They most likely coult do it in Windows as well, at least NT/2000/XP, but I am not sure whether Win95 has something like that.</strong><hr></blockquote>
I don't think they would need to do it in Windows. My guess is that Adobe's frameworks provide adequate threading under NT/2k/XP but not under X due to Carbon.
To counter the threading issue in X, bumping up priority would help. Regardless, I've found Photoshop/Illustrator/InDesign under X to not thread in a manner that I find acceptable. They're fine apps, but they routinely bind up unnecessarily.
Adobe's apps aren't unique in this respect, by any means. I find that Office and Filemaker are the same way. Carbon apps seems to be at a distinct disadvantage when competing for system resources. Overall, the system is more responsive, but the individual app performance tends to suffer.
For PS and Ill, I'm sure that swap isn't the problem as my files rarely exceed 50MB - 10MB is more common for me. My InDesign files can run considerably larger than my PS files, but ID seems to handle things considerably better. Of course, ID is better able to partition off the rendering than PS or Ill, so that makes a certain amount of sense to me.
I remember hearing at the time that IBM's engineering group felt that a SIMD vector unit was counter to the philosophy of RISC and therefore shouldn't be added to the PowerPC. Once the G4 was delivered and they had a chance to play with it the were "impressed" and started to change their tune. Since then more and more news of IBM SIMD units has been cropping up. So I lay the blame for this rift largely at IBM's feet for not having the vision to see that alternative computing models can deliver serious performance improvements. Now, of course, IBM has come around 180 degrees and is working on things like the "Cell" project, their own SIMD, and they have done the Gekko for Nintendo.</strong><hr></blockquote>
Yeah, I remember that now. Thanks for the refresh. There was far less cooperation than the term "alliance" suggested. Each partner had an agenda to advance its own ideas and products at the expense of the others. It may be that Apple had no choice but to go with MOT after IBM dug in its heels, or it could be that IBM got ornery only after Apple started cozying up with MOT. I hope Steven Levy writes a book about it someday, otherwise we may never know. At this point, I guess, it really doesn't matter. IBM is needed as a full partner again, and MOT still has a lot of expertise to add, as well. Everyone needs to kiss and make up and get on with the job of getting some first-class chips out the door. It looks like that is exactly what has been going on over the last year or so, so I remain an optimist.
That certainly could be, but I'm fairly certain it isn't. I'll do some real investigation tomorrow.</strong><hr></blockquote>
Hmm.... so this might be a reason why Apple chooses Photoshop then for the bake-off stuff eh? Well who knows. Interesting that it looks like we are getting Firewire 2 with the new tower revisions -- this bodes well for the "true motherboard" that you've heard about moki if I remember correctly.
"For the past month or so, Apple Demo Days have been going on in France, promoting the XServe. Apparently, the Apple managers there have been a bit talkative, and have leaked eerily similar info... Here's what has been told:
The G5 is being evaluated right now, and the G4 will further be modified. The G5 isn't stated for this year, perhaps the end of 2003. It does exist though, and works fine right now, however the manufacturing process is far from ready, much too costly and low yield percentage. (confirmed by many different sources) The G5 will be manufactured in France (Grenoble), the factory is not finished yet.
The G4's evolution will not be dramatic, however the motherboards will significantly change with HyperTransport (some say the XServe already implements HT) and better cache structures.
The next mobos will not support 9, so it's pretty much the end of the line for it...
The kernel following that of Jag will be clusterable at a system level...
The site has been quite reliable for a while now, since they rarely indulge in rumormongering... So take it FWIW.
The good news being that since the G5 is still far away (further than MWSF), Apple has no reason whatsoever to introduce stop-gap solutions at NY, except if the mobos aren't ready.
[ 06-21-2002: Message edited by: SYN ]"
Posted by Syn. This thread is dry. Dorsal's disappeared (again...c'mon guy...put up or shut up...)
I don't expect Moki to say anymore than he has. It's clear the big 'iron' isn't going to appear this year and a 'jumped' up G4 in massive bandwidth early to mid-next year.
If the 'Syn' info' is true...I'm bloody depressed.
