Apple launches the PowerMac G5

[Mr. Nice Guy]

Originally posted by Asher

I'm at work on a dialup. I can check it out when I get home because it's in my browser history there.


Don't have the link offhand, but it wouldn't be hard to find.
Intel gives general roadmaps for when they plan to retire cores. They plan to retire Prescott in September 04 or thereabouts, and they also say Prescott scales past 5GHz. So you put two and two together.

Hell, even right now people are getting their hotter, 130nm Northwood cores up to 3.6-3.8GHz with the stock Intel cooler. Prescott is on 90nm with strained silicon and stuff, will run much cooler and scale much higher.

And the Yamhill stuff are more than just rumours. Intel sources confirmed the existence of it, then they denied it, and the official stance right now is "No comment." Not only that, but it makes business sense and fits into Intel's history, not to mention the high-res die photos of Prescott (from Intel themselves!) which show the presence of 64-bit components.

It's simply unofficial because Intel doesn't want to talk about it.

Intel also denied that HyperThreading was present in all P4s, and just recently enabled. But the rumours and same Intel sources said it was so, as did die photos, and it turns out they were absolutely correct.

Um, exacty. Intel doesn't like to comment, just like Apple. So I don't see how there could be an official roadmap released by Intel regarding the Prescott and Yamhill.

There are many Mac rumor sites out there as well, and they don't always get it right, but they do sometimes. So just because these sites were right about Hyperthreading, doesn't mean they'll be right about Prescott and Yamhill.

[Asher]

Originally posted by Agathon
The whole thing looks like crap.

That's your opinion. And if it is, skin it and stop complaining.

Wow, I had a transparent terminal in 10.0

Never mind that 10.0 came out 2 years after Win2K, and never mind that 10.0's was done in software emulation rather than hardware acceleration.

It's better.

That's rich.
Why is it better? Do tell me, I'm interested to see what a philosophy major knows about Quartz vs GDI+...

I wish I could, but it won't be out for TWO YEARS!!!!

I wish I could try it for myself, but it won't be out for TWO YEARS!!!!

Is this out yet, or do I have to wait TWO YEARS!!!!!

It's out now if you don't mind Alpha software.
Or you could wait ~8-12 months for the public beta.

Bizarre. The speed has improved notably, even on my machine which is now a bit long in the tooth (new one in late 2004 I think).

That's because, as I've said already, Apple didn't do hardware acceleration at first. Windows has had it since Win2K, Apple is slowly adding it to OS X and people are lapping it up and thinking "how great!".

This is laughable. Windows XP costs more than OS X, but you will get ichat AV free with 10.3. Only a skinflint would pay the $30 upgrade - it's better value to get the whole thing.

Windows XP's upgrade is certainly much cheaper than OS X's.

One could ask where the decent junk mail filter is in Outlook Express.

SpamInspector. Way better than MozillaMail's, Mail's, etc.

And the only reason they developed it is that MS made such a poor job of IE on the Mac. Face it, nobody who knows anything has used IE as their default browser on OS X. There were always better competitors available (Omniweb, Chimera and now Safari).

No, they made Safari because MS told Apple they were halting stand-alone IE development almost a year ago. That is to say, no more Apple versions or downloadable Windows versions.

Apple paniced, started looking out for their own browser.

Which they had to do since IE performed so poorly.

You've no idea what you're talking about -- look into it first. It's diplomatic, MS was terminiating standalone IE development, which meant Apple's IE developers got the axe and the program was cancelled.

Which Jobs admitted in his keynote. Who cares anyway? Panther's fast user switching is way cooler than XP's.

You sound like a jr. high kid...
Why don't you actually tell us why Panther's fast user switching is 'way cooler'?

Had it since 10.2.2

HFS's kludged journalling can count I guess, but Apple still has a lot of work to do. 25-30% performance hit on writes with journalling enabled? Jeez.

Anyway, this is all bull****. You are comparing a version of OS X which will probably ship in November 2003 (perhaps earlier) with an OS that won't ship until at least 2005 and probably 2006.

Um. You're the one that brought it up, Einstein.

[Mr. Nice Guy]

oops. double post.

[Asher]

Originally posted by Cloud9
Both of these articles say that Intel refuses to comment, so I don't see how there could be an official roadmap for Prescott released by Intel.

Oh. Well, then, it must not be true because you don't think it's so...

At IDF, Intel puts up big PowerPoint slides with roadmaps. They don't discuss most features, especially ones like Yamhill because they want to push IA-64 sales, and Yamhill will likely detract from those.

