possible its on TSMC N3E which would be 1st for Intel has they have never used CPU manufactured outside(unless you can take N6 SOC tile e-cores ). Other question is whether its monolithic or tiles/chiplet based?
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
8 Comments
Back to Article
shabby - Wednesday, September 20, 2023 - link
Intel has ambition 😂lemurbutton - Thursday, September 21, 2023 - link
The problem will always be that Intel has to use 5-10x the power just to match the efficiency of Apple Silicon. No matter what node they're using. It's the design philosophy that is the issue.drwho9437 - Friday, September 22, 2023 - link
Well that isn't true at all...Thread director and efficiency cores are basically the same as why Apple silicon is efficient. When M1 came all the cores in Intel CPUs were high performance and the same. Apple as the advantage of tighter control of power management and the kernel that addresses their hardware than Intel.
If you just mean x86 decode that is a small overhead in fraction of a die at this point. That was a huge overhead for a 486 but there are orders of magnitude more transistors now.
I don't think Intel will win on power vs Apple because of the software aspect but the gap should close to maybe 10% or something.
JayNor - Saturday, September 23, 2023 - link
Intel could gain a power advantage in Intel 20A vs TSM, since they are introducing both PowerVIA and GAA. TSM is behind on these.goatfajitas - Friday, September 22, 2023 - link
Apple? Thats an ARM processor. Not the same thing at all. The problem with Intel lately is that they use 1.5-2x the power just to match the efficiency of AMD Silicontipoo - Friday, September 22, 2023 - link
I'm very curious about Meteor Lake right now. If it delivers 50% better efficiency than raptor lake, it should bring the efficiency fight right back to AMD in slim laptops. Just doesn't scale up to desktop class yet, but laptops sell way more honestly.And then only spending one gen on Intel 4 before already moving onto 20a. Next few years will be very interesting for Intel.
wanderer66 - Saturday, October 14, 2023 - link
Such predictive certainties always make for enjoyable reading. Please, do continue.trivik12 - Thursday, September 21, 2023 - link
Any chance of asking Intel the question as to what node Lunar Lake is on. Its weird they mentioned 20A for Arrow Lake and 18A for Panther Lake but nothing about Lunar Lake. Its possible its on TSMC N3E which would be 1st for Intel has they have never used CPU manufactured outside(unless you can take N6 SOC tile e-cores ). Other question is whether its monolithic or tiles/chiplet based?