POWER ISA Contributed To Open-Source, OpenPOWER Joining The Linux Foundation

[coder]

As far as South Korea goes, Samsung are designing their own ARM CPUs.

I'm pretty sure they still depend on ARM for software and support. If Japan banned ARM from dealing with South Korea, like what happened with Huawei, I think they'd still suffer some issues. And that's not even considering their use of Mali GPUs (and their closed-source driver).

[coder]

That is going to be different soon, as they do co-develop something with AMD as announced some weeks ago.

The time-horizon on that is probably about 2 years. So, that means for about the next 4-5 years, they're still going to need to support Mali-based products.

[pal666]

They still rely upon ARM for the graphics.

And that's not even considering their use of Mali GPUs (and their closed-source driver).

they've bought amd graphics recently

[coder]

They are also responsible for much of the new vector capability ( the acronym escapes me). The thing that excites me is a desktop or even laptop ARM processor that fully implements that vector standard. Not everybody can afford a Fujitsu super computer but many would like an order of magnitude increase in FP processing on their desktop.

Use your GPU.

Say what you want about Intel iGPUs, but they're the only truly mass market GPU to have a 2:1 ratio of fp32 to fp64. That's a lot of fp64 performance that's probably sitting right in your desktop or laptop, untouched. And it's more than you're going to get from some ARM vector extensions.

If you want yet more fp64, you can still pick up a new Radeon VII. Not for much longer, though.

[coder]

Power comes from the commercial Unix world and reflects that. That means there's an emphasis on:

• Per-socket and per-core throughput, especially with multithreading and large caches, which tend to benefit OLTP applications.
• Scalability. P9 scales gluelessly to 16-socket systems.
• I/O. P9 was the first major processor to support PCIe4, and it also supports OpenCAPI and NVlink directly for talking to accelerators.
• Bandwidth. Enterprise-class Power9 goes to 230GB/s of memory bandwidth per socket today, and that's being increased to 650GB/s with the new Power9 AIO chips coming out next year. These are large numbers.
• Openness. This is a pretty new thing, but the entire OpenPower firmware stack is open-source. (The enterprise Power firmware stack is not; those are different machines with higher price tags.)

There are still some downsides, historically.

• Cost. Power is priced rather well now, but historically carried a significant price premium over commodity hardware.
• Weak ecosystem. Lots of things have assembly optimizations for x86. Fewer do for Power. As a result, performance on random open-source apps can be decidedly hit-or-miss.
• Few hardware options. Raptor is doing an amazing job, with reasonable pricing and a true commitment to openness... but it's not like you can just order a Proliant with a Power9 in it.

IMO, you're missing the point. You're just talking about its history and products, but none of that is specific to its ISA, which is the thing that's actually now open and free to use.

So, in my mind, the operative questions are whether someone would go with RISC-V or POWER, and why. I could imagine the answer is somewhat dependent on what they want to do with it, but that also makes it more interesting.

[Dawn]

IMO, you're missing the point. You're just talking about its history and products, but none of that is specific to its ISA, which is the thing that's actually now open and free to use.

So, in my mind, the operative questions are whether someone would go with RISC-V or POWER, and why. I could imagine the answer is somewhat dependent on what they want to do with it, but that also makes it more interesting.

You may find my post further down the same page to be of interest.

[tuxd3v]

They still rely upon ARM for the graphics.

In the Graphics sector its a lot easier..
They can still go with Imagination PowerVr( Or Maybe NOT, Taiwan dispute with UK? ), or with ThinkSilicon Graphics( the one already partnering with SiFive.. ), this one most probably..

Nothing will prevent Japan from blocking South Korea( if they really want too, to protect SONY.. )..
China Huawei,
Also develops its ARM CPUs, but seems that ARM can block them...

The Mobile Market, with cpu cores( from ARM ),
Seems the problem for China/South Korea, both are really good on this markets..
I believe that China will try a RISC-V/( SiliconMotion Graphics or ThinkSilicon Nema) design on mobile market, if it feels completely against the wall..

[coder]

There are only 5 Countries in the world, that can trash International Law, US/Russia/China/UK/France.

US did it with ARM against China, and the world was watching it..

You think that had anything to do with permanent UN Security Council membership? And what's this about international law? What international law was broken?

And it wasn't all Chinese companies - just Huawei. Technically, not even a state-owned enterprise.

It will hurt ARM a lot( which is nice for the US, bad for Japan/UK ),

You're over-nationalizing this. It was also bad for the US, since ARM has a lot of US-based employees and customers. And most of the custom ARM cores were designed in the US, by the likes of Qualcomm, Apple, Nvidia, Cavium, Ampere Computing, and probably others.

So, I'd say anything bad for ARM is a net-negative for the US.

It will give MIPS a big chance in the US,

I'm skeptical, but open minded. So, go ahead and make the case for MIPS, if you want.

It will definitely bring up RISC-V has a way to escape bully situations,
Every single country now Understands that technology is a very Important thing, for them to be Independent..

I think it was already clear from the US restricting exports of Xeons for Chinese supercomputers, about 5 years ago.

And even International Law,
That before was considered, as written in Stone,

Um, in what reality? Have you heard of the WTO? That established a court system in which trade disputes could be decided. If laws and trade agreements were never violated, there wouldn't be any need for these courts, would there?

[coder]

You may find my post further down the same page to be of interest.

Yes, very good. Thanks for taking the time. Those seem like some good insights.

saturation math, integer rotate (this is a big one! it's supposedly coming in the BitManip extension but AFAIK that's not finalized yet, several years after RV started taking off.)

AFAIK, saturation math is useful mostly for signal processing, no? So, I don't mind if it's relegated to an extension.

Bit rotate - is that useful for anything besides encryption? If so, being in an extension won't be of much use (for non-embedded), since code compiled for the base ISA won't be able to utilize it. Same for popcnt, clz, and whatever other goodies BitManip adds.

It doesn't make a lot of the rookie mistakes from early RISCs (hi/lo SPRs, delay slots)

SPR = Special Purpose Registers? What for - holding the results of multiplies, or something?

And delay slots are annoying, if you have to write assembly. I think they're mostly an in-order CPU thing. If you have OoO and especially speculative execution, then the CPU will find other work to do there.

[tuxd3v]

If I was Ampere (or someone like them), I would take that POWER9, run right over to TSMC and get it on a new node toot sweet. Use that assembly agreement with Lenovo and start marketing product.

I Understand the 'tout suit' Idea..

US is trying to block China from ARM/Google,
Taiwan, belongs to China...
Even tough that right now, the US/UK are trying hard to take if back... why trying so fast, with so Anxiety with ONG's over the place secret meetings and such?

Since UK/US/Japan are trying to block China from ARM/Google, China can still block US Companies from access to TSMC..!
On contrary of the past... nor UK nor US can afford a new Militarily War against China today..

So the term 'tout suit',
I think is 100% correct, because things are starting to become 'complicated'( in every direction.. ) on that side of the Globe...