News Intel next-generation Lunar Lake CPUs launching in Q3, Arrow Lake in Q4 — mobile chips claimed to be 1.4x faster than Qualcomm's X Elite processors

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Better battery life, sure. There is no way that x86 can get anywhere near arm in terms of battery life - especially with windows being the way they are. Ain't happening.
x86 isn't forced to run windows, linux is a thing, heck you can run (free)dos on it if you want long battery life.
Also they usually measure this by having a clean install that won't do anything at all other than run the app they want to test, so their battery life for consuming media will be 100% only media with nothing else running ever.
My point being that windows won't change battery life since it won't be allowed to do anything during measuring.
 
  • Like
Reactions: rtoaht
Better battery life, sure. There is no way that x86 can get anywhere near arm in terms of battery life - especially with windows being the way they are. Ain't happening.
I feel performance and power go hand in hand. the problem for Intel is that they NEED to show that they have the absolute fastest performance somehow. That's why they created power hogs like the 13900/ 14900 that is pushed to the extreme, and burdened by 16! E-cores. With Meteor Lake, I think Intel back pedaled to focus on efficiency. Now the thing is whether they choice to focus on efficiency or forced to go with the efficiency narrative, I am not sure. But in my experience, the battery life is an improvement, but at the expense of performance.
 
I feel performance and power go hand in hand. the problem for Intel is that they NEED to show that they have the absolute fastest performance somehow. That's why they created power hogs like the 13900/ 14900 that is pushed to the extreme, and burdened by 16! E-cores. With Meteor Lake, I think Intel back pedaled to focus on efficiency. Now the thing is whether they choice to focus on efficiency or forced to go with the efficiency narrative, I am not sure. But in my experience, the battery life is an improvement, but at the expense of performance.
It's not hard to increase your perceived efficiency. They can launch their new cpus with 150w power limits and boom, they'll get double the efficiency. Heck, they can re-release their 14900 with a 150w power limit and boom, marvelous efficiency.

The hard part is to increase actual efficiency, which is measured at iso power. I think neither zen 5 or arl will be a big upgrade over zen 4 and rpl, they are already incredibly efficient chips if you don't push a gazillion watts into them.
 
While the whole ARM's race thing means nothing to me, I am happy to see some real competition in this industry. Go get 'em Intel.... Go get 'em Qualcomm.... And whom ever else wants to play the bigger badder faster less power draw game. The race is on!
Cheers to all the racers willing to compete 🍻.
 
  • Like
Reactions: bit_user
AMD isn't even designing different chips for cloud and lap/desktop now with only x86
That's blatantly false, if you just look at their APUs. Yes, their chiplets are shared between high-end laptop, desktop, and servers, but the APUs are something they do primarily for laptops and only port to desktop as a secondary market.

and you are dreaming of them doing it for arm...
There seems to be some misunderstanding, because what I meant by that statement is that if AMD invested the R&D into building ARM cores, they could monetize them in both the laptop & cloud markets, which are both places where the importance of perf/W is paramount. My statement had nothing to do about whether or how the CPUs are customized for each market.

Going arm would mean that they would have to pay yet another party for the license
So does Qualcomm and Apple, but it doesn't stop them. If paying a license fee is what it takes to stay competitive in those markets, AMD will obviously do it.

while they barely have enough funds to make x86 (and gpus) alone.
Given AMD even started on the K12 ARM project, back when they were teetering on the brink of bankruptcy, they obviously can manage to make both x86 and ARM CPUs. Jim Keller still believes it was a mistake for them to mothball that project, which I doubt he'd say if he agreed that it was financially untenable.
 
Edit: This doesn't have anything to do with the potential performance, but "Sound Wave" is a terrible name for a CPU arch that has little if any extra to offer in terms of sound waves. It was even a lame transformer,
Disagree.

