- Joined
- May 18, 1997
- Messages
- 55,634
Nuvia is going to their lunch.
Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
Sigh. Fine. So I dug up the ISA REFERENCE guide.No. That Wikipedia article is wrong and re-linking to it is not going to change anything.
ARMv8 introduced two architectural states: Aarch64 and Aarch32. Aarch64 only can execute instructions from the A64 set (what I have called ARM64). Aarch32 only can execute instructions from the A32 and T32 sets (what I have collectively called ARM32).
The first armv8 mobile cores (e.g., Apple Cyclone and Cortex A57) implemented support for both ARM64 and ARM32 instructions sets, the first for the new 64bit code and the other for all the legacy code. Apple has dropped ARM32 in its last cores and ARM plans to do it by the year 2022
https://www.androidauthority.com/arm-64-bit-1165264/
A vendor choosing to drop support for part of an instruction state has nothing to do with the ISA. Marvell exiting the generic core market, however, shows that there's not enough business in that market to justify their continuing to pursue it. Part of the problem with ARM in the enterprise space is the fragmentation of the supplier base. While they're all supposed to be compatible, any engineer knows that's not going to be 100% true.The armv8 server cores didn't have to support tons of legacy code because ARM servers were just starting. So engineers implemented only ARM64 in the server chips. E.g. the TX2 core doesn't supports ARM32
https://en.wikichip.org/wiki/cavium/thunderx2#CN99xx
That is just what Marwell is saying you in the link that I posted before. Their TX2 and TX3 cores only support the new ARM64 ISA. The cores don't have to support the old ISAs and the legacy 32bit ARM code. That is the reason why the TX3 core uses about 30% less die area than an equivalent Zen3 core (both in 7nm):
Again, no.ARM engineers implemented 64bit as a new separate ISA instead as an extension to 32bit, because the long term plan was to offer the technical possibility to drop support for legacy stuff in the future. AMD engineers implemented 64bit as a cheap extension to 32bit and this is the reason why any modern 64bit x86 core from either AMD or Intel has to support legacy instructions sets back to 16 bits.
x86 wasn't designed for that - the GPU portion of that list is what you should care about, because that's what was heading to dominate during that timeframe - we all knew it was coming, even back in the mid-2000s when it was still Power on a lot of them and we were futzing around with Blue Gene and the like. But that's neither here nor there - ARM currently is on top, and that will change in 2-3 years like it always does (my bet is back to another custom ASIC of some kind, GPU or otherwise, although it might be hosted on ARM).Not so fast. I posted in this thread a figure with the evolution of architectures in the TOP500 list and x86 needed 10 years to dominate the list. ARM is expected to need a similar amount of time.
Which companies? Paypal is heavily on x86 - they've talked and presented at VMworld about their environment in the past, both when they were ebay and now separate. Links? Show me a non-hyper scalar that has done a mass migration to ARM, please - or even a significant (15%+) investment in ARM servers.The efficiency gain depends of what is being compared, but the gap is real and the reason why companies are migrating to ARM servers. I have some old graphs from Paypal with the reduction in "Power Consumption Per Year" resulting from the migration to ARM servers.
Keep reading the rest of what he posted - he goes into that too. The power advantage is minimal because hte CPU is only a small part of the overall server equation, but you know what? Lets set that aside, because at the end of the day, for your average Fortune 500 company, it doesn't really matter.He is making it in the first part that you quoted: "None of this is true. There are plenty of RISC alternatives to Intel, like SPARC, POWER, and MIPS, and none of them ended up having a power efficiency advantage."
I repeat again, this is no the older RISC vs CISC debate. ARM64 has a power efficiency advantage because it has been designed to be clean and efficient.
