<strong>I have a feeling that a 7455 is Apollo on a 0.18 SOI process. I think Apollo 7460 will be on 0.13 SOI, but basically they're the same chip. Yes we have Apollo!</strong><hr></blockquote>
I don't what's to be happy about if this is apollo. if it is then that means apple must wait to the G5 for anything faster than 133Mhz bus
Well more MHz can be sqeezed out of the G4 when it moves to 130nm. And 1GHz may not be the top speed for the 7455. When the 7450 came out, top speed was 733. Then they bumped it to 866 later on. Add SOi and you can bump it to 1GHz. As Motorola gets better at fabbing these 7455's then I don't see why 1.13, 1.2 and 1.26GHz versions are not in the pipeline on 180nm. That's if the process change to 130nm doesn't come sooner rather than later, then we may be looking at 1.3-1.5GHz G4's (by then, though, I think g5's will be out).
<strong>Well more MHz can be sqeezed out of the G4 when it moves to 130nm. And 1GHz may not be the top speed for the 7455. When the 7450 came out, top speed was 733. Then they bumped it to 866 later on. Add SOi and you can bump it to 1GHz. As Motorola gets better at fabbing these 7455's then I don't see why 1.13, 1.2 and 1.26GHz versions are not in the pipeline on 180nm. That's if the process change to 130nm doesn't come sooner rather than later, then we may be looking at 1.3-1.5GHz G4's (by then, though, I think g5's will be out).</strong><hr></blockquote>
motorola TODAY is able to produce 1.08Ghz G4's given the fact that SOI adds 25% to the former 867 Mhz. -> therefor: enough 1Ghz G4s for the dualies...
if you add 30% for the change to 130nm you get 1.4 Ghz - that's all...
how long will the change to 130 nm take?
if i think about it. if they put all those new technologys to the G5 and add the goodies of the BookE etc the G5 will be a mind blowing chip....
Falcon: Yes. Or, more correctly, you'd have the option of clocking the chip that much higher. You could also run it at an unchanged clockspeed and get roughly 30% lower energy consumption/heat dissipation.
Motorola is the first chipmaker to bring SOI to the desktop.
[...]
The 1GHz PowerPC runs at 1.6 volts and will consume 15 watts to 17 watts of power, Motorola executives said. Its predecessor, the 533MHz PowerPC 7450, ran at 1.8 volts and consumed 14 watts of power.
[...]
The new 733MHz PowerPC 7445, an even lower-power version of the 7455, runs at 1.3 volts and consumes about 10 watts.
<hr></blockquote>
(Emphasis mine.)
You have to admit, that's pretty damned impressive.
SOI requires some layout changes that a process shrink doesn't necessarily entail. Based on the new Moto documents, the 7455 seems to include a few other minor improvements (cache locking & twice as many BAT registers). My guess is that the performance improvements, beyond the expected clock rate improvement, are more from compiler improvements since the 7450 results were first published. In the new Apple machines the faster L3 probably helps a bit as well. I think this is the Apollo and we'll see a migration to a smaller process in the next few months. I don't think the iMac is using this processor yet, but it'll probably migrate to it as soon as the number of chips being produces gets high enough and the 7450/7451 will go out of production. The Apple systems don't use DDR memory (just cache) because the 7455 doesn't support a faster memory bus. The low power version is impressive and should show up in the portables in the next few months, but I'd bet that the 7455 has production priority.
The Apollo was always promised to be a low-power / higher speed version of the G4+ intended for the consumer machines. Clock rates should continue to climb, but I'd expect them to top out at 1.5 - 1.8 GHz by the end chip's end of life. To compete in non-vector performance computing applications with high-end Athlon / P4 machines Apple will need the G5. The latest Athlon/P4 machines will also eclipse the 7455's vector performance.
Addendum: on further reflection perhaps the expected nomenclature of 7460 is accurate, and this is just a minor update to the 7541. Add SOI to the current process in order to get speeds higher and power consumption lower, maximizing what can be obtained from the current process. 7460 may be a bigger jump, including changes to the memory subsystem (DDR) and process (-> 1.5 GHz) to gain substantial performance improvements.
<strong>This is not the (original) Apollo. The technically specifications officially announced by Moto. (some times ago) were:
- Name: 7460
- L2 cache size 512 KB
- low-power/heat-dissipation
So this is a tweaked 7451 on a new process (SOI). Maybe they simply lowered their goals in order to deliver. Or G5 is faster progressing than expected while the Apollo slipped behind schedule? But then: What about new PowerBooks? 550 and 667 MHz is not competative.</strong><hr></blockquote>
I don't think they ever gave a name, and I know they never said that the L2 would be 512k. The Apollo was always supposed to be a 7450 on a new process (+ maybe a little tweaking). I was under the impression that the new process would be .13u + SOI, but looking back over the press releases and news clips over the years, I think it might have been .18u all along.
Having said that, it makes a lot more sense for MOT to release a .18u SOI rather than jumping to .13u SOI. The former is only a single process change whereas the latter is 2 process changes. It's always wiser to change things one at a time.
On another note, AMD has been working with MOT on SOI and 0.13u. AMD started fabbing .0.18u SOI in Dec 2001, and MOT must have
The low power version is impressive and should show up in the portables in the next few months, but I'd bet that the 7455 has production priority.</strong>
Apple can use the 800 MHz 7455 fine. I do believe they use the 7451 in the current Powerbooks, not the 7441. A 667 and 800 MHz Powerbook by the end of March seems like a reasonable upgrade to me.
<strong>The Apollo was always promised to be a low-power / higher speed version of the G4+ intended for the consumer machines.</strong>
The Apollo was never promised for anything. All it is is a G4+ architecture fabbed on a 0.18u SOI process. It's uses are purely up to the manufacturer.
<strong>To compete in non-vector performance computing applications with high-end Athlon / P4 machines Apple will need the G5. The latest Athlon/P4 machines will also eclipse the 7455's vector performance.</strong>
In integer, they are fine with a 20 to 30% clock rate gap with the Athlon and 30 to 40% gap with the P4. In floating point, they'll need a second FPU unit or higher MHz. So a G5 maybe needed there. But in vector performance, it is very debatable if the G4 needs to close the gap at all.
As long as Moto fabs a G4 on HiP 7 (0.13u), there will be faster G4 chips in 2002. If Moto ships a 0.13u G4 (7460) with support for a DDR processor bus, Apple will be able to use DDR DRAM. So there can be a lot of life in it left.
I view any all G5 rumors as fantasy right now. There has zero solid information on it. It's almost if it doesn't exist.
The low power version is impressive and should show up in the portables in the next few months, but I'd bet that the 7455 has production priority.</strong>
Apple can use the 800 MHz 7455 fine. I do believe they use the 7451 in the current Powerbooks, not the 7441. A 667 and 800 MHz Powerbook by the end of March seems like a reasonable upgrade to me.
<strong>The Apollo was always promised to be a low-power / higher speed version of the G4+ intended for the consumer machines.</strong>
The Apollo was never promised for anything