Discussion ARM unveils Cortex A78/X1

[Tabalan]

Hi,
this topic is about yesterdays ARM big announcement - Cortex A78 (which was expected by everyone) and Cortex X1 (which was surprise, I couldn't find any leaks or gossips about new, big core from ARM, closest thing I could find are some extremply sparse info about next year Mattlehorn core). Link to AnandTech article:

Arm's New Cortex-A78 and Cortex-X1 Microarchitectures: An Efficiency and Performance Divergence

www.anandtech.com

What surprised me is the fact that ARM decided to unveil Cortex X1 now. Currently, in mobile you have low performance Cortex A55, high performance A78 (A77 is described as "Leadership performance and efficiency for 5G mobile solutions ") and even higher performance X1. From my point of view, we lack medium core, which cxould be used in mid range SoC (Snapdragon 600, Kirin 600 and so on). I'm aware that Cortex A78 couldn't have same performance as A77 (but lower area and power consumption, shifting into middle core), because it would force ARM clients into noticeably bigger and more power hungry X1. This (pretty much forcing into X1) could negatively reflect on ARM-client relationship in future.
Next year ARMv9.0 should be announced, so, honestly, I expected new cores (huge, medium, small) to be released then. I think new small core will be mandatory, because there might be some problems with big.LITTLE combining ARMv8.2 and v9.0 cores (at least I assume so).

Another things, it's interesting to whether manufacturers will decide to go for X1 or stick with A78. I think all of them might jump on X1 train to not be left behind in benchmarks ("...but new Exynos is so slow, new Kirin/Snapdragon is 20% faster") with tri-cluster solutions (A78 would act as medium core, X1 would replace current A77 with higher clock speed), while mid-high (Kirin 800, SD700) series will adopt A78 + A55 combo.

What are your thoughts on this?
<<I'm just tech enthusiast, so pretty sure I'm wrong somewhere. Point it out or disregard it >>

Edit:
"What’s really shocking here is how close Arm would be getting to Intel and AMD’s current best desktop systems in terms of performance. If both incumbent x86 vendors weren’t already worried about Arm’s yearly rate of improvement over the last few generations, they should outright panic at these figures if they actually materialize – and I do expect them to materialize. "
Love this part. Considering vast majority of people use browser, Office and some minor programmes, is it pretty much safe to say we reached the point where we can shift from x86 to ARM? Yes, pro software and games would stay on x86 for now, but everyone else could migrate to ARM without any problem.

 

[soresu]

Next year ARMv9.0 should be announced, so, honestly, I expected new cores (huge, medium, small) to be released then. I think new small core will be mandatory, because there might be some problems with big.LITTLE combining ARMv8.2 and v9.0 cores (at least I assume so).

There's something to be said for staggering changes, rather than overly confusing consumers and spooking partners by a whole raft of big changes at once.

 

[soresu]

Yes, pro software and games would stay on x86 for now

Switch is changing that paradigm for games, and I saw a recent article by what sounded like a port focused company very bullish about future Switch ports of AAA games.

Part of the problem for pro software is often the things that go with it.

ie Cintiqs for drawing/sculpting, specialised input devices and keyboards for video editing and grading - not to mention that handling audio and video editing usually is much easier with a sizable setup where ARM doesn't really offer much of an advantage, excepting the possibility of fanless silent operation which can be done on some x86 setups anyway.

Much like servers, pro software is slow to change so I would expect it to come last after more games.

 

[dacostafilipe]

"What’s really shocking here is how close Arm would be getting to Intel and AMD’s current best desktop systems in terms of performance. If both incumbent x86 vendors weren’t already worried about Arm’s yearly rate of improvement over the last few generations, they should outright panic at these figures if they actually materialize – and I do expect them to materialize. "

With all the money thrown at ARM, this will certainly materialise in the near future. But until ARM supasses the x86 boys by a lot (in most applications!) there's no need to panic.

Much like servers, pro software is slow to change so I would expect it to come last after more games.

I don't know. Most game studios, specially AAA ones, normally only target platforms with a lot of possible sales. It will take time for ARM to "invade" desktops and attracts those game studios.

 

[soresu]

I don't know. Most game studios, specially AAA ones, normally only target platforms with a lot of possible sales. It will take time for ARM to "invade" desktops and attracts those game studios.

I don't have the sales figures for the AAA console titles on Switch, but Switch itself sells like hotcakes - I make no bones about such titles on ARM laptop/desktop platforms though.

