News Qualcomm announces Nuvia-powered PC chip - competitive with Apple M series

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

NTMBK

Lifer
Nov 14, 2011
10,208
4,940
136

Image_from_iOS__1_.jpg


I'm excited to see this pan out. If we can get CPUs as good as the M1 on PCs, I'm all for it. More competition is good!
 

Joe NYC

Golden Member
Jun 26, 2021
1,893
2,190
106
Now I remember (so, 'I remember it well' was off, lol!). Jobs was a cagey ah heck!

The "reality distortion field" that pro-Jobs groupies elevated to leadership style / vision, of making that seemed impossible possible.

Detractors said it simply lying - about sub-par performance of Apple Macs, back in the day, to the Apple-brainwashed masses.
 

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
Charlie over at Semiaccurate has an article (with the meat of it behind a paywall as usual) about how the Nuvia stuff is going at Qualcomm, and he doesn't seem to think it is going well at all.

There was another article on the same day about the 8cx Gen 3 where he mentions in passing "More on that in a separate article but if you are expecting the Nuvia cores to fix the Qualcomm compute efforts, you are in for a rude awakening". The separate article is the one mostly behind the paywall.

I've read his stuff since he was at the Inquirer (the former UK technology site not the US supermarket rumor rag) many years ago and while his record isn't perfect he usually knows what he's talking about - he was sounding the alarm about Intel's process problems years before anyone else for instance. He is sometimes a bit over the top (like the "too dangerous to deploy" tagline in the 8cx Gen 3 article) but even if he's sometimes off on the details I've never seen him wrong overall when he attacks something full force like he is/has been with Qualcomm ARM PCs up to this point, and like it appears he is with the Nuvia effort as well.

Not sure what is going on there but it sounds like there may be reason for concern if you're expecting that Qualcomm is going to be able to compete toe to toe with Apple in the near future.
 

NTMBK

Lifer
Nov 14, 2011
10,208
4,940
136
Charlie over at Semiaccurate has an article (with the meat of it behind a paywall as usual) about how the Nuvia stuff is going at Qualcomm, and he doesn't seem to think it is going well at all.

There was another article on the same day about the 8cx Gen 3 where he mentions in passing "More on that in a separate article but if you are expecting the Nuvia cores to fix the Qualcomm compute efforts, you are in for a rude awakening". The separate article is the one mostly behind the paywall.

I've read his stuff since he was at the Inquirer (the former UK technology site not the US supermarket rumor rag) many years ago and while his record isn't perfect he usually knows what he's talking about - he was sounding the alarm about Intel's process problems years before anyone else for instance. He is sometimes a bit over the top (like the "too dangerous to deploy" tagline in the 8cx Gen 3 article) but even if he's sometimes off on the details I've never seen him wrong overall when he attacks something full force like he is/has been with Qualcomm ARM PCs up to this point, and like it appears he is with the Nuvia effort as well.

Not sure what is going on there but it sounds like there may be reason for concern if you're expecting that Qualcomm is going to be able to compete toe to toe with Apple in the near future.

Ah damn. Charlie has a pretty solid track record, so there's probably a good reason why he's concerned. Hopefully he's got this one wrong, but I'm less excited about it than I was...
 

Joe NYC

Golden Member
Jun 26, 2021
1,893
2,190
106
Charlie over at Semiaccurate has an article (with the meat of it behind a paywall as usual) about how the Nuvia stuff is going at Qualcomm, and he doesn't seem to think it is going well at all.

There was another article on the same day about the 8cx Gen 3 where he mentions in passing "More on that in a separate article but if you are expecting the Nuvia cores to fix the Qualcomm compute efforts, you are in for a rude awakening". The separate article is the one mostly behind the paywall.

I've read his stuff since he was at the Inquirer (the former UK technology site not the US supermarket rumor rag) many years ago and while his record isn't perfect he usually knows what he's talking about - he was sounding the alarm about Intel's process problems years before anyone else for instance. He is sometimes a bit over the top (like the "too dangerous to deploy" tagline in the 8cx Gen 3 article) but even if he's sometimes off on the details I've never seen him wrong overall when he attacks something full force like he is/has been with Qualcomm ARM PCs up to this point, and like it appears he is with the Nuvia effort as well.

Not sure what is going on there but it sounds like there may be reason for concern if you're expecting that Qualcomm is going to be able to compete toe to toe with Apple in the near future.

