HP To Launch Linux++ Operating System Next Year

[coder]

2. The IA-64 didn't fail due to patent issues. It failed due to 3 different reasons:
A. It was late and uber expensive from day one.
B. It had a *very* complex instruction set (VLIW) that left all the heavy lifting to the compilers which in-turn, never got it right. (Intel own ICC was late, and underperforming).
C. Intel planned to kill the x86 line by keeping it in 32bits, forcing the market to far-more-expensive IA64. AMD released the Athlon64, forcing Intel to add 64bit support to the x86_64 line, effectively killing the IA64.

Initial delays and pricing didn't kill it, because that chip was only targeted for the server market, anyhow. The problem was that it never gained traction, in that market, so the never had the volume to support lower pricing. Its poor x86 emulation performance didn't help matters.

Performance wise, it actually was competitive. It took a couple generations for its potential to show, but by then the x86 chips had a process node (and therefore clock speed) advantage. Intel couldn't justify using their latest fab when it meant detracting from x86, which was their bread and butter, while Itanium adoption wasn't happening. But if you go back and compare IA64 performance to x86 chips that were built on the same process node, it's definitely competitive.

I consider AMD64 to be more of a footnote. If that were Itanium's biggest problem, that wouldn't have stopped it. But that just shows how important it is that you have competition (or at least the potential) within a given ISA. In a way, you really made my point for me. The big customers understood this, slowing adoption and the whole thing never really got off the ground. IMO, it's little coincidence that ARM, one of the most open ISAs, is now ascendant. Nobody wants to be locked into an ISA that can only be built by one chip maker.

And to bring this post back on topic, the same will happen for any proprietary computing technologies HP might introduce. Best case scenario for them is probably that Oracle supports it and they recapture some piece of the data center market, for a while.

[gilboa]

Initial delays and pricing didn't kill it, because that chip was only targeted for the server market, anyhow. The problem was that it never gained traction, in that market, so the never had the volume to support lower pricing. Its poor x86 emulation performance didn't help matters.

Performance wise, it actually was competitive. It took a couple generations for its potential to show, but by then the x86 chips had a process node (and therefore clock speed) advantage. Intel couldn't justify using their latest fab when it meant detracting from x86, which was their bread and butter, while Itanium adoption wasn't happening. But if you go back and compare IA64 performance to x86 chips that were built on the same process node, it's definitely competitive.

I consider AMD64 to be more of a footnote. If that were Itanium's biggest problem, that wouldn't have stopped it. But that just shows how important it is that you have competition (or at least the potential) within a given ISA. In a way, you really made my point for me. The big customers understood this, slowing adoption and the whole thing never really got off the ground. IMO, it's little coincidence that ARM, one of the most open ISAs, is now ascendant. Nobody wants to be locked into an ISA that can only be built by one chip maker.

And to bring this post back on topic, the same will happen for any proprietary computing technologies HP might introduce. Best case scenario for them is probably that Oracle supports it and they recapture some piece of the data center market, for a while.

I beg to differ IA64 was supposed to *replace* the x86 Xeon line, covering high-end servers, mid-end servers and high-end workstations (I actually had one).
Keep in mind that at the time, 64bit was limited to high-end RISC chips, and Intel never planned to add it to the soon-to-be-released Pentium 4 CPU (Xeon family included).
It was also a meant to finally kill both AMD and VIA, as both could no longer produce IA64 compatible chips (both are currently protected by licensing agreements from the mid-80s').

However, Intel never managed to get the platform performing up-to-spec. The first 733Mhz Itanium Merced core had severe issues out-performing a 1.7Ghz Pentium 4 Willamette core. In theory, Intel could slowly force the market into adopting IA64 by keeping P4 32bit-only, but the Atlhon64/Opteron left them not choice. (And the lackluster performance of the McKinley/Madison cores didn't really improve things)
Sure, by 2006 the performance was acceptable, but by then both the Xeon and the Opteron shoved it into a very small market segment (Super-high-end servers) and most of the Itanium server manufacturers left the architecture.

BTW, comparing the IA64 to "the machine" is invalid (hence I disagree with the statement that the IA64 failed because it was closed).
The Itanium was never limited to HP (we had a two Dell machines) - most of the big names had Itanium based servers. It had support for all the main OSs (Windows NT - 2K8, Linux, *Unix, FreeBSD and VMS). Unlike "the machine", the IA64 was an open architecture.