I tend to trust Keller. Now tack on 1TB of ram, drives, case fans/etc, and show me a 15%+ overall power improvement. If you can't, I'm not interested as a buyer, because the effort of the changeover will never generate an ROI. Even anand commented on performance per dollar, not performance per watt. Small TDP differences in the processor (I will gladly grant that ARM consumes less power than x86 - that was never my argument) won't make a significant difference to overall operational expenses. Go to my last point to understand why.TX2 in a 16nm node is as efficient as Naples in 14nm. Using the same node TX2 had been more efficient than Naples. Something similar happens when comparing TX3 to Rome. The ARM core needs about 30% less transistors and this translates into less power consumption. Jim Keller gave a talk explaining why ARM64 was more efficient than x86 and why K12 core was going to be better than Zen. Anandtech has an early analysis of the new N1 cores and of their efficiency gap with x86:
Ok, so once again - someone NOT a hyper-scalar. The markets are two different things, and Graviton is AWS only.I hope that they can soon test efficiency of some Neoverse chip.
Companies have been migrating to ARM no matter what you think about it. Paypal uses ARM servers, Microsoft is using ARM servers...
Moreover, the ARM-powered AWS instances are growing very fast.
EDIT: The Wikipedia page comparing ISAs is better than the other page you cite.
https://en.wikipedia.org/wiki/Comparison_of_instruction_set_architectures
As you can check, there is a single entry for x86 in the table, because AMD64 is an extension to x86-32. However, there are three entries for ARM in the table because A32, T32, and A64 are three different ISAs.
What do either the British and Chinese governments have to do with this, even remotely?Mark it down. Nvidia will never actually acquire ARM. The British government, the chinese government, and various regulatory agencies will step in. This will never happen.
Regulatory approval.What do either the British and Chinese governments have to do with this, even remotely?
British government has a lot of control over ARM, it being based there (the original ARM systems were tied to BBC projects, in fact - sorta). ARM also has a major branch in China, which has (naturally) resulted in the Chinese taking a certain interest in it (especially given the recent sanctions; this is a processor they can build since they can just "take" the designs as needed). Brits more than the Chinese have a chance of stopping it - ARM could always just cut off their Chinese arm (pun intended) and the Chinese would be out of the conversation, but the Brits - they control HQ. And treaty/legal agreements there, if the Brits don't approve... it won't happen.What do either the British and Chinese governments have to do with this, even remotely?
They need regulator approval from... the US, the UK, the EU and China.What do either the British and Chinese governments have to do with this, even remotely?
The one caveat on the British side is the current wave of nationalism going on there - they have a lot of pride in ARM (it's their CPU, dammit!), and I could see the MPs blocking it just because they don't want to give that up to the states. It's one of the few things of that nature that are totally British (current Japanese ownership aside). They could be dicks just to be dicks, in other words. Agreed on the Chinese/EU though. And the US.They need regulator approval from... the US, the UK, the EU and China.
US regulation is a joke and the US is rubbing their hands together that all major CPU ISA would be US owned. (which means if they say Nvidia no more ARM license's for Chinese companies... that is the way it will be) The UK let the previous sale of ARM go through so I can't see them standing in the way this time either. The EU has no teeth and will just go along... unless they see an opportunity to extract a toll which is likely. China however is likely going to fight this tooth and fucking nail. Personally I say the likely hood of this going through is 50/50. China may have the leverage to put the kybosh on this. Obviously they are not going to roll over and let the US have the ability to turn off the export of ALL standard CPU ISAs.
I see a Chinese ARM spin off with Nvidia losing billions of $$. A move like ARM ISA theft would sink any outside involvement in chinese fabs and chip production.Interesting conversation, thanks to all that are chiming in.
What prevents China from just stealing the ARM ISA like they do everything else if the sale goes through? Some sort of tariff threat maybe?
Reality: NadaInteresting conversation, thanks to all that are chiming in.
What prevents China from just stealing the ARM ISA like they do everything else if the sale goes through? Some sort of tariff threat maybe?
Interesting conversation, thanks to all that are chiming in.
What prevents China from just stealing the ARM ISA like they do everything else if the sale goes through? Some sort of tariff threat maybe?
You don't think they already have?Interesting conversation, thanks to all that are chiming in.
What prevents China from just stealing the ARM ISA like they do everything else if the sale goes through? Some sort of tariff threat maybe?
They need regulator approval from... the US, the UK, the EU and China.
