Why did IBM start at the number 970 for this chip? Did they permanently waist the numbers 910 through 960?
Not that I have any inside info from IBM, but I can venture a guess.
In the early and mid 80's everyone was talking about how the Minicomputer (the leading company was "Digital" or "DEC") was going to take over the computing world and push IBM out of business. A minicomputer is sort of midrange between a mainframe and a micro-computer. While that never happened and Digital finally was overtaken by Compaq, IBM did feel a pressure to put out a mini-computer.
IBM considered several alternatives: a Unix server, a proprietary solution or a low-cost mainframe. As usual for IBM, they chose to do all three. Thus was born a computer for running AIX (unix solution. Later incarnations were called RS/6000 and pSeries), the AS/400 (proprietary solution, later called iSeries) and the 970 (a mini-mainframe)
The 970 came out in the late 80's and was a uniprocessor with 2.7 MIPS and up to 16 megabytes of memory. Contemporary "real" mainframes had up to 16 processors with up to 20 MIPS processors. Modern mainframes have up to 16 processors with up to 400 MIPS per processor.
The 970 was also unique in that it was the first mainframe chip made with the CMOS process and cooled by air, so it took a fraction of the space and a fraction of the power of its water-cooled brethren. Only in 1994 did IBM switch its entire line to an air-cooled CMOS process
Anyway, the 970 name has already been used once to signify a low-cost, small footprint alternative. Maybe that's why.
There are 1xx, 4xx, 5xx, 6xx, 7xx and 8xx PowerPC-processors, so its natural that the most powerful of all will be the 9xx. The most powerful PowerPCs to date are the 7457 and 8560, and IBM just might prepare for a x6x from Motorola and they wanted a head start to the x7x, hence the 970.
I really don't worry about the numbers running out even if they start so late in the series running from 0 til 9. The 750 have evolved quite well and they havn't even changed the number.. 750FX is the latest and it have gained a whole lot of good stuff along the way from just being 750.
The Power5-derivative just might be called 970FX for all we know, but that's not like considering it'd be using a completely new core compared to 970. 980 seems reasonable, but so does 9700. Are there any official word from IBM what it will be called or are just everyone just assuming it will be called 980?
Yeah, Apple might just have G4-powered landers roaming Mars right now!
Why do you think it's a so far fetched idea for Apple to have OSX today running on computers sporting Power-processors? If I were Apple and I planned to move to a 64-bit PowerPC platform I'd use the only 64-bit PowerPC plattform available to get a head start in the developmet, and these have been available for several years now.
I really expect from programmers like Avie to just drool over the opportunity to port OSX to pSeries IBM-servers. Just for the fun of it! I'm not an expert in such matters but how hard can it be, really?
I really expect from programmers like Avie to just drool over the opportunity to port OSX to pSeries IBM-servers. Just for the fun of it! I'm not an expert in such matters but how hard can it be, really?
That depends. Just getting it to work would mean some changes in the kernel, probably some drivers. Think a five-person team working 2-3 months.
Getting the most out of the different processor would probably involve more major changes and optimization. You might also want to change the compiler to make better optimized code. Two 5-person teams working for 6 months.
Not a great effort, but considerably more than what you'd do just for the fun of it. Apple would not do it unless they were considering adding a midrange server to their lineup.
Getting the most out of the different processor would probably involve more major changes and optimization. You might also want to change the compiler to make better optimized code. Two 5-person teams working for 6 months.
I don't think they'll have to do any major changes or optimization to make osX run on the POWER4 as the POWER ISA is a superset of the PowerPC ISA and is completely compatible with the PowerPC. Most PowerPC code will run on the POWER4 without any problems at all.
I think Apple has been playing around with different processors for a while and since the POWER series from IBM is a "PowerPC" why wouldn't they at least have some test systems in their labs. I'm certain that Apple have talked to IBM for a few years of taking over as a supplier for Apples high-end processors. When IBM told Apple that they were *going* to make a low-cost version of the 64 bit POWER4, I'm sure the first thing Apple would do was to try to build some 64 bit systems to do testing on. And what other candidate did they have than the - pause - POWER4.
