Jump to content
22 posts in this topic

Recommended Posts

---Edit

NO LONGER TRUE SSE2 CAN BOOT GUI

------sportman

 

 

Bad news for most of us: we'll need a SSE3 powered CPU to start the GUI. I arrived at this conclusion after disassembling the code being executed when I get those Illegal instruction messages when trying to run graphical stuff like WindowServer -daemon, open and so on. The instruction being executed when the exception happens is fisttpll, a PNI (Prescott New Instructions) one. Then, no blue background for us at the moment.

 

sse3needed8dj.th.gif

 

Latest versions of Bochs emulate SSE3 but it's just too slow and regardless it doesn't even boot the Darwin CD, so it can't help us at this point. We could take code from it to emulate SSE3 in some way at kernel level, but that could take weeks... or months.

 

By the way, having a DVD image wouldn't change anything, so probably you don't actually need it nor the missing files at the moment if you don't have a SSE3 CPU.

 

Regards, blex0r

 

PS: If you're getting the blue bg with a non-sse3 computer please tell us. Try running WindowServer -debug.

Link to comment
https://www.insanelymac.com/forum/topic/70-sse3-needed-to-run-the-gui/
Share on other sites

Do all P4s support SSE3? Mine's a 2.4Ghz with a 533 bus. I think it's a 478 socket.

 

You can guess it by issuing sysctl hw.optional.sse3 at Darwin's command line. You can also use cpu-z on Windows or just cat /proc/cpuinfo in Linux. However, I'm pretty sure your CPU doesn't support SSE3 at all as only Prescott and above do. That includes Prescott P4s, Pentium D, Celeron D. Venice Athlon64's also support SSE3 although they lack HyperThreading, but that doesn't seem to matter.

 

Regards, blex0r

Motherf***er. -.-

 

Well, it gives me another reason to start saving up for that Pentium D 820 I've been eyeballin.

 

Overclocks to 3.6 w/AS and air cooling, Dual Core, SSE 3. :D

 

I'm normally a AMD guy, but the AMD Dual Cores are way TOO F***IN EXPENSIVE for about the same performance. Espically now that the Intel 820s use the AMD64 instruction sets. :D

With an Athlon64 non sse3 i get the WindowServer mouse pointer without blue bg. Please tell us that there's a way to bypass it...please.

 

Edit: i'm quoting some lines from buildyourownmac.com

 

The processors will be Intel chips using the standard x86 instruction set (IA-32), with at least the SSE and SSE2 vector extensions to replace Altivec. SSE3 will also be used, but Apple’s Universal Binary Programming Guidelines (pdf) lead us to believe that they will ship some machines without it, since SSE3 support is listed as an optional compiler setting, and is not on by default.
With an Athlon64 non sse3 i get the WindowServer mouse pointer without blue bg. Please tell us that there's a way to bypass it...please.

 

Edit: i'm quoting some lines from buildyourownmac.com

 

The processors will be Intel chips using the standard x86 instruction set (IA-32), with at least the SSE and SSE2 vector extensions to replace Altivec. SSE3 will also be used, but Apple’s Universal Binary Programming Guidelines (pdf) lead us to believe that they will ship some machines without it, since SSE3 support is listed as an optional compiler setting, and is not on by default.

 

That pointer-without-bg situation is the best we can get without a SSE3-enabled cpu. There's no way to bypass it apart from emulating SSE3 at kernel level or recompiling those binaries whose source code we don't have. We could patch that particular line but there are going to be a lot of SSE3 instructions all across the code as OS X86 has been compiled with the SSE3 flag enabled.

 

And about your quote, perhaps in the final version of OS X for Intel processors SSE3 won't be mandatory, but the only Intel Mac currently available supports SSE3 and it's a beta version so they don't actually need to support non-SSE3 cpus as they don't have any. When they will do, the only thing they need to do is recompiling with the SSE3 flag turned off.

 

Regards, blex0r

{censored}, dunno posted that last post I posted the specs on my server doh!!.