That's funny, I raised it (employees leaving) as a possibility just a few messages ago (about 2 weeks ago) purely as a speculation, just based on the fact that these people already did it once, and cashed in big time, why not do it again?

BTW, I wasn't aware of Charlie being at Inquirer. I just remember Mike Magee as the most prolific writer there.
 

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
That's funny, I raised it (employees leaving) as a possibility just a few messages ago (about 2 weeks ago) purely as a speculation, just based on the fact that these people already did it once, and cashed in big time, why not do it again?

BTW, I wasn't aware of Charlie being at Inquirer. I just remember Mike Magee as the most prolific writer there.

Its unlikely any key employees are leaving already, they will have a lockup period as part of the acquisition deal. Leaving before that expires (typically 24 to 36 months) would cost them millions of dollars.

If there are problems, it is more likely it has to do with taking a server focused core and translating it to something focused for mobile/laptop. Or maybe turf battles between the existing Qualcomm SoC team and the newcomers from Nuvia.
 

Thala

Golden Member
Nov 12, 2014
1,355
653
136
If there are problems, it is more likely it has to do with taking a server focused core and translating it to something focused for mobile/laptop. Or maybe turf battles between the existing Qualcomm SoC team and the newcomers from Nuvia.

Thats not an issue. It was even Nuvia claiming, that server cores and mobile cores have many properties in common - in particular the performance and power properties.
 
Last edited:

MadRat

Lifer
Oct 14, 1999
11,909
228
106
If they are aiming for Apple M1 wouldn't this be more of a chromebook venture? My guess is they also have a partnership for automotive entertainment systems. There is so much more they could do in that realm than what is currently in cars, just have to control costs and there is less pressure on performance.
 

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
If they are aiming for Apple M1 wouldn't this be more of a chromebook venture? My guess is they also have a partnership for automotive entertainment systems. There is so much more they could do in that realm than what is currently in cars, just have to control costs and there is less pressure on performance.

Chromebooks don't need Apple M1 performance (they are mostly used in schools, who not only don't need high end gaming performance they probably considering not having it a feature rather than a bug) nor do they need its battery life (kids don't put in 12+ hours a day in front of a screen)

Chromebook competes at the bottom end of the laptop market price wise to be able to sell to school districts and less wealthy customers, so even if they can achieve M1 level performance they would use a cut down version. Heck, if Apple was competing in the Chromebook market I'd argue that not only would they use iPhone SoCs rather than M1s, they wouldn't even use the latest. They'd use an older cheaper one like the base iPad and Apple TV do.
 

Mopetar

Diamond Member
Jan 31, 2011
7,797
5,899
136
I think it's just a matter of perspective. Qualcomm will probably have a better core than the rest of the market as far as commodity ARM SoCs are concerned. Few companies do any significant custom design so it's not hard for them to be ahead there. It's hard to imagine them overtaking Apple in the same way that Apple did with respect to the rest of the market. From that perspective they fall short.

Depending on how you want to look at it all you can come away with two very different takes.
 

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
I think it's just a matter of perspective. Qualcomm will probably have a better core than the rest of the market as far as commodity ARM SoCs are concerned. Few companies do any significant custom design so it's not hard for them to be ahead there. It's hard to imagine them overtaking Apple in the same way that Apple did with respect to the rest of the market. From that perspective they fall short.

Depending on how you want to look at it all you can come away with two very different takes.


Let's take it as a given that Qualcomm will beat ARM designed cores, and thus have a performance advantage over companies that choose not to design their own (or could try but would fail at it like Samsung did) like Mediatek, Samsung and Google.

If CPU power/performance was all that mattered then Qualcomm would be the clear winner, but people care about other stuff like GPU performance (a potential advantage for Nvidia, should they enter the ARM SoC market) or "AI" performance (which Google is banking on as a differentiator) or price (which Mediatek has been making a killing on recently) or "I know what phone I want so I'm fine with whatever it comes with" which Samsung is relying on to push Exynos across their whole line and drop Qualcomm as they won't need US spec CDMA for much longer since Verizon is about to drop it.
 

Roland00Address

Platinum Member
Dec 17, 2008
2,196
260
126

Interesting interview up on the main site today. It worries me that the Senior Director for PCs is so dead set on fanless designs. Fans give you useful thermal headroom and higher performance.
Not me.

For performance laptops, nucs, etc sure.