Having said all that, I do agree that even if the machine redefines the computing world, unless they plan to turn it into an open architecture, it will not gain traction beyond the super-high-end-server-market.

- Gilboa

[coder]

I beg to differ IA64 was supposed to *replace* the x86 Xeon line, covering high-end servers, mid-end servers and high-end workstations (I actually had one).

No disagreement there.

It was also a meant to finally kill both AMD and VIA, as both could no longer produce IA64 compatible chips (both are currently protected by licensing agreements from the mid-80s').

Exactly my point. Now, if you're a big customer, why would you buy a bunch of IA64 systems and software, locking yourself into paying whatever Intel felt like changing for future generations? Many would go so far as to opt for a technically inferior solution, just to avoid that sort of vendor lock-in.

The Itanium was never limited to HP

No, but it was limited to Intel. I was analogizing HP's "the machine" (stupid name) to Intel's IA64.

Unlike "the machine", the IA64 was an open architecture.

Which was so heavily patented that no one could ever clone it without Intel's explicit blessing (and whatever license fees they felt like charging, that particular day).

[gilboa]

No disagreement there.

Exactly my point. Now, if you're a big customer, why would you buy a bunch of IA64 systems and software, locking yourself into paying whatever Intel felt like changing for future generations? Many would go so far as to opt for a technically inferior solution, just to avoid that sort of vendor lock-in.

No, but it was limited to Intel. I was analogizing HP's "the machine" (stupid name) to Intel's IA64.

Which was so heavily patented that no one could ever clone it without Intel's explicit blessing (and whatever license fees they felt like charging, that particular day).

ARM is heavily patented, and it never stopped it from slowly taking over the world.
In general, customer usually don't care who produces the chip inside their computer, as long as the pricing and performance is acceptable and as long as they are not forced to buy it from one source (E.g. IBM PS2).
One you have customers, ISVs will come rushing in (E.g. Android).
IA-64 was late, slow and expensive; customers didn't come so the ISVs decided to skip it.

In general, "The Machine" will have two major issues: 1. proprietary (and closed) hardware 2. propriety OS and software.
Unless they redefined computing as we know it, this will not end well....

- Gilboa

[grndzro]

A fair point (and not the first time I've heard this particular complaint).

Analogies are rarely perfect. This usage of the term is clearly analogizing a development to the discontinuous nature of a quantum leap - not its magnitude. A better way to describe it would be "revolutionary", as opposed to "evolutionary".

Now, when people talk about a "quantum" of something, they mean the smallest discrete unit of it.

A quantum particle being one of the smallest particles when used referring to size means smallest.

However a quantum leap refers to quantum entanglement of quantum particles which transcend space, ultimately meaning huge leap in progress.

[coder]

ARM is heavily patented, and it never stopped it from slowly taking over the world.

First, I don't know to what extent the nature of ARMs patents prevent someone from making an ISA-compatible CPU as Intel's did with IA64. Second, ARM is in the business of licensing its IP. When you make an ARM CPU, you're not competing with ARM - you're a customer!

as long as the pricing and performance is acceptable and as long as they are not forced to buy it from one source (E.g. IBM PS2).

But that's my point. You could only buy IA64 CPUs from Intel, and anyone who ever cloned them would have to pay whatever license fee Intel wanted to charge. So, you're entirely at the mercy of Intel.

Once upon a time, this was just how things were done. A big company bought XYZ mainframes, because there was no standard ISA. Sure, there were some VAX clones, but that was just a precursor of things to come. With x86, the corporate IT market got a taste of the benefits of commoditization and open competition and the understandably didn't want to go back.

One you have customers, ISVs will come rushing in (E.g. Android).

But customers and software is a chicken & egg problem. One of Itanium's only inroads was with Java apps, because they didn't require any porting.

In general, "The Machine" will have two major issues: 1. proprietary (and closed) hardware 2. propriety OS and software.
Unless they redefined computing as we know it, this will not end well....

Totally agree.

[coder]

ARM is heavily patented, and it never stopped it from slowly taking over the world.

First, I don't know to what extent the nature of ARMs patents prevent someone from making an ISA-compatible CPU as Intel's did with IA64. Second, ARM is in the business of licensing its IP. When you make an ARM CPU, you're not competing with ARM - you're just one of their customers!

as long as the pricing and performance is acceptable and as long as they are not forced to buy it from one source (E.g. IBM PS2).