There are rumors for the Mac on many sites out there, too. These Mac sites don't always get it right, but they do sometimes. Just because reports regarding Hyperthreading were right the last time, doesn't mean they'll be right about the Prescott and Yamhill.

Cloud9! Hello!

How do you counter the pictures of it being present on Prescott????

[Agathon]

Originally posted by Asher

That's your opinion. And if it is, skin it and stop complaining.

Yeah, that's going to help with the stupid layout.

Why is it better? Do tell me, I'm interested to see what a philosophy major knows about Quartz vs GDI+...

Anyone can see that it simply looks better.

It's out now if you don't mind Alpha software.
Or you could wait ~8-12 months for the public beta.

Or I could wait a few months and by a full working version of Panther. No contest really.

That's because, as I've said already, Apple didn't do hardware acceleration at first. Windows has had it since Win2K, Apple is slowly adding it to OS X and people are lapping it up and thinking "how great!".

I'm sure they have their reasons.

Windows XP's upgrade is certainly much cheaper than OS X's.

But it's not really an upgrade is it. It's still XP. Take a look at Panther and then at 10.0. See the radical differences.

SpamInspector. Way better than MozillaMail's, Mail's, etc.

I suppose this is a non MS add on.

[QUOTE]No, they made Safari because MS told Apple they were halting stand-alone IE development almost a year ago. That is to say, no more Apple versions or downloadable Windows versions.

Apple paniced, started looking out for their own browser.[QUOTE]



What a load. Apple panicked, so they released a browser that worked about 5 times as well as IE.

You've no idea what you're talking about -- look into it first. It's diplomatic, MS was terminiating standalone IE development, which meant Apple's IE developers got the axe and the program was cancelled.

Bull. People have been complaining from day one that IE sucks. Apple was pissed off because they knew that browser performance could be much better on OS X.

Your theory would have some merit, if only Safari wasn't light years better than IE.

You sound like a jr. high kid...
Why don't you actually tell us why Panther's fast user switching is 'way cooler'?

Haven't seen it, have you?

HFS's kludged journalling can count I guess, but Apple still has a lot of work to do. 25-30% performance hit on writes with journalling enabled? Jeez.

I guess I'll have to wait TWO YEARS!!!!!!!

Um. You're the one that brought it up, Einstein.

Yes I did, because the comparison works in my favour. You are stuck with XP for two or three years; by that time Apple will have streaked ahead.


And I see that the new expose feature has basically rendered most of your UI complaints impotent.

Want to see all windows open? Drag the mouse to the requisite screen corner (in adherence to Fitts law ).

Want to see only the windows associated with the selected app? Drag the mouse to the requisite screen corner.

Want to get all the windows off screen? Do the same.

Want to do this on XP. Prepare to read cut off labels in the taskbar while us Apple heads can actually see the open windows themselves.

[Mr. Nice Guy]

Originally posted by Asher

I've no idea what you're talking about -- did you not see the IBM PDF I linked to?

Not only does it give very detailed G5 specs, it gives the official SPEC benchmarks.

What am I assuming?

Sorry for the delay. This stupid Windows ME is giving me probs.

I'm on a slow dial-up connection and don't have the patience to download the report, but regardless of that. . .

Apple has posted it's offical SPEC benchmarks as well and you have refuted them. I don't think it's a good idea to take the benchmarks from the companies that worked on these products. Might as well rely on a reliable third party like Spec.org to do the testing.

[Mr. Nice Guy]

Originally posted by Asher

Oh. Well, then, it must not be true because you don't think it's so...

At IDF, Intel puts up big PowerPoint slides with roadmaps. They don't discuss most features, especially ones like Yamhill because they want to push IA-64 sales, and Yamhill will likely detract from those.

Um, exactly.

Cloud9! Hello!

How do you counter the pictures of it being present on Prescott????

The pictures could be fake, just like a lot of pics that are posted on Mac rumor sites.

Oh, and by the way, I couldn't tell by looking; I'm not a computer scientist like you.

[Asher]

Originally posted by Agathon
Yeah, that's going to help with the stupid layout.

You're really not good at this are you.
I was responding to your complaint about the graphics.

Anyone can see that it simply looks better.

Yeah, precisely what I thought -- you haven't a clue what Quartz or GDI+ are if you can "see" them.

I suppose this is a non MS add on.

Of course, but why does it matter? You just mentioned freeware as a way to customize your desktop, but the second someone else mentions it...

MS wants you to get Outlook 11.

Bull. People have been complaining from day one that IE sucks. Apple was pissed off because they knew that browser performance could be much better on OS X.

