it's easy to believe that bad scheduling based on G4 specs would lead to really really poor performance by the 970. Relatively speaking, of course.
My understanding is that the 970 has very aggressive out-of-order execution capabilities, so it should be able to deal reasonably well with poorly scheduled code.
I see... A torrent of before and after Panther benchmarks!!!
Although the article makes no mention of them, or perhaps because it makes no mention of them, I have a feeling that if any Powerbook model will get revved to the 970 then it will be closer to the Panther release than to WWDC. Otherwise if a new G4 15" PB is debuted right along side the new PMs, well, that seals the fate of 970 PBs until the end of the year...
Heh, "buy this computer! It will get really really fast in like 2 months... I promise "
Most of this compiler/scheduler stuff is way over my head, but I'll tell you this: If there is any real world performance penalty running the 970 under Smeagol, then Apple should wait. They won't get a second chance to make a first impression (bake-off).
Calling Programmer! We need you to weigh in on this.
No, I thought Apple is continuing its letter-number code. Those revised Powermacs that Thinksecret reported that raised way too much speculation was P58, yes?
Well, the 970 PMs are a new generation therefore from P to Q.
Most of this compiler/scheduler stuff is way over my head, but I'll tell you this: If there is any real world performance penalty running the 970 under Smeagol, then Apple should wait. They won't get a second chance to make a first impression (bake-off).
Calling Programmer! We need you to weigh in on this.
The 970 will be faster than the G4 at the same clock rate, and (depending on what you are doing) potentially much faster. Software that is then recompiled specifically for the 970 will get faster yet although its not clear how much faster -- typically previous generational changes have given a 15-20% performance improvement from just recompiling. Panther (and all of Apple's iApps) will probably be completely recompiled for the new processor, whereas the Jaguar patch will just have the small driver & kernel changes necessary to enable the OS to run on the 970.
If they are running a specific bake-off they could probably use a recompiled version of Photoshop (or whatever).
Further performance improvements are not going to be related to the 970's "64-bitness" as seemed to be implied above. They instead result from the processor's architecture.
Not buying one just because it isn't running at full speed on day 1 is just silly. No radically new hardware is running at its full potential the day it ships. Even the current G4 towers aren't fully utilized, if they were then Panther couldn't deliver any performance improvements as it apparently does. Have no fear, however, the demonstration that Steve does will highlight the machine's performance -- Steve wouldn't have it any other way.
maybe these new machines will be only for close apple developers to try along with the pre release of panther to work the bugs out. I mean wwdc is for developers not public consumer announcements. If steve is gonna do show speed tests at wwdc hes gonna use panther.
maybe these new machines will be only for close apple developers to try along with the pre release of panther to work the bugs out. I mean wwdc is for developers not public consumer announcements. If steve is gonna do show speed tests at wwdc hes gonna use panther.
And everyone at WWDC has been promised a preview release of Panther so anyone taking it home from there won't be using 10.2.x...
The 970 will be faster than the G4 at the same clock rate, and (depending on what you are doing) potentially much faster. Software that is then recompiled specifically for the 970 will get faster yet although its not clear how much faster -- typically previous generational changes have given a 15-20% performance improvement from just recompiling. Panther (and all of Apple's iApps) will probably be completely recompiled for the new processor, whereas the Jaguar patch will just have the small driver & kernel changes necessary to enable the OS to run on the 970.
If they are running a specific bake-off they could probably use a recompiled version of Photoshop (or whatever).
Further performance improvements are not going to be related to the 970's "64-bitness" as seemed to be implied above. They instead result from the processor's architecture.
Not buying one just because it isn't running at full speed on day 1 is just silly. No radically new hardware is running at its full potential the day it ships. Even the current G4 towers aren't fully utilized, if they were then Panther couldn't deliver any performance improvements as it apparently does. Have no fear, however, the demonstration that Steve does will highlight the machine's performance -- Steve wouldn't have it any other way.
Thanks Programmer, that backs up everything I've heard about the 970. Any idea why Rothenberg seems to want to talk it down? Is he just confused over the importance of 64 bit, or does he just not understand the issues involved?
I wouldn't be too worried about the "on par with G4" comment. There are alot of issues to resolve and even if the new PowerMacs aren't optimized by WWDC, I don't mind having a few months of public tweaking by early-adopters. It means that what ships in September will be ready to go!
I think Apple also needs to give something in the summer to schools and companies with Fall fiscal years to see, so that orders can be made in anticipation of Panther.
