Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

SoftBank set to sell UK’s Arm Holdings to Nvidia for $40B

ft.com

Post reply on HN
Print view View original

221–230 of 268 posts

Re: SoftBank set to sell UK’s Arm Holdings to Nvidia for $40B

#221

Post by zozbot234 »

adrian_b wrote: post #199
acidbaseextract wrote: post #183

Earlier quoted context omitted.

Why is RISC-V bad? You spend paragraphs ripping on it without actually explaining why it's bad.

This has been discussed many times on many forums on the Internet. The summary is that RISC-V is inefficient because it requires more instructions to do the same work as other ISAs and it does not have any advantage to compensate for this flaw. Those extra instructions appear especially in almost all loops and the most important reason is that RISC-V has a worse set of addressing modes than the the vacuum-tube comput…

> Nevertheless, that is wrong for 2 reasons, instruction fusion cannot reduce the larger code size due to the inefficient instruction encoding

RISC-V includes a compressed extension that makes instruction encoding competitive or better than x86(!), and with none of the drawbacks of ARM's Thumb modes.

Re: SoftBank set to sell UK’s Arm Holdings to Nvidia for $40B

#222

Post by Bayart »

noir_lord wrote: post #32
rvz wrote: post #16

Earlier quoted context omitted.

Unfortunately, you are correct. This is what Nvidia was looking for since out of the big 3 US semiconductor manufacturers, it doesn't have its own IP on CPUs. Now it will soon be able to control ARM to its advantage against the competition. This will indeed push the urgency for RISC-V to be worth looking at in the next few years to be used by hardware vendors soon. But it will be a long wait. As for the UK now realiz…

The techies here including myself did, it was a deal the government should have looked into. It was a crown jewel of a company, hard to imagine the French government letting it slide or the US government letting a intel be sold to a foreign company.

>hard to imagine the French government letting it slide or the US government letting a intel be sold to a foreign company

You mean something like GE buying off Alstom Énergie using corruption and government pressures ? Or the Chinese getting a piece of all our ports and nuclear tech ?

Re: SoftBank set to sell UK’s Arm Holdings to Nvidia for $40B

#223

Post by adrian_b »

zozbot234 wrote: post #221
adrian_b wrote: post #199

Earlier quoted context omitted.

This has been discussed many times on many forums on the Internet. The summary is that RISC-V is inefficient because it requires more instructions to do the same work as other ISAs and it does not have any advantage to compensate for this flaw. Those extra instructions appear especially in almost all loops and the most important reason is that RISC-V has a worse set of addressing modes than the the vacuum-tube comput…

> Nevertheless, that is wrong for 2 reasons, instruction fusion cannot reduce the larger code size due to the inefficient instruction encoding RISC-V includes a compressed extension that makes instruction encoding competitive or better than x86(!), and with none of the drawbacks of ARM's Thumb modes.

Compression methods reduce the size of the code, but that does not matter when assessing the efficiency of the base instruction encoding.

If you would apply the same compression methods to a more compact original encoding then the compressed code would be even smaller.

Competing ISAs, such as ARM (Thumb), MIPS (nanoMIPS) and POWER also have compressed encoding variants.

Re: SoftBank set to sell UK’s Arm Holdings to Nvidia for $40B

#224

Post by jka »

adrian_b wrote: post #223
zozbot234 wrote: post #221

Earlier quoted context omitted.

> Nevertheless, that is wrong for 2 reasons, instruction fusion cannot reduce the larger code size due to the inefficient instruction encoding RISC-V includes a compressed extension that makes instruction encoding competitive or better than x86(!), and with none of the drawbacks of ARM's Thumb modes.

Compression methods reduce the size of the code, but that does not matter when assessing the efficiency of the base instruction encoding. If you would apply the same compression methods to a more compact original encoding then the compressed code would be even smaller. Competing ISAs, such as ARM (Thumb), MIPS (nanoMIPS) and POWER also have compressed encoding variants.

Are there any benchmarks (in terms of code size, runtime performance, energy effiency, ...) available for OpenPOWER vs RISC-V?

(and if not, what would be some of the metrics to objectively compare the architectures?)

Re: SoftBank set to sell UK’s Arm Holdings to Nvidia for $40B

#225

Post by opencl »

microtonal wrote: post #213
phonon wrote: post #144

Earlier quoted context omitted.

