Maybe Motorola isn't going anywhere, but these time lines do not look terribly favorable for near term upgrades if Apple is going to continue to use Moto's G4's in the consumer machines.
"A PowerPC processor made for performance in 90-nm SOI CMOS-and for economic efficiency on 300-mm-diameter wafers-should be in manufacture at Crolles in 2004."
"For now, work is under way at Motorola's Dan Noble Center (Austin) to bring the process online at Motorola's 200-mm MOS-13 fab. A test line for SOI chips should be running in Crolles sometime this summer."
[B]True, but the G4 needs expensive L3 cache to boost performance, let alone any speculation on yield issues.
I`m not sure if the L3 ram is realy so extreme expensive.
Perhaps sombody can elaborate on the price!
But, as you stated, they're still left using @ best a 200MGz FSB, thus aggravating an already excruciatingly, painfully, woefully choked FSB. Their answer, support for even more, up to 4MB/processor, very expensive L3 cache.
How much faster would be G4 with DDR FSB. We can only speculate, but remember the time as AMD Athlon implemented a DDR FSB - At Tom`s Harware Guide, they made intensive benchmarking about and found out a 10% performance increase.
Now think about it - you must redesign all parts conected with FSB or simpy get a pin to pin replacement in form of 7457 without any additional investment.
You hit the nail on the head with the statement,"any time soon". The last information from Motorola had the MPC7457-rm(re: using Rapid I/O) as only "proposed".
If and when this cpu appears it won't be before 2005, where will IBM be in this time frame. I like Motorola, and they have apparantly done what they could for Apple in getting the MPC7455 up to an astonishing 1.42GHz, but I hold no hope for miracles from them. Maybe collaboration w/ Phillips and SMT will speed any development, but the future of Apple's consumer machines looks fairly grim if they stay w/ Motorola cpu's.
If Moto is to continue with desktop cpu then for sure with Rapidio. The direction was mentioned by Moto in many publications. Moto took enough time with Rapidio and made it right. There are paralel and serial standards. HT on the other hand made a lot of compromises. There is a good reading at: http://www.rapidio.org/data/press/diverging_reqs.pdf
The problem is that Motorola isn't yet shipping any RapidIO processors -- they are way behind schedule. The 7457-RM makes a lot of sense for Motorola in their embedded business, but now that Apple is clearly using HyperTransport its a little less clear that it makes sense for Apple to use. Better for Apple would be a 7457-HM which would communicate with the I/O chipset using HyperTransport, plus have an on-chip memory controller. The best and easiest might actually be a 7457-M which just brings the memory controller onto the G4 but retains the MPX bus for communicating with the I/O chipset. This would replace the L3 cache interface with a full memory subsystem but retain the current bus interface... a decent solution since most of the traffic is removed from the MPX bus.
The problem is that Motorola isn't yet shipping any RapidIO processors -- they are way behind schedule....
Saying "way behind"
is being very kind.
he-he
I would think Motorola's biggest concern with future processors right now is with the MPC8540/MPC8560, both of which have absolutely no use as desktop cpu's. These processors will really make a large impact in the embedded space.
Come on Motorola, I am rooting for you, just a bit disappointed the last 2 1/2 years. A MPC7457-rm is very compelling, but not in 2005.
"Will Strauss, president of market research firm Forward Concepts (Tempe, Ariz.), said Apple CEO Steve Jobs had been unhappy for some time with the Motorola MPUs' high-end performance. Some of the most senior Motorola Semiconductor Products Sector executives have told tales of middle-of-the-night phone calls from Jobs, who would rant about Motorola's alleged inability to match Jobs' clock-speed expectations. Jobs also would complain about Motorola's alleged inability to supply Apple with MPUs in sufficient quantities, sources said."
"middle of the night phone calls" gotta laugh, wish those conversations were downloadable from the iTunes Music Store.
