An email in Apple's newly formed Xgrid mailing list references the "XServe G5" ("i cannot start up when a XServe G5 is hooked") supposedly from an Xgrid tester.
[edit: this was a hoax]
[edit: this was a hoax]
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
G5 Xserve Reference
- Thread starter MacRumors
- Start date
- Sort by reaction score
G5 XServe release?
Are there any rumors (reliable or not) suggesting a release date for the G5 XServe?
Are there any rumors (reliable or not) suggesting a release date for the G5 XServe?
good news: xserve g5 probobly in late-ish development
bad news: its still has issues
edit: needed to make late>late-ish
bad news: its still has issues
edit: needed to make late>late-ish
It's just a matter of time.... I hope they don't delay the launch for MWNY. That would be silly..
Actually "16 G5's per Xserve" is possible
depends on what G5 IBM sends over...
They are putting 4 chips on each POWER5 MCM and could definitely put many more 970's ;(or 980's)?
Suppose you had a 5 GHz POWER5 equivalent IBM 980 module.
4 chips per MCM and 4 MCM's per Xserve. If you used liquid cooling like they are considering for the large notebooks, then 10
Xserves per rack :
4 x 4 x10 x 14.4GFlops x 2.5 = 5.7 Teraflops
41 1U's would be ~ 22 Teraflops per rack
That is five times more computing power more than what Virginia Tech is putting in a very big closet.
This stuff is getting very interesting.
100 Petaflops is considered "very near" the equivalent to human true mental computing capacity.
Maybe software is the only real issue.
<---wants a VA Tech data link at 10GB/s
depends on what G5 IBM sends over...
They are putting 4 chips on each POWER5 MCM and could definitely put many more 970's ;(or 980's)?
Suppose you had a 5 GHz POWER5 equivalent IBM 980 module.
4 chips per MCM and 4 MCM's per Xserve. If you used liquid cooling like they are considering for the large notebooks, then 10
Xserves per rack :
4 x 4 x10 x 14.4GFlops x 2.5 = 5.7 Teraflops
41 1U's would be ~ 22 Teraflops per rack
That is five times more computing power more than what Virginia Tech is putting in a very big closet.
This stuff is getting very interesting.
100 Petaflops is considered "very near" the equivalent to human true mental computing capacity.
Maybe software is the only real issue.
<---wants a VA Tech data link at 10GB/s
The post to the Xgrid list does exist --
http://lists.apple.com/mhonarc/xgrid-users/msg00070.html
If you mean that you don't believe Mr. Leduc actually has an Xserve G5, well, I'm skeptical too, but I'm sure there must be prototypes somewhere by now...
hehe I ment to say MWSF. Doh!Originally posted by Ambrose Chapel
yeah, especially since there may never be another one!
If you assume the post is legit (which I don't, btw), it doesn't say anything about when an XSG5 may become available -- it sounds like he claims to be having issues when it is "connected" (to what? Xgrid?). So it sounds like either Xgrid or XSG5 aren't ready for prime time, but the other may be. On the other hand, they may be slated to be introduced together.
But yeah, I'm not buyin' it.
But yeah, I'm not buyin' it.
Re: The brain CPU
We can process and render full motion video instantaneously. Think of a scene, any scene, real or imagined... run through it in your head. That's processing and rendering a movie as much as anything in a G5 Xserve could be ;-).
We just don't have a way to export the data, we don't know what format it's in, etc. But our brains can already do what computers do effortlessly and instantaneously.
Should be interesting to see what happens when computers do start reaching processing capabilities on par or beyond that of the human brain. Will they just "wake up" at some point (as Robert A. Heinlein explains in 'The Moon is a Harsh Mistress'), or will they remain unintelligent machines?
Food for thought.
But this is very off-topic. Not sure if the post is real or not, but I think it's simply logical that a G5 Xserve is in the works. The sooner the better!
So long as they work, lol.
How are the Xserves selling anyway? I used to be in IT and still know some people who are, they are taking Apple more seriously than ever before (a year and a half ago they'd laugh if somebody said Macintosh in their presence... now they seem to have an intrigued curiosity). I'm wondering if anybody in IT is actually BUYING Mac servers now.
lol - that's because our brains don't have the codec.
We can process and render full motion video instantaneously. Think of a scene, any scene, real or imagined... run through it in your head. That's processing and rendering a movie as much as anything in a G5 Xserve could be ;-).
We just don't have a way to export the data, we don't know what format it's in, etc. But our brains can already do what computers do effortlessly and instantaneously.
Should be interesting to see what happens when computers do start reaching processing capabilities on par or beyond that of the human brain. Will they just "wake up" at some point (as Robert A. Heinlein explains in 'The Moon is a Harsh Mistress'), or will they remain unintelligent machines?
Food for thought.
But this is very off-topic. Not sure if the post is real or not, but I think it's simply logical that a G5 Xserve is in the works. The sooner the better!
So long as they work, lol.
How are the Xserves selling anyway? I used to be in IT and still know some people who are, they are taking Apple more seriously than ever before (a year and a half ago they'd laugh if somebody said Macintosh in their presence... now they seem to have an intrigued curiosity). I'm wondering if anybody in IT is actually BUYING Mac servers now.
8-Way limit on 970
The 970 has 3 coherent processor interconnects and is thus limited to an 8-way configuration (2^3=8). If you want more you need to go to a grid configuration or use Power4's or Power5's.
By the way, here is a copy of the now infamous post:
The 970 has 3 coherent processor interconnects and is thus limited to an 8-way configuration (2^3=8). If you want more you need to go to a grid configuration or use Power4's or Power5's.
By the way, here is a copy of the now infamous post:
Wow, one rumored product is rumored to have problems working with another rumored product.
