Dual 3.07ghz G5, the clock speed would edge out the pentium4 and the speed, dear god, the speed!!! this will never happen at wwdc lest hell freezes over
Call me crazy but I think the iMac has one more speed bump left in it before it goes to G5. Apple's desktop sales are slipping and it's not because people are waiting for a 64-bit iMac.
G4 or G5, a new form factor has to come. This is why sales are dropping.
EDIT: Alos, regarding the iPod, Apple would be STUPID to make the iPod smaller. It's already small enough, and people who need something smaller can get the mini. Maybe the internal stuff in the iPod will shrink, but that will leave more room for BATTERY.
I agree with HOM, the fact that the G5 would have migrated so quickly to Apple's consumer desktop would be of tremendous interest to developers. Suddenly the customer base for their G5 optimized code more than doubles.
True, but irrelevant. Unless it also introduced something that developers would have to account for somehow, Apple could release an iMac G5 at any time and developers would be tremendously interested.
What matters is not whether developers could possibly be interested in it - after all, there are probably developers interested in the next game in the Escape Velocity series - but whether it's something that impacts development. An iMac G5 wouldn't impact development. It could impact market size and sales, but WWDC isn't about that.
I'm glad you have such high regards for the new FX's. Lets hope that IBM has things squared away for the higher clock rate devices that we are all waiting on.
But that is about rumors what I'd really like from you is more info on your XServe. How does it perform, what type of modeling are you doing on it, eitc & etc. Also a little more technical, are you using this as a compute server and if so how are you managing resources? In other words MPI or something like that involved in your work?
Thanks
Dave
Dave, I am using a dual G5 Xserve (2 Ghz) in conjunction with a dual 2.0 GHz PowerMac G5 to run a slightly modified Rutgers Ocean Model. The model is written in FORTRAN 90, and I use XLF. I do distributed runs with LAM-MPI. LAM-MPI works very well on the Macs and is actively developed. MPICH on the mac is absolutely horrible.
The model is not very memory intensive, but, very computational expensive (double-precision floating point). I get very good performance; however, I don't have a direct comparison for you. Using 4 G5 CPU's, it is just as fast (or faster) than a 16 CPU Baewolf cluster that I tried it on; however, don't assume too much from this as that cluster had a variety of Pentium III and IV machines at varying speeds, motherboards, ethernet cards, etc. I have not tried to compare against Xeons or the 64-bit Athlon (I don't have access to any of those).
Anyhow, over the continuous model run I did for 12 days (on just the Xserve), the internal CPU temperatures (as given by the Server Monitor program) never got above 145F and the case stayed around 130F. Again, this was with each CPU at 85-95% usage. Pretty impressive, I think.
Well thats good news. How do you know they have been producing the 970fx for the last month with no flaws? Any url or is this just word of mouth?
This is more from looking at Xserve deliveries. The Apple deliveries of the 2.0GHz really ramped up around three-four weeks ago and I promptly received mine (after having been told the delay was until the end of May). For Apple to have filled most of the Xserve orders now means that IBM must have worked it out over a month ago.
I just noticed that Appleinsider and others have now written about this:
-This will include more Unix libraries. Perhaps CigWin api's (wishful thinking).
-Also migrating design with code with Interface builder will be a lot easier (perhaps closer to what .net does).
-Support for more languages: ppc assembly (has support now but could be better), algorithm coloring (hopefull on this), PHP.
Given that Cygwin is for Windows only, I see no point of having it in something for the Mac. I'm hoping for code-completion for Java and CocoaJava (which is rumoured to be coming). Oh, and removing the bug where if you accidentally click build twice Xcode crashes.
The model is not very memory intensive, but, very computational expensive (double-precision floating point). I get very good performance; however, I don't have a direct comparison for you. Using 4 G5 CPU's, it is just as fast (or faster) than a 16 CPU Baewolf cluster that I tried it on; however, don't assume too much from this as that cluster had a variety of Pentium III and IV machines at varying speeds, motherboards, ethernet cards, etc. I have not tried to compare against Xeons or the 64-bit Athlon (I don't have access to any of those).
Thanks for the response, it is always good to see how people in the field are doing with Apple's technology. Once Apple has production of their hardware stabilized they need to promote this capability extensively.
Quote:
Anyhow, over the continuous model run I did for 12 days (on just the Xserve), the internal CPU temperatures (as given by the Server Monitor program) never got above 145F and the case stayed around 130F. Again, this was with each CPU at 85-95% usage. Pretty impressive, I think.
One has to wonder how much head room we now have for higher clock rates. Especially in a machine with the G5 Towers cooling capability.
Given that Cygwin is for Windows only, I see no point of having it in something for the Mac. I'm hoping for code-completion for Java and CocoaJava (which is rumoured to be coming). Oh, and removing the bug where if you accidentally click build twice Xcode crashes.
Because it would be possible to develop windows applications on a mac using linux libraries that would work in both *nix and win worlds.
Comments
Babelfish translation:
http://croquer.free.fr
Originally posted by Michael Grey
Call me crazy but I think the iMac has one more speed bump left in it before it goes to G5. Apple's desktop sales are slipping and it's not because people are waiting for a 64-bit iMac.
G4 or G5, a new form factor has to come. This is why sales are dropping.
