Arm is canceling Qualcomm's chip design license • Hacker News Board

Re: Arm is canceling Qualcomm's chip design license

#381

Post by GeekyBear » Wed, Oct 23, 2024, 9:27 PM UTC

throwup238 wrote: post #376

GeekyBear wrote: post #339

Earlier quoted context omitted.

> throwing their partners under the bus I found it telling that every single smartphone vendor refused to license Qualcomm's proprietary tech for smartphone to satellite messaging. > In a statement given to CNBC, Qualcomm says smartphone makers “indicated a preference towards standards-based solutions” for satellite-to-phone connectivity https://arstechnica.com/gadgets/2023/11/qualcomm-kills-its-c...

Anyone only ever uses Qualcomm chips because they have a gun to their head. Usually that gun is the latest wireless standard like 4g or 5g.

4g and 5g are open standards.

It was their now mostly irrelevant CDMA patents that Qualcomm used as a weapon against device makers.

> Many carriers (such as AT&T, UScellular and Verizon) shut down 3G CDMA-based networks in 2022 and 2024, rendering handsets supporting only those protocols unusable for calls, even to 911.

https://en.m.wikipedia.org/wiki/Code-division_multiple_acces...

In my opinion, Qualcomm's abuse of their CDMA patents is the reason that zero device makers were willing to get on board with a new Qualcomm proprietary technology.

Re: Arm is canceling Qualcomm's chip design license

#382

Post by derstander » Wed, Oct 23, 2024, 9:31 PM UTC

wang_li wrote: post #352

coder543 wrote: post #296

Earlier quoted context omitted.

Hardware drivers certainly can be annoying, but a hobbyist struggling to bring big GPUs’ hardware drivers to a random platform is not at all indicative of how hard it would be for a company with teams of engineers. If NVidia wanted their GPUs to work on Raspberry Pi, then it would already be done . It wouldn’t be an issue. But NVidia doesn’t care, because that’s not a real market for their GPUs. Most OEMs don’t have…

> If NVidia wanted their GPUs to work on Raspberry Pi, then it would already be done. It wouldn’t be an issue. But NVidia doesn’t care, because that’s not a real market for their GPUs. It's weird af that Geerling ignores nVidia. They have a line of ARM based SBCs with GPUs from Maxwell to Ampere. They have full software support for OpenGL, CUDA, and etc. For the price of an RPi 5 + discreet GPU, you can get a Jetson…

> It's weird af that Geerling ignores nVidia.

“With the PCI Express slot ready to go, you need to choose a card to go into it. After a few years of testing various cards, our little group has settled on Polaris generation AMD graphics cards.

Why? Because they're new enough to use the open source amdgpu driver in the Linux kernel, and old enough the drivers and card details are pretty well known.

We had some success with older cards using the radeon driver, but that driver is older and the hardware is a bit outdated for any practical use with a Pi.

Nvidia hardware is right out, since outside of community nouveau drivers, Nvidia provides little in the way of open source code for the parts of their drivers we need to fix any quirks with the card on the Pi's PCI Express bus.”

Reference = https://www.jeffgeerling.com/blog/2024/use-external-gpu-on-r...

I’m not in a position to evaluate his statement vs yours, but he’s clearly thought about it.

Re: Arm is canceling Qualcomm's chip design license

#383

Post by Kirby64 » Wed, Oct 23, 2024, 9:35 PM UTC

GeekyBear wrote: post #381

throwup238 wrote: post #376

Earlier quoted context omitted.

Anyone only ever uses Qualcomm chips because they have a gun to their head. Usually that gun is the latest wireless standard like 4g or 5g.

4g and 5g are open standards. It was their now mostly irrelevant CDMA patents that Qualcomm used as a weapon against device makers. > Many carriers (such as AT&T, UScellular and Verizon) shut down 3G CDMA-based networks in 2022 and 2024, rendering handsets supporting only those protocols unusable for calls, even to 911. https://en.m.wikipedia.org/wiki/Code-division_multiple_acces... In my opinion, Qualcomm's abuse of…

They’re open standards, sure, but there is almost no competition in the cellular modem space. Intel tried and failed spectacularly. Apple bought the scraps of the Intel modem business and still hasn’t released their own modems after… 5 years? Cellular tech is hard and has a lot of esoteric requirements. If you want a cell modem for a smartphone, you essentially buy either the latest and greatest from Qualcomm or you buy something from one of the Chinese companies (Huawei, Mediatek) which has its own set of problems.

