[Gee4orce]

http://www.macworld.co.uk/news/top_n...fm?NewsID=6795

of course, there's a technobabble excuse, but the long and short of it is that Virtual PC is not going to appear for a while on the G5.

[biscuit]

Well the G5 is a very different beast to the G3 and the G4, let's not forget that. It's also not surprising that VPC used every last feature of the G3s and G4s to eek out decent performance.

But the fact that it's only happened to VPC, now that's a little suspicious...

biscuit

[videian28]

doesn't surprise me one bit

would not doubt that virtual pc 6.0 would work, but 6.1 (from microsoft) won't. or the fact that they were VERY aware that this would happen before they bought them out.


i am so sick of microsoft

[JLL]

Originally posted by biscuit:
But the fact that it's only happened to VPC, now that's a little suspicious...

Why? VPC are in many situations talking directly to the hardware - that was also why VPC4 (and earlier) didn't run in Classic.

[osiris]

ok, now that RealPC is officially dead, VPC won't be seen on a G5 for a while. A long while.

It doesn't take a fortune teller to figure out which evil-doer is behind this one.

[voodoo]

Originally posted by JLL:
Why? VPC are in many situations talking directly to the hardware - that was also why VPC4 (and earlier) didn't run in Classic.

Well nerdboy - nobody's talking about that.

[biscuit]

OK, so it's not really that suspicious. I was just trying to keep the conspiracy theory alive...

I'd have to agree with you JLL, it doesn't surprise me that VPC had close links to the hardware that are broken by the G5. As I mentioned, they are very different from Moto's chips.

Now the real question is, when do we see an update, if at all? Personally I think there will be one, since VPC is potentially a very important product for MS's Mac business unit. They also hired a bunch of Connectix developers with the take-over, so the signs are good. What's not so good is the statement that lots of it needs re-writing; how much manpower do they have to put into this thing?

That RealPC thing is crazy indeed.

biscuit

[Will Brocklebank]

Yeah - I loved the fact that they bought the company and then expect to be congratulated for hiring some of the expert developers from Connectix who built it in the first place. Well done M$oft - brilliant.

Imagine what would happen if M$haft had taken over the running of NASA missions - "Hell, we'll take Frank here but I am sure that we can make a new shuttle with these old egg boxes and washing-up bottles so no need for the other guys with thick glasses. It's not rocket science."

I NEED VPC 6 and I need it to run on a G5 otherwise it is dog slow - now I have to wait until 2010...

uuuuuggggghhhhh.

[Eug]

What is it with you guys?

It's not as if Microsoft bought VPC 3 years ago and had time to rewrite it. And furthermore, I doubt Apple even gave Connectix/Microsoft a dual G5 to play with early on.

If you want to blame anyone, blame Apple for being so secretive.

Anyways, I'm thinking 2004 for a VPC update.

[mrmister]

I agree with Eug--i hate Microsoft, but people are getting soft in the head with their conspiracy theories.

[P]

I'm willing to give M$ the benefit of the doubt here. The feature they're talking about is little-endian mode, which the G5 seems to lack.

Processors can work in either big-endian or little-endian mode. Intel uses little-endian - the rest of the world uses big-endian, more or less. Big-endian means that the most significant bit is written first when writing to memory, little-endian has the least significant bit first. If you want to write one thousand two hundred and thirty four, in big endian you'd write 1234, while in little-endian you'd write 4321 - but only in memory. Since the same processor that wrote that info is also going to be the one that reads it, it doesn't really matter, right?

This problem is when a number spills over more than one byte. A 32-bit processor handles four bytes in a go. The word "virtual PC!!" written by a little-endian processor will read "triv lau!!CP" when read by a big-endian one.

A PowerPC can run in either mode, but you have to reboot to switch. The G3 and the G4 were exceptions to this - you could sort of run them in little-endian without a reboot. This makes writing the actual Pentium emulator much easier, and that very central emulator will have to be re-written to work on a G5 - or MS has to come up with a nifty workaround in the Windows side of things.

Note that emulators written for older PowerPCs (601, 603 and 604) did not work this way. Softwindows worked on those, so it might work on a G5 (after being updated for OS X).

[sniffer]

Thanks for your explanation P. It kind off make more sense now.

[Eug]

P,

So does that mean a major rewrite of much of the core code? If so, methinks my 2004 prediction might have to be late 2004 or later.