US regulation is a joke and the US is rubbing their hands together that all major CPU ISA would be US owned. (which means if they say Nvidia no more ARM license's for Chinese companies... that is the way it will be) The UK let the previous sale of ARM go through so I can't see them standing in the way this time either. The EU has no teeth and will just go along... unless they see an opportunity to extract a toll which is likely. China however is likely going to fight this tooth and fucking nail. Personally I say the likely hood of this going through is 50/50. China may have the leverage to put the kybosh on this. Obviously they are not going to roll over and let the US have the ability to turn off the export of ALL standard CPU ISAs.
Don't forget the sanctions and embargo though - I'm not sure that Huawei has the latest ISA updates, or could easily get them legally. Not saying they couldn't get them otherwise, mind you - just not the normal way...Well they haven't really stolen any CPU ISAs.... the companies there making x86 have had proper licenses. (one was stopped by the US, when they forced AMD to end that deal). Their home grown ISAs for their super computers are as I understand it pretty unique... they may share some Alpha like elements, but so does every other CPU in the world really.
As I type this... I think I would perhaps revise my own forecasted chance of this going though to 90/10. What I sort of forgot about is huawei has a ARM Architecture license.... they may be having production problems thanks to the US. However the US can't do shit about them using the ARM ISA... they basically have the exact same deal Apple does. So if China can ramp up fabrication they can make ALL the ARM Cpus they like forever no matter what any other country says. So more then possible they don't really care all that much. They might put up a show and in the end let it go through after the new US administration throws them a few bones as compensation on the back end.
They do they have an Architecture license. Its no different from the ones Apple and Samsung hold. China is covered unless Nvidia tried to screw everyone over by massively extending the ISA... which isn't impossible.Don't forget the sanctions and embargo though - I'm not sure that Huawei has the latest ISA updates, or could easily get them legally. Not saying they couldn't get them otherwise, mind you - just not the normal way...
As for sanctions against Huawei..... I somehow doubt they survive this year. Perhaps they don't lift them in January... but come on Biden isn't going to continue.
you are joking right? all apple products ran on PowerPC for many years after they switched away from the /\/\otorola 68k processors
As of July 2020 17% MacOS.
https://www.statista.com/statistics/218089/global-market-share-of-windows-7/
Compared to the powerpc days of 1997
https://www.cnet.com/news/apple-market-share-sinks-again/#:~:text=Apple also slipped from this,the same quarter of 1997.
3.3%
So your comparison is... well quite frankly... not very valid is it?
I am a bit concerned about the URL.As of July 2020 17% MacOS.
https://www.statista.com/statistics/218089/global-market-share-of-windows-7/
As of July 2020 17% MacOS.
https://www.statista.com/statistics/218089/global-market-share-of-windows-7/
Compared to the powerpc days of 1997
https://www.cnet.com/news/apple-market-share-sinks-again/#:~:text=Apple also slipped from this,the same quarter of 1997.
3.3%
So your comparison is... well quite frankly... not very valid is it?
I'm not referring to gaming consoles obviously.then your inital question was stupid, you asked when the last time a whatever dollar company switched to/from one of the platforms on my list (clearly referring to apple)... one of which was PowerPC which apple switched to/from... and if you want to go with a 1,5T company microsoft switched from x86 to PowerPC and back to x86 with the xbox/xbox360/xbone
so you still waiting?
Sigh. Fine. So I dug up the ISA REFERENCE guide.
An important feature of the Armv8 architecture is backwards compatibility, combined with the freedom for optimal implementation in a wide range of standard and more specialized use cases. The Armv8 architecture supports: • A 64-bit Execution state, AArch64. • A 32-bit Execution state, AArch32, that is compatible with previous versions of the Arm architecture.
https://developer.arm.com/documentation/ddi0487/latest/
The Arm architecture has evolved significantly since its introduction, and Arm continues to develop it. Eight major versions of the architecture have been defined to date, denoted by the version numbers 1 to 8. Of these, the first three versions are now obsolete. The generic names AArch64 and AArch32 describe the 64-bit and 32-bit Execution states: AArch64 Is the 64-bit Execution state, meaning addresses are held in 64-bit registers, and instructions in the base instruction set can use 64-bit registers for their processing. AArch64 state supports the A64 instruction set. AArch32 Is the 32-bit Execution state, meaning addresses are held in 32-bit registers, and instructions in the base instruction sets use 32-bit registers for their processing. AArch32 state supports the T32 and A32 instruction sets
And page A1-38:
Transferring control between the AArch64 and AArch32 Execution states is known as interprocessing. The PE can move between Execution states only on a change of Exception level, and subject to the rules given in Interprocessing on page D1-2400. This means different software layers, such as an application, an operating system kernel, and a hypervisor, executing at different Exception levels, can execute in different Execution states.
