Multiple Personality

  • Comments posted to this topic are about the item Multiple Personality

  • Generally no chip can be all things to all people, but if vendors could be bothered providing their software on alternative architectures that would be cool. (only if they bothered making all the special optimizations though and not just a straight port)

    Maybe I want a DB that doesn't need to scale too much but needs some heavy crypto lifting, that would probably come out pretty well on the new SUN T2 if it was written to take advantage of all the on chip optimized crypto. But code quality is bad enough these days without trying to produce different versions for even the top 5 chip architectures.

    What would be better (for the customer at any rate) is if the concept of ODBC and SQL standards were more pervasive, complete and performant in all vendor products so that I could choose the DB software and chip architecture that suits the project I'm working on and know that the SQL I write and the wire protocol I communicate with server on could be pointed at a different box with different software and chip and just work. The performance profile might be different obviously but I'm sick of software that will only run on Oracle or SQL server, I want it to run on what is best for me and the skills I have available to deploy, not the skills of the vendor who wrote the application that I want.

    What makes you think IA64 version is going to disappear soon??

  • I think it would be a waste of time and resources to port Windows to other chipsets but, more importantly, it would probably lead to a "dumbing down" of the underlying OS code to the lowest common denominator.

    For example, one of the core reasons that Windows crashes is poorly written device drivers etc. that have to run in kernel mode and can therefore clobber the kernel memory space (this is why MS introduced the device driver signing programme - to avoid other people's poorly written code bringing down Windows and making everyone believe that Windows itself is unstable). There is a way to completely avoid this on x86/x64 architectures by making all device drivers and other non-OS code run in a different processor mode. So then you would have the OS in one mode, device drivers etc. in another and, lastly, user programs in yet another.

    This would likely eliminate the dreaded "blue screen of death" once and for all.

    So why isn't this the case? Because the Alpha chip only had two processor modes.

    So, we still have to put up with crashes that could otherwise be seamlessly handled because of the need to support a now defunct processor architecture.

    No, I think this kind of thing ends up impoverishing or overcomplicating the OS.

    Malcolm
    DB Ghost - Build, compare and synchronize from source control = Database Change Management for SQL Server
    www.dbghost.com

  • Totally agree - Windows should stay where it is, on the x86 chipset. I wouldn't mind seeing SQL Server ported to Linux though.

    CDP.

    Chris.

  • While I agree that competition is good for the market, however I also feel that The windows OS should stay where it is. From a market perspective I can't see MS realizing any profit in broadening their offering to additional chip sets, as I think it could hurt their placement.

    Think of the current incompatibilities of the just the various versions of SQL server. See http://www.sqlservercentral.com/Forums/Topic463445-146-2.aspx for more 64-bit Gotchas as the forum post was titled... If they can't even get all of the peripheral parts of SQL Server up to 64bit do we really expect them to be able to put out a complete version for powerPC or anything else?

    Let them fix what they've got first, the additional overhead of developers it would take to roll out all of their apps for the various chipsets would be immense. Then start thinking about all of the other smaller software houses out there kicking out custom apps for this and that. They'd have to begin to write not just for windows or X now they have to write for Windows on x86, x64, ia64, sparc, powerPC? Perhaps that would be good and something like silverlight/java might work well in that case, but... I just don't see it as a realistic occurrence.

    -Luke.

    To help us help you read this[/url]For better help with performance problems please read this[/url]

  • I agree that competition is a good thing. I'll chime in with some others that having linux compete with Windows helps keep Microsoft in line.

    On the other hand, limiting hardware can allow software to reach higher potentials. Take the game console industry - when you can choose (and hence limit) the hardware for graphics processing, IO, etc., you can tune everything for fantastic performance. There is actually discussion about whether PC gaming will go away someday, because it's getting stupid to purchase a $2000 game pc when you can purchase a $500 desktop for business and a $400 box for gaming. Limiting windows to intel and intel compatable copies would allow the system to be tuned for better and better performance.

    So, my vote is: Stay right where we are. The only reason for considering a move would be instability at Intel. If the whole thing were based on AMD, I'd be considering my options right now.

    Considering a release to Linux: The question is - Does Microsoft think competition is good for them? Right now, the reason I don't buy Linux is that it doesn't run as much software as Windows does (It's also not as dependable as everyone makes it out to be; a story for another day). The software that it has is generally of lower quality. If Microsoft takes the lead and starts providing software for Linux, then they're shooting themselves in the foot by giving up a key advantage over other operating systems.