I don't think they'll have to do any major changes or optimization to make osX run on the POWER4 as the POWER ISA is a superset of the PowerPC ISA and is completely compatible with the PowerPC. Most PowerPC code will run on the POWER4 without any problems at all.
No, they don't have to do something major to make it run, but to get the most out of it, they'll probably have to optimize alot.
And I too believe that Apple must have had some test-systems with POWER4-systems inhouse, since getting their OS to run on the POWER4 is a very similiar issue to getting it running on a 970. To sit and wait for the 970 doesn't sound like the greatest plan to me
No, they don't have to do something major to make it run, but to get the most out of it, they'll probably have to optimize alot.
And I too believe that Apple must have had some test-systems with POWER4-systems inhouse, since getting their OS to run on the POWER4 is a very similiar issue to getting it running on a 970. To sit and wait for the 970 doesn't sound like the greatest plan to me
Low end servers for IBM start around $20,000 or so.
Not quite. IBM entry level RS/6000 B50 UNIX-server actually start at $3600. For that price it's vastly under powered and would be eaten performance wise by an new iMac (or perhaps even an iBook) any day, considering it sports a 604e processor clocked at 375 MHz.
The entry level pSeries machine, the 610, costs just over $6800 and it has a 333 MHz Power3 processor. Too expensive also.
So, that's what IBM consideres an entry level server.
The Power5 would not just go into low end servers, the'd be part of blade clusters as well, and those just can't cost $20 000 per module, that'd just be absurd. IBM is taking on Itanium servers here, and they just can't overcharge like they're used to. And really.. if they want to sell Linux-servers in large numbers, and why wouldn't they? They have really no reason to sell these machines at ten time the price that HP would.
Getting the most out of the different processor would probably involve more major changes and optimization. You might also want to change the compiler to make better optimized code. Two 5-person teams working for 6 months.
Your forget that IBM is investing 1 billion dollars a year in Linux development. We know that IBM and Red Hat are optimizing gcc for PowerPC, Altivec and 64-bitness, so this isn't a battle Apple have to do alone.
Why do you think it's a so far fetched idea for Apple to have OSX today running on computers sporting Power-processors? If I were Apple and I planned to move to a 64-bit PowerPC platform I'd use the only 64-bit PowerPC plattform available to get a head start in the developmet, and these have been available for several years now.
I really expect from programmers like Avie to just drool over the opportunity to port OSX to pSeries IBM-servers. Just for the fun of it! I'm not an expert in such matters but how hard can it be, really?
Considering that Apple has known the 970 is coming for 2-3 years now, and that it is basically a desktop-POWER4, it would be an obvious move to have bought a POWER4 machine or two as soon as they were available and port MacOSX to it. This would let them evaluate whether the 970 would make decent Mac well before it was actually available... and if they were impressed (like there is any question of that!) then that would give them a leg up on delivering Panther.
Considering that Apple has known the 970 is coming for 2-3 years now, and that it is basically a desktop-POWER4, it would be an obvious move to have bought a POWER4 machine or two as soon as they were available and port MacOSX to it. This would let them evaluate whether the 970 would make decent Mac well before it was actually available... and if they were impressed (like there is any question of that!) then that would give them a leg up on delivering Panther.
Is it possible that having the POWER4 in the labs would help them towards developing the companion chips and the motherbord for the 970?
Are there any similarities except from the ISA and 64-bitness that would help them begin the developement of the new PowerMac hardware before they'd actually have the 970 at hand?
The Power4 uses a different external interconnect architecture to the 970. Two basic differences are that the memory controller is on chip, and the bus is different.
I don't understand why people are concerned that Apple wouldn't have a MLB ready in time. Apple would surely have access to development info and prototype since the 970 project started. Apple has tonnes of engineering talent. The 970 is possibly the most important project at Apple since the iMac, and probably as important as Mac OS X.