 

[Tabalan]

@soresu
Wikipedia suggests that pretty much all best selling titles on Switch are typical Nintendo games:

List of best-selling Nintendo Switch video games - Wikipedia

en.wikipedia.org

Here you have numbers for Witcher 3 (2015, so 2 years before Switch release):

• PC: ~12 million
• PS4: ~ 10.8 million
• Xbox One: ~ 4.3 million
• Nintendo Switch: ~ 700,000
• Total: ~28.3 million

I believe porting AAA game to Switch might be challenging thing do so also because it's SoC (Cortex A57 + Cortex A53, Maxwell GPU, 20->16 nm process node) is borderline outdated and offers performance subpar to mid range smartphones.

@NeoLuxembourg
Right now, if I had ARM system on X1 with decent GPU, I could make 75% of my family to migrate from x86 to ARM overnight. Just like this. That's simply because there is already everything they need in ARM ecosystem (browser, office, basic apps, CAD viewing apps). If Intel and AMD would loose those casual users (plus many office PCs), it'd deeply hurt both companies.

 

[soresu]

@soresu
Wikipedia suggests that pretty much all best selling titles on Switch are typical Nintendo games:

List of best-selling Nintendo Switch video games - Wikipedia

en.wikipedia.org

Here you have numbers for Witcher 3 (2015, so 2 years before Switch release):

• PC: ~12 million
• PS4: ~ 10.8 million
• Xbox One: ~ 4.3 million
• Nintendo Switch: ~ 700,000
• Total: ~28.3 million

I believe porting AAA game to Switch might be challenging thing do so also because it's SoC (Cortex A57 + Cortex A53, Maxwell GPU, 20->16 nm process node) is borderline outdated and offers performance subpar to mid range smartphones.

@NeoLuxembourg
Right now, if I had ARM system on X1 with decent GPU, I could make 75% of my family to migrate from x86 to ARM overnight. Just like this. That's simply because there is already everything they need in ARM ecosystem (browser, office, basic apps, CAD viewing apps). If Intel and AMD would loose those casual users (plus many office PCs), it'd deeply hurt both companies.

Not a giant surprise when most of the current AAA console title ports were released years later, it would be more interesting to see sales for brand new games on Switch apples to apples with other consoles and PC at the same time.

 

[dacostafilipe]

@NeoLuxembourg
Right now, if I had ARM system on X1 with decent GPU, I could make 75% of my family to migrate from x86 to ARM overnight. Just like this. That's simply because there is already everything they need in ARM ecosystem (browser, office, basic apps, CAD viewing apps). If Intel and AMD would loose those casual users (plus many office PCs), it'd deeply hurt both companies.
[/QUOTE]

There's no need to wait for X1. There are powerful tablets / smartphones available today that have everything you listed and a multitude of games to play.

I mean, when I'm not on my desktop/laptop, I mostly only use ARM: SmartTV, iPhone, iPad, Watch, Car, ...

For all other markets (DIY, gamers, workstations, servers, ...) x86 is here to stay, for now (!)

 

[Tabalan]

@NeoLuxembourg
Right now, if I had ARM system on X1 with decent GPU, I could make 75% of my family to migrate from x86 to ARM overnight. Just like this. That's simply because there is already everything they need in ARM ecosystem (browser, office, basic apps, CAD viewing apps). If Intel and AMD would loose those casual users (plus many office PCs), it'd deeply hurt both companies.

There's no need to wait for X1. There are powerful tablets / smartphones available today that have everything you listed and a multitude of games to play.

I mean, when I'm not on my desktop/laptop, I mostly only use ARM: SmartTV, iPhone, iPad, Watch, Car, ...

For all other markets (DIY, gamers, workstations, servers, ...) x86 is here to stay, for now (!)
[/QUOTE]
I know there are powerful smartphones and tablets on ARM SoC, but I rather need NUC and laptop. Did quick search and in Poland it's pretty much impossible to buy Chromebook with ARM SoC, everything uses Intel/AMD. Including some dual core Celeron or A4-9000 series crap. Took a look google.com store and a) this website is atrocious is you want to find out something about Chromebook spec, b) everything uses AMD/Intel CPU c) they don't sell to Poland, so RIP.
NUC - better availability, more models to choose from, but all on Intel/AMD.

If I could buy NUC/laptop with solid SoC (newer ?SD 700 or 800 series, Kirin 800 or 900 series), I'd do it with pleasure (assuming it could run ChromOs or other relatively user friendly OS).

@soresu
It's very hard to find AAA game, which was released on same date and publisher shows sales for each platform. Witcher 3 was the closest I got.

