Is 64-bit the key?

coscos
Posted:
in Future Apple Hardware edited January 2014
While I don't have any rumors to report, I would like to pose a question to the Appleinsider community that really hasen't been discussed at length as of yet.



While I don't think that anyone would disagree that the 970 (or similar varient) will almost certinly boost Apple's mind and market share, the introduction of a 64-bit processor does introduce new posibilities that haden't been available to Apple in the past. (I'll explain further as i finish this buildup)



Some have discussed the posibility of incorporating an emulator into OS X so as to be able to run Windows applications, the obvious disadvantage to this is that along with file compatibility comes slow speed.



The question I have been pondering is whether or not IBM could bundle Intel's and/or AMD's 64-bit architectures into the power4's 64 bit archatecture, thus giving native processor compatibility. Assuming that could be done, that would settle the hardware issue.



Secondly, IBM could license code to Apple from OS/2 (which, if you remember, was binary compatible with Windows). If Apple included this code (along with Cocoa Carbon and Java) as one of its native APIs all while retaining an Aqua UI not unlike the way Java applications appear in OS X, then we would have total native compatibility with Windows.



While some might fear that people will give up on writing code in Cocoa, my response would be... "who cares."



If the API was supported natively in both hardware and software (meaning no speed degridation when compared to everyday PCs) what does it matter?



I'd venture to guess that application written with the Cocoa API would probably outperform their x86 counterparts, which of course would give reason for developers to continue using it and thus allowing Apple to maintain control.



It would seem to me that a transition to 64-bit is the key...



Anyways... I just wanted to get your feedback... what does everybody think of this... (and please don't give the, "who wants emulated code" response... because this was already addressed in my philosophy....

Comments

  • Reply 1 of 8
    There is no sign of interest in this kind of a hardware approach. If there were, however, it wouldn't require a 64-bit PowerPC core to do it, the long instruction cracking style pipeline is the more relevant architectural feature -- just attach an x86 decoder and away you go. There are a few issues (like byte ordering and segmentation) but they could probably be resolved. The amount of circuitry to do this wouldn't be huge. The real problem is that it wouldn't result in an excellent x86, and it would negatively effect the price/performance of the PowerPC at least slightly. I just don't see IBM being interested in building such a chip.



    I also have a hard time seeing this idea as being good for Apple. They could probably make the Win32 software run decently on MacOSX, but it would still be Win32 software (i.e. some of the inherent nature of it would show through). And if developers could support both platforms by doing this, then most would and the Mac platform would suffer for it. User would then look at the platforms and see that the same software running on both, but running better under Microsoft's OS and buy that.



    I don't think it'll fly.
  • Reply 2 of 8
    [quote]Originally posted by COS:

    <strong>Secondly, IBM could license code to Apple from OS/2 (which, if you remember, was binary compatible with Windows). If Apple included this code (along with Cocoa Carbon and Java) as one of its native APIs all while retaining an Aqua UI not unlike the way Java applications appear in OS X, then we would have total native compatibility with Windows.</strong><hr></blockquote>



    Two issues:



    a) WinOS/2 was Windows-compatible up to and including Windows 3.11 only, i.e. no Win32 APIs. As such, it would probably be only slightly more useful than a MS-DOS 6.22 compatibility layer.



    b) A lot of people think that WinOS/2 was in fact [i]the[/] predominant reason that OS/2 never really caught on. Because it worked so well, software developers lost any incentive to do native OS/2 applications (after all, people could just run the Windows version in WinOS/2). Customers, on the other hand, didn't really see the point in buying OS/2 just for running Windows apps (since hardly any native OS/2 apps existed), so adoption was low, making it even harder for software developers to justify writing native OS/2 software, and so on.



    Bye,

    RazzFazz
  • Reply 3 of 8
    stoostoo Posts: 1,490member
    I dimly remember a hybrid PowerPC/x86 CPU in the past bein dropped: anyone else have similar recollections?
  • Reply 4 of 8
    outsideroutsider Posts: 6,008member
    Yes , the PowerPC 615 was dropped and it never went beyond prototype stage.
  • Reply 5 of 8
    mr. memr. me Posts: 3,221member
    [quote]Originally posted by COS:

    <strong>....



    Some have discussed the posibility of incorporating an emulator into OS X so as to be able to run Windows applications, the obvious disadvantage to this is that along with file compatibility comes slow speed.



    The question I have been pondering is whether or not IBM could bundle Intel's and/or AMD's 64-bit architectures into the power4's 64 bit archatecture, thus giving native processor compatibility. Assuming that could be done, that would settle the hardware issue.





    ....</strong><hr></blockquote>



    And another thing.... This is a major reason for the failure of IA-64 to replace IA-32. Had Intel and HP reduced HP-PA to a single chip and left the IA-32 emutation to software, Willamette, McKinley, and the other IA-64 chips would be in 90% of computers sold today. Instead, Intel chose to preserve IA-32 in hardware....
  • Reply 6 of 8
    drudru Posts: 43member
    [quote]Originally posted by RazzFazz:

    <strong>Two issues:



    a) WinOS/2 was Windows-compatible up to and including Windows 3.11 only, i.e. no Win32 APIs. As such, it would probably be only slightly more useful than a MS-DOS 6.22 compatibility layer.

