Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

immovableobject

macrumors newbie
Jun 15, 2005
5
0
EFI Licensing

GregA said:
I don't know much about EFI, just what I quickly found on the web.

Firstly - "Universally" hated may be overstating it! I've found very few criticisms of the new standard actually.

Secondly, EFI requires that the graphics card has UGA firmware otherwise it falls back to VGA.
http://www.kernelthread.com/publications/firmware/
http://www.ami.com/support/doc/EFI-FAQ.pdf

As far as requiring a license to use FAT, do all EFI implementations require FAT? Even if they did, it appears MS has given permission for use of FAT for EFI implementations.

http://www.intel.com/technology/efi/agreesource_draft.htm

There were two points on VGA. One was that a framebuffer is always required. Why do servers need a framebuffer? They don't. It's a needless piece of hardware. The other is that no debugging is possible on framebuffer-based ROMS. All the messages go to the framebuffer, once it's initialized. Generally, that's late in the boot process, so any early debugging messages (such as bad memory) are lost. I hate the 3-beeps that indicate memory problems. There needs to be better diagnostics than that.

Regarding the FAT Licensing, Microsoft will license FAT, but it can not be used in open source. So, what about OpenDarwin? Will it be able to boot on a inteltosh?

I know Linux boots on EFI, however, I don't know if the EFI command line is supported for Linux.
 

GregA

macrumors 65816
Mar 14, 2003
1,249
15
Sydney Australia
immovableobject said:
There were two points on VGA. One was that a framebuffer is always required. Why do servers need a framebuffer? They don't. It's a needless piece of hardware. The other is that no debugging is possible on framebuffer-based ROMS. All the messages go to the framebuffer, once it's initialized. Generally, that's late in the boot process, so any early debugging messages (such as bad memory) are lost. I hate the 3-beeps that indicate memory problems. There needs to be better diagnostics than that.

Regarding the FAT Licensing, Microsoft will license FAT, but it can not be used in open source. So, what about OpenDarwin? Will it be able to boot on a inteltosh?

I know Linux boots on EFI, however, I don't know if the EFI command line is supported for Linux.
I read the VGA limitation in a criticism of the current BIOSes, not for EFI which uses UGA.

Linux already boots on Itanium chips which use EFI - so either FAT is one option within EFI (see my original question), or MS allows it. Besides, doesn't EFI come with the hardware, not the OS?

Remember there is Intel's framework and then what the implementations do with it. Command line is an option for an EFI interface AFAIK, as is mouse and graphics.
 

mischief

macrumors 68030
Aug 1, 2001
2,921
1
Santa Cruz Ca
Hmm... It sounds as if there's a "standard" implementation of EFI and then there's the technology on it's own; eg: There's the EFI that Intel's selling as "EFI" and then there's the technology of the same name that Intel came up with to replace BIOS.

Sounds more and more from this (remarkably composed) discussion as if Apple will work very closely with Intel to produce a hybrid using EFI's configurability but expressing Open Firmware's intelligence of design.
 

GregA

macrumors 65816
Mar 14, 2003
1,249
15
Sydney Australia
mischief said:
Sounds more and more from this (remarkably composed) discussion as if Apple will work very closely with Intel to produce a hybrid using EFI's configurability but expressing Open Firmware's intelligence of design.
There is certainly a difference between how Phoenix and AMI implement EFI (and how MS is involved?) and what Apple might choose to do with EFI.
 

shamino

macrumors 68040
Jan 7, 2004
3,443
271
Purcellville, VA
Rocketman said:
I will simply point out to all Apple zealots (and collectors) that the release of this system represents an official Apple release of a CPU with MacOS installed. Thus and therefore it must remain on all future hardware lists.
Not at all.

This is a prototype development system. It is only available to developers to pay for membership in the ADC Select or Premier programs. And they're rentals, not for sale - they have to be returned (at the end of 2006, IIRC.)

There is no reason whatsoever that the OS shipped with the commercial x86 boxes will have to be in any way compatible with these units.
 

shamino

macrumors 68040
Jan 7, 2004
3,443
271
Purcellville, VA
GregA said:
I read the VGA limitation in a criticism of the current BIOSes, not for EFI which uses UGA.
It's not a limitation on current BIOS's either. Every one I've seen still allows configuration for text-only displays or for no display. Most users don't really care, however. These settings come pre-configured for whatever display hardware the computer comes shipped with.

The lack of serial-console support is significant, but there's no reason at all that it couldn't be added, should some motherboard maker want to include it. Maybe it would be more interesting to ask why PC customers are not demanding this feature from their suppliers.
 

mischief

macrumors 68030
Aug 1, 2001
2,921
1
Santa Cruz Ca
shamino said:
Not at all.

This is a prototype development system. It is only available to developers to pay for membership in the ADC Select or Premier programs. And they're rentals, not for sale - they have to be returned (at the end of 2006, IIRC.)

There is no reason whatsoever that the OS shipped with the commercial x86 boxes will have to be in any way compatible with these units.