 

[dacostafilipe]

If I could buy NUC/laptop with solid SoC (newer ?SD 700 or 800 series, Kirin 800 or 900 series), I'd do it with pleasure (assuming it could run ChromOs or other relatively user friendly OS).

There are enough guides online on how to create a NUC-like system based on Raspberry Pi or similar boards, for really cheap. Except for games, this will run most of the software you listed above.

As for ARM Laptops, what goes against a Tablet + Keyboard case?

 

[Roland00Address]

There are enough guides online on how to create a NUC-like system based on Raspberry Pi or similar boards, for really cheap. Except for games, this will run most of the software you listed above.

A 28nm 1.5 ghz cortex A72 would be nothing a-like a
A 07nm 2.8 ghz cortex A77.

 

[moinmoin]

All the Switch talk reminds me that ARM's decision to call the high end design X1 is really unfortunate considering Nvidia already got a Tegra X1 out for ages (using Cortex A57/A53).

 

[soresu]

All the Switch talk reminds me that ARM's decision to call the high end design X1 is really unfortunate considering Nvidia already got a Tegra X1 out for ages (using Cortex A57/A53).

Not particularly, there are only so many alphabetical monikers to denote a super high end class.

S could have been another option, but that has its own problems far more relevant, namely it's association with Samsung devices.

By comparison I've often mixed Apple Axx SoC's with Cortex A IP's.

Not to mention that Cortex X1 will not be the name of the product, just one (albeit important) element in a future SoC.

Given Tegra Orin is already slated to use A78 I wouldn't expect X1 to be used on its successor.

 

[soresu]

A 28nm 1.5 ghz cortex A72 would be nothing a-like a
A 07nm 2.8 ghz cortex A77.

By some quick math at least 3.2x greater performance including IPC and clock.

That's mostly counting integer though so it's a more conservative estimate.

 

[AkulaMD]

A 28nm 1.5 ghz cortex A72 would be nothing a-like a
A 07nm 2.8 ghz cortex A77.

I'm sure the single thread performance is much much better with A77, perhaps ~2X clock for clock?

Would be great if mainstream SBC like Odroid, Khadas, Raspberry Pi etc come with 7nm A77.

 

[dacostafilipe]

A 28nm 1.5 ghz cortex A72 would be nothing a-like a
A 07nm 2.8 ghz cortex A77.

New SoC outperforming an old SoC. Who would have known? /s

 

[Tabalan]

I'm sure the single thread performance is much much better with A77, perhaps ~2X clock for clock?

Would be great if mainstream SBC like Odroid, Khadas, Raspberry Pi etc come with 7nm A77.

I think Cortex A78 would be superior as it offers higher IPC then A77 (good for users), lower power consumption at same performance (can use cheaper cooling, less throttling, good for users and manufacturer) and it uses less area (good for manufacturer). Even at same process node A78 should offer lower cost for manufacturer and higher performance for user (vs A77).
Tho, I bet we'll see Cortex A77/78 NUC or SBC only after few years, just like we did with Cortex A72/73.

 

[AkulaMD]

I think Cortex A78 would be superior as it offers higher IPC then A77 (good for users), lower power consumption at same performance (can use cheaper cooling, less throttling, good for users and manufacturer) and it uses less area (good for manufacturer). Even at same process node A78 should offer lower cost for manufacturer and higher performance for user (vs A77).
Tho, I bet we'll see Cortex A77/78 NUC or SBC only after few years, just like we did with Cortex A72/73.

Agree with you on the A78 performance advantage over A77, would be very happy to get my hands on it. But considering the fact that the best mainstream SBC (e.g Odroid N2) is only running on A73 even today, i wouldn't hold my breath for A78 even after a few years down the road.

 

[NTMBK]

All the Switch talk reminds me that ARM's decision to call the high end design X1 is really unfortunate considering Nvidia already got a Tegra X1 out for ages (using Cortex A57/A53).

Can't wait for Nvidia to launch a Tegra A57 using the Cortex X1.

 

[Tabalan]

Agree with you on the A78 performance advantage over A77, would be very happy to get my hands on it. But considering the fact that the best mainstream SBC (e.g Odroid N2) is only running on A73 even today, i wouldn't hold my breath for A78 even after a few years down the road.

I think ARM might lower IP licensee fees for older uarch, so that's why now we see A72/73. At least that makes sense for me, so they still earn some $, while manufacturers get relatively high performance core for less bucks than A77/78 (considering only fees). So we might have chance to see A78 down the road, but I assume, by then, we should be preparing for something like Cortex A82 in mobile market.