 

Anyway I'm ok cause I got an Intel P4 550 @4.1Ghz [Yes 4.1!!!] running smoothly away in my box. It cost a bomb in the UK and I got it running with the new Asus P5WD2 [Which is amazing!!!]

hi, im a newbie to this forum and subject as well, but i have been using pearpc with os x, that aside, do any of you think it would be possible to create a kernel module or something of the likes to intercept the SSE3 commands and trade them off with another(probably slower) assembly routine? im not great at programming and am not terribly familiar with *nix systems to know where to start

 

edit: i remember finding a program once that if installed in dos gave a 486 certain pentium features such as MMX if i remember correctly, never tested it, but maybe its possible :D

  • 2 weeks later...

I'm new to this forum but I have taken a lot of interest in the OSX86 topic (as has just about everyone with a PC who has worked on OSX at some point)

I would have to agree with blex0r.

 

I can't imagine the SSE3 requirement will be in the final x86 build.

And about your quote, perhaps in the final version of OS X for Intel processors SSE3 won't be mandatory, but the only Intel Mac currently available supports SSE3 and it's a beta version so they don't actually need to support non-SSE3 cpus as they don't have any. When they will do, the only thing they need to do is recompiling with the SSE3 flag turned off.

 

A lot of the talk(that I'm sure all of you have heard) is that Apple may have gone with Intel due to its low power/high performance notebook processors. However, no one seemed to mention that the Pentium M does not support the SSE3 instruction set (as far as I am aware). And there is also a lot of hype that Intel will be going to the desktop with their Pentium M's in the future.

 

While pointing out this speculation doesn't really offer a solution it does add validity to the opinion blex0r offered.

With an Athlon64 non sse3 i get the WindowServer mouse pointer without blue bg. Please tell us that there's a way to bypass it...please.

 

Edit: i'm quoting some lines from buildyourownmac.com

 

The processors will be Intel chips using the standard x86 instruction set (IA-32), with at least the SSE and SSE2 vector extensions to replace Altivec. SSE3 will also be used, but Apple’s Universal Binary Programming Guidelines (pdf) lead us to believe that they will ship some machines without it, since SSE3 support is listed as an optional compiler setting, and is not on by default.

 

That pointer-without-bg situation is the best we can get without a SSE3-enabled cpu. There's no way to bypass it apart from emulating SSE3 at kernel level or recompiling those binaries whose source code we don't have. We could patch that particular line but there are going to be a lot of SSE3 instructions all across the code as OS X86 has been compiled with the SSE3 flag enabled.

 

And about your quote, perhaps in the final version of OS X for Intel processors SSE3 won't be mandatory, but the only Intel Mac currently available supports SSE3 and it's a beta version so they don't actually need to support non-SSE3 cpus as they don't have any. When they will do, the only thing they need to do is recompiling with the SSE3 flag turned off.

 

Regards, blex0r

Both Intel's and AMD's processors support SSE3 now. I'm not certain about Intel, but in AMD, they have SSE accross the line (Their new revision E "Venice" cores have SSE3, and scale from the 3000+ up to the high end). Since everybody buying CPUs these days are probably getting SSE3 chips, why hasn't anybody tried this on an SSE3 chip? I mean, don't all Prescott P4s support SSE3? Those have been on sale for MANY months.

It would suck if SSE3 was a must and it would make sense to force the taking advantage of such a feature since all P4s now come with the SSE3 instructions. What catches my attention however is that the Pentium M does NOT support SSE3. This would leave the APPLE laptops with the reliance of the G4 OR Intel is planning on releasing a new PM in the future sporting SS3.

Guest bugmenot
It would suck if SSE3 was a must and it would make sense to force the taking advantage of such a feature since all P4s now come with the SSE3 instructions.  What catches my attention however is that the Pentium M does NOT support SSE3. This would leave the APPLE laptops with the reliance of the G4 OR Intel is planning on releasing a new PM in the future sporting SS3.

 

Dual core pentium m's should be out by the end of the year with sse3 right?

×
×
  • Create New...