[quote]The G5 isn't stated for this year, perhaps the end of 2003. <hr></blockquote>
Argh. This is one rumor I hope is wrong. If Apple can't deliver a G5, I certainly hope that they can squeeze a lot more speed out of the G4. Anything less than 1.3 GHZ machines at MWNY and I will be very disappointed.
<strong>"For the past month or so, Apple Demo Days have been going on in France, promoting the XServe. Apparently, the Apple managers there have been a bit talkative, and have leaked eerily similar info... Here's what has been told:
The G5 is being evaluated right now, and the G4 will further be modified. The G5 isn't stated for this year, perhaps the end of 2003. It does exist though, and works fine right now, however the manufacturing process is far from ready, much too costly and low yield percentage. (confirmed by many different sources) The G5 will be manufactured in France (Grenoble), the factory is not finished yet.
The G4's evolution will not be dramatic, however the motherboards will significantly change with HyperTransport (some say the XServe already implements HT) and better cache structures.
The next mobos will not support 9, so it's pretty much the end of the line for it...
The kernel following that of Jag will be clusterable at a system level...
The site has been quite reliable for a while now, since they rarely indulge in rumormongering... So take it FWIW.
The good news being that since the G5 is still far away (further than MWSF), Apple has no reason whatsoever to introduce stop-gap solutions at NY, except if the mobos aren't ready."
I don't expect Moki to say anymore than he has. It's clear the big 'iron' isn't going to appear this year and a 'jumped' up G4 in massive bandwidth early to mid-next year.
If the 'Syn' info' is true...I'm bloody depressed.
End of 2003?
Yeesh. (Shakes head.)</strong><hr></blockquote>
I can't really comment on a lot of that, or at least I am not going to, but before people start whining that the G5 isn't ready yet maybe you should wait for the next evolutions of the G4. The G4 isn't actually that bad a chip it just hasn't been scaled as quickly as one would like.
<strong>"For the past month or so, Apple Demo Days have been going on in France, promoting the XServe. Apparently, the Apple managers there have been a bit talkative, and have leaked eerily similar info... Here's what has been told:
The G5 is being evaluated right now, and the G4 will further be modified. The G5 isn't stated for this year, perhaps the end of 2003. It does exist though, and works fine right now, however the manufacturing process is far from ready, much too costly and low yield percentage. (confirmed by many different sources) The G5 will be manufactured in France (Grenoble), the factory is not finished yet.
The G4's evolution will not be dramatic, however the motherboards will significantly change with HyperTransport (some say the XServe already implements HT) and better cache structures.
The next mobos will not support 9, so it's pretty much the end of the line for it...
The kernel following that of Jag will be clusterable at a system level...
The site has been quite reliable for a while now, since they rarely indulge in rumormongering... So take it FWIW.
The good news being that since the G5 is still far away (further than MWSF), Apple has no reason whatsoever to introduce stop-gap solutions at NY, except if the mobos aren't ready.
[ 06-21-2002: Message edited by: SYN ]"
Posted by Syn. This thread is dry. Dorsal's disappeared (again...c'mon guy...put up or shut up...)
I don't expect Moki to say anymore than he has. It's clear the big 'iron' isn't going to appear this year and a 'jumped' up G4 in massive bandwidth early to mid-next year.
If the 'Syn' info' is true...I'm bloody depressed.
I doubt that this info is true, because in Grenoble ( more exactly in Crolles) , it's not a factory which is scheduled but an joint venture research center in order to improve the fab process (from 0,9 to under).
If the G5 is ready i doubt that it will need 18 months in order to have good yields, 6 months even for mot is sufficiant.
Maybe that just means the end of a bootable OS 9. Wasn't there a rumor that a future release of 9 would not work as a standalone OS, but rather a stripped-down emulation environment that would be much more efficient (less CPU use) but still allow you to run legacy software?
<strong>Maybe that just means the end of a bootable OS 9. Wasn't there a rumor that a future release of 9 would not work as a standalone OS, but rather a stripped-down emulation environment that would be much more efficient (less CPU use) but still allow you to run legacy software?</strong><hr></blockquote>
...and they called it Classic. Actually, I heard the same thing a while ago, that new mobo wouldn't support OS9 boot, but would still support Classic.