Is it possible that having the POWER4 in the labs would help them towards developing the companion chips and the motherbord for the 970?
Are there any similarities except from the ISA and 64-bitness that would help them begin the developement of the new PowerMac hardware before they'd actually have the 970 at hand?
No, it wouldn't help with hardware development but it would help with software. The change to 64-bit, a deeply pipelined out-of-order processor, a different cache architecture, etc all can make quite a difference to the software development and the sooner they figure those issues out, the better.
It's supposed to be everything the Power4 was, just a whole lot faster and far more scaleable; in other words, without much modification, it's supposed to fit just fine in everything from blades, and all the way up to the big iron.
The 980 is actually what is being reported as being the blade CPU.
It's supposed to be everything the Power4 was, just a whole lot faster and far more scaleable; in other words, without much modification, it's supposed to fit just fine in everything from blades, and all the way up to the big iron.
The 980 is actually what is being reported as being the blade CPU.
No, it wouldn't help with hardware development but it would help with software. The change to 64-bit, a deeply pipelined out-of-order processor, a different cache architecture, etc all can make quite a difference to the software development and the sooner they figure those issues out, the better.
Is there any chance that big software houses like Adobe, Quark and MS have had POWER4 systems to test software on, or do you think they'll keep the 64-bit project in-house until they have finished their own hardware prototypes. We have all heard rumors about the sealed rooms that were supposed to hide secret apple hardware.
My point is: Do software developers need the hardware before release to iron out bugs and incompabilities (stupid question I know) and it they do, how much time would they need to test new hardware like this.
And, if the 970 is being tested at various software houses around the world by now, how tight lipped could these people really be???
Comments
Originally posted by Locomotive
Why did IBM start at the number 970 for this chip? Did they permanently waist the numbers 910 through 960?
Not that I have any inside info from IBM, but I can venture a guess.
In the early and mid 80's everyone was talking about how the Minicomputer (the leading company was "Digital" or "DEC") was going to take over the computing world and push IBM out of business. A minicomputer is sort of midrange between a mainframe and a micro-computer. While that never happened and Digital finally was overtaken by Compaq, IBM did feel a pressure to put out a mini-computer.
IBM considered several alternatives: a Unix server, a proprietary solution or a low-cost mainframe. As usual for IBM, they chose to do all three. Thus was born a computer for running AIX (unix solution. Later incarnations were called RS/6000 and pSeries), the AS/400 (proprietary solution, later called iSeries) and the 970 (a mini-mainframe)
The 970 came out in the late 80's and was a uniprocessor with 2.7 MIPS and up to 16 megabytes of memory. Contemporary "real" mainframes had up to 16 processors with up to 20 MIPS processors. Modern mainframes have up to 16 processors with up to 400 MIPS per processor.
The 970 was also unique in that it was the first mainframe chip made with the CMOS process and cooled by air, so it took a fraction of the space and a fraction of the power of its water-cooled brethren. Only in 1994 did IBM switch its entire line to an air-cooled CMOS process
Anyway, the 970 name has already been used once to signify a low-cost, small footprint alternative. Maybe that's why.
I really don't worry about the numbers running out even if they start so late in the series running from 0 til 9. The 750 have evolved quite well and they havn't even changed the number.. 750FX is the latest and it have gained a whole lot of good stuff along the way from just being 750.
The Power5-derivative just might be called 970FX for all we know, but that's not like considering it'd be using a completely new core compared to 970. 980 seems reasonable, but so does 9700. Are there any official word from IBM what it will be called or are just everyone just assuming it will be called 980?
Originally posted by Clive
Yeah, Apple might just have G4-powered landers roaming Mars right now!
Why do you think it's a so far fetched idea for Apple to have OSX today running on computers sporting Power-processors? If I were Apple and I planned to move to a 64-bit PowerPC platform I'd use the only 64-bit PowerPC plattform available to get a head start in the developmet, and these have been available for several years now.