    <hr></blockquote></strong>



    Yes and no. IBM went further. They developed the Open32 API set which was a native implementation of the most used Win32 APIs. Lotus ported Smartsuite to OS/2 using them. I think the Netscape port used them as well but I couldn't swear to it--it's been a *long* time. Open source efforts have taken this further allowing OS/2 to run Word and other Windows applications (last I knew this was project Odin and working with the Linux Wine effort to further compatibility). In case those aren't enough ways to run Win32 apps, there's now Virtual PC for OS/2. Anyway... I figure about ohh 0% of this is useful to Apple. I mean, do they really need to support *another* API (Open32, which by now is too dated to keep up PLUS they'd need a software x86/whatever emulator unless the old rumored PowerPC 615 experiment of hardware-based x86 emulation were to emerge; maybe not insane looking at AMD's tech)?



    IBM did develop a Mach-based OS/2 for their PowerPC systems. They took Mach 3.0, churned it into the IBM Microkernel 1.0 (I don't remember 2.0 surfacing) research project and spat back some (IIRC) pipe and memory enhancements that were rolled into Mach 4.0. This version too ran DOS & Windows via a built-in x86 emulator.



    [quote]<strong>b) A lot of people think that WinOS/2 was in fact [i]the[/] predominant reason that OS/2 never really caught on. Because it worked so well, software developers lost any incentive to do native OS/2 applications (after all, people could just run the Windows version in WinOS/2). Customers, on the other hand, didn't really see the point in buying OS/2 just for running Windows apps (since hardly any native OS/2 apps existed), so adoption was low, making it even harder for software developers to justify writing native OS/2 software, and so on.</strong><hr></blockquote>



    It was a contributing factor. Perhaps worse was OS/2 being locked out of preloads in a serious fashion. Apps did exist. StarOffice, DeScribe Word Processor and other apps were born on OS/2 but it was more the exception than the rule.



    IBM had a hard time convincing developers to support OS/2 even when it went so far as to pay for development. A number (Lotus, Borland, WordPerfect, etc.) all followed Microsoft (shipped Word & Excel) into OS/2 1.x app development and had a bitter taste left in their mouth when it was abandoned for Windows. They were left high and dry needing to play catchup in something of a 'bait-and-switch'.



    Even with IBM throwing money at ISVs, applications simply weren't materializing. Framemaker was scraped in late beta. QuickTime never got out of beta either. Corel kept making promises to bring their graphics products to parity with Windows offerings but those promises were redefined until, eventually, it was irrelevent. Lotus, despite being OWNED by IBM, refused to use OpenDoc and took years to finally ship SmartSuite.



    Where are the companies that followed around the bully Microsoft all those years? Corel? Borland? For those still around, what kind of market share to they enjoy? If they'd told Microsoft "no" and supported IBM's OS/2 they may well have been in better shape. Instead they fell into a trap they couldn't escape as Microsoft benefited not only from an endless supply of cash from its PC tax (per CPU "licensing" that meant you paid even if you didn't actually ship DOS on your computer), as well as API control, "insider" knowledge of Windows and so on.



    Imagine how bad Apple would have it if Classic had been developed by a rival and OS X had been developed by someone else and both had to compete? They're having hard enough time getting users to migrate but had to produce Carbon to get ISVs to even play ball period.



    Most Carbon ports, Apple's included, are pathetic (less so than OS/2 "Mirrors" ports from Windows 3.x which comprised far too many apps) but it's getting better. As far as executions go, Apple's doing pretty well with the move to OS X. They have no need, no use for falling into a trap of chasing Microsoft APIs. Besides, Mac users would loathe those apps ever bit as much as they loathe the un-Macness of Mozilla or the old Office 6.x.



    Apple's better off staying the course. However, if there's technology in OS/2 PowerPC edition that might be useful it's probably already been in their hands from the Taligent venture, or OpenDoc (in the case of SOM) or publically available like JFS and IBM's tweaks to Mach. Don't forget Ellen Hancock looked at OS/2 as a potential platform when she killed Copland. It wasn't only Be that was competing with NeXT. They even looked at the NT core (minus Win32). They were fortunate to have so many choices already shipping on a PowerPC system. With the exception of NeXT. Funny how that won the day.



    [ 11-13-2002: Message edited by: dru ]</p>
  • Reply 7 of 8
    sdw2001sdw2001 Posts: 18,015member
    [quote] \t



    Some have discussed the posibility of incorporating an emulator into OS X so as to be able to run Windows applications, the obvious disadvantage to this is that along with file compatibility comes slow speed. <hr></blockquote>



    There was once a rumor where Apple had already developed the ability to run almost any Windows app natively....with basically zero performance loss.



    The thinking was that it was killed because there would then be no reason to develop for the Mac anymore.
  • Reply 8 of 8
    nevynnevyn Posts: 360member
    [quote]Originally posted by dru:

    <strong>

    Even with IBM throwing money at ISVs, applications simply weren't materializing.</strong><hr></blockquote>



    I didn't see any mention of the various x.x.1 type 'upgrades' from Dozeland giving OS2 fits. MS 'Oh, not our fault' yada yada. It was very irritating.



    Everyone I convinced to actually sit down in front of OS2 was amazed at the stability, then left because they needed to read Word documents. Sad.
Sign In or Register to comment.