G4 by Motorola a flop for 4 years due to manufacturing problems
G3 by Motorola has manufacturing problems, high end G3 manufactured and designed by IBM
"G2" aka 604 by Motorola has manufacturing problems, high end 604E manufactured and designed by IBM
68040 by Motorola can not compete with 486, the replacing 68050 and 68060 are delayed and forces Apple to migrate to PPC.
CPUs is a "perishable" product, they go from fresh to stale quite quickly. To be able to manufacture in time is very very important. That Motorola had a "G5" protoype at 2 GHz two years ago or that their roadmaps state things does not make a single Macintosh any faster, manufactured CPUs does
Wellcome IBM, I like to see Apple with a "big blue" backgroud
I'm not anti-Motorola, and I'd venture to guess most people here aren't anti-Motorola. However, Motorola has, for whatever reason, not developed the G4 in a timely manner and has yet to introduce their flagship processors, the MPC8540's and MPC8560's. These were to be available the second half of 2002, then beginning 2003, and yet the embedded guys still wait(re: yes I know these aren't intended for desktops).
Most people here would be very happy, if not ecstatic over a G5 from Motorola that included items from their Roadmap; multicore superscaler, Rapid I/O, Ocean Switched fabric, 0.13µm die, etc. But this seems like it may not even appear in a desktop version(re: it was only listed as PROPOSED in a Motorola pdf making the internet rumor circulation departments).
If Motorla can't get their next generation processors using Rapid I/O out relatively quickly, Apple may not have to worry about getting any cpu's from them because they may be out of the semiconductor business. We'll see how their agreements w/ Phillips and SMT pan out, hope they do, because it can only help Apple to have at least 2 suppliers for their chips.
I'm not at all anti-Motorola, but I certainly haven't been impressed by their management or fab capabilities. For a long time they had a strong processor design team, but many (most?) of those were hired away by Intel about 5 years ago (mostly due to mis-management at Motorola).
Personally I think we (and Apple) would be better off with two healthy & capable PowerPC providers. Apple by itself is not big enough, currently, to sustain this but it would increase the PowerPC's other markets and might bring increased attention to the processor family from the Linux crowd.
Comments
Apple plucks 64-bit IBM processor for Power Mac
Maybe Motorola isn't going anywhere, but these time lines do not look terribly favorable for near term upgrades if Apple is going to continue to use Moto's G4's in the consumer machines.
"A PowerPC processor made for performance in 90-nm SOI CMOS-and for economic efficiency on 300-mm-diameter wafers-should be in manufacture at Crolles in 2004."
"For now, work is under way at Motorola's Dan Noble Center (Austin) to bring the process online at Motorola's 200-mm MOS-13 fab. A test line for SOI chips should be running in Crolles sometime this summer."
[B]True, but the G4 needs expensive L3 cache to boost performance, let alone any speculation on yield issues.
I`m not sure if the L3 ram is realy so extreme expensive.
Perhaps sombody can elaborate on the price!
But, as you stated, they're still left using @ best a 200MGz FSB, thus aggravating an already excruciatingly, painfully, woefully choked FSB. Their answer, support for even more, up to 4MB/processor, very expensive L3 cache.
How much faster would be G4 with DDR FSB. We can only speculate, but remember the time as AMD Athlon implemented a DDR FSB - At Tom`s Harware Guide, they made intensive benchmarking about and found out a 10% performance increase.
Now think about it - you must redesign all parts conected with FSB or simpy get a pin to pin replacement in form of 7457 without any additional investment.
Read this: DDR-SDRAM Has Finally Arrived
http://www6.tomshardware.com/motherb...030/index.html
You hit the nail on the head with the statement,"any time soon". The last information from Motorola had the MPC7457-rm(re: using Rapid I/O) as only "proposed".
