Discussion ARM vs Qualcomm: The Lawsuit Begins

[FlameTail]

Are you sure about that?

View attachment 113545

Pretty sure Qualcomm's ALA includes both ARMv8 and ARMv9. It was mentioned in a court document.

Oryon Gen 3 (Pegasus) will be ARMv9, according to rumours.

Qualcomm vs. Arm trial, Day 5 – Qualcomm wins! – Tantra Analyst

www.tantraanalyst.com

With Day 5, the trial wraps up.

 

[lakedude]

Pretty sure Qualcomm's ALA includes both ARMv8 and ARMv9. It was mentioned in a court document.

Yeah I don't know, what I'm seeing is that Qualcomm has an ARMv9 TLA not ALA. Got a source? I'd like to know one way or the other.

Ian was talking about this, maybe I'll rewatch his video...

 

[FlameTail]

Yeah I don't know, what I'm seeing is that Qualcomm has an ARMv9 TLA not ALA. Got a source? I'd like to know one way or the other.

Ian was talking about this, maybe I'll rewatch his video...

https://storage.courtlistener.com/recap/gov.uscourts.ded.79892/gov.uscourts.ded.79892.409.0.pdf

Pegasus

 

[Raqia]

People need to stop hanging on this because it just isn't viable.

The sheer amount of native ARM code in the smartphone space was enough to create an unassailable bulwark against Intel breaking into it with x86 options, and I don't imagine that the situation has become any less problematic for competing ISAs since.

In the Mac or WoA markets the problem is even worse, because the code bases people want ported to them tend to be beefier than most smartphone/tablet apps to the point that we still don't have native ARM ports for a whole raft of major applications on WoA.

It would take Apple switching to RISC-V and enforcing the change, and frankly I just don't see that as likely at this point.

Obviously using a royalty free ISA would be ideal, but it's not such a problem that Apple are losing sleep over it, especially not when they clearly have enough pull with ARM to get something like SME created in the ISA to align with their own additions.

Qualcomm already replaced microcontrollers in Snapdragon with RISC-V since the 865 much to ARM's chagrin. These may not be large, user facing cores, but they have considerable experience with it already.

As for mobile, most users download from app stores where apps are written in higher level languages running on top of runtimes meant to abstract away the underlying architecture. Android on RISC-V is maturing as well:

Android and RISC-V: What you need to know to be ready

You can build, test, and run the Android support for RISC-V on your own machine.

opensource.googleblog.com

AnandTech Forums: Technology, Hardware, Software, and Deals

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

www.anandtech.com

I think if they were able to more or less pull off Windows on ARM by '24 (with efforts first materializing in '18) which is a much more complex eco-system with decades of warts, they'll definitely be able to pull off RISC-V for mobile (a bread and butter source of revenue rather than an ancillary project) by the time their contract with ARM needs to be renegotiated in the 2030's.

 

[Tup3x]

Qualcomm already replaced microcontrollers in Snapdragon with RISC-V since the 865 much to ARM's chagrin. These may not be large, user facing cores, but they have considerable experience with it already.

As for mobile, most users download from app stores where apps are written in higher level languages running on top of runtimes meant to abstract away the underlying architecture. Android on RISC-V is maturing as well:

Android and RISC-V: What you need to know to be ready

You can build, test, and run the Android support for RISC-V on your own machine.

opensource.googleblog.com

AnandTech Forums: Technology, Hardware, Software, and Deals

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

www.anandtech.com

I think if they were able to more or less pull off Windows on ARM by '24 (with efforts first materializing in '18) which is a much more complex eco-system with decades of warts, they'll definitely be able to pull off RISC-V for mobile (a bread and butter source of revenue rather than an ancillary project) by the time their contract with ARM needs to be renegotiated in the 2030's.

You picked the first result from Google but in reality things aren't going that smoothly. It's not going to be a viable option anytime soon for Android.

RISC-V support in Android just got a big setback

The Android Common Kernel is about to remove support for the RISC-V architecture, which is a big setback for RISC-V support in Android.

www.androidauthority.com

 

[soresu]

Qualcomm already replaced microcontrollers in Snapdragon with RISC-V since the 865 much to ARM's chagrin. These may not be large, user facing cores, but they have considerable experience with it already.

As for mobile, most users download from app stores where apps are written in higher level languages running on top of runtimes meant to abstract away the underlying architecture. Android on RISC-V is maturing as well:

Android and RISC-V: What you need to know to be ready

You can build, test, and run the Android support for RISC-V on your own machine.

opensource.googleblog.com

AnandTech Forums: Technology, Hardware, Software, and Deals

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

www.anandtech.com

I think if they were able to more or less pull off Windows on ARM by '24 (with efforts first materializing in '18) which is a much more complex eco-system with decades of warts, they'll definitely be able to pull off RISC-V for mobile (a bread and butter source of revenue rather than an ancillary project) by the time their contract with ARM needs to be renegotiated in the 2030's.

