IBM is Moving OpenPower Foundation To The Linux Foundation (techcrunch.com) 46
IBM makes the Power Series chips, and as part of that has open sourced some of the underlying technologies to encourage wider use of these chips. The open source pieces have been part of the OpenPower Foundation. Today, the company announced it was moving the foundation under The Linux Foundation, and while it was at it, announced it was open sourcing several other important bits. From a report: Ken King, general manager for OpenPower at IBM, says that at this point in his organization's evolution, they wanted to move it under the auspices of the Linux Foundation. But IBM didn't stop there. It also announced that it was open sourcing some of the technical underpinnings of the Power Series chip to make it easier for developers and engineers to build on top of the technology. Perhaps most importantly, the company is open sourcing the Power Instruction Set Architecture (ISA). These are "the definitions developers use for ensuring hardware and software work together on Power," the company explained.
King sees open sourcing this technology as an important step for a number of reasons around licensing and governance. "The first thing is that we are taking the ability to be able to implement what we're licensing, the ISA instruction set architecture, for others to be able to implement on top of that instruction set royalty free with patent rights," he explained. The company is also putting this under an open governance workgroup at the OpenPower Foundation.
King sees open sourcing this technology as an important step for a number of reasons around licensing and governance. "The first thing is that we are taking the ability to be able to implement what we're licensing, the ISA instruction set architecture, for others to be able to implement on top of that instruction set royalty free with patent rights," he explained. The company is also putting this under an open governance workgroup at the OpenPower Foundation.
Is that wise? (Score:1)
People tend to remember more recently that Microsoft acquiring GitHub.
Re:Is that wise? (Score:4, Insightful)
Does IBM remember that Microsoft acquired The Linux Foundation in 2016 ?
No, probably because they didn't acquire the Linux Foundation.
Re: (Score:1, Offtopic)
Re: Is that wise? (Score:2)
Re: (Score:2)
Once the embrace step has been initiated, it's safe to assume that the other two will follow. At least, that's my interpretation of his post.
Re: (Score:2)
Re: (Score:1)
https://www.linuxfoundation.or... [linuxfoundation.org]
Look at all these acquirers! It is a wonder there is any Linux Foundation left!
Re: (Score:2)
While the original declaration that MS 'bought' the foundation is overstated, the volume of 'members' is symptomatic of the reality of the modern Linux foundation: it's mostly a place for corporations to use as they see fit for branding purposes.
Being a Linux Foundation member or project doesn't speak much to the relative importance or even necessarily committing to purely open source. It is simply a statement of how much money some corporate interest was willing to spend to promote it.
It's basically a mea
RISC-V is moving mountains? (Score:4, Interesting)
Wasn't the POWER arch a crown jewel of IBM mainframe dominance for the least 25+ years? Apple used a reduced version of the POWER3 and POWER 4 instruction set in their G3 and G4 PPC processors (and also the short-lived G5) for many years before switching to Intel in '06/'07.
Interesting to see RISC-V gaining traction and suddenly IBM is open sourcing POWER out of the blue (no pun intended)
Re: (Score:2)
No,POWER has nothing to do with mainframes, and never did.
Re: (Score:3)
No,POWER has nothing to do with mainframes, and never did.
Completely wrong. IBM's current mainframe CPU is based on POWER. Different layout, different instruction set, but same functional units behind the scenes.
Re: (Score:2)
You have no idea what you are talking about. The z14 processor https://en.wikipedia.org/wiki/... [wikipedia.org] is NOT POWER. And neither were any of the previous generations. I don't know where you get your supposed information, but it is totally incorrect.
'Functional units' do not describe POWER (or mainframe). They may make up a particular implementation, but they are NOT the architecture.
Re: (Score:2)
You have no idea what you are talking about. The z14 processor https://en.wikipedia.org/wiki/ [wikipedia.org]... is NOT POWER. And neither were any of the previous generations.
I didn't say it was. Learn to read. I said it was based on POWER.
'Functional units' do not describe POWER (or mainframe). They may make up a particular implementation, but they are NOT the architecture.
The architecture is both the functional units AND how they are glued together. But learn to read anyway.
Re: (Score:2)
and processors are based on logic, so what? It is possible to leverage expertise into multiple otherwise unrelated projects. Just accept that you are wrong and move on.
POWER was IBM's workstation RISC architecture, not the "crown jewel" of their mainframe architecture for the last 25+ years. Basic reading comprehension would tell you how wrong that assertion is, yet you've chosen to get on the wrong side of the argument and then argue pedantry and hurl personal insults. Sad.