BY THE WAY, any one ever thought of quad processors? Imagine Four 1.2 Ghz Processors on a single computer.(I know there are no 1.2 Ghz, yet, it is just common knowledge)
What would have to be changed to make way for this? It seems like a good idea to hold people off until a G5 came? <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
BY THE WAY, any one ever thought of quad processors? Imagine Four 1.2 Ghz Processors on a single computer.(I know there are no 1.2 Ghz, yet, it is just common knowledge)
What would have to be changed to make way for this? It seems like a good idea to hold people off until a G5 came? <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" /> </strong><hr></blockquote>
People keep asking this -- better memory bandwidth would be required. The current MPX bus isn't fast enough to support 4 fast G4s efficiently.
the g4 has 1 GB/s bandwidth with a 133 Mhz bus does that mean a 166 Mhz bus will scale to 1.3 GB/s or is there some overhead.
The 1.7 Ghz Xeon has a 3.2 GB/s theoretical Bus limit but only can obtain about 1.7 GB/s for Integer and FP streams. Is there similar info available for the G4 MPX bus??
"I can't really comment on a lot of that, or at least I am not going to,"
Oh really? ? Do tell. Apparently the site concerned even had the name of the 'Apple Man' that said it. Which adds some substance to this rumour. Though his name is mysteriously no longer there..? (The Apple Mafia discreetly wading in?)
"...but before people start whining that the G5 isn't ready yet maybe you should wait for the next evolutions of the G4. The G4 isn't actually that bad a chip it just hasn't been scaled as quickly as one would like."
John McCarmack himself said that Apple's top chip had trouble hanging with a Pentium 3 at 1 gig.
It was a 733 at the time, I think. Since then the disparity has got much worse. It'll take more than a DDR dual 1.2 G4 to convince me Apple's taking it's overpriced 'power'Mac range seriously. If it's on hypertransport perhaps? If there's a dramatic mhz swing? IF there's an add fpu to keep the lonely fpu happy?
Hey, I'd love to be proved wrong.
"A good friend will bail you out of jail, but your best friend will be sitting next to you saying: 'that was f@#%ing awesome!' "
<strong>John McCarmack himself said that Apple's top chip had trouble hanging with a Pentium 3 at 1 gig.</strong><hr></blockquote>
This had more to do with memory bandwidth issues, especially where interaction with AGP devices was concerned. The Intel & AMD chips have something called write combining which helps with communicating with graphics chips on the AGP bus. The 7450 had that added, but I believe some issues still existed in the chipset. It may have been addressed in January's machines, but I'm not sure.
MPX's performance on a 133 MHz bus is about 850 MB/sec maximum sustained with Apple's latest chipset. It should scale linearly to 166 MHz, I think.
It seems everyone has a pet hardware gizmo that is a must have, but what can really be expected?
If Apple is buying up all the video production software companies and they are "Going After Hollywood" what sort of hardware would be needed to beat the pants off of Intel based hardware? Are XServes usable as render farm components? <img src="graemlins/oyvey.gif" border="0" alt="[No]" /> I propose a cheap AMD/Intel is better for that (render farms), so what sort of desktop would be needed from Apple?
It makes no sense to get agressive in video production without having agressive showing in video production speed. So, what is the minimum hardware that can beat anything else on the scene in that market? (Hardware guys please speculate!)
I'm not expecting such a machine in July, but it would make sense to have something ready after Jaguar's release.
yeah, Apple buys up all this software that people who use a lot of NON-macs use. Then, they convert it to Mac ONLY, so people are forced to move to make for a new version and better support and new features.
Also, the mac is like an expensive car. It, like the other cars, can get you to point B from A. Yet, the expensive car does it better with more style and the route it takes it always faster and nicer
Comments
<strong>
They most likely coult do it in Windows as well, at least NT/2000/XP, but I am not sure whether Win95 has something like that.</strong><hr></blockquote>
I don't think they would need to do it in Windows. My guess is that Adobe's frameworks provide adequate threading under NT/2k/XP but not under X due to Carbon.