Soundwaveboxart.JPG


only the cassettes were any good.
Yeah, I liked Laserbeak and Ravage. Even their re-imagining in the Transformers live action movie was pretty good, IMO.

There are a ton of better names they could have chosen like Orange Julius, Ginko Leaf or Rex the Dog. Pretty much anything would have been better.
Sorry, that's not correct. Sound wave is easy to say, not terribly long, not trademarked (at least, if you're not explicitly referring to the transformer), and has a "cool" factor that a lot of names don't.
 
l would! What would be the point of that?

At least Intel gains some benefit by using their own fabs. For AMD, they would have almost no points of differentiation.
Are you aware of any work with modern Arm designs by AMD? I can't think of any so I assumed that AMD, like Intel, would leverage that for some of their other markets where designing a full custom core doesn't make financial sense. It doesn't seem unreasonable to get some first hand engineering work done with a canned core before diving in the deep end.

What market do you think AMD would be spending the money on designing a custom Arm core for?
 
Are you aware of any work with modern Arm designs by AMD?
No, but given that AMD has much more at stake in x86 than last time, I expect they would keep any ARM projects Top Secret.

Think of it from the perspective of a corporate buyer or cloud operator, especially one who still uses Intel. If you're considering a switch to AMD, how would you feel about adopting their x86 products when you learn that they're less than 100% committed to them? You don't want to switch from Intel to AMD, only to have to switch back in a few years, if it turns out that AMD takes their foot off the gas and really switches focus to ARM.

People like Terry would lean into this and stoke the FUD about AMD's commitment to x86, and now AMD is having to play damage control. The result would likely be AMD's x86 revenues faltering well in advance of the point when they have any ARM products on the market to pick up the slack. AMD really can't afford be perceived as having split focus or loyalty.

like Intel, would leverage that for some of their other markets
Link?

What market do you think AMD would be spending the money on designing a custom Arm core for?
Look at where ARM is blowing up, lately: Laptops and cloud! Those are the main bread baskets for AMD. Other than Xilinx, they're not super big in embedded/IoT.
 
No, but given that AMD has much more at stake in x86 than last time, I expect they would keep any ARM projects Top Secret.

Think of it from the perspective of a corporate buyer or cloud operator, especially one who still uses Intel. If you're considering a switch to AMD, how would you feel about adopting their x86 products when you learn that they're less than 100% committed to them? You don't want to switch from Intel to AMD, only to have to switch back in a few years, if it turns out that AMD takes their foot off the gas and really switches focus to ARM.
So you are saying that AMD should pull a sneaky on those people, string them along make them buy amd x86 and then have them be forced to switch to amd arm (or intel) ,so your position is that amd should outright lie to their customers.
People like Terry would lean into this and stoke the FUD about AMD's commitment to x86, and now AMD is having to play damage control. The result would likely be AMD's x86 revenues faltering well in advance of the point when they have any ARM products on the market to pick up the slack. AMD really can't afford be perceived as having split focus or loyalty.
Thankfully I don't even have to spread any FUD, you already did a great job of it.
Everything you say beyond the first sentence is pure FUD ,but then again this is what happened before zen released as well, so it's not that uncertain.
 
  • Like
Reactions: rtoaht
So you are saying that AMD should pull a sneaky on those people,
There's nothing unusual about companies being secretive about unreleased products. For quite a while, Apple was famous for this.

string them along make them buy amd x86
There's nothing intrinsically dishonest about it. They're freely making their own decision, based on launched products and announced roadmaps.

and then have them be forced to switch to amd arm (or intel)
I think the fear would be much worse than the reality, here. AMD really can't afford to turn its back on x86, at least until their ARM revenue overtakes it. You can make your own guess about that timeframe, but I'm sure it won't happen overnight. For all we know, it might never happen for ARM, but actually RISC-V (or something else).

so your position is that amd should outright lie to their customers.
Way to prove me right, BTW!
: )
 
The DPU Intel made for Google (E2000) uses N1 cores, and AMD bought Pensando around that time who makes similar products, but I don't think they've put out a new one since acquisition.
Look at where ARM is blowing up, lately: Laptops and cloud! Those are the main bread baskets for AMD. Other than Xilinx, they're not super big in embedded/IoT.
I don't really see AMD spending that kind of money for just laptops, especially if any of the leaks regarding laptop CPU design are accurate. It's certainly not outside of the realm of possibilities given that Microsoft seems to be taking Arm seriously finally.