Re: (Score:1)
and processors are based on logic, so what? It is possible to leverage expertise into multiple otherwise unrelated projects. Just accept that you are wrong and move on.
If I were wrong, I would admit it and move on. But I'm not. The Z series is made from bits and pieces of POWER, glued together differently. Just accept that you know fuck-all and fuck off.
Re: (Score:2)
Provide any credible source for your statement.
Do IBM designed POWER and Z processor CHIPs share some design? Certainly. Does that mean Z is 'bits of POWER glued together'? Hell no. Why not just claim POWER is bits of 4004 'glued together', cause you know, AND gates. There is not one shred of POWER ARCHITECTURE in a Z microprocessor. If you are going to claim otherwise, prove it or shut up. I'll make it easy for you: here is the POWER architecture spec - point out the pieces you will find in a Z proc
Re: (Score:2)
So, you can build a CISC processor (Z14), from bits and pieces of a RISC processor (POWER). Got it.
Re: (Score:2)
"So, you can build a CISC processor (Z14), from bits and pieces of a RISC processor (POWER). Got it."
Literally all x86 processors since the AMD K5 and the Intel P54c have been internally RISCy. They have an x86 decoder which decomposes x86 instructions into architecture micro-operations. And since IBM created POWER, all of their chips have been based on it. AS/400 used POWER chips with a hardware decoder to make them compatible with legacy processors. The earlier Z-series processors were based on POWER so i
Re: (Score:2)
Constantly repeating the same nonsense over and over does not make it true. No Z architecture processor has ever been 'based on POWER'. None. Zero. Nada. How do I know? I have been in mainframe engineering at IBM for almost 40 years. Yes, a given Z processor and a given POWER processor may share a common FPU (for instance). That does not mean the Z processor is 'based on POWER' any more than the POWER chip is 'based on Z'. It means that both chips are 'based on a common FPU'. And that common FPU ha
Re: (Score:2)
Constantly repeating the same nonsense over and over does not make it true. No Z architecture processor has ever been 'based on POWER'.
Here is your citation: [realworldtech.com]
Now I don't actually believe that you ever worked in mainframe engineering, let alone on the Z/architecture, since you clearly do not know what you are talking about. Shit, did you even work for IBM?
Yes, a given Z processor and a given POWER processor may share a common FPU (for instance).
I am not talking about coprocessors, I'm talking about the basic functional units that ma
Re: (Score:2)
Can you fucking read? It was a COUSIN to POWER, not 'based on', and it SHARED BUILDING BLOCKS which is exactly what I said. And the FPU is not a 'coprocessor', it IS one of the 'basic building blocks', along with the integer unit, DFPU, Instruction Fetch, and some other stuff.
Re: (Score:2)
POWER is what is bringing in the big bucks, because they sell assloads of them. Z system sales have trended downwards fairly consistently (save for literally a couple of good years) since the nineties. Since POWER is where the money is coming from, POWER is what's driving the direction. POWER took a hit in profit this year... due to investment in the next generation of POWER.
If you don't get that POWER is driving development at IBM, you don't get anything at all. Every other IBM architecture is either liter
Re: (Score:2)
You are completely wrong.
Re: (Score:2)
Re: (Score:3)
POWER chips run the p-series (AIX) and i-series (OS400, now called simply 'i') minicomputers, not the mainframes. There's also a POWER-based workstation called Talos-II
https://www.raptorcs.com/TALOSII/ [raptorcs.com]
IBM would *NEVER* open-source mainframe processors.
Re: (Score:3)
Just lookie at the prices: regular workstation "Starting at $6,064.00" (for a bare-bones system), while 4-core dev board starts at $2,924.99.
So the cheapest dev board costs more than the whole new 64-way x86 box I recently bought. Or, 100 times more than some desktop-capable 64-bit Allwinners you can use for basic porting/testing tasks.
No wonders no one uses the architecture.
Re: (Score:3)
No wonders no one uses the architecture.
Almost every large company uses the architecture.
Re: (Score:2)
Economy of scale.
Re: (Score:3)
That 4 core is a full system as well. There's CPU+MB bundles listed beneath it and individual parts below that. I do admit the prices for everything but the processor seem a bit inflated, for example charging $17.50 for the tool you need to affix the heatsink is a bit obnoxious.