PowerVR, Matrox, Trident, Bitboys Oy, 3Dlabs, Rendition, Intel740...

SiS also had 3D graphics chips. Tseng Labs was probably one of the earliest victims of the 3D GPUs wars, having developed but never completed the ET6300. PowerVR still exists and is seems successful in the mobile market (having provided Apple's GPUs until A10).

Imgtech/PowerVR was successful in the mobile market until Apple stopped buying their GPUs. Now they have practically no mobile customers though they're still reasonably successful in the automotive market.

Re: SoftBank set to sell UK’s Arm Holdings to Nvidia for $40B

#226

Post by adrian_b »

ksec wrote: post #172
adrian_b wrote: post #167

Earlier quoted context omitted.

I agree with what you said, but there is a reason for wanting to converge on ARM. There exists no good alternative. ARMv8.2 or newer is a very well designed ISA, while RISC-V is a very bad ISA and I would hate to be forced to use it. OpenPOWER is a far better ISA than RISC-V, but unfortunately most developers do not have any experience with POWER and they have the wrong belief that POWER is some antique ISA while RIS…

I wish I could upvote your comment a thousand times! Thoughts on OpenPOWER vs ARM v8.2 in terms of ISA? And for those who may be interested in OpenPOWER, https://github.com/antonblanchard/microwatt

ARMv8 was a clean design not constrained by compatibility with the past and it was created by people having a lot of experience with the implementation of ISAs in hardware and in software tools.

Therefore there is no surprise that it is an efficient ISA.

The only significant flaw in its first version was the lack of atomic instructions, but that was corrected in the subsequent versions.

The 32-bit POWER was a very nice ISA, but it was not designed to be extendable to 64-bit. It had blocks of the encoding space reserved for future extensions, but various details of the instruction word formats depended on the fact that the size of the registers was 32 bit.

When POWER was extended to 64-bit, much earlier than IBM expected, i.e. only 5 years after the introduction of the 32-bit variant, the extension was constrained because IBM has chosen to not have a mode switch like ARM but they have chosen to make a compatible ISA extension, i.e. which has the original POWER ISA as an instruction subset.

This has constrained the instruction encodings, so the 64-bit POWER ISA has some parts that seem more clumsy that in ARMv8 and the result is that programs for POWER are usually slightly larger than their ARMv8 equivalents. However, the hardware implementation effort for equivalent performance levels should be very similar for POWER and ARM and significantly less for both than for x86.

POWER also had a compressed encoding variant, but that was implemented in very few chips. Now the latest ISA variant has introduced 2 instruction word lengths, i.e. both 64-bit and 32-bit long instructions, instead of just 32-bit long instructions.

This allows the embedding of large immediate constants in the instructions, which is an important advantage of x86 vs. traditional RISC ISAs. This might help to reduce the sizes of many POWER programs.

Re: SoftBank set to sell UK’s Arm Holdings to Nvidia for $40B

#227

Post by dippersauce »

gridlockd wrote: post #211
gwd wrote: post #202

Earlier quoted context omitted.

> Now, an opportunity to become a technical peer to Intel at the CPU level? To control their own destiny? > They'd be insane not to take this opportunity. It made financial sense for ARM's board to sell to SoftBank; it makes financial sense for SoftBank to sell to NVIDIA; it makes financial and strategic sense for NVIDIA to buy ARM; it makes financial and strategic sense for NVIDIA to radically change the way the ARM…

This is a premature judgement. If NVIDIA treats ARM poorly, that benefits alternative ISAs and may ultimately undermine NVIDIAs investment. If NVIDIA treats ARM well, then that benefits the platform as a whole. The reality is going to be somewhere in the middle. Market equilibrium does not imply maximum benefit for any one interest group. "Consumers" or "end-users" are often conflated with "society".

Looking through the doomsayers in this thread, this is the result I believe to be most likely. If ARM goes belly up through gross mismanagement, then the laurels will be taken up by someone else. I think another aspect to consider is the perpetual license agreements ARM holds with several other businesses. I think this may muzzle Nvidia to some extent.

Re: SoftBank set to sell UK’s Arm Holdings to Nvidia for $40B

#228

Post by mamon »

the_other wrote: post #212
baybal2 wrote: post #75

Earlier quoted context omitted.

