Who cares what software it runs, the interesting thing here is the memory: http://en.wikipedia.org/wiki/Memristor. If they manage to bring this to market, it'll really turn around their reputation as a company that's lost its R&D roots and become just another clone manufacturer. That's the bigger strategic move here: creating product differentiation using technology your competitors don't have.
I care if it runs a Unix-like operating system. As an engineer, my life is made so much easier when standard Unix stuff is available. And I'm rather miserable when I have to rewrite something to account for the absence of standard Unix stuff.
For end-users, a Unix-like system can be perfectly usable. OS X runs on Darwin, which has slightly less Unix support than I'd prefer, but still waaaay more so than Windows. Ubuntu is, in my opinion, just as easy for the non-technical to learn as Windows or OS X.
Given the benefits to developers, and the absence of a downside for end-users, I see no reason not to root for Unix on The Machine.
Now, it doesn't have to use the Linux kernel, GNU utilities, or any of the other familiar building blocks. It just has to act like systems that do.
Memristor's aren't the only interesting thing--they also said it would use new optical interconnect techology, which would contribute much of the speed and energy efficiency improvements.
I don't think this is really a plan to destroy Windows. Sounds more like they're just working on the open source OSs because it's easier to do while the Machine is still largely theoretical.
If it does ever come to market, I'm sure they'll end up working with Microsoft in some form to create something together. But that requires all sorts of legal business agreements, which you can completely avoid in the early stages by working on own-brand or open-source systems.
Are these going to be expensive server computers? I would think so for the first several years. They must be high-margin machines for HP to throw money at it.
Anyway, what kind of a market would there be for Windows in this market? There's a Windows supercomputer OS but it don't think it's used much, for example.
Well if it's as revolutionary as the hype suggests, they could be looking at all market segments. Given that they're working on an Android version, they must at the very least be thinking of its possibilities on mobile.
But you're right on the the server front. If that's the way they're going, then I doubt Windows will get much of a look-in (unless MS make it themselves).
EDIT:
I guess a lot of this boils down to whether they'll be going with a standard open system like exists in the current PC market. Or whether they'll be locking the hardware like Apple. If it's the former, then anyone can make an OS to fit.
I don't want a "brand new operating system based on Linux". I just want it to be Linux.
When companies say stuff like this it usually just means their gonna help themselves to all of the great free stuff everyone contributed and then make their machine a thorn in the community's side with all kinds of binary blob drivers that only work with one kernel/release.
From a familiarity point of view so do I but progress in operating systems and languages hasn't moved on much for decades. Linux is the pragmatic solution for now and for some time to come but if my kids' kids are still using vim on a unix-like system to create software in a c-like language for an Intel compatible CPU architecture I am happy to entertain the possibility that we may have fucked up somewhere.
We badly need a Xerox Parc again doing stuff because they can and not because they must. So much technology came out of places like that and Bell Labs and we seem to have lost that.
Linux is a member of a long lineage of operating systems that assume centralized processors connected by a slow bus to volatile memory that only temporarily holds information that needs to be backed up by other devices connected to buses much slower than memory. Every OS in current use is based on these ideas that haven't changed much since the 60's.
If you want, you may imagine Smalltalk running directly on metal with no OS underneath it, with all your data being just part of the image you work with. This is the kind of machine that this makes possible. This was, of course, done in the 80's with technology that was barely sufficient.
I think HP has a bigger vision here. They aren't looking to build a new car, only this car has electric batteries, they want to build a plane. They announced two OS projects: a modified version of Linux that runs on this new machine, and an entirely new operating system based on fundamentally different assumptions. They only care about Linux as a stopgap, because it allows people to start using this machine while they come up with something better.
HP-UX? Hps unix variant. It had some good features and some odd issues. One of my least favorite OSs.
HPUX did have an add-on (Real Time Extensions?) to allow you to disable processor interrupts on select processors providing almost realtime mainframe(superdome) computing (some system calls were verboten in this mode). Of course a crash would mean a reboot. Grouping processes to processor sets (psrsets) was kind of neat.
So they did try and its not like they didn't have experience with OS development before. Starting with Linux is probably a better bet for them.
HPUX is light years ahead of Linux on workload management, seriously. HP have an amazing portfolio of OSs, HPUX, OpenVMS, Domain/OS, NonStop, etc. Linux is the lowest common denominator, in 2014 it's barely where these were in 1994...
I'd argue that's the problem with Linux, its trying very much to be all things, to all people, rather than a really good desktop system, a really good server system or a really good mobile system.
I've never used a piece of software from HP that I've thought was of any quality. We use several of their enterprise development tools where I work, due to an ancient contract and some political contrivances, and nobody is happy with them.
And don't even get me started on their printer ride-alongs. It's just a printer, HP. I don't need 10 programs to use it.