To counter the threading issue in X, bumping up priority would help. Regardless, I've found Photoshop/Illustrator/InDesign under X to not thread in a manner that I find acceptable. They're fine apps, but they routinely bind up unnecessarily.
Adobe's apps aren't unique in this respect, by any means. I find that Office and Filemaker are the same way. Carbon apps seems to be at a distinct disadvantage when competing for system resources. Overall, the system is more responsive, but the individual app performance tends to suffer.
For PS and Ill, I'm sure that swap isn't the problem as my files rarely exceed 50MB - 10MB is more common for me. My InDesign files can run considerably larger than my PS files, but ID seems to handle things considerably better. Of course, ID is better able to partition off the rendering than PS or Ill, so that makes a certain amount of sense to me.
<strong>
I remember hearing at the time that IBM's engineering group felt that a SIMD vector unit was counter to the philosophy of RISC and therefore shouldn't be added to the PowerPC. Once the G4 was delivered and they had a chance to play with it the were "impressed" and started to change their tune. Since then more and more news of IBM SIMD units has been cropping up. So I lay the blame for this rift largely at IBM's feet for not having the vision to see that alternative computing models can deliver serious performance improvements. Now, of course, IBM has come around 180 degrees and is working on things like the "Cell" project, their own SIMD, and they have done the Gekko for Nintendo.</strong><hr></blockquote>
Yeah, I remember that now. Thanks for the refresh. There was far less cooperation than the term "alliance" suggested. Each partner had an agenda to advance its own ideas and products at the expense of the others. It may be that Apple had no choice but to go with MOT after IBM dug in its heels, or it could be that IBM got ornery only after Apple started cozying up with MOT. I hope Steven Levy writes a book about it someday, otherwise we may never know. At this point, I guess, it really doesn't matter. IBM is needed as a full partner again, and MOT still has a lot of expertise to add, as well. Everyone needs to kiss and make up and get on with the job of getting some first-class chips out the door. It looks like that is exactly what has been going on over the last year or so, so I remain an optimist.
<strong>
That certainly could be, but I'm fairly certain it isn't. I'll do some real investigation tomorrow.</strong><hr></blockquote>
Hmm.... so this might be a reason why Apple chooses Photoshop then for the bake-off stuff eh? Well who knows. Interesting that it looks like we are getting Firewire 2 with the new tower revisions -- this bodes well for the "true motherboard" that you've heard about moki if I remember correctly.
[ 06-21-2002: Message edited by: ouroboros ]</p>
The G5 is being evaluated right now, and the G4 will further be modified. The G5 isn't stated for this year, perhaps the end of 2003. It does exist though, and works fine right now, however the manufacturing process is far from ready, much too costly and low yield percentage. (confirmed by many different sources) The G5 will be manufactured in France (Grenoble), the factory is not finished yet.
The G4's evolution will not be dramatic, however the motherboards will significantly change with HyperTransport (some say the XServe already implements HT) and better cache structures.
The next mobos will not support 9, so it's pretty much the end of the line for it...
The kernel following that of Jag will be clusterable at a system level...
all this is here
<a href="http://www.macbidouille.com/niouzcontenu.php?date=2002-06-21#2827"; target="_blank">http://www.macbidouille.com/niouzcontenu.php?date=2002-06-21#2827</a>;
The site has been quite reliable for a while now, since they rarely indulge in rumormongering... So take it FWIW.
The good news being that since the G5 is still far away (further than MWSF), Apple has no reason whatsoever to introduce stop-gap solutions at NY, except if the mobos aren't ready.
[ 06-21-2002: Message edited by: SYN ]"
Posted by Syn. This thread is dry. Dorsal's disappeared (again...c'mon guy...put up or shut up...)
I don't expect Moki to say anymore than he has. It's clear the big 'iron' isn't going to appear this year and a 'jumped' up G4 in massive bandwidth early to mid-next year.
If the 'Syn' info' is true...I'm bloody depressed.
End of 2003?
Yeesh. (Shakes head.)
Lemon Bon Bon
[ 06-21-2002: Message edited by: Lemon Bon Bon ]</p>
Argh. This is one rumor I hope is wrong. If Apple can't deliver a G5, I certainly hope that they can squeeze a lot more speed out of the G4. Anything less than 1.3 GHZ machines at MWNY and I will be very disappointed.