Re: Arm is canceling Qualcomm's chip design license

#384

Post by starspangled » Wed, Oct 23, 2024, 9:45 PM UTC

markhahn wrote: post #285

starspangled wrote: post #93

Earlier quoted context omitted.

Just the impact of making this move will have a chilling effect, regardless of the long term outome. ARM Ltd wants to position itself as the ISA. It is highly proprietary of course, but the impression they want to give is that it is "open" and freely available, no lock-in, etc. This really brings the reality back into focus that ARM controls it with an iron fist, and they're not above playing political games and sidi…

"Chilling effect" implies that we should want ARM to succeed. IMO we need to question the premises of the current IP ecosystem. Obviously, the principles of open-source are quite the opposite to how ARM licenses IP. (Afaik, ARM also licenses ready-to-go cores, which is very different from what Q is getting.) It's easy to see how RISC-V avoids the conflict of interest between owning the ISA and licensing specific impl…

> "Chilling effect" implies that we should want ARM to succeed.

It really doesn't.

I agree an actual open ISA is far preferable, ARM is not much different than x86.

Re: Arm is canceling Qualcomm's chip design license

#385

Post by GeekyBear » Wed, Oct 23, 2024, 10:38 PM UTC

Kirby64 wrote: post #383

GeekyBear wrote: post #381

Earlier quoted context omitted.

4g and 5g are open standards. It was their now mostly irrelevant CDMA patents that Qualcomm used as a weapon against device makers. > Many carriers (such as AT&T, UScellular and Verizon) shut down 3G CDMA-based networks in 2022 and 2024, rendering handsets supporting only those protocols unusable for calls, even to 911. https://en.m.wikipedia.org/wiki/Code-division_multiple_acces... In my opinion, Qualcomm's abuse of…

They’re open standards, sure, but there is almost no competition in the cellular modem space. Intel tried and failed spectacularly. Apple bought the scraps of the Intel modem business and still hasn’t released their own modems after… 5 years? Cellular tech is hard and has a lot of esoteric requirements. If you want a cell modem for a smartphone, you essentially buy either the latest and greatest from Qualcomm or you…

Samsung, Huawei and MediaTek SOCs implement the same open 4g/5g standards.

Apple's modem is said to be shipping this coming spring in the newest iPhone SE iteration.

Google's Pixel phone lineup has used Samsung's modems for generations now.

Re: Arm is canceling Qualcomm's chip design license

#386

Post by Kirby64 » Wed, Oct 23, 2024, 10:44 PM UTC

GeekyBear wrote: post #385

Kirby64 wrote: post #383

Earlier quoted context omitted.

They’re open standards, sure, but there is almost no competition in the cellular modem space. Intel tried and failed spectacularly. Apple bought the scraps of the Intel modem business and still hasn’t released their own modems after… 5 years? Cellular tech is hard and has a lot of esoteric requirements. If you want a cell modem for a smartphone, you essentially buy either the latest and greatest from Qualcomm or you…

Samsung, Huawei and MediaTek SOCs implement the same open 4g/5g standards. Apple's modem is said to be shipping this coming spring in the newest iPhone SE iteration. Google's Pixel phone lineup has used Samsung's modems for generations now.

There’s so many sub varieties of the standards. You’re making a gross oversimplification that they’re “the same”. Compare intel vs Qualcomm modems that were released on iPhones. They were “the same” standard but the Qualcomm modems were notably faster in testing. Maybe they’re all at parity these days, but it’s pretty hard to do a fair comparison.

Re: Arm is canceling Qualcomm's chip design license

#387

Post by wbl » Wed, Oct 23, 2024, 10:46 PM UTC

rangestransform wrote: post #378

fidotron wrote: post #300

Earlier quoted context omitted.

Brucehoult mentions it further down. > This time last year they were all over the RISC-V mailing lists, trying to convince everyone to drop the "C" extension from RVA23 because (basically confirmed by their employees) it was not easy to retrofit mildly variable length RISC-V instructions (2 bytes and 4 bytes) to the Aarch64 core they acquired from Nuvia. At the same time, Qualcomm proposed a new RISC-V extension that…

Not a cpu designer, but aren’t variable length instructions a big part of why x86 decoders take up so much area, and also increases branch predictor complexity?

So much is relative: on a giant OoO core it's barely noticeable. In the case of x86 the range of lengths is extreme: a simple two or four choice is much simpler.