#1. RISC-V is basically taking the microcontroller market, that is in no doubt, but that's a long way from serious penetration into the consumer electronics market.

Android is not a great yard stick for measuring whether an ISA has gained ground given x86 flopped catastrophically with it despite the insane amount of contra revenue Intel put into making it happen, and MIPS barely even caused a noticeable ripple.

#2. Given the current state of native major software ports for WoA I'd say it's definitely nowhere near "pulled off" as yet, the fact that Adobe hasn't ported most of their creative cloud apps yet is especially glaring.

That may change in the near future, but for now it's definitely not accurate.

In fact as far as AAA games go it's not accurate at all.

 

[Nothingness]

I think if they were able to more or less pull off Windows on ARM by '24 (with efforts first materializing in '18) which is a much more complex eco-system with decades of warts, they'll definitely be able to pull off RISC-V for mobile (a bread and butter source of revenue rather than an ancillary project) by the time their contract with ARM needs to be renegotiated in the 2030's.

MS has been working on several variants of their OS for Arm for almost 30 years. Of course the first variants were a seriously stripped down version which likely had little to do with the full-fledged Windows, but they surely gained experience.

 

[Raqia]

You picked the first result from Google but in reality things aren't going that smoothly. It's not going to be a viable option anytime soon for Android.

RISC-V support in Android just got a big setback

The Android Common Kernel is about to remove support for the RISC-V architecture, which is a big setback for RISC-V support in Android.

www.androidauthority.com

Is cherry picking the second result much better? Google hasn't announced any changes to their plans to get Android RISC-V compatible:

Google, however, has denied this means it is abandoning its efforts. "Android will continue to support RISC-V," the company claims in a statement made in response to the article highlighting the patches. "Due to the rapid rate of iteration, we are not ready to provide a single supported image for all vendors. This particular series of patches removes RISC-V support from the Android Generic Kernel Image (GKI)," which it must be noted isn't quite the same as support being entirely discontinued.

Google Removes RISC-V Support From the Android Common Kernel — But Denies Abandoning the ISA

Company blames the "rapid rate of iteration" for its removal, but says "Android will continue to support RISC-V."

www.hackster.io

 

[Raqia]

#1. RISC-V is basically taking the microcontroller market, that is in no doubt, but that's a long way from serious penetration into the consumer electronics market.

Android is not a great yard stick for measuring whether an ISA has gained ground given x86 flopped catastrophically with it despite the insane amount of contra revenue Intel put into making it happen, and MIPS barely even caused a noticeable ripple.

#2. Given the current state of native major software ports for WoA I'd say it's definitely nowhere near "pulled off" as yet, the fact that Adobe hasn't ported most of their creative cloud apps yet is especially glaring.

That may change in the near future, but for now it's definitely not accurate.

In fact as far as AAA games go it's not accurate at all.

RE:

1) I think Intel was horrifically mismanaged for a very long time and its hubris brought its downfall. Intel wasn't competitive in the mobile space because they targeted the space as a lazy after-thought during the height of its dominance rather than as a central focus i.e. where the hockey puck will be a few years down the road where they should have put the majority of its resources.

2) I think you're cherry-picking specific app categories to try to make a point, but in terms of customer support and industry adoption, '24 is an inflection point for sure with a surging trajectory. Sales have been good despite the nay-sayers and other CEOs cherry picking reports. Pricing is competitive and build quality is actually superior to x86 entrants due to the less expensive, better integrated and less battery hungry / cooler SoC.

The browser is indeed the most important cross-platform interface layer for most consumer use cases today and this finally saw Google support this year, and also reasonably good x86-64 support finally came in from Microsoft in the form of Prism / ARM64EC ABI and recently announced AVX2 support. Hardware was delayed directly due to ARM's interference in the process (many weeks were spent on the removal of ARM related IP from Phoenix according to court documents), and the resulting design was repurposed from a Server oriented design without power gating features for consumers. However, at 4nm it still sports better PPA than Intel at 3nm for less money too. At 3nm, I think they'll be more than competitive with M4 class big cores from what they're projecting.

 

[soresu]

and also reasonably good x86-64 support finally came in from Microsoft in the form of Prism / ARM64EC ABI and recently announced AVX2 support

You misspelled emulation 😑

It's better than nothing, but:

#1. It's imperfect compatibility at best.

#2. Emulation comes at a cost to power and/or performance vs native code.