POWER is overkill for a desktop, but it's remarkably space and energy efficient for the sheer volume of virtual guests you can comfortably shove into it compared to trying to get the same out of x86 server hardware. For a higher u
Re: (Score:2)
No wonders no one uses the architecture.
I still use my Xbox 360, FWIW.
Re: (Score:2)
There's lots of them in cars!
https://en.wikipedia.org/wiki/... [wikipedia.org]
IBM would NEVER open-source mainframe processors? (Score:1)
IBM would *NEVER* open-source mainframe processors.
In the 1970s IBM had anti-trust problems and unbundled the mainframe OS from the machine architecture. This was because a company named Amdahl was making Instruction-set compatible mainframes and had complained because the OS was bundled with the hardware. This ruling let customers purchase an Amdahl mainframe that worked just like an IBM mainframe (only better in some estimates) and run IBM's OS on it.
So IBM has probably learned and would not open-source the mainframe architecture today, but it did happen
Re: (Score:2)
They didn't open-source the mainframe architecture - the Amdahl machines were clean-room re-implementations of the IBM architecture. IBM's response to being forced to unbundle the OS was to start requiring microcode changes for OS updates. Part of the reason the Amdahl clones performed better was that they used hard-coded functionality rather than microcoded implementations. IBM could make trivial microcode changes that would require new hardware for Amdahl to follow.
Re: (Score:3)
Wasn't the POWER arch a crown jewel of IBM mainframe dominance for the least 25+ years?
Sort of. RS6k and AS/400 were POWER-based, while the mainframe CPU was based on POWER's internals, but glued together very differently.
Apple used a reduced version of the POWER3 and POWER 4 instruction set in their G3 and G4 PPC processors (and also the short-lived G5) for many years before switching to Intel in '06/'07.
All PowerPC processors are based on a POWER processor. The PPC601 (original PowerPC) actually implemented the entire POWER instruction set at the time. All other PowerPC processors implemented only a subset.
Re: (Score:2)
PowerPC isn't really a subset of POWER, it's a slightly-incompatible evolution. The most obvious change is in multiply/divide instructions (they use GPRs only on PPC, while POWER has a special multiply/divide register called MQ, like the MIPS and SPARC did), and removal of ALU operations that require selection logic to save die space (mostly saturating maths). PPC also has a different MMU. The 601 is a POWER chip with additional instructions implemented to support user-mode PPC code, while still implemen
Re: (Score:3)
The G3 and G4 (and G5) were not Apple processors, they were processors Apple used that were designed by IBM (or decedents of IBM designs). Furthermore, these processors were not "reduced versions" of POWER instruction sets, they were PowerPC processors, and Apple was merely a consumer of them, a move that proved unwise.
Re: (Score:2)
The G3 and G4 (and G5) were not Apple processors, they were processors Apple used that were designed by IBM (or decedents of IBM designs).
Apple collaborated on the design of the G4 and G5 (maybe G3 too?) though they're only a design partner, and not designers themselves. They're Apple-flavored
Re: (Score:2)
"AIX + i5/OS go against IBM's current strategy"
They'd be foolish not to offer those options in their current strategy.
IME, AS400/iSeries customers tend to be fiercely loyal - with good reason, and wouldn't be impressed if it was replaced with some vague promises of "cloud".
Re: (Score:1)
RISC-V? Bzzt. No. You're looking for a processor called "Rome". RISC-V has no foothold in the markets served by POWER9 in particular. Rome, on the other hand . . .
Re: (Score:3)
RISC-V isn't much of a threat to POWER. RISC-V is gaining popularity where previously people would have used ARM, i.e. embedded as part of other chips or where they need a low power CPU that doesn't have to have great performance.
To scale up to anything like the level of performance that POWER offers would require a massive amount of development work RISC-V. It's not going to happen.
Even ARM, where there are people working on getting the most performance from it, doesn't really compete with POWER.
Take a look (Score:2)
at the Top 500 list in case you haven't recently https://www.top500.org/lists/2... [top500.org]
Re: (Score:2)
Of course, the list shows that 15 out of 500 are IBM.
Six of those are actually based on currently available POWER platform (a lot of leftover BlueGene systems and a couple of 5 year old Intel based systems.
Of those six, all of them are more about the NVIDIA part than the IBM part. The processors in those installations are just facilitating moving data between the GPU space and the traditional compute space.
The POWER architecture is in trouble in general. IBM has gone through several rounds of trying to ma