This lines up eventually well for holiday media blitz's ... let's hope!
Also would all of these changes to the mobo may also increase downloads to iPods? I thought I read that download rates were not slowed down by FireWire.
The point is that there is alot to potentially crow about when a change like this is made and even if the 970 is a bit hampered, I don't care. This is for DEVELOPERS to see and they will get the Panther beta! That's what we want. Then by the Fall we'll be cheering, "here come the snail ads!!!" Too bad Intel took away the bunny suit guys.
Have no fear, however, the demonstration that Steve does will highlight the machine's performance -- Steve wouldn't have it any other way.
Agreed.
I'm confident we're about to be impressed. If the machine runs at half speed even until Panther...you're still getting a 1.8 G4 on a whoppa of a bus...with altivec cut loose and an extra fpu. (You see any G4s lookin' dat good?) If I wasn't broke after forking out for two design collections and Final Cut Pro 3, I'd be on that bus. Personally, I'm waiting until after Sept. Will or iron, eh?
Heh, I can use my wife's 600 mhz G3 until then. It still seems like lightspeed compared to my old Powermac 604e clone!
I don't think we all quite realise what Steve Jobs is about to hit us with in 2 weeks... Despite all the 'confirmation'...it's all rather quiet and understated from IBM/Apple. It's deafeningly quiet.
The point is that there is alot to potentially crow about when a change like this is made and even if the 970 is a bit hampered, I don't care. This is for DEVELOPERS to see and they will get the Panther beta! That's what we want. Then by the Fall we'll be cheering, "here come the snail ads!!!" Too bad Intel took away the bunny suit guys.
I wouldn't be too worried about the "on par with G4" comment. There are alot of issues to resolve and even if the new PowerMacs aren't optimized by WWDC, I don't mind having a few months of public tweaking by early-adopters. . .
I think the author just doesn't understand that the 970 will easily out perform a G4 running the same 32 bit applications. I doubt there will need to be any tweaking or tuning to see a big difference right away.
The real question is, How much faster will the 970 run AOL?
3-4 times faster depending on the version corresponding the color of smeagol panther pink or green 64 bitness off cetering the 32 bit version which would be realsed maybe in quarter 3 of the year before next.
Comments
Boy, Apple will be walking on a wire explaining 32-bits versus 64-bits. Maybe it'd be best not to mention it other than to developers...
Screed
Originally posted by Kickaha
it's easy to believe that bad scheduling based on G4 specs would lead to really really poor performance by the 970. Relatively speaking, of course.
My understanding is that the 970 has very aggressive out-of-order execution capabilities, so it should be able to deal reasonably well with poorly scheduled code.
I see... A torrent of before and after Panther benchmarks!!!
Although the article makes no mention of them, or perhaps because it makes no mention of them, I have a feeling that if any Powerbook model will get revved to the 970 then it will be closer to the Panther release than to WWDC. Otherwise if a new G4 15" PB is debuted right along side the new PMs, well, that seals the fate of 970 PBs until the end of the year...
Screed
Originally posted by Ti Fighter
Heh, "buy this computer! It will get really really fast in like 2 months... I promise
Most of this compiler/scheduler stuff is way over my head, but I'll tell you this: If there is any real world performance penalty running the 970 under Smeagol, then Apple should wait. They won't get a second chance to make a first impression (bake-off).
Calling Programmer! We need you to weigh in on this.
Well, the 970 PMs are a new generation therefore from P to Q.
Screed
Originally posted by Ensign Pulver
Most of this compiler/scheduler stuff is way over my head, but I'll tell you this: If there is any real world performance penalty running the 970 under Smeagol, then Apple should wait. They won't get a second chance to make a first impression (bake-off).
Calling Programmer! We need you to weigh in on this.
The 970 will be faster than the G4 at the same clock rate, and (depending on what you are doing) potentially much faster. Software that is then recompiled specifically for the 970 will get faster yet although its not clear how much faster -- typically previous generational changes have given a 15-20% performance improvement from just recompiling. Panther (and all of Apple's iApps) will probably be completely recompiled for the new processor, whereas the Jaguar patch will just have the small driver & kernel changes necessary to enable the OS to run on the 970.
If they are running a specific bake-off they could probably use a recompiled version of Photoshop (or whatever).
Further performance improvements are not going to be related to the 970's "64-bitness" as seemed to be implied above. They instead result from the processor's architecture.