[waffffffle]

Originally posted by P:
Note that emulators written for older PowerPCs (601, 603 and 604) did not work this way. Softwindows worked on those, so it might work on a G5 (after being updated for OS X).

SoftWindows -> RealPC. We all know how well that project is coming along...

[Wiskedjak]

Totally silly. Microsoft doesn't care if you're buying an Apple machine or a PC, just so long as you buy Windows. VPC allows M$ to sell Windows to people who would otherwise never purchase it. Therefore, it is in M$'s best interest to keep VPC active.

[Person Man]

Originally posted by Eug:
P,

So does that mean a major rewrite of much of the core code? If so, methinks my 2004 prediction might have to be late 2004 or later.

Not necessarily. They could start with the code that they used for 601-604 processors and optimize it for the G5, since only the G3/G4 had that special mode.

But yes, it would mean a major rewrite of the emulated x86 processor chip, which is essentially the core code for a PC emulator

[Gee4orce]

Originally posted by Wiskedjak:
Totally silly. Microsoft doesn't care if you're buying an Apple machine or a PC, just so long as you buy Windows. VPC allows M$ to sell Windows to people who would otherwise never purchase it. Therefore, it is in M$'s best interest to keep VPC active.

...except of course if you absolutely need to run a piece of Windows software - if you can't use Virtual PC your only option is to buy a real PC, in which case you might was well ditch the Mac alltogether. I'd count that as a win for Microsoft, and a loose for Apple, wouldn't you ?

If you want to blame anyone, blame Apple for being so secretive.

Ahhh. It's all clear now - it's Apple's fault. I should have known.

[Person Man]

Originally posted by Gee4orce:
...except of course if you absolutely need to run a piece of Windows software - if you can't use Virtual PC your only option is to buy a real PC, in which case you might was well ditch the Mac alltogether.

Uh, no.

You buy a PC *just* for that application. The rest of the time you use your Macintosh. No need to have your computer crash all the time during normal use, just on that one application

[sushiism]

"if you can't use Virtual PC your only option is to buy a real PC, in which case you might was well ditch the Mac alltogether. I'd count that as a win for Microsoft, and a loose for Apple, wouldn't you ?"
erm you buying virtual pc is already a win for microsoft, do they make any more money off you if you buy a copy of windows xp with a real pc or a virtual one? Nope not really a sale is a sale, the only person benifitiing really is dell or whatever shitty company you buy it off, apple wouldnt loose out unless you never bought another mac

[KidRed]

My reasons for conspiracy thinking- Why would Apple design a next gen processor which doesn't support a major app? Wow, if they did, why would they be beta testing Panther to work with VPC? Hmm, sounds like Apple wants to make sure VPC works well on Panther, but wait, what about the G5s? Oh, that's right IBM and Apple decided not to impliment a hardware feature that basically alienates thousands of users and possible switchers from using a G5 to run window apps?

Ah no. This stinks of M$ crafty, shady greediness.

[Hi I'm Ben]

I'm not sure how microsoft makes anymore money if you buy a pc or use virtual pc. Eitherway you have to pay for the software liscenses that you use on the PC. They should know most mac people are ignorant and would never ever use a PC anyway unless they had to. So why would they bother trying to make someone switch just to use an app, they won't. They know that apple will keep coming out with new software to avoid that from happening. Selling a good working copy of Virtual PC is just cash in the bank.

Originally posted by sushiism:
"if you can't use Virtual PC your only option is to buy a real PC, in which case you might was well ditch the Mac alltogether. I'd count that as a win for Microsoft, and a loose for Apple, wouldn't you ?"
erm you buying virtual pc is already a win for microsoft, do they make any more money off you if you buy a copy of windows xp with a real pc or a virtual one? Nope not really a sale is a sale, the only person benifitiing really is dell or whatever shitty company you buy it off, apple wouldnt loose out unless you never bought another mac

[biscuit]

Well although I encouraged the conspiracy theory earlier, I have to say this sounds like a bit of an oversight.

Apple didn't design the G5, IBM did. Unless I'm missing something, IBM won't really care about this little endian mode thingy, since they're using it for servers and the like and they know what they're running. Moto may have included it to help in the embedded market where all kinds of funny software is used I'm sure. So it gets left out of the G5 because no one at IBM sees the point in it, or perhaps because it's harder to implement than we all think (it's a 64-bit chip remember).

