64-Bit Jaguar and 970 PowerMacs

[nagromme]

The article says that without Panther, the 970 will run with the performance of a G4. (But at slightly higher MHz than we have now, and with a faster system bus.)

I have to question that. I can't see Apple releasing 970 Macs, yet dilluting their impact by having only minor speed increases until a sofware update months later. If the impact of Panther were THAT important, I'd expect Apple to wait for it. But every other report seems to suggest that Panther is NOT needed for big speed gains on the 970.

 

[fred]

Well....I believe it would probably be wiser to wait until Panther is ready rather than give bashers the chance to benchmark the systems running Smeagol and find them lacking..... So much for all the BS going around that stated that OS X was 64 bit ready and the work required would be trivial...

 

[szark]

Well, don't forget that Apple has to put support into the operating system for the new hardware. (new CPU, new chipset, etc.) That's why it needs a new build of Jaguar.

My question: will the preview release of Panther being given to developers at WWDC run on the new machines?

 

[shadowfax]

Originally posted by scem0
hmmmmmmm from cats to MTV award show winners.....

Well, I am thoroughly convinced that apple will be releasing new PowerMacs at the WWDC. Now let's hope they go past 1.8....

i thought you would make a comment about them naming this project after you, gollum.

i wonder why everyone thinks they are going to release 64 bit processors before their 64 bit OS. it seems to me that it would simplify thingss to wait.

 

[chetwilliams]

Originally posted by Laurent
I think that the Jaguar update won't be required, but Apple want to offer something 64-bit optimized so that the difference between G4 an 970 be greater...

The "Smeagol" 10.2.7 update will serve two purposes:
1. Support the new 970 processor -- this will be a relatively small change to ensure that the 970 runs all code in 32-bit mode properly; it also might include some more extensive changes to support >4GB memory
2. Support the new motherboard -- more extensive changes to support the new motherboard chipset and all of its goodies (hopefully all of the following: fast bus, USB2, Serial-ATA, AGP 8X)

 

[andyduncan]

Originally posted by nagromme
The article says that without Panther, the 970 will run with the performance of a G4. (But at slightly higher MHz than we have now, and with a faster system bus.)

That actually sounds pretty good to me.

I have to question that. I can't see Apple releasing 970 Macs, yet dilluting their impact by having only minor speed increases until a sofware update months later.

You mean like, for instance, releasing Powermacs that ran the vast majority of mac code slower than the 68k machines they replaced? It was the 9500 that was the first powerpc mac to actually emulate 68k code faster than the Quadras ran it natively. And that was more than a few months later.

Besides, as has been mentioned by myself and several other people, you have to have the machines in the hands of developers before the apps get optimized.

Id say the timing sounds perfect.

 

[jettredmont]

Scary ...

Sources said Apple's goal for Smeagol is to deliver Mac OS X performance at least "on par" with what Jaguar could achieve on Motorola G4 chips running at the same speed

ON PAR with the G4?!?

My goodness, I hope that was just thrown in as an easily-achievable goal, kinda like "Make MS Word run at least as fast as on a 386SX-16 running Windows 3.0 with 2MB of RAM ..."

 

[richard5mith]

Re: Scary ...

Originally posted by jettredmont
ON PAR with the G4?!?

My goodness, I hope that was just thrown in as an easily-achievable goal, kinda like "Make MS Word run at least as fast as on a 386SX-16 running Windows 3.0 with 2MB of RAM ..."

Well realise that means make it run like OS X would on a 1.8Ghz G4, which would still be faster than ever.

The difference is that now you have the promise of even better performance in a couple of months with the OS X upgrade and then again with Panther. And all without needing a new machine.

 

[phanguye]

i would assume that the reason they choose "Smeagol" as the code name is because Smeagol is what Gollum is before he gets the precious and even when he has it but before it totally consumes him...

"Smeagol" the jaguar version will have the precious but will not totally be what is should be (will not be consumed) until it gets panther

that is just my guess

 

[eric_n_dfw]

64 bit ready rant

Did those of you bitching about the lack of OS X being 64 bit ready even read the article?

my emphasis added
Part of the issue with OS compatibility lies with Mac OS X's compiler, GCC 3, which lacks scheduling support for the PowerPC 970. Apple and IBM are reportedly working to add 970-specific support to the latest version, GCC 3.3. The development effort is proceeding well, sources report, but the compiler isn't yet ready for full use.

While Smeagol will be built using GCC 3.1, Apple plans to compile Mac OS X 10.3 with GCC 3.3.