I use an HP EliteBook 8570w every day. It's an excellent developer box. It's very fast, reliable, and easy to upgrade with off the shelf components and it's a very good value. Even with 24GB of RAM and 500GB SSD drive it's cheaper than a comparable Macbook and I can add another drive and increase the memory if I want; I also made these upgrades as I needed them. It's heavy and not as attractive as a Macbook or other aluminum framed notebooks (HP has those too though).
HP's claim isn't rediculous if you consider that they're not developing memristor tech to replace flash storage. They want to unify mass storage drives with DRAM. You do need an O.S. that can be partially rewritten, because the system can remain in the state where you switched it off. Richard Stanley Williams is the name to Google if you want insight into their roadmap.
I think it is more of a dig on academic operating systems research, which some say hasn't made any major breakthroughs since the 1980's, possibly because we have been stuck on the same architecture for so long.
IMO, the focus on Windows vs UNIX/Linux or HP vs MS is really missing the point though. If you build a computer as they are suggesting, using non-volatile memory without a deep memory hierarchy, and with an optical interconnect that enables low-latency memory access across a datacenter, then that changes a huge number of the assumptions that led to the operating systems we have now. In order to really take advantage of such a machine, you really need to invent new operating systems. You can hack Linux to run on it, which is what they said they are doing, but you will lose some of the benefits of the hardware because the architecture assumptions are baked into the fundamental design of existing OSes. File systems are one example that is easy to think about. Virtual memory. But consider anything that an OS does and how this new architecture changes it. Consider loading executables to start a process, for example. If there is no difference between memory and disk, then you don't even need to load executables into memory. All files would be memory-mapped in some huge global address space. Programs could jump from memory in one machine to a whole different machine (VM migration? HA?). It really changes everything. We would need new R&D, some of it very fundamental, in order to understand and build systems to take full advantage of such technology. That's all HP is saying in this announcement.
Oh, the old comparison of Windows NT and VMS. I never thought that was fair by a long shot. Are there some actual real evidence of NT/VMS similarities other than some of the original team members?
EDIT: So you "really don't like Microsoft much". Maybe that's all I need to hear.
I was going to point out some, but mhurron did a much better job. One of the odd things I remember is that VMS had a version of Diskeeper that predates the NT version.
I don't like Microsoft, but I have no issue in recognizing their technical merits when they have some. I'm using a Microsoft keyboard right now and was, for a very long time, an MCSD. What people misinterpret as an anti-Microsoft bias is, usually, the effect of a wider perspective than the average Microsoft enthusiast.
Windows, like Linux, assumes a certain architecture: one or more processors connected to volatile memory that holds transient data and much slower persistent storage that is required to preserve a consistent state for long periods.
You don't see "Windows for D-Wave computers" and you won't see "Windows for HP's The Machine".
Any new OS will have to, at least, have a POSIX API if it wants to be used from day one, but, from that point, HP is free to invent.
This is link bait. HP is just doing what we used to do until the 90's, making an OS tied to the hardware to optimize things. They have done this before.
Banks are still heavily run off of mainframes and IBM Power servers that aren't standard x86 running Windows or GNU/Linux.
This isn't a plan to destroy Windows, I expect most of the market that would be interested in "the machine" is currently running a mainframe or some unix variant/clone right now.
Seems like a pretty sensational article. Using open-source technology is not an attempt to "destroy Windows," and I wouldn't classify memristors as "new technology," since they've been around since 2008 [1].
I bet they've been waiting to do this for quite a while. Why? Because this is exactly what Microsoft did to them with Surface, two years ago. HP clearly hasn't forgiven Microsoft for that stunt, and since then it has only tried to strengthen the relationship with Google and put more ChromeOS/Android devices on the market.
Look at Lenovo which is actually innovating instead of resting on its laurels like HP and churning out the same cookie cutter crap year after year and expecting profit to stay up. The Surface is a red herring, actual innovation is still rewarded, for example see the Lenovo Yoga line of laptops.
The head of HP Labs explained this in his talk. He said it is called "the machine" because 1) they are researchers, not marketing people, and 2) they started thinking of it as not just another server, but a whole new architecture that scales from phones to data centers (as in, THE machine, the only one you need).
Competition is good, if its not HP it will be someone else but they are at least going in the right direction. We have to break free from Windows and move to something much much better.
How would you rebuild Windows into something that runs on a machine where there is no volatile memory and no distinction between memory and persistent storage?
Whatever results from this would only be called Windows for marketing reasons.
Wait, the same HP that tossed aside PalmOS? The same HP that gave us HP-UX? Wait stop, I am laughing too much, the coffee shot out of my nose.
Glad to see you pushing the memristor, but really, have any of the board actually, like, you know, used an HP laptop recently? Mine waa awesome as a backup oven, the overheating at last useful, but otherwise?