But if I have 28 or 40 watt hours of battery* I want it to be as efficient as possible. It is not worth it being 20% faster if you sacrifice way more than 20% of battery life. Adding fans does not just take more juice on the actual device, but it encourages engineers to make the wrong efficiency trade offs.

*Edit: Dr Ian Cutress mentioned a specific laptop with a 52 watt hour battery. The precise amount does not matter most of the ultraportables I like are in that 28 to 40 range, 52 to 100 is performance territory / extra weight and cost. Of course more battery life is always good but if the battery is 30% bigger I am not sure it is worth to try to squeeze less than 30% of performance out of it.
 
Last edited:

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136

Interesting interview up on the main site today. It worries me that the Senior Director for PCs is so dead set on fanless designs. Fans give you useful thermal headroom and higher performance.

In order to be successful with ARM laptops, Qualcomm has to differentiate themselves from the mass of x86 laptops. They won't be able to compete head to head in performance (even if they win like Apple did shortly after announcement, the win won't last long and the x86 translation hit is much worse on Windows than Mac)

So how do you differentiate? You do fanless designs, which x86 laptops can't do unless they are dog slow. You may want the "headroom", but most consumers don't care because they don't need higher performance. They'd rather have something silent.
 

soresu

Platinum Member
Dec 19, 2014
2,617
1,812
136
and the x86 translation hit is much worse on Windows than Mac
I think the lesser transition hit has a lot to do with a couple of Firestorm's features.

1) The memory management being capable of mimicking x86 ISA.

2) Some instruction set called LSE2 (Large System Extensions 2). Ryan Houdek, one of the devs for FEX-emu identifies this as having significant perf benefits for running x86 code on ARM64, it was not in Cortex X1, but it can be found in its Neoverse sibling V1/Zeus aswell as the newer Cortex A710 and X2 CPU cores, so we shall see how much better SoC's based on them do with x86 binary translation.

A quick note, it seems like the FEX-emu guys are doing a bang up job of following in the footsteps of Rosetta 2 while coupling their work to DXVK when possible, worth checking out.
 

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
I think the lesser transition hit has a lot to do with a couple of Firestorm's features.

1) The memory management being capable of mimicking x86 ISA.

2) Some instruction set called LSE2 (Large System Extensions 2). Ryan Houdek, one of the devs for FEX-emu identifies this as having significant perf benefits for running x86 code on ARM64, it was not in Cortex X1, but it can be found in its Neoverse sibling V1/Zeus aswell as the newer Cortex A710 and X2 CPU cores, so we shall see how much better SoC's based on them do with x86 binary translation.

A quick note, it seems like the FEX-emu guys are doing a bang up job of following in the footsteps of Rosetta 2 while coupling their work to DXVK when possible, worth checking out.


No the main reason for Rosetta 2 doing better is because it is static translation rather than dynamic. That's very hard to do, and was really only made possible by the fact that Apple was able to retire 32 bit ARM and all the old APIs that went along with it, and basically all Mac applications are built using Apple's tools. They have probably been preparing the way for this for over half a decade.

Contrast with Microsoft's task of translating x86 and they have to worry about both 32 bit and 64 bit code, and decades old APIs are still fully supported, and tons of code is built with non Microsoft tools. Static translation is probably impossible for all practical purposes, unless you decide working on say 9 out of 10 applications and crashing or producing wrong results on the other 1/10th is fine.

Stuff like supporting the x86 memory model was only necessary because Apple wanted to do static translation. If you do it dynamically you can add barriers where necessary (which it usually isn't) to account for ARM's looser memory model. This is like a fraction of a percent difference.
 

Thala

Golden Member
Nov 12, 2014
1,355
653
136
(even if they win like Apple did shortly after announcement, the win won't last long and the x86 translation hit is much worse on Windows than Mac)

This is not true at all. Relative x64 emulation performance under Windows is very close to Rosetta despite we are comparing JIT vs. static translation. x86 emulation is somewhat slower, but Apple does not have x86 emulation at all to compare to.
To give some numbers x64 emulation performance under Windows is between 50% and 75% while Rosetta achieves 60%-85% native performance. In my book this makes the Windows emulator even more impressive as it is pure JIT without memory model support in HW.
In both cases this is significantly better than any other solution like Qemu or Box64 under Linux. That having said, Box64 outperforms Qemu.
 
Last edited:
  • Like
Reactions: soresu