Now the only problem I see is that MS must have had PowerMac G5s for ages, they must have noticed this little problem before now. Why it never came up is anybody's guess. MS and/or Apple may have wanted to keep it quiet for lots of reasons.

BTW I know nothing about chips, or indeed programming so this is all made-up out of my rumor-filled head. The only question I have is; will the absence of this feature severely hamper VPC performance on a G5 even when it's been re-written?

biscuit

[Wiskedjak]

Originally posted by KidRed:
Why would Apple design a next gen processor which doesn't support a major app?

Why would Apple design a next gen OS which doesn't support a major app ... or several of them? Could you use OS9 versions of Adobe products and Quark on OSX before those software developers came up with OSX versions? And, how long was it before Quark had an OSX version? Perhaps Quark is involved in a conspiracy as well ...?

[RooneyX]

Instead of arguing about conspiracy theories or whether Microsoft benefits more if you have a real PC instead of a virtual one, could someone try to run an older version of Virtual Pc from version 5 to 6.1 to see whether they work on the G5 prior to Microsoft's acquisition?

[Mediaman_12]

Originally posted by biscuit:
Now the only problem I see is that MS must have had PowerMac G5s for ages, they must have noticed this little problem before now. Why it never came up is anybody's guess. MS and/or Apple may have wanted to keep it quiet for lots of reasons.

Microsoft only recently bought up VPC.
Microsoft will have probably had pre-production G5's, but Connectix (the makers of VPC and a few disk repair app's) more then likely didn't. It's likely that Microsoft found out about the VPC+G5 incompatibility as soon as they where able to run it on there test machine, They tried a few tweaks with the code to get it to run, and failed, realizing it would require a major re-wright of the core of the app to get it working.
At least they where honest and came out with a direct statement, saying it wasn't going to work. Unlike the 'Real PC' people who didn't have anything and suckered a few people into buying the old version, by taking about free upgrades.

[voodoo]

Originally posted by Mediaman_12:
Microsoft only recently bought up VPC.

They bought VPC last February. Over 6 months ago.

[macmike42]

As much as absolutely detest Microsoft, you guys have got to give them a break, folks. Consider these points. VirtualPC is an emulator. Not only that, but it is an extremely fast emulator, considering how extraordinarily different the Pentium and PowerPC chipsets are. Honestly, compare it to any x86-on-PPC emulator out there, and VPC will throroughly trounce the competition every time. Secondly, it is now moving to a massively re-designed architecture on which an essential feature is missing, one that has been relied upon most likely since the G3 first came out. Little-endian mode is essential to the process of passing values freely back and forth between emulated and native code. If you can just set a flag and have it work for the duration of your program, and that flag suddenly ceases to exist, you are in for a seriously grueling find-and-replace session, at best. A more likely secnario is that you will have to go through your code and review every assignment operation (use of "="), replacing anything that broke with a call to a conversion function or macro. You would also have to review every copy operation. Finally you would have to hunt for every situation in which native code directly accesses memory "owned" by emulated code and vice-versa. The skill required to do all these things properly and successfully without incurring a substantial performance hit is formidable. I am a programmer, and if I was faced with this kind of re-write, I would either give up, or ask for a raise.

[ZackS]

Originally posted by videian28:
i am so sick of microsoft

Hey, remember when they bought out one of the last remaining good Mac game companies, Bungie, to make games for the XBox when a PC studio like Valve (makers of Half Life) would have worked just as well? Every day I realize more and more how much I truly hate the Redmond Giant. It's not just Apple zealotry, it is pure, seething, unadulterated HATRID.

[ZackS]

Originally posted by macmike42:
As much as absolutely detest Microsoft, you guys have got to give them a break, folks. Consider these points. VirtualPC is an emulator. Not only that, but it is an extremely fast emulator, considering how extraordinarily different the Pentium and PowerPC chipsets are. Honestly, compare it to any x86-on-PPC emulator out there, and VPC will throroughly trounce the competition every time. Secondly, it is now moving to a massively re-designed architecture on which an essential feature is missing, one that has been relied upon most likely since the G3 first came out. Little-endian mode is essential to the process of passing values freely back and forth between emulated and native code. If you can just set a flag and have it work for the duration of your program, and that flag suddenly ceases to exist, you are in for a seriously grueling find-and-replace session, at best. A more likely secnario is that you will have to go through your code and review every assignment operation (use of "="), replacing anything that broke with a call to a conversion function or macro. You would also have to review every copy operation. Finally you would have to hunt for every situation in which native code directly accesses memory "owned" by emulated code and vice-versa. The skill required to do all these things properly and successfully without incurring a substantial performance hit is formidable. I am a programmer, and if I was faced with this kind of re-write, I would either give up, or ask for a raise.