EDIT: Alos, regarding the iPod, Apple would be STUPID to make the iPod smaller. It's already small enough, and people who need something smaller can get the mini. Maybe the internal stuff in the iPod will shrink, but that will leave more room for BATTERY.
Originally posted by Ensign Pulver
I agree with HOM, the fact that the G5 would have migrated so quickly to Apple's consumer desktop would be of tremendous interest to developers. Suddenly the customer base for their G5 optimized code more than doubles.
True, but irrelevant. Unless it also introduced something that developers would have to account for somehow, Apple could release an iMac G5 at any time and developers would be tremendously interested.
What matters is not whether developers could possibly be interested in it - after all, there are probably developers interested in the next game in the Escape Velocity series
Originally posted by wizard69
Hi Atom
I'm glad you have such high regards for the new FX's. Lets hope that IBM has things squared away for the higher clock rate devices that we are all waiting on.
But that is about rumors what I'd really like from you is more info on your XServe. How does it perform, what type of modeling are you doing on it, eitc & etc. Also a little more technical, are you using this as a compute server and if so how are you managing resources? In other words MPI or something like that involved in your work?
Thanks
Dave
Dave, I am using a dual G5 Xserve (2 Ghz) in conjunction with a dual 2.0 GHz PowerMac G5 to run a slightly modified Rutgers Ocean Model. The model is written in FORTRAN 90, and I use XLF. I do distributed runs with LAM-MPI. LAM-MPI works very well on the Macs and is actively developed. MPICH on the mac is absolutely horrible.
The model is not very memory intensive, but, very computational expensive (double-precision floating point). I get very good performance; however, I don't have a direct comparison for you. Using 4 G5 CPU's, it is just as fast (or faster) than a 16 CPU Baewolf cluster that I tried it on; however, don't assume too much from this as that cluster had a variety of Pentium III and IV machines at varying speeds, motherboards, ethernet cards, etc. I have not tried to compare against Xeons or the 64-bit Athlon (I don't have access to any of those).
Anyhow, over the continuous model run I did for 12 days (on just the Xserve), the internal CPU temperatures (as given by the Server Monitor program) never got above 145F and the case stayed around 130F. Again, this was with each CPU at 85-95% usage. Pretty impressive, I think.
Originally posted by emig647
Well thats good news. How do you know they have been producing the 970fx for the last month with no flaws? Any url or is this just word of mouth?
This is more from looking at Xserve deliveries. The Apple deliveries of the 2.0GHz really ramped up around three-four weeks ago and I promptly received mine (after having been told the delay was until the end of May). For Apple to have filled most of the Xserve orders now means that IBM must have worked it out over a month ago.
I just noticed that Appleinsider and others have now written about this:
<http://www.appleinsider.com/article.php?id=451>
Originally posted by emig647
XCode 2.0
-This will include more Unix libraries. Perhaps CigWin api's (wishful thinking).
-Also migrating design with code with Interface builder will be a lot easier (perhaps closer to what .net does).
-Support for more languages: ppc assembly (has support now but could be better), algorithm coloring (hopefull on this), PHP.
Given that Cygwin is for Windows only, I see no point of having it in something for the Mac. I'm hoping for code-completion for Java and CocoaJava (which is rumoured to be coming). Oh, and removing the bug where if you accidentally click build twice Xcode crashes.
Originally posted by atomicham
The model is not very memory intensive, but, very computational expensive (double-precision floating point). I get very good performance; however, I don't have a direct comparison for you. Using 4 G5 CPU's, it is just as fast (or faster) than a 16 CPU Baewolf cluster that I tried it on; however, don't assume too much from this as that cluster had a variety of Pentium III and IV machines at varying speeds, motherboards, ethernet cards, etc. I have not tried to compare against Xeons or the 64-bit Athlon (I don't have access to any of those).
Thanks for the response, it is always good to see how people in the field are doing with Apple's technology. Once Apple has production of their hardware stabilized they need to promote this capability extensively.
Anyhow, over the continuous model run I did for 12 days (on just the Xserve), the internal CPU temperatures (as given by the Server Monitor program) never got above 145F and the case stayed around 130F. Again, this was with each CPU at 85-95% usage. Pretty impressive, I think.
One has to wonder how much head room we now have for higher clock rates. Especially in a machine with the G5 Towers cooling capability.
Originally posted by beamso
Given that Cygwin is for Windows only, I see no point of having it in something for the Mac. I'm hoping for code-completion for Java and CocoaJava (which is rumoured to be coming). Oh, and removing the bug where if you accidentally click build twice Xcode crashes.
Because it would be possible to develop windows applications on a mac using linux libraries that would work in both *nix and win worlds.
Originally posted by emig647
Because it would be possible to develop windows applications on a mac using linux libraries that would work in both *nix and win worlds.
Is Qt not adequate to the task?
Heck, you can do this with CodeWarrior's PowerPlant, too. It's not a "Linux library" but it's a mature cross-platform application kit.
Originally posted by Amorph
Is Qt not adequate to the task?
Heck, you can do this with CodeWarrior's PowerPlant, too. It's not a "Linux library" but it's a mature cross-platform application kit.
Qt would work too. Actually I'd rather prefer Qt but there isn't a whole lot of documentation out there for it...