 

[soresu]

I think ARM might lower IP licensee fees for older uarch, so that's why now we see A72/73. At least that makes sense for me, so they still earn some $, while manufacturers get relatively high performance core for less bucks than A77/78 (considering only fees). So we might have chance to see A78 down the road, but I assume, by then, we should be preparing for something like Cortex A82 in mobile market.

The question is, what is Matterhorn?

The super wide or the ordinary big core it is based on?

Having said that, just because X1 got chonky I don't expect this to be a continuing theme across the Xn series - especially if Sophia Antipolis are the ones making the next one given their record in efficiency.

At the very least the relative difference in big to huge core size shouldn't change much I think.

As for IP prices lowering, we've had mid range A72/73 for years now - I'd expect to see A75+ going forward.

The last Amazon Fire TV Stick 4K pretty much replaced the 2nd gen base Fire TV model (albeit with A53 little core instead of big A72), so I would expect a new FTVS 4K probably using Amlogic S905X4 with A55 and AV1 decoding for a solid boost, and a year later the next gen FTV Cube with a A76 powered SoC to replace the current with A73.

Going forward it sounds like Google might be joining the interactive streamer market with their next Chromecast Ultra which is supposed to be a full Android TV device with a Google Assistant remote rather than just a dumb streaming Google Cast device - which should make installing and updating Kodi a lot easier than on a Fire TV product.

I would probably expect the nu Chromecast Ultra to use Amlogic S905X4 too, as Google is one of the main benefactors of AV1 after all, perhaps we might even see a nu base Chromecast which does AV1 as well - that would mean a huge uptick in AV1 use very quickly.

 

[Tabalan]

Good question, what is Matterhorn. I assume only ARM and maybe some people close to it know. ARM is pretty good with keeping secrets, there are hardly any rumours. 2 weeks ago I'd say Matterhorn is ARM's 1st huge core, but well...

As for X series - I have opposite feeling. Imo, it might stick to being big, wide core, which can compete with Apple big core. But with ARMv9.0 next year ARM can pretty much do complete reset of naming/positioning of cores (so we get proper medium core). Damn, ARM market is imo so much more exciting than x86.

 

[soresu]

As for X series - I have opposite feeling. Imo, it might stick to being big, wide core, which can compete with Apple big core. But with ARMv9.0 next year ARM can pretty much do complete reset of naming/positioning of cores (so we get proper medium core). Damn, ARM market is imo so much more exciting than x86.

I didn't mean X2-3 wouldn't be huge, I meant that it will be proportionally big relative to the difference between X1 and A78.

ie on the next node it's not going to balloon in size, because area growth like that can't be sustained anyway, and power certainly can't.

ARM is more exciting, but that is part and parcel of their more open licensing model - x86 could be like this, though I suspect that Intel doesn't want any more competition.

 

[Tabalan]

I didn't mean X2-3 wouldn't be huge, I meant that it will be proportionally big relative to the difference between X1 and A78.

ie on the next node it's not going to balloon in size, because area growth like that can't be sustained anyway, and power certainly can't.

ARM is more exciting, but that is part and parcel of their more open licensing model - x86 could be like this, though I suspect that Intel doesn't want any more competition.

We will not get next node (3 nm) for next gen ARM cores, it'll still be 5 nm (or its' upgrades, like 4nm). Hard to tell what ARM is preparing for next year, they could do almost anything.

Agree on open model part, but I'm afraid we'll see less and less variety in ARM core design. Samsung and Qualcomm already pulled plug on theirs custom CPU designs. In smartphone market we'll be left with ARM Cortex (plus it's tweaked versions from different manufacturers) and Apple.

 

[soresu]

Agree on open model part, but I'm afraid we'll see less and less variety in ARM core design. Samsung and Qualcomm already pulled plug on theirs custom CPU designs. In smartphone market we'll be left with ARM Cortex (plus it's tweaked versions from different manufacturers) and Apple.

Don't forget nVidia, the use of A78 on Tegra Orin suggests Carmel didn't make the cut.

We will not get next node (3 nm) for next gen ARM cores, it'll still be 5 nm (or its' upgrades, like 4nm). Hard to tell what ARM is preparing for next year, they could do almost anything.

Don't bet on it, mobile SoC's have consistently pushed the state of the art on process nodes ahead of other market segments.

Besides TSMC's 3nm will not be nearly as different as Samsung's as it is targeted to remain finFET instead of moving to MBCFET as Samsung will on 3nm.