I really expect from programmers like Avie to just drool over the opportunity to port OSX to pSeries IBM-servers. Just for the fun of it! I'm not an expert in such matters but how hard can it be, really?
Originally posted by Henriok
I really expect from programmers like Avie to just drool over the opportunity to port OSX to pSeries IBM-servers. Just for the fun of it! I'm not an expert in such matters but how hard can it be, really?
That depends. Just getting it to work would mean some changes in the kernel, probably some drivers. Think a five-person team working 2-3 months.
Getting the most out of the different processor would probably involve more major changes and optimization. You might also want to change the compiler to make better optimized code. Two 5-person teams working for 6 months.
Not a great effort, but considerably more than what you'd do just for the fun of it. Apple would not do it unless they were considering adding a midrange server to their lineup.
Originally posted by synp
Getting the most out of the different processor would probably involve more major changes and optimization. You might also want to change the compiler to make better optimized code. Two 5-person teams working for 6 months.
I don't think they'll have to do any major changes or optimization to make osX run on the POWER4 as the POWER ISA is a superset of the PowerPC ISA and is completely compatible with the PowerPC. Most PowerPC code will run on the POWER4 without any problems at all.
I think Apple has been playing around with different processors for a while and since the POWER series from IBM is a "PowerPC" why wouldn't they at least have some test systems in their labs. I'm certain that Apple have talked to IBM for a few years of taking over as a supplier for Apples high-end processors. When IBM told Apple that they were *going* to make a low-cost version of the 64 bit POWER4, I'm sure the first thing Apple would do was to try to build some 64 bit systems to do testing on. And what other candidate did they have than the - pause - POWER4.
Just my thoughts - no inside information
[Edit: Formatting]
Originally posted by NETROMac
I don't think they'll have to do any major changes or optimization to make osX run on the POWER4 as the POWER ISA is a superset of the PowerPC ISA and is completely compatible with the PowerPC. Most PowerPC code will run on the POWER4 without any problems at all.
No, they don't have to do something major to make it run, but to get the most out of it, they'll probably have to optimize alot.
And I too believe that Apple must have had some test-systems with POWER4-systems inhouse, since getting their OS to run on the POWER4 is a very similiar issue to getting it running on a 970. To sit and wait for the 970 doesn't sound like the greatest plan to me
Originally posted by r-0X#Zapchud
No, they don't have to do something major to make it run, but to get the most out of it, they'll probably have to optimize alot.
And I too believe that Apple must have had some test-systems with POWER4-systems inhouse, since getting their OS to run on the POWER4 is a very similiar issue to getting it running on a 970. To sit and wait for the 970 doesn't sound like the greatest plan to me
I completely agree with you r-0X#Zapchud
Originally posted by jante99
Low end servers for IBM start around $20,000 or so.
Not quite. IBM entry level RS/6000 B50 UNIX-server actually start at $3600. For that price it's vastly under powered and would be eaten performance wise by an new iMac (or perhaps even an iBook) any day, considering it sports a 604e processor clocked at 375 MHz.
The entry level pSeries machine, the 610, costs just over $6800 and it has a 333 MHz Power3 processor. Too expensive also.
So, that's what IBM consideres an entry level server.
The Power5 would not just go into low end servers, the'd be part of blade clusters as well, and those just can't cost $20 000 per module, that'd just be absurd. IBM is taking on Itanium servers here, and they just can't overcharge like they're used to. And really.. if they want to sell Linux-servers in large numbers, and why wouldn't they? They have really no reason to sell these machines at ten time the price that HP would.
Originally posted by synp
Getting the most out of the different processor would probably involve more major changes and optimization. You might also want to change the compiler to make better optimized code. Two 5-person teams working for 6 months.