<strong>"For the past month or so, Apple Demo Days have been going on in France, promoting the XServe. Apparently, the Apple managers there have been a bit talkative, and have leaked eerily similar info... Here's what has been told:
The G5 is being evaluated right now, and the G4 will further be modified. The G5 isn't stated for this year, perhaps the end of 2003. It does exist though, and works fine right now, however the manufacturing process is far from ready, much too costly and low yield percentage. (confirmed by many different sources) The G5 will be manufactured in France (Grenoble), the factory is not finished yet.
The G4's evolution will not be dramatic, however the motherboards will significantly change with HyperTransport (some say the XServe already implements HT) and better cache structures.
The next mobos will not support 9, so it's pretty much the end of the line for it...
The kernel following that of Jag will be clusterable at a system level...
all this is here
<a href="http://www.macbidouille.com/niouzcontenu.php?date=2002-06-21#2827"; target="_blank">http://www.macbidouille.com/niouzcontenu.php?date=2002-06-21#2827</a>;
The site has been quite reliable for a while now, since they rarely indulge in rumormongering... So take it FWIW.
The good news being that since the G5 is still far away (further than MWSF), Apple has no reason whatsoever to introduce stop-gap solutions at NY, except if the mobos aren't ready."
I don't expect Moki to say anymore than he has. It's clear the big 'iron' isn't going to appear this year and a 'jumped' up G4 in massive bandwidth early to mid-next year.
If the 'Syn' info' is true...I'm bloody depressed.
End of 2003?
Yeesh. (Shakes head.)</strong><hr></blockquote>
I can't really comment on a lot of that, or at least I am not going to, but before people start whining that the G5 isn't ready yet maybe you should wait for the next evolutions of the G4. The G4 isn't actually that bad a chip it just hasn't been scaled as quickly as one would like.
[ 06-21-2002: Message edited by: Telomar ]</p>
<strong>"For the past month or so, Apple Demo Days have been going on in France, promoting the XServe. Apparently, the Apple managers there have been a bit talkative, and have leaked eerily similar info... Here's what has been told:
The G5 is being evaluated right now, and the G4 will further be modified. The G5 isn't stated for this year, perhaps the end of 2003. It does exist though, and works fine right now, however the manufacturing process is far from ready, much too costly and low yield percentage. (confirmed by many different sources) The G5 will be manufactured in France (Grenoble), the factory is not finished yet.
The G4's evolution will not be dramatic, however the motherboards will significantly change with HyperTransport (some say the XServe already implements HT) and better cache structures.
The next mobos will not support 9, so it's pretty much the end of the line for it...
The kernel following that of Jag will be clusterable at a system level...
all this is here
<a href="http://www.macbidouille.com/niouzcontenu.php?date=2002-06-21#2827"; target="_blank">http://www.macbidouille.com/niouzcontenu.php?date=2002-06-21#2827</a>;
The site has been quite reliable for a while now, since they rarely indulge in rumormongering... So take it FWIW.
The good news being that since the G5 is still far away (further than MWSF), Apple has no reason whatsoever to introduce stop-gap solutions at NY, except if the mobos aren't ready.
[ 06-21-2002: Message edited by: SYN ]"
Posted by Syn. This thread is dry. Dorsal's disappeared (again...c'mon guy...put up or shut up...)
I don't expect Moki to say anymore than he has. It's clear the big 'iron' isn't going to appear this year and a 'jumped' up G4 in massive bandwidth early to mid-next year.
If the 'Syn' info' is true...I'm bloody depressed.
End of 2003?
Yeesh. (Shakes head.)
Lemon Bon Bon
[ 06-21-2002: Message edited by: Lemon Bon Bon ]</strong><hr></blockquote>
I doubt that this info is true, because in Grenoble ( more exactly in Crolles) , it's not a factory which is scheduled but an joint venture research center in order to improve the fab process (from 0,9 to under).
If the G5 is ready i doubt that it will need 18 months in order to have good yields, 6 months even for mot is sufficiant.
Very unlikely.