Re: Arm is canceling Qualcomm's chip design license

#388

Post by brucehoult » Wed, Oct 23, 2024, 11:18 PM UTC

Narishma wrote: post #358

cesarb wrote: post #304

Earlier quoted context omitted.

> Yeah, it's a little harder than for pure fixed-width Arm64, but it is massively massively easier than for amd64. For those who haven't read the details of the RISC-V ISA: the first two bits of every instruction tell the decoder whether it's a 16-bit or a 32-bit instruction. It's always in that same fixed place, there's no need to look at any other bit in the instruction. Decoding the length of a x86-64 instruction…

Why do they use two bits for it? Do they plan to support other instruction lengths in the future?

So that there are 48k combinations available for 2-byte instructions and 1 billion for 4-byte (or longer) instructions. Using just 1 bit to choose would mean 32k 2-byte instructions and 2 billion 4-byte instructions.

Note that ARMv7 uses a similar scheme with two instruction lengths, but using The first 4 bits from each 2-byte parcel to determine the instruction length. It's quite complex, but the end result is 7/8 (56k) 2-byte instructions are possible and 1/8 (512 million) 4-byte instructions.

IBM 360 in 1964 thru Z-System today also uses a 2-bit scheme to choose between 2-byte instructions with 00 meaning 2-bytes (16k instructions available), 01 or 10 meaning 4-bytes (2 billion instructions available), and 11 meaning 6-bytes (64 terra instructions available).

Re: Arm is canceling Qualcomm's chip design license

#389

Post by GeekyBear » Wed, Oct 23, 2024, 11:20 PM UTC

Kirby64 wrote: post #386

GeekyBear wrote: post #385

Earlier quoted context omitted.

Samsung, Huawei and MediaTek SOCs implement the same open 4g/5g standards. Apple's modem is said to be shipping this coming spring in the newest iPhone SE iteration. Google's Pixel phone lineup has used Samsung's modems for generations now.

There’s so many sub varieties of the standards. You’re making a gross oversimplification that they’re “the same”. Compare intel vs Qualcomm modems that were released on iPhones. They were “the same” standard but the Qualcomm modems were notably faster in testing. Maybe they’re all at parity these days, but it’s pretty hard to do a fair comparison.

By all means, point to some official statement showing that Google cannot market the Pixel phone as supporting the 4g/5g standards due to Google's use of Samsung modems.

Re: Arm is canceling Qualcomm's chip design license

#390

Post by jecel » Wed, Oct 23, 2024, 11:31 PM UTC

rangestransform wrote: post #378

fidotron wrote: post #300

Earlier quoted context omitted.

Brucehoult mentions it further down. > This time last year they were all over the RISC-V mailing lists, trying to convince everyone to drop the "C" extension from RVA23 because (basically confirmed by their employees) it was not easy to retrofit mildly variable length RISC-V instructions (2 bytes and 4 bytes) to the Aarch64 core they acquired from Nuvia. At the same time, Qualcomm proposed a new RISC-V extension that…

Not a cpu designer, but aren’t variable length instructions a big part of why x86 decoders take up so much area, and also increases branch predictor complexity?

The RISC-V creators were aware of the issues when they designed the C extension with 16 bit instructions to be able to compete with ARM's Thumb. So the bottom two bits of an instruction are enough to distinguish between 16 bit and 32 bit encodings (the standard has a scheme for future longer instructions, but hardware that doesn't implement them can ignore that).

This means that if a RISC-V reads a 16 byte block from instruction memory, it only has to look at 8 pairs of bits. This would require 8 NAND gates plus 8 more NANDs to ignore the top half of any 32 bit instructions. That is 4x(8+8)=64 transistors.

The corresponding circuit for x86 would be huge.

But note that this just separates the instructions. You still have to decode them. Most simple RISC-V implementations have a circuit that transforms each 16 bit instruction into the corresponding 32 bit one, which is all the rest of the processor has to deal with. Here are the sizes of some such circuits:

Hazard 3: 733 NANDs (used in the Raspberry Pi RP2350)

SERV: 532 NANDs (serial RISC-V)

Revive: 506 NANDs (part of FPGAboy)

You would need 8 such circuits to handle the maximum 16 bit instructions in a 16 byte block, and then you would need more circuits to decode the resulting 32 bit instructions. So the 16 NANDs to separate the variable length instructions is not a problem like it is for other ISAs.

The problem with 16 bit instructions for small RISC-V implementations is that now 32 bit instructions will not always be aligned with 32 bit words. Having to fetch an instruction from two separate words adds circuits that can be a large fraction of a small design.