Not buying one just because it isn't running at full speed on day 1 is just silly. No radically new hardware is running at its full potential the day it ships. Even the current G4 towers aren't fully utilized, if they were then Panther couldn't deliver any performance improvements as it apparently does. Have no fear, however, the demonstration that Steve does will highlight the machine's performance -- Steve wouldn't have it any other way.
Originally posted by Ti Fighter
maybe these new machines will be only for close apple developers to try along with the pre release of panther to work the bugs out. I mean wwdc is for developers not public consumer announcements. If steve is gonna do show speed tests at wwdc hes gonna use panther.
And everyone at WWDC has been promised a preview release of Panther so anyone taking it home from there won't be using 10.2.x...
Originally posted by Ti Fighter
wwdc is for developers not public consumer announcements.
Not anymore.
Originally posted by Programmer
The 970 will be faster than the G4 at the same clock rate, and (depending on what you are doing) potentially much faster. Software that is then recompiled specifically for the 970 will get faster yet although its not clear how much faster -- typically previous generational changes have given a 15-20% performance improvement from just recompiling. Panther (and all of Apple's iApps) will probably be completely recompiled for the new processor, whereas the Jaguar patch will just have the small driver & kernel changes necessary to enable the OS to run on the 970.
If they are running a specific bake-off they could probably use a recompiled version of Photoshop (or whatever).
Further performance improvements are not going to be related to the 970's "64-bitness" as seemed to be implied above. They instead result from the processor's architecture.
Not buying one just because it isn't running at full speed on day 1 is just silly. No radically new hardware is running at its full potential the day it ships. Even the current G4 towers aren't fully utilized, if they were then Panther couldn't deliver any performance improvements as it apparently does. Have no fear, however, the demonstration that Steve does will highlight the machine's performance -- Steve wouldn't have it any other way.
Thanks Programmer, that backs up everything I've heard about the 970. Any idea why Rothenberg seems to want to talk it down? Is he just confused over the importance of 64 bit, or does he just not understand the issues involved?
I think Apple also needs to give something in the summer to schools and companies with Fall fiscal years to see, so that orders can be made in anticipation of Panther.
This lines up eventually well for holiday media blitz's ... let's hope!
Also would all of these changes to the mobo may also increase downloads to iPods? I thought I read that download rates were not slowed down by FireWire.
The point is that there is alot to potentially crow about when a change like this is made and even if the 970 is a bit hampered, I don't care. This is for DEVELOPERS to see and they will get the Panther beta! That's what we want. Then by the Fall we'll be cheering, "here come the snail ads!!!" Too bad Intel took away the bunny suit guys.
Have no fear, however, the demonstration that Steve does will highlight the machine's performance -- Steve wouldn't have it any other way.
Agreed.
I'm confident we're about to be impressed. If the machine runs at half speed even until Panther...you're still getting a 1.8 G4 on a whoppa of a bus...with altivec cut loose and an extra fpu. (You see any G4s lookin' dat good?) If I wasn't broke after forking out for two design collections and Final Cut Pro 3, I'd be on that bus. Personally, I'm waiting until after Sept.
Heh, I can use my wife's 600 mhz G3 until then. It still seems like lightspeed compared to my old Powermac 604e clone!
I don't think we all quite realise what Steve Jobs is about to hit us with in 2 weeks... Despite all the 'confirmation'...it's all rather quiet and understated from IBM/Apple. It's deafeningly quiet.
Lemon Bon Bon
The point is that there is alot to potentially crow about when a change like this is made and even if the 970 is a bit hampered, I don't care. This is for DEVELOPERS to see and they will get the Panther beta! That's what we want. Then by the Fall we'll be cheering, "here come the snail ads!!!" Too bad Intel took away the bunny suit guys.
Smoken!
Lemon Bon Bon
Originally posted by MacGregor
I wouldn't be too worried about the "on par with G4" comment. There are alot of issues to resolve and even if the new PowerMacs aren't optimized by WWDC, I don't mind having a few months of public tweaking by early-adopters. . .
I think the author just doesn't understand that the 970 will easily out perform a G4 running the same 32 bit applications. I doubt there will need to be any tweaking or tuning to see a big difference right away.
Originally posted by hjordis71
The real question is, How much faster will the 970 run AOL?
3-4 times faster depending on the version corresponding the color of smeagol panther pink or green 64 bitness off cetering the 32 bit version which would be realsed maybe in quarter 3 of the year before next.