But that's my point. You could only buy IA64 CPUs from Intel, and anyone who ever cloned them would have to pay whatever license fee Intel wanted to charge. So, you're entirely at the mercy of Intel.

Once upon a time, this was just how things were done. A big company bought XYZ mainframes, because there was no standard ISA. Sure, there were some VAX clones, but that was just a precursor of things to come. With x86, the corporate IT market got a taste of the benefits of commoditization and open competition and the understandably didn't want to go back.

One you have customers, ISVs will come rushing in (E.g. Android).

But customers and software is a chicken & egg problem. One of Itanium's only inroads was with Java apps, because they didn't require any porting.

In general, "The Machine" will have two major issues: 1. proprietary (and closed) hardware 2. propriety OS and software.
Unless they redefined computing as we know it, this will not end well....

Totally agree.

[gilboa]

First, I don't know to what extent the nature of ARMs patents prevent someone from making an ISA-compatible CPU as Intel's did with IA64. Second, ARM is in the business of licensing its IP. When you make an ARM CPU, you're not competing with ARM - you're a customer!


But that's my point. You could only buy IA64 CPUs from Intel, and anyone who ever cloned them would have to pay whatever license fee Intel wanted to charge. So, you're entirely at the mercy of Intel.

Once upon a time, this was just how things were done. A big company bought XYZ mainframes, because there was no standard ISA. Sure, there were some VAX clones, but that was just a precursor of things to come. With x86, the corporate IT market got a taste of the benefits of commoditization and open competition and the understandably didn't want to go back.


But customers and software is a chicken & egg problem. One of Itanium's only inroads was with Java apps, because they didn't require any porting.

Totally agree.

To be honest, we are splitting hairs over theory. We will never know if the IA-64 ever had a chance to kick the Xeon and (later) the Opteron out of the server make, had Intel managed to get it out on time and with the promised performance advantage over the x86. Upon release it was slow, expensive and late. In short, brain dead.

Question is, assuming that HP will deliver as promised - do they have a chance at capturing the market.
In my view, it all depends on what Linux++ means.
If Linux++ is a set of (OSS) libraries and kernel modules that can be added to all major existing enterprise Linux', they may have a chance.
If its a closed OS that partially built around Linux, their chances are slim.

- Gilboa

[droste]

Is anybody here actually reading the things in the article and the source? HP made quite clear what Linux++ is and what they want to do with it?

A working prototype of The Machine should be ready by 2016, says Bresniker. However, he wants researchers and programmers to get familiar with how it will work well before then. His team aims to complete an operating system designed for The Machine, called Linux++, in June 2015. Software that emulates the hardware design of The Machine and other tools will be released so that programmers can test their code against the new operating system. Linux++ is intended to ultimately be replaced by an operating system designed from scratch for The Machine, which HP calls Carbon.

It's a modified linux (+distro) to emulate new hardware + their new OS on traditional hardware and they don't want to have success with it. It's just there so programmers (note: not users!) can get familiar with the new architecture and will be replaced by "Carbon" an operating system designed from scratch by HP.
They want to have success with their new hardware/architecture (based on memristors) and Carbon!

[gilboa]

Is anybody here actually reading the things in the article and the source? HP made quite clear what Linux++ is and what they want to do with it?

It's a modified linux (+distro) to emulate new hardware + their new OS on traditional hardware and they don't want to have success with it. It's just there so programmers (note: not users!) can get familiar with the new architecture and will be replaced by "Carbon" an operating system designed from scratch by HP.
They want to have success with their new hardware/architecture (based on memristors) and Carbon!

I beg to differ.
Unless HP can get a *lot* of people using Linux++, developer adoption rate will be *very* low, killing any chance of Carbon OS gaining momentum before its even born.
Let alone the fact that developing a full stack proprietary OS is *very* hard (At the very least, beyond the basics [kernel, user-land] you'll also have to post Java, Python, PHP, C/C++ [glibc?/gcc?] and a huge stack of libraries.).

... And this, assuming that Carbon OS is *really* a full OS written from scratch and not large number of kernel modules + user land libraries glued around Linux / BSD / HP-unix.

- Gilboa