The fact that they can build Smeagol on GCC 3.1 proves that the code is at least clean enough to run on the chip. 10.3 will give the 64 bit code and GCC 3.3 will alow it to run at more optimized speeds.

How about reading the article the next time.

 

[eric_n_dfw]

Re: Scary ...

Originally posted by jettredmont
ON PAR with the G4?!?

My goodness, I hope that was just thrown in as an easily-achievable goal, kinda like "Make MS Word run at least as fast as on a 386SX-16 running Windows 3.0 with 2MB of RAM ..."

Sounds like very conservative estimates - especially if the GCC 970 scheduling optimizations get put in place. I, for one, am just dieing to see the AltiVec unit(s) take off when they get a taste of that DDR FSB!

 

[jettredmont]

Originally posted by rjstanford
Of course, there's only one problem. Can anyone confirm the gcc issues? Is this really so significantly different? If so, that would mean that software built for the G4 series wouldn't run on the 970 without modification, which is something that I have a hard time believing (from previous statements and from IBM history).

Your G4 software will not work on a 970 if your software happens to be an OS kernel.

The gcc modifications are kernel-level. OS X apps will not require G4-970 modification from all indications.

Do note that IBM is on record saying that there are "minor" changes required to the OS for the 970 relative to the G3/G4 lines. "Minor" to the hardware vendor rarely means "minor" to the software provider ...

 

[mathiasr]

Re: Scary ...

Originally posted by jettredmont
ON PAR with the G4?!?

My goodness, I hope that was just thrown in as an easily-achievable goal, kinda like "Make MS Word run at least as fast as on a 386SX-16 running Windows 3.0 with 2MB of RAM ..."

You know internaly the 970 is pretty different from the 7450, for instance it has only 2 integer execution units compared to 4 for the G4e, and the AltiVec unit will probably behave differently too.

Running code that has been optimized for the G4 on the 970 could hurt, and the article pointed out that a suitable GCC 3.3 was not finished...

 

[andyduncan]

Re: 64 bit ready and willing

Originally posted by eric_n_dfw
The fact that they can build Smeagol on GCC 3.1 proves that the code is at least clean enough to run on the chip.

How does that prove anything?

 

[ParadisePete]

Smeagol

Originally posted by trebblekicked
and the smeagol name is priceless.

There already is a Smeagol.

 

[ddtlm]

richard5mith:

Well realise that means make it run like OS X would on a 1.8Ghz G4, which would still be faster than ever.

But still miserably behind the curve set by the competition. If Apple releases a single 1.8ghz PPC970 with single 1.8ghz G4 performance and expects people to be excited... well maybe people will be, but noone who thought about it for very long. A new version of the OS will mean little for performance I might add, since its the applications that need to be fast, not the OS.

andyduncan:

You mean like, for instance, releasing Powermacs that ran the vast majority of mac code slower than the 68k machines they replaced? It was the 9500 that was the first powerpc mac to actually emulate 68k code faster than the Quadras ran it natively. And that was more than a few months later.

The PPC970 does not emulate 32-bit PPC instructions so your example does not relate. The entire cause for concern is that the code that is out there may not be sufficiently optimized for the PPC970, which would be very dissapointing if true. To put it mildly, the G4 is not a benchmark that the 970 should be proud of defeating.

jettredmont:

My goodness, I hope that was just thrown in as an easily-achievable goal, kinda like "Make MS Word run at least as fast as on a 386SX-16 running Windows 3.0 with 2MB of RAM ..."

Well said.

eric_n_dfw:

I, for one, am just dieing to see the AltiVec unit(s) take off when they get a taste of that DDR FSB!

You are assuming that a decent chipset exists, hopefully not only supporting dual channel DDR-400 but also doing so well. I'm really not sure that Apple is up to that task... perhaps they'll buy someone else's chip... but we shall eventually find out. Anyway, the bandwidth probably won't much increase the things the G4 did best, just increase the range of things that go quickly.

 

[Stike]

I can explain the Smeagol name...

Comp Lab:
Developer1: "Mr. Jobs, here is the new prototype of the PowerMac 970 with Jaguar finally running on it!" (sweats)
SJ: "Okay, leave me alone, I will test it."
(exit Developer)
In front of the Comp Lab:
Developer1: "He is in there since 48 hours. I wonder if he is alright.."
Developer2: "Lets take a look then!" (opens door)
SJ can be seen in a completely dark room, only illuminated by the pale screen light.
SJ: "Get out!! Leave the lights out! I need morE testing! MorrE! moOOorrree!"
(Developer2 instantly closes door)
SJ: "My precioussss...!!"