If and when this cpu appears it won't be before 2005, where will IBM be in this time frame. I like Motorola, and they have apparantly done what they could for Apple in getting the MPC7455 up to an astonishing 1.42GHz, but I hold no hope for miracles from them. Maybe collaboration w/ Phillips and SMT will speed any development, but the future of Apple's consumer machines looks fairly grim if they stay w/ Motorola cpu's.
If Moto is to continue with desktop cpu then for sure with Rapidio. The direction was mentioned by Moto in many publications. Moto took enough time with Rapidio and made it right. There are paralel and serial standards. HT on the other hand made a lot of compromises. There is a good reading at: http://www.rapidio.org/data/press/diverging_reqs.pdf
Rooster
Originally posted by Programmer
The problem is that Motorola isn't yet shipping any RapidIO processors -- they are way behind schedule....
Saying "way behind"
is being very kind.
he-he
I would think Motorola's biggest concern with future processors right now is with the MPC8540/MPC8560, both of which have absolutely no use as desktop cpu's. These processors will really make a large impact in the embedded space.
Come on Motorola, I am rooting for you, just a bit disappointed the last 2 1/2 years. A MPC7457-rm is very compelling, but not in 2005.
Originally posted by neumac
This from last week's EE Times
Apple plucks 64-bit IBM processor for Power Mac
Best quote of the day is from this article.
"Will Strauss, president of market research firm Forward Concepts (Tempe, Ariz.), said Apple CEO Steve Jobs had been unhappy for some time with the Motorola MPUs' high-end performance. Some of the most senior Motorola Semiconductor Products Sector executives have told tales of middle-of-the-night phone calls from Jobs, who would rant about Motorola's alleged inability to match Jobs' clock-speed expectations. Jobs also would complain about Motorola's alleged inability to supply Apple with MPUs in sufficient quantities, sources said."
"middle of the night phone calls" gotta laugh, wish those conversations were downloadable from the iTunes Music Store.
G4 by Motorola a flop for 4 years due to manufacturing problems
G3 by Motorola has manufacturing problems, high end G3 manufactured and designed by IBM
"G2" aka 604 by Motorola has manufacturing problems, high end 604E manufactured and designed by IBM
68040 by Motorola can not compete with 486, the replacing 68050 and 68060 are delayed and forces Apple to migrate to PPC.
CPUs is a "perishable" product, they go from fresh to stale quite quickly. To be able to manufacture in time is very very important. That Motorola had a "G5" protoype at 2 GHz two years ago or that their roadmaps state things does not make a single Macintosh any faster, manufactured CPUs does
Wellcome IBM, I like to see Apple with a "big blue" backgroud
But what I'm thinking, and others are saying, is that IF Motorola shuts up and puts up, I'll welcome the competition to IBM.
Originally posted by Ompus
You anti-Motorola guys
I'm not anti-Motorola, and I'd venture to guess most people here aren't anti-Motorola. However, Motorola has, for whatever reason, not developed the G4 in a timely manner and has yet to introduce their flagship processors, the MPC8540's and MPC8560's. These were to be available the second half of 2002, then beginning 2003, and yet the embedded guys still wait(re: yes I know these aren't intended for desktops).
Most people here would be very happy, if not ecstatic over a G5 from Motorola that included items from their Roadmap; multicore superscaler, Rapid I/O, Ocean Switched fabric, 0.13µm die, etc. But this seems like it may not even appear in a desktop version(re: it was only listed as PROPOSED in a Motorola pdf making the internet rumor circulation departments).
If Motorla can't get their next generation processors using Rapid I/O out relatively quickly, Apple may not have to worry about getting any cpu's from them because they may be out of the semiconductor business. We'll see how their agreements w/ Phillips and SMT pan out, hope they do, because it can only help Apple to have at least 2 suppliers for their chips.
Personally I think we (and Apple) would be better off with two healthy & capable PowerPC providers. Apple by itself is not big enough, currently, to sustain this but it would increase the PowerPC's other markets and might bring increased attention to the processor family from the Linux crowd.