    Forgive me for being a bit of a Pollyanna, but right now the balance appears just about right. Microsoft has enough "little dogs" barking at its heels to keep Microsoft in line. If they get too lazy, the competition could catch up. Microsoft if forced to press on for better quality and features. The only problem? Expense. So, if someone in a third world country wants to get Ubuntu for free, they can get it. Pretty nice.

    ___________________________________________________
    “Politicians are like diapers. They both need changing regularly and for the same reason.”

  • Why port Windows to another chipset? They just have to create a virtual machine that can install and run windows.

  • Virtual machines would kill the performance too much. It would work for testing, but not in a production environment.

    Even if MS ported this stuff to other chipsets I doubt anyone would buy it. Most the little guys sell systems that cost significantly less than an AMD/Intel system. If I am going budget server, I am not going to waste money on software licenses that I would need to renew or maintain. Just get a linux distro, install Apache, MySQL and phpMyAdmin and call it a day.

    Side note: I would hardly call AMD an Intel "imitator." I would argue (along with most gamers and small business owners) that AMD builds a superior product and sells if for significantly less than Intel. Plus the two companies have cross-licensing agreements allowing each partner to use the other partner's patented technology without charge after a certain amount of time. Innovation is a two-way street.

  • I am thrilled that Windows is staying put on the intel chip set. I would hate to see such a badly design/patch-up bloated ware being ported to other capable platform.

    rant mode on

    And please, do not get me started out monopolizing the market!!! MS is the master at being a monopoly and it is playing the victim card(with respect to the internet searching ) because MS themself has been unsuccessful in trying to tie people in to use their own offering. I would bet anything that should today MS is the #1 in internet search; they would claim that it is because of their forthsight/business knowhow/innovation/IP that make it happen and not because of their monopolized tie-in practices.

    rant mode off:)

  • Do you remember the days when NT supported more than one chipset? It used to run on Intel IA-32, MIPS, Alpha, PowerPC, SPARC, Intel i860, and Intel i960 (http://en.wikipedia.org/wiki/Windows_NT).

  • mwacht, are you talking about SQL Server, or Windows in general?

    Virtualization works just fine for most applications. I don't know about SQL specifically (and I don't want any of my production SQL servers running in a VM), but that statement does not apply in the context of the article (Windows running on other architectures).

    VMWare for one has a great enterprise virtualization product that performs well for most apps - you just have to make sure you aren;t runing the edge case.

  • ChrisMoix (4/4/2008)


    Do you remember the days when NT supported more than one chipset? It used to run on Intel IA-32, MIPS, Alpha, PowerPC, SPARC, Intel i860, and Intel i960 (http://en.wikipedia.org/wiki/Windows_NT).

    Yes, I do. I still have the installation CDs to prove it. 🙂 But the comments remain: pig does not fly(ok. may in a far distance future where we can inject pigeon DNA into the pig to them to grow wings and plus we can increase its IQ so that they would learn to fly with the new wings) 😛

    Seriously, a port of SQL itself to a different platform makes more sense than the entire windows os.

  • Or do we just concede to Intel and it's imitator AMD?

    I would like to ask Steve who made the x64 instruction set that everything is moving to in the PC world? Intel decided to go with a nice RISC instruction set with IA64, but AMD created a x86 compatible

    upgrade

    with the x64 instruction set.

    That being said, the main thing to keep in mind is that Microsoft is a company, and companies go for profit. Running Windows on more instruction sets means increased cost for questionable returns, so it would make more sense for the to go with what they know. I think more options would be better, as it would increase competition amongst CPU vendors, but I just don't think MS has any reason to do that, and if I were them, I wouldn't either.

    Eric

  • :unsure: Looks at article again...

    My bad--yea I was thinking about SQL Server running on a virtual Windows install, running on another platform.

    As for Windows OS on a VM--I agree it makes is a waste to port Windows itself to other chipset.

  • DEC ported VMS to the Alpha and that used 4 "modes" (kernel, exec, super and user) so implementing 4 "rings" on Alpha for NT shouldn't have been a problem. I suspect the MIPS was the problem child that only had prinvileged and unprivileged CPU modes.

    In any event, while I dislike monopolies, I can't see the dominance of the Intel/AMD architecture being challenged in the near future, so there's little point in porting Windows (or SQL server) to another platform.

    Of course, there is some (small) case for porting SQLserver to run on Linux (for the server market) and then maybe to other hardware, but Windows? No chance!

    Derek

Viewing 15 posts - 1 through 15 (of 20 total)

You must be logged in to reply to this topic. Login to reply