Cloud it's mostly about being able to customize for the workload which is why all the big cloud providers are making their own chips. I don't think that's ever going to go back to a third party provider at scale. I believe the sole provider of high end chips in that market is Ampere and everyone else either went under or left the market. I'm not sure where the room would be for such a chip from AMD unless they were going to try to use it to cut into the 8S market Intel has a relative lock on.
 
The DPU Intel made for Google (E2000) uses N1 cores, and AMD bought Pensando around that time who makes similar products, but I don't think they've put out a new one since acquisition.
...
Cloud it's mostly about being able to customize for the workload which is why all the big cloud providers are making their own chips. I don't think that's ever going to go back to a third party provider at scale.
Oops. You just proved my point. Intel made a customized chip for Google with ARM-designed cores, so if AMD had competitive ARMv9-A cores of their own, why do you think they wouldn't offer to integrate them into customized silicon for the cloud providers? Whether those guys are using ARM-designed or AMD-designed cores, it's still someone else's design they're licensing and it's still AArch64 ISA.

I believe the sole provider of high end chips in that market is Ampere and everyone else either went under or left the market.
Marvel is reportedly integrating their cores in custom silicon. Anyway, competitive CPU design is an expensive business and it's one that AMD has proven they can sustain. Plus, they already have their own platforms. So, I really don't see the logic that they couldn't play in the ARM market as well as they do in x86.
 
Oops. You just proved my point. Intel made a customized chip for Google with ARM-designed cores, so if AMD had competitive ARMv9-A cores of their own, why do you think they wouldn't offer to integrate them into customized silicon for the cloud providers? Whether those guys are using ARM-designed or AMD-designed cores, it's still someone else's design they're licensing and it's still AArch64 ISA.
There's a significant cost difference between custom designing an entire chip and using one that is already done...
Marvel is reportedly integrating their cores in custom silicon. Anyway, competitive CPU design is an expensive business and it's one that AMD has proven they can sustain. Plus, they already have their own platforms. So, I really don't see the logic that they couldn't play in the ARM market as well as they do in x86.
While AMD isn't lacking for money why do you downplay the cost of custom designing an entire chip? They're still a publicly traded company so if their R&D costs spiked it would be really obvious, and if they didn't that would mean taking away from the core business. Neither one of which is something that would go unnoticed.
 
There's a significant cost difference between custom designing an entire chip and using one that is already done...
If you mean "custom core", then yes. However, the K12 allegedly wasn't much more than an ARM frontend bolted on Zen1. As I've already said, if AMD could even embark on such a project, while in such a precarious situation, then it stands to reason they could easily do it now.

While AMD isn't lacking for money why do you downplay the cost of custom designing an entire chip?
I don't, but why do you downplay the existential risk ARM poses to AMD's long term future? The ARM market is growing. They need to go where their customers and partners are headed.

They're still a publicly traded company so if their R&D costs spiked it would be really obvious,
Disagree. Also, their R&D costs have been ramping up. You wouldn't realistically see such a clear signal in that.

We can disagree on this. Time will tell. We've both been wrong, before.
 
  • Like
Reactions: thestryker
Better battery life, sure. There is no way that x86 can get anywhere near arm in terms of battery life - especially with windows being the way they are. Ain't happening.
Wait…are you under the impression that the ARM ISA has some sort of innate efficiency benefits? ARM chips are more efficient simply because of architectural choices, not the ISA.