Your forget that IBM is investing 1 billion dollars a year in Linux development. We know that IBM and Red Hat are optimizing gcc for PowerPC, Altivec and 64-bitness, so this isn't a battle Apple have to do alone.
Originally posted by Henriok
Why do you think it's a so far fetched idea for Apple to have OSX today running on computers sporting Power-processors? If I were Apple and I planned to move to a 64-bit PowerPC platform I'd use the only 64-bit PowerPC plattform available to get a head start in the developmet, and these have been available for several years now.
I really expect from programmers like Avie to just drool over the opportunity to port OSX to pSeries IBM-servers. Just for the fun of it! I'm not an expert in such matters but how hard can it be, really?
Considering that Apple has known the 970 is coming for 2-3 years now, and that it is basically a desktop-POWER4, it would be an obvious move to have bought a POWER4 machine or two as soon as they were available and port MacOSX to it. This would let them evaluate whether the 970 would make decent Mac well before it was actually available... and if they were impressed (like there is any question of that!) then that would give them a leg up on delivering Panther.
Originally posted by Programmer
Considering that Apple has known the 970 is coming for 2-3 years now, and that it is basically a desktop-POWER4, it would be an obvious move to have bought a POWER4 machine or two as soon as they were available and port MacOSX to it. This would let them evaluate whether the 970 would make decent Mac well before it was actually available... and if they were impressed (like there is any question of that!) then that would give them a leg up on delivering Panther.
Is it possible that having the POWER4 in the labs would help them towards developing the companion chips and the motherbord for the 970?
Are there any similarities except from the ISA and 64-bitness that would help them begin the developement of the new PowerMac hardware before they'd actually have the 970 at hand?
I don't understand why people are concerned that Apple wouldn't have a MLB ready in time. Apple would surely have access to development info and prototype since the 970 project started. Apple has tonnes of engineering talent. The 970 is possibly the most important project at Apple since the iMac, and probably as important as Mac OS X.
Barto
Originally posted by NETROMac
Is it possible that having the POWER4 in the labs would help them towards developing the companion chips and the motherbord for the 970?
Are there any similarities except from the ISA and 64-bitness that would help them begin the developement of the new PowerMac hardware before they'd actually have the 970 at hand?
No, it wouldn't help with hardware development but it would help with software. The change to 64-bit, a deeply pipelined out-of-order processor, a different cache architecture, etc all can make quite a difference to the software development and the sooner they figure those issues out, the better.
The 980 is actually what is being reported as being the blade CPU.
Originally posted by RBR
It's supposed to be everything the Power4 was, just a whole lot faster and far more scaleable; in other words, without much modification, it's supposed to fit just fine in everything from blades, and all the way up to the big iron.
The 980 is actually what is being reported as being the blade CPU.
What blades?
Originally posted by Programmer
No, it wouldn't help with hardware development but it would help with software. The change to 64-bit, a deeply pipelined out-of-order processor, a different cache architecture, etc all can make quite a difference to the software development and the sooner they figure those issues out, the better.
Is there any chance that big software houses like Adobe, Quark and MS have had POWER4 systems to test software on, or do you think they'll keep the 64-bit project in-house until they have finished their own hardware prototypes. We have all heard rumors about the sealed rooms that were supposed to hide secret apple hardware.
My point is: Do software developers need the hardware before release to iron out bugs and incompabilities (stupid question I know) and it they do, how much time would they need to test new hardware like this.
And, if the 970 is being tested at various software houses around the world by now, how tight lipped could these people really be???
Originally posted by NETROMac
What blades?
Don't believe IBM has any blade servers running risc chips.
Originally posted by trailmaster308
Don't believe IBM has any blade servers running risc chips.
And if there were, the 970 would be the candidate for such a product (PR from IBM Germany) NOT the 980, which is not even on IBM's roadmaps yet.
Originally posted by RBR
The 980 is actually what is being reported as being the blade CPU.
No it isn't. There are no reports of 980 at all, just speculation. The Power5 and the 970 on the other hand are being reported for blade use.