Now that is bull. People complain about IE sucks just like people complain that Windows sucks. The people who complain tend to be Linux/*nix users and/or Mac users.

As for performance issues, IE is faster on some pages and Safari on others. Apple selectively shows the benchmarks from the pages that run faster on Safari, and moronic users jump for joy and look no further. Google for some real benchmarks.

Your theory would have some merit, if only Safari wasn't light years better than IE.

Well, can't argue with that.

Haven't seen it, have you?

Why can't you actually answer with content, just once?
I'm getting a tad sick of all of your rhetoric...

I guess I'll have to wait TWO YEARS!!!!!!!

Does this make any sense?
XP has journalling now...has had it since Win2K.

Yes I did, because the comparison works in my favour. You are stuck with XP for two or three years; by that time Apple will have streaked ahead.

Oh my God...you can't argue like this at your job, right? It's embarassing.
Let's just go through what you just did:
* Compare OS 10.3 to Longhorn
* ***** at me for comparing OS 10.3 to Longhorn
* Admit you did compare OS 10.3, then repeat your stupid comments about "Apple streaking ahead while you wait", when in reality, if you looked at the additions to 10.3, Apple is catching up to what MS did in 2001 with XP...

And I see that the new expose feature has basically rendered most of your UI complaints impotent.

But you told me my UI complaints weren't valid!
Apparently Apple saw otherwise. Good on them.

Maybe you should listen to me more, Apple certainly does.

Want to do this on XP. Prepare to read cut off labels in the taskbar while us Apple heads can actually see the open windows themselves.

While it's nice Apple finally added the "show desktop" functionality that's been in Windows since Win95, it's also a bit weird to do the whole cascade thing. The only reason it's necessary is because Apple doesn't label the taskbar entries.

[Asher]

Does anybody know what they're talking about here?

Originally posted by Cloud9
Um, exactly.

I'm glad you agree with me! So why're you arguing?

The pictures could be fake, just like a lot of pics that are posted on Mac rumor sites.

I feel like I'm repeating myself a lot.
The pictures came from Intel at IDF2003.

Oh, and by the way, I couldn't tell by looking; I'm not a computer scientist like you.

Computer Scientists do software, Computer Engineers do hardware.

But apparently you can't read either -- because the site clearly explains it.

Apple has posted it's offical SPEC benchmarks as well and you have refuted them.

Those aren't official SPEC benchmarks, they're PR benchmarks. The G5's official SPEC benchmarks come from IBM, and they're in the PDF I posted.

I don't think it's a good idea to take the benchmarks from the companies that worked on these products. Might as well rely on a reliable third party like Spec.org to do the testing.

SPEC.org doesn't test.
SPEC.org posts the results from the companies who test them. Notice the "submitters" on SPEC.org are companies such as "IBM", "Dell Computers", "Intel Corporation", etc?

[Mr. Nice Guy]

Re: Does anybody know what they're talking about here?

Originally posted by Asher

I'm glad you agree with me! So why're you arguing?

Um, following the logic, you agreed with me, not I agreed with you.

I feel like I'm repeating myself a lot.
The pictures came from Intel at IDF2003.

How do you know?

Those aren't official SPEC benchmarks, they're PR benchmarks. The G5's official SPEC benchmarks come from IBM, and they're in the PDF I posted.


SPEC.org doesn't test.
SPEC.org posts the results from the companies who test them. Notice the "submitters" on SPEC.org are companies such as "IBM", "Dell Computers", "Intel Corporation", etc?

Oh, you missed something completely. Look at the revision date on this page. This was done the day the PowerPC 970 was introduced; it's an older benchmark so why hasn't it been posted on the Spec.org site yet? And if it's an older benchmark, then that means that it was done with slower Power 970--probably a 1.4GHz. And was this a DP test? I haven't had the patience to download the full report yet; guess I'll just have to trickle it. How big is the PDF anyway?

[Asher]

Re: Re: Does anybody know what they're talking about here?

Originally posted by Cloud9
How do you know?

Because hundreds of press were there when it was shown?

Oh, you missed something completely. Look at the revision date on this page. This was done the day the PowerPC 970 was introduced; it's an older benchmark so why hasn't it been posted on the Spec.org site yet?

I didn't miss that. It's not on SPEC.org because IBM wasn't allowed to submit it until the NDA went up on the G5. It would spoil Apple's little announcement on how "fast" it was.

And was this a DP test?

AGAIN -- the SPEC base benchmarks are uniprocessor only.

How big is the PDF anyway?