G_News
Power Macs are used by companies, schools and individuals with money to buy up-to-date software, or licences where they get the latest software.
If Apple continues Mac OS 9, that means software for Mac OS 9 will exist.
Apple is going strong, and killing Mac OS 9 now means Quark will have more incentive for getting off their ass.
Barto
<strong>Maybe that just means the end of a bootable OS 9. Wasn't there a rumor that a future release of 9 would not work as a standalone OS, but rather a stripped-down emulation environment that would be much more efficient (less CPU use) but still allow you to run legacy software?</strong><hr></blockquote>
...and they called it Classic. Actually, I heard the same thing a while ago, that new mobo wouldn't support OS9 boot, but would still support Classic.
Keep the boot off, yet Classic still working.
BY THE WAY, any one ever thought of quad processors? Imagine Four 1.2 Ghz Processors on a single computer.(I know there are no 1.2 Ghz, yet, it is just common knowledge)
What would have to be changed to make way for this? It seems like a good idea to hold people off until a G5 came? <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
<strong>That would be a a smart idea.
Keep the boot off, yet Classic still working.
BY THE WAY, any one ever thought of quad processors? Imagine Four 1.2 Ghz Processors on a single computer.(I know there are no 1.2 Ghz, yet, it is just common knowledge)
What would have to be changed to make way for this? It seems like a good idea to hold people off until a G5 came? <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
People keep asking this -- better memory bandwidth would be required. The current MPX bus isn't fast enough to support 4 fast G4s efficiently.
The 1.7 Ghz Xeon has a 3.2 GB/s theoretical Bus limit but only can obtain about 1.7 GB/s for Integer and FP streams. Is there similar info available for the G4 MPX bus??
"I can't really comment on a lot of that, or at least I am not going to,"
Oh really?
"...but before people start whining that the G5 isn't ready yet maybe you should wait for the next evolutions of the G4. The G4 isn't actually that bad a chip it just hasn't been scaled as quickly as one would like."
John McCarmack himself said that Apple's top chip had trouble hanging with a Pentium 3 at 1 gig.
It was a 733 at the time, I think. Since then the disparity has got much worse. It'll take more than a DDR dual 1.2 G4 to convince me Apple's taking it's overpriced 'power'Mac range seriously. If it's on hypertransport perhaps? If there's a dramatic mhz swing? IF there's an add fpu to keep the lonely fpu happy?
Hey, I'd love to be proved wrong.
"A good friend will bail you out of jail, but your best friend will be sitting next to you saying: 'that was f@#%ing awesome!' "
I like that. I think that's funny.
lemon bon bon
<strong>John McCarmack himself said that Apple's top chip had trouble hanging with a Pentium 3 at 1 gig.</strong><hr></blockquote>
This had more to do with memory bandwidth issues, especially where interaction with AGP devices was concerned. The Intel & AMD chips have something called write combining which helps with communicating with graphics chips on the AGP bus. The 7450 had that added, but I believe some issues still existed in the chipset. It may have been addressed in January's machines, but I'm not sure.
MPX's performance on a 133 MHz bus is about 850 MB/sec maximum sustained with Apple's latest chipset. It should scale linearly to 166 MHz, I think.
<a href="http://www.anandtech.com/showdoc.html?i=1483&p=12"; target="_blank">http://www.anandtech.com/showdoc.html?i=1483&p=12</a>;
If Apple is buying up all the video production software companies and they are "Going After Hollywood" what sort of hardware would be needed to beat the pants off of Intel based hardware? Are XServes usable as render farm components? <img src="graemlins/oyvey.gif" border="0" alt="[No]" /> I propose a cheap AMD/Intel is better for that (render farms), so what sort of desktop would be needed from Apple?
It makes no sense to get agressive in video production without having agressive showing in video production speed. So, what is the minimum hardware that can beat anything else on the scene in that market? (Hardware guys please speculate!)
I'm not expecting such a machine in July, but it would make sense to have something ready after Jaguar's release.
Please comment!
[ 06-21-2002: Message edited by: iCode ]</p>
Also, the mac is like an expensive car. It, like the other cars, can get you to point B from A. Yet, the expensive car does it better with more style and the route it takes it always faster and nicer