But they are friggin Microsoft. With their resources, if they wanted to, they could have gotten that job finished within a week. They just don't care and they LOVE having something to hang over the heads of Mac users.

[MindFad]

Just warez the damn thing when it gets updated for the G5.

[Gul Banana]

Hello! Misinformation, I shall stab you in the eye with a rusty fork.

• There is no conspiracy. Don't be idiots.
• The G5 has little-endian mode.
• The G5 does NOT have pseudo-little-endian mode, which is where it pretends to a specific process to be little-endian even though the memory is actually stored big-endian.
• VPC used this heavily for performance reasons; having to use true-little-endian mode will only provide a small performance hit, though it will complicate things for the programmers.
•_Because of this, VPC will not run on the G5 until it is updated to make these changes. They will be difficult and complex; not so much so that it'll take forever, but enough so that I would have been surprised if Microsoft had a G5 version ready in time.

[biscuit]

Originally posted by Gul Banana:
• There is no conspiracy. Don't be idiots.

Aww...you've gone and spoilt all our fun...

Thanks for the other info though, that's exactly what I wanted to hear. If I understand you correctly then the re-written VPC will not suffer a gigantic performance drop, although it will be a while in coming. Sounds OK to me.

biscuit

[Catfish_Man]

Originally posted by macmike42:
As much as absolutely detest Microsoft, you guys have got to give them a break, folks. Consider these points. VirtualPC is an emulator. Not only that, but it is an extremely fast emulator, considering how extraordinarily different the Pentium and PowerPC chipsets are. Honestly, compare it to any x86-on-PPC emulator out there, and VPC will throroughly trounce the competition every time. Secondly, it is now moving to a massively re-designed architecture on which an essential feature is missing, one that has been relied upon most likely since the G3 first came out. Little-endian mode is essential to the process of passing values freely back and forth between emulated and native code. If you can just set a flag and have it work for the duration of your program, and that flag suddenly ceases to exist, you are in for a seriously grueling find-and-replace session, at best. A more likely secnario is that you will have to go through your code and review every assignment operation (use of "="), replacing anything that broke with a call to a conversion function or macro. You would also have to review every copy operation. Finally you would have to hunt for every situation in which native code directly accesses memory "owned" by emulated code and vice-versa. The skill required to do all these things properly and successfully without incurring a substantial performance hit is formidable. I am a programmer, and if I was faced with this kind of re-write, I would either give up, or ask for a raise.

Agreed. That's going to be hell on earth for their programmers.

[Gee4orce]

...and of course, we all know that Microsoft employs the best programmers. I mean, just look at the quality of their code.

What I was getting at in my earlier post is that for a lot of the buyers of Virtual PC they only need it to run one or two legacy apps that are not available on the Mac - something like Access or some Visual Basic app that's been developed in-house. If they can't run these apps, then they may as well buy a PC instead of a Mac - let's face it, most Mac apps have PC versions available. That, ultimately, is a win for Microsoft - moreso than just another Virtual PC licence - because it probably means $$$ on other Microsoft Windows software, not non-Microsoft Mac software.

I trust Microsoft about as far as I could throw an lead elephant. These are the guys who pulled IE on Mac when Apple released Safari (boo-hoo - competition !), and bought out Bungie (who were well on the way to releasing the worlds best FPS on the Mac, first).

Virtual PC will be Microsoft's new axe to hold over Mac users' heads (now that we seem to be getting some independence from Office).

[JLL]

Originally posted by Gee4orce:
These are the guys who pulled IE on Mac when Apple released Safari (boo-hoo - competition !),

Stopping IE has absolutely nothing to do with safari - they didn't release a major update in three years.

IE is dumped on Windows too and they are focusing on MSN and the built-in HTML core in Longhorn.