The IBM one? ~2.5MB

[Mr. Nice Guy]

Re: Re: Re: Does anybody know what they're talking about here?

Originally posted by Asher
Because hundreds of press were there when it was shown?

Granted that, can you prove these rumors surrounding the Prescott beyond a shadow of a doubt? Remember, I can't interpret the pictures.

By the way, I've been to other sites that have reported on the Prescott and they say nothing about it being 64-bit.

I didn't miss that. It's not on SPEC.org because IBM wasn't allowed to submit it until the NDA went up on the G5. It would spoil Apple's little announcement on how "fast" it was.


AGAIN -- the SPEC base benchmarks are uniprocessor only.

Ok, I finally got the PDF downloaded. According to the report, this test was done with a single 1.8GHz PPC 970 processor and scored fairly well: 937int and 1051fp repectively. Now imagine what would happen if you bumped that up to 2.0GHz and added another processor. Also, imagine that, instead of a 1GHz bus for the whole system, that each individual processor has a 1GHz bus!!! That's what Apple says on their site; we'll have to see about that one.

[Agathon]

Re: Re: Re: Re: Does anybody know what they're talking about here?

Originally posted by Cloud9

Ok, I finally got the PDF downloaded. According to the report, this test was done with a single 1.8GHz PPC 970 processor and scored fairly well: 937int and 1051fp repectively. Now imagine what would happen if you bumped that up to 2.0GHz and added another processor. Also, imagine that, instead of a 1GHz bus for the whole system, that each individual processor has a 1GHz bus!!! That's what Apple says on their site; we'll have to see about that one.

Yes this is the problem. Anyway Pentiums can't be DPed as far as I know.

[Asher]

Re: Re: Re: Re: Does anybody know what they're talking about here?

Originally posted by Cloud9
Granted that, can you prove these rumors surrounding the Prescott beyond a shadow of a doubt? Remember, I can't interpret the pictures.

Well, let's look at the facts.
Integer registers double from 32-bit to 64-bit.
There are now 2 ALUs instead of 1 (ALUs handle integer arithmetic & logic instructions), but the second one lacks an address generator. They're also positioned right next to eachother. This means that the second one is merely an extension of the first. Now why would that be? If you add 2 32-bit ALUs together, what do you get? A 64-bit ALU.

By the way, I've been to other sites that have reported on the Prescott and they say nothing about it being 64-bit.

Just like how if you look at most sites about the original Pentium 4, they won't mention a thing about HyperThreading.

Prescott won't have the 64-bit extensions enabled, so unless it's a hardcore site like chip-architect (it's by and for processor designers...), I bet it won't mention Yamhill because it's irrelevant. Just like the presence of HyperThreading on every single P4 made is irrelevant when it's only enabled for the most recent ones.

Ok, I finally got the PDF downloaded. According to the report, this test was done with a single 1.8GHz PPC 970 processor and scored fairly well: 937int and 1051fp repectively. Now imagine what would happen if you bumped that up to 2.0GHz and added another processor.

1.8->2.0 GHz isn't much of a speedbump.
Assuming perfect scalability (which basically never happens), a 2.0GHz PPC 970 would get 1030int and 1156 fp. Which is really good, just not as good as the P4s that've been out for 6 months already, and certainly no competition for Prescott.

As for the second processor, the vast majority of applications see no benefit from it. Only applications written to work with multiple threads do, or if you run two programs at once. Even then, you're likely to only see a 30-40% improvement at a maximum.

And still, you can also get dual Xeons and Opterons.

Also, imagine that, instead of a 1GHz bus for the whole system, that each individual processor has a 1GHz bus!!! That's what Apple says on their site; we'll have to see about that one.

That's AMD's HyperTransport technology. That's right, AMD.

Opteron has that as well, it's nothing special.

[Asher]

Re: Re: Re: Re: Re: Does anybody know what they're talking about here?

Originally posted by Agathon
Yes this is the problem. Anyway Pentiums can't be DPed as far as I know.

Xeons and Opterons can.

Xeons are the workstation/high-end Pentium 4/5s, Opterons are the workstation/high-end Athlon 64s.

In fact, you can get 32-way Xeons and 8-way Opterons if you so choose.

The reason you very rarely find dual processors in most consumer PCs is, well, you don't need it. The PC's CPUs are so fast by themselves they're unnecessary. Apple had to do dual G4s as a pathetic attempt to keep up, and they're doing the dual 2GHz right now to try to capture the audio nuts/graphics nuts that have started jumping ships in hordes to faster PCs.

Which isn't that impressive still, since you can get 2/4-way Xeons as well.