I've been here too long, haven't I...
I've been here too long, haven't I...
Re: Re: The brain CPU
Hmm, we are way off topic, but try that little thought exercise again:
Imagine a man walking down a busy city street - now how much detail are you actually seeing of the cars, the shop windows etc, the man's face and clothing, without slowing to a near freeze frame or zooming in to a partial view and "rendering" the details up as it were?
I think the brain is fudging over the details to give the impression of more bandwidth than it actually has. A kind of detail compression codec that prioritises movement over details (Good for hunters and threat assessment). Nowhere near the information required for full motion video.
Or is it just MY brain
Hmm, we are way off topic, but try that little thought exercise again:
Imagine a man walking down a busy city street - now how much detail are you actually seeing of the cars, the shop windows etc, the man's face and clothing, without slowing to a near freeze frame or zooming in to a partial view and "rendering" the details up as it were?
I think the brain is fudging over the details to give the impression of more bandwidth than it actually has. A kind of detail compression codec that prioritises movement over details (Good for hunters and threat assessment). Nowhere near the information required for full motion video.
Or is it just MY brain
You can't compare computational power of your brain with a computer because they are qualitatively different forms of thought.
more "brain cpu " calculations...
8-way only huh?
Is that direct interconnect or can you use a memory/processor crossbar switch?
Also heat issues may make a 16 cpu Xserve a "zServe"...i.e. an IBM custom box..not an Apple system. However it could use a 64 bit Java2 chip and/or other os's. Black zServe AIX box. White zServe True64 or Linux box. Blue zServe a MacOS X or XI box.
100 Petaflops is based on 10 Trillion neurons per brain, neural "noise clock" between 100 and 10,000 usecs, and 20-50:1 data compression using genetic algorithms or similar. The OS would be based on a Doug Lenat type program (OPENCYC iII or something) and use 256 bit emulators for each neuron (1024 bit or 4096 bit emulation possible). This would allow for chemical memory modeling with 3D templates downloading at usec speed ..and even a Russian "PSI model" for outer reality influences in the nsec range. The Russians have measured some 'brain noises' at over 400 MHz..and could be a type of stimulus-response to outer EM type "networks"..quantum noise etc..
Even a quad Xserve would still be better off with a liquid system.
Cray/SGI might build one for Apple if IBM wasn't interested.
Hey Virginia Tech...you guys gonna let us rent time on your "grid system" or not?
<--gee now what
8-way only huh?
Is that direct interconnect or can you use a memory/processor crossbar switch?
Also heat issues may make a 16 cpu Xserve a "zServe"...i.e. an IBM custom box..not an Apple system. However it could use a 64 bit Java2 chip and/or other os's. Black zServe AIX box. White zServe True64 or Linux box. Blue zServe a MacOS X or XI box.
100 Petaflops is based on 10 Trillion neurons per brain, neural "noise clock" between 100 and 10,000 usecs, and 20-50:1 data compression using genetic algorithms or similar. The OS would be based on a Doug Lenat type program (OPENCYC iII or something) and use 256 bit emulators for each neuron (1024 bit or 4096 bit emulation possible). This would allow for chemical memory modeling with 3D templates downloading at usec speed ..and even a Russian "PSI model" for outer reality influences in the nsec range. The Russians have measured some 'brain noises' at over 400 MHz..and could be a type of stimulus-response to outer EM type "networks"..quantum noise etc..
Even a quad Xserve would still be better off with a liquid system.
Cray/SGI might build one for Apple if IBM wasn't interested.
Hey Virginia Tech...you guys gonna let us rent time on your "grid system" or not?
<--gee now what
HOAX
It's totally a hoax.
1) they guy posts HEAVILY NDA'ed info on an open list. He'd never get close to a G5 xServe without having this amply explained to him. He also posts what appears to be an NDA-Agreement number which would have had his xServe pulled faster than you can say... something fast.
2) This was actually pointed out to him on the list... don't talk about NDA'ed products on an open list.
3) He post 2 more notes on the supposed xServe G5 today, even after being warned by list members. Still Apple doesn't come down on him.
4) He'd have to be from a rather important client to beta test an xServe G5 yet his email if from some generic German portal... bluewin.ch Where is the company info? domain? sig?
5) In his last post (still from bluewin.ch) his .sig claims
"Christian Leduc
Apple. Lounge
(Future ;-) Localizing Webmaster Apple.ch / at
Switzerland
cleduc@solnet.ch
(soon: cleduc@lounge.apple.com)"
Can you say wishful thinking? His .sig brags about all the jobs and domains he will have.. some day.
total hoax. I'm surprised he hasn't been banned from the xGrid list yet.
It's totally a hoax.
1) they guy posts HEAVILY NDA'ed info on an open list. He'd never get close to a G5 xServe without having this amply explained to him. He also posts what appears to be an NDA-Agreement number which would have had his xServe pulled faster than you can say... something fast.
2) This was actually pointed out to him on the list... don't talk about NDA'ed products on an open list.
3) He post 2 more notes on the supposed xServe G5 today, even after being warned by list members. Still Apple doesn't come down on him.
4) He'd have to be from a rather important client to beta test an xServe G5 yet his email if from some generic German portal... bluewin.ch Where is the company info? domain? sig?
5) In his last post (still from bluewin.ch) his .sig claims
"Christian Leduc
Apple. Lounge
(Future ;-) Localizing Webmaster Apple.ch / at
Switzerland
cleduc@solnet.ch
(soon: cleduc@lounge.apple.com)"
Can you say wishful thinking? His .sig brags about all the jobs and domains he will have.. some day.
total hoax. I'm surprised he hasn't been banned from the xGrid list yet.