When battery constrained mobile markets are the main focus of WoA this is not a trivial issue.

This is precisely why Apple set a firm date to phase out Rosetta2 support, to force devs to get their posteriors in motion sooner rather than later.

Infortunately this isn't something Microsoft can really do - best they or others can do is throw money at the problem and hire extra programmers for each dev to speed up their efforts, aka the nVidia approach.

 

[Tup3x]

Is cherry picking the second result much better? Google hasn't announced any changes to their plans to get Android RISC-V compatible:

The point is, it's not going to be viable platform for Android anytime soon.

 

[igor_kavinski]

Qualcomm wins legal battle over Arm — chipmaker didn't violate Arm's chip licensing agreement

But Arm could appeal.

www.tomshardware.com

Boring. That was too quick. Not much drama.

Qualcomm Statement on Trial Verdict Win | Qualcomm

www.qualcomm.com

 

[soresu]

Boring. That was too quick. Not much drama.

Real trials are a lot of build up and then over pretty quick.

The courts have a lot more to do than hand hold squabbling corporations.

 

[Doug S]

What does it mean? I am sure Arm will appeal? It will go all the way to Supreme Court. But in the interim QCOM will keep releasing new products. Can they also adopt ArmV9?

You can't appeal just because you think "the jury decided wrong". Courts give great deference to juries, so unless some sort of misconduct can be claimed and proven the jury verdict won't get thrown out.

So they are left with appeals based on something they think was wrongly entered/not entered into evidence, or some testimony that was objected to. Even then, the appeals court would have to believe that 1) the judge erred in his ruling on that and 2) it may have changed the jury's ruling - so it would have to something pretty big. That's one of the reasons you always see a lot of action around the judge's decisions to enter exhibits into evidence, or objections to things that might sound routine. They want it in the record that they said "this is wrong" in case they decide to base their appeal on it. Because if you don't say anything at trial the bar for basing an appeal on it is WAY higher, and requires extenuating circumstances.

Whether it is worth appealing will depend on whether they actually have anything material upon which to build such an appeal, that an appeals court might believe is significant enough it may change the jury's verdict. Because now that ARM has lost, there is not really any value for them in delay. If they don't have anything real to appeal on I expect them to talk a bit about how they are disappointed and reserve their right to appeal but not do so.

 

[lakedude]

So now what?

ARM canceled Qualcomm's license, is that automatically not an issue now?

Qualcomm can make Oyron 2 chips at their v8 ALA rate?

What about the counter-suit?

 

[Tuna-Fish]

ARM canceled Qualcomm's license, is that automatically not an issue now?

ARM justified it with "Qualcomm violated their license." A jury just found that no, they did not. If ARM doesn't just quietly make that go away, Qualcomm suing ARM about it is the easiest contract violatipn suit anyone has ever litigated.

 

[Thibsie]

ARM justified it with "Qualcomm violated their license." A jury just found that no, they did not. If ARM doesn't just quietly make that go away, Qualcomm suing ARM about it is the easiest contract violatipn suit anyone has ever litigated.

Agreed. Arm thought they were smarter.
Bad for them.

 

[DrMrLordX]

ARM is not in a good position. It appeared that this lawsuit was an attempt to squeeze Qualcomm for cash as a response to the failure of the nVidia buyout. Instead ARM blew a bunch of cash on legal fees and came up empty-handed.

 

[NTMBK]

Arm will seek a new trial against Qualcomm

Only one of the three original charges can be relitigated.

www.tomshardware.com

As expected, this will rumble on for a while

 

[poke01]

Arm will seek a new trial against Qualcomm

Only one of the three original charges can be relitigated.

www.tomshardware.com

As expected, this will rumble on for a while

This doesn't mean much, the appeal courts will likely upload the original ruling. ARM is wasting its time and money

 

[LightningZ71]

They (ARM) likely have little choice. They chose this path and need to see it through. If they don't, they can face shareholder lawsuits over it.

 

[Win2012R2]

This doesn't mean much, the appeal courts will likely upload the original ruling. ARM is wasting its time and money

It will have to be a new trial for that deadlocked question, not an appeal - appeals will be after unless they settle but at this stage Qualcomm got zero desire to do it

 

[igor_kavinski]

QC's gonna take over the world after all.

Get vigilant, AMD.

 

[gdansk]

Get vigilant, AMD.

AMD has nothing to lose from Qualcomm. No one buys their mobile parts anyway. It's Intel that stands to lose the sales it depends on. But only if WoA can be made not DoA. And that's a big if.

 

[adroc_thurston]

QC's gonna take over the world after all.

Get vigilant, AMD.

QC is a comically inept company at things not being parasites to a giant RF patent chest.