You can change between states on the same processor.
In addition, the Armv8-A architecture allows different levels of AArch64 and AArch32 support, for example:
- AArch64 only designs.
- AArch64 designs that also support AArch32 operating systems/virtual machines.
- AArch64 support with AArch32 at (unprivileged) application level only.
A vendor choosing to drop support for part of an instruction state has nothing to do with the ISA. Although I do appreciate that ARM gives you that option - it does let the vendors build optimized CPUs.
Which companies? Paypal is heavily on x86 - they've talked and presented at VMworld about their environment in the past, both when they were ebay and now separate. Links?
I tend to trust Keller. Now tack on 1TB of ram, drives, case fans/etc, and show me a 15%+ overall power improvement. If you can't, I'm not interested as a buyer, because the effort of the changeover will never generate an ROI. Even anand commented on performance per dollar, not performance per watt.
Win10 runs on ARM; but you’re right on the applications. Recompile and reachitecture aren’t always easywe have been hearing this for decades now that ARM is the next big thing...yeah no because they do not have the ability to run windows and it's applications at an acceptable level
I think almost everyone here would much prefer if RISC-V filled that roll.Though seriously do you think that ARM is the answer to the ever encroaching x86 limit due to silicons eventual process limits?
Is that from some source? Why not? AMD can do what they want with ARM, produce and sell chips as they please, no different than Apple.I was hoping AMD would do an ARM chip as they were so close before, Now that Nvidia bought ARM - forget about it. its not going to happen.
IMOIs that from some source? Why not? AMD can do what they want with ARM, produce and sell chips as they please, no different than Apple.
Ok
Not 100% sure this is true, but it would not surprise me at all if they did have an ARM-based design in the works, they are trying to aim for Console dominance and getting something in there to get NVidia out of the Nintendo space would be their next objective.well lookie here. I had forgotten about the C7 info leaked last year.
AMD Ryzen C7
AMD Ryzen C7 specifications:
- CPU
- 2x Gaugin_Pro Mobile Core @ 3.0 GHz (Cortex-X1 based)
- 2x Gaugin_Mobile Core @ 2.6 GHz (Cortex-A78 based)
- 4x Arm Cortex-A55 clocked @ 2.0 GHz
- GPU – AMD Radeon RDNA 2 Mobile with 4x compute units @ 700 Mhz that claims to be 45% faster than Adreno 650 GPU found in Qualcomm Snapdragon 865 SoC, and supports real-time hardware ray-tracing and variable rate shading.
- Memory I/F – 4x 16-bit LPDDR5 @ 2750 MHz
- Storage – UFS 3.1
- Video Output – Up to 2K @ 144 Hz with HDR10+ support
- Connectivity
- Cellular – 5G via MediaTek 5G UltraSave modem with 2CC carrier aggregation and dual SIM support
- 802.11 b/g/n/ac/ax/ad/ay WiFi 6
- Bluetooth 5.1
- GNSS – GPS, Beidou, Galileo, GLONASS
- NFC
- USB – USB 3.1
- Process – 5nm TSMC
From what I understand of the M1, it does have some sort of Translation layer so not all the x86 to arm instructions are emulated just a subset of them. If Apple can do it I would have to assume that AMD could as well, question becomes do they have the time to sink the resources into a serious ARM product at this time. I would really hate to see AMD's hot streak come to an end because they stretched themselves too thin on too many projects at once.Being that Apple, AMD and a few others can create any design supporting the ARM 64 instruction set, something that will probably be around for decades, if AMD can or if they will, a hybrid processor that does both X64/Arm64 effectively is not off the table. A transition processor. As for users, I don't think most care what instruction set is being used as long as it does what one wants and is better than the previous device they used. In other words it does not matter too much to users if ARM64 or X64.