Thank you!! I've been saying this to deaf ears since the keynote. Developper-specific custom protos will not necessarilly reflect the production machines' security and mobo design except in fairly general ways. They will also not run anywhere near as fast as production machines because they're ad-hoc compared to the machines they're being used to develop for.

Most industries will use this same technique to test things. Auto makers will use old chassis to test new engines that will never be seen in those cars, Software manufacturers often have to code for unreleased chips through cumbersome emulation on old ones. Aerospace companies will fit new engines to older aircraft that just happen to have the right size frame for the engine with no intention of resurrecting the old design with the new engine.

This is just how it's done folks! Apple is not releasing to it's developers it's Alpha production protos so get it out of your heads!!! Sheesh! :rolleyes:
 

Sam0r

macrumors regular
Jun 15, 2005
199
0
Birmingham, UK
If apple have been developing OSX for x68 since its birth, then it must run on other hardware.

You really think that that exact same hardware setup thats in the development mac's was used to compile/code the x86 build of Tiger?

I'm betting they'll go with OF, They have to really.
 

shamino

macrumors 68040
Jan 7, 2004
3,443
271
Purcellville, VA
Sam0r said:
If apple have been developing OSX for x68 since its birth, then it must run on other hardware.
It's obvious that they used generic (probably PC-type) hardware for their internal development over the past 5 years.

And I'm sure the code is generic enough to run on anything. After all, Darwin has been ported to everything, and OpenStep (the roots of the OS X stuff) has also been ported to everything.

But this says nothing about what the commercially shipping Intel-Macs will be like, and it says nothing about how much work it will be to hack a commercial release of MacOS-Intel to boot on ordinary PC's. You can be certain that Apple will be adding code to prevent booting on non-Apple equipment, and it almost certainly won't be something that will be easy to reverse engineer.

My guess is that they'll use a customized core-logic chipset instead of a PC-standard one. (Remember that they've been designing these kinds of chipsets for years for the PPC systems.)

In doing so, they'll compile-in kernel drivers for this chipset and leave out the drivers that would be needed for the chipsets used by their prototype systems. So any third party that wants to hack it into a generic PC will have to write his own chipset drivers - which is not an easy thing to do.
Sam0r said:
I'm betting they'll go with OF, They have to really.
They don't have to. And it really doesn't matter to most people.

As long as the commonly-used functions (like Target Disk mode, Firewire booting, pressing C to boot CDs, etc.) exist in the new ROMs, the internal architecture of it will only matter to evangelists, hackers, and a very small number of developers.
 

Huked on Fonick

macrumors 6502
Jul 16, 2002
300
0
1 Loop
Who cares!!! as long as they dont start useing dogcows to do the job we should be set. Its apple im sure it will work Good(enough) at least....Maybe even slightly better.

P.S. Dogcows are actully better suited for this job than one may first think. Being only 2 Dimentions they can only travel forward and backwards making them perfect for I/O.

Ok ill shut up now.
 

jsalzer

macrumors 6502a
Jan 18, 2004
607
0
The Happy DogCow

Sweet - hearing the cry of angry Mac users everywhere who miss the Happy Mac and the DogCow, Apple puts Clarus in charge of the I/O and introduces a new icon at startup - the Happy DogCow.

If Clarus can't find your hard disk, he is shown searching for his frisbee in one direction while it is sitting behind him.

;)
 

mangoarts

macrumors newbie
Mar 24, 2005
15
0
Los Angeles
New BIOS

Intel keeps reminding us that these processors are not referred to as dual-core at Intel but instead multi-core. A processor is not all that Apple will use in their new machines. Lets not forget about EFI, Vanderpool (VT), and TCP. EFI is the next gen bios replacement that will make OS <-> mainboard interaction much simpler for OS developers. Vanderpool will allow us to run more than one OS at a time. I’m not talking emulation ala VMware, I am talking about more than one full blown OS runing at the same time. Can you see a Mactel box that lets you transition from you M$ apps to Mac OS X apps by actually letting you run the other OS at full speed. TCP ( Trusted Computing Platform ) is what I believe Apple will use to make sure that Mac OS X will only run their hardware. I think they will put some shared keys in the TCP. IBM is using this not on their laptops for password security, We shall see.
 

GregA

macrumors 65816
Mar 14, 2003
1,249
15
Sydney Australia
mangoarts said:
Vanderpool will allow us to run more than one OS at a time. I’m not talking emulation ala VMware, I am talking about more than one full blown OS runing at the same time. Can you see a Mactel box that lets you transition from you M$ apps to Mac OS X apps by actually letting you run the other OS at full speed.
Anyone know how this works?

I've read
1. that the Freescale dual-core G4 can run multiple OSes... ONE on each core. Quite limiting IMO (I don't want my MacOS running at half speed). Is this what Intel is doing?
2. that the OSes need to support it (simply to avoid stepping on each other) (can't remember where I read this)

Anyone know what Vanderpool will mean to Mac users?
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.