---

BTW, if we go with the alphabetical order in codenames like in OS X (6L60 and so on) and the old MDD PMacs are P57/58, we can imagine that Q37 is a step beyond...!

 

[andyduncan]

Originally posted by ddtlm
The PPC970 does not emulate 32-bit PPC instructions

I never said it did, I only said that Apple has in the past released machines with new architectures that aren't always as fast as people think they should be on day one, or even day 365. And I think that relates perfectly.

The entire cause for concern is that the code that is out there may not be sufficiently optimized for the PPC970, which would be very dissapointing if true.

Of course its true, its a different chip. How could it possibly be otherwise? Even different versions of the G4 need compiler optimizations to acheive their full potential. Remember when the G4 powermacs jumped into the 800 range? The chips were less efficient for a number of reasons, but many of those were due to the compilers being optimized for the previous generation (half generation). We're talking about a whole new chip that shares only a portion of the ISA with the processors that the compilers have been tweaked for. There's bound to be optimization issues.

edit: mis-matched tag

 

[JLL]

Re: I can explain the Smeagol name...

Originally posted by Stike
BTW, if we go with the alphabetical order in codenames like in OS X (6L60 and so on)

6L60 is a build number - not a code name.

 

[JLL]

Re: Scary ...

Originally posted by jettredmont
ON PAR with the G4?!?

My goodness, I hope that was just thrown in as an easily-achievable goal, kinda like "Make MS Word run at least as fast as on a 386SX-16 running Windows 3.0 with 2MB of RAM ..."

Perhaps you didn't notice the At Least part.

And he could mean that it would run at least as fast as a 1.8GHz G4 with a 900MHz FSB

 

[ddtlm]

andyduncan:

Of course its true, its a different chip.

You overlooked my use of "sufficiently." Clearly the fact that the PPC970 is a different chip does not imply that existing code is not "sufficiently" opimized. The Athlon, for example, does great on old x86 code. I think it would be very dissapointing if the PPC970 couldn't beat the IPC of a G4 on existing code. I have been very dissapointed with the gutlessness of G4's compared to x86 compeditors on code that I have written.

 

[beatle888]

Re: Pre-announce at WWDC seems questionable

Originally posted by fpnc
I don't think Apple would announce or show new PPC970-based Macs at WWDC and then wait a month or more before they actually ship. This isn't going to be a simple speed bump or change in appearance. When the PPC970 __is__ announced it will immediately make obsolete every G4-based PowerMac that is still unsold or on store shelves.

its already been reported that PowerMac stock is extremely low. its time. but still. im gonna want more power than what they'll have to offer. however it is progress.

 

[andyduncan]

Originally posted by ddtlm
You overlooked my use of "sufficiently."

Bah.

Clearly the fact that the PPC970 is a different chip does not imply that existing code is not "sufficiently" opimized. The Athlon, for example, does great on x86 code.

And yet there are still compiler optimizations that can bring out more speed for athlon systems, just look at some of the good linux software. In addition, the athlon was designed to do well on x86 code that had been optimized for intel peculiarities, while the power4 was targeting it's own market. Sure the 970 is targeting the same market as the g4, but why redesign the chip just so someone doesn't have to recompile on the next point release? It runs, just not as well as it could. and its not the end of the world if we have to wait for the next version of gcc before the compiler starts putting instructions in a 970-friendly order.

I think it would be very dissapointing if the PPC970 couldn't beat the IPC of a G4 on existing code. I have been very dissapointed with the gutlessness of G4's compared to x86 compeditors on code that I have written.

Well then I guess there's a sufficiently probable chance that you'll be disappointed.

 

[shadowfax]

Originally posted by ddtlm
I have been very dissapointed with the gutlessness of G4's compared to x86 compeditors on code that I have written.

LOL
LOL
LOL

sorry, that's pathetic. you should learn to write G4-optimized code before you get all disappointed. it still is a pretty competitive chip with the right software.

to be disappointed that a chip won't outperform another chip with an OS that is specifically written for that chip and not for the new one is kinda stupid, too, IMO.

 

[joost538]

the info is accurate..

at least the part about gcc3.3 is accurate.


from the GCC3.3 page at gnu.org:

The following changes have been made to the PowerPC port:
Support for IBM Power4 processor added.