> You're blaming Nvidia. You should be blaming SoftBank and Masayoshi Son: bunch of fucking hucksters misleading people and making bad investments. You shouldn't be blaming bad investors. They pay for their decisions with their, or more often somebody's else, money. Blame the collective ownership of means of production, which in form of public companies have dominated the Western world for the last 30+ years. When pe…

> You shouldn't be blaming bad investors. They pay for their decisions with their, or more often somebody's else, money. > Blame the collective ownership of means of production, which in form of public companies have dominated the Western world for the last 30+ years. Can you unpack these two points? I don’t understand. It sounds like you’re saying “don’t blame the owners (because investors ARE owners); blame the own…

I think GP is arguing that public companies, which are often blamed for "shortcomings of capitalism" are in fact not capitalistic at all. "Collective ownership of means of production" was postulated by Karl Marx as one of the pillars of communism.

Publicly traded companies allow shareholders to limit their risk - when things start to go wrong you sell your shares to a bigger fool, and it's not your problem anymore. Also, as shareholders are not personally responsible for wrongdoings of their companies it also limits the risk to the buyer.

In short: publicly traded companies are enablers for moral hazard that we constantly observe in financial markets.

Re: SoftBank set to sell UK’s Arm Holdings to Nvidia for $40B

#229

Post by adrian_b »

jka wrote: post #224
adrian_b wrote: post #223

Earlier quoted context omitted.

Compression methods reduce the size of the code, but that does not matter when assessing the efficiency of the base instruction encoding. If you would apply the same compression methods to a more compact original encoding then the compressed code would be even smaller. Competing ISAs, such as ARM (Thumb), MIPS (nanoMIPS) and POWER also have compressed encoding variants.

Are there any benchmarks (in terms of code size, runtime performance, energy effiency, ...) available for OpenPOWER vs RISC-V? (and if not, what would be some of the metrics to objectively compare the architectures?)

It is very difficult to make a non-biased comparison between different ISAs.

If you just compile some benchmark programs for 2 different architectures and you look at the program sizes and the execution times, the differences might happen to be determined mostly by the quality of the compilers, not by the ISAs, in which case you could reach a wrong conclusion.

Many years ago, on one occasion I have spent many months working at the porting of a real-time operating system between Motorola 68k and 32-bit POWER. At another time I have also spent a couple of months with the porting of many device drivers between 32-bit POWER and 32-bit ARM and Thumb.

Such projects required a lot of examination of the code generated by compilers for the target architectures and also a lot of time spent with writing some optimized assembly sequences for a few parts of the code that were critical for the performance.

After spending so much time, i.e. weeks or months, with porting some large program, whose performance you understand well, between 2 ISAs, you may be reasonably confident of having a correct comparison of them.

If you want to reach a conclusion in a few hours at most, it is unlikely to be able to find an unbiased benchmark.

RISC-V is however a special case. Even if I have never spent time with implementing any program for it, after having experience with assembly programming for more than a dozen ISAs, when I see that almost any RISC-V loop may require up to a double number of instructions compared to most other ISAs, then I do not need more investigations to realize that reaching the same level of performance with RISC-V will require more complex hardware than for other ISAs.

Also, when comparing ISAs, I place a large weight on how good those ISAs are at GMPbench, i.e. at large number arithmetic. In my experience with embedded system programming large integer operations are useful much more frequently than traditional RISC ISA designers believe.

While x86 has always been very good at GMPbench, many traditional RISC ISAs suck badly, because they lack either good carry handling instructions or good double-word multiply/divide/shift instructions.

RISC-V also seems to have particularly bad multi-word operation support.

Re: SoftBank set to sell UK’s Arm Holdings to Nvidia for $40B

#230

Post by ece »

rbirkby wrote: post #170
mohankumar246 wrote: post #85

Along with cpu's ARM also makes mobile gpu ip (Mali), nvidia as well in their tegra. I would be surprised if this deal went through.

Mali was an ARM acquisition. It makes sense for NVidia to smash Mali and RTX together to allow Android manufacturers to compete against Apple. Apple built their GPU studio from ex-Imagination staff and will introduce 3 GPUs over the next year: Sicilian, Tonga, Lifuka to support their mobile and desktop plans. The question is whether ARM will sub license this combined GPU tech, or if it will be NVidia silicon only.

I think Android manufacturers need SoCs with 16MB L3 shared across the CPU+GPU+DSP+NN like the A13 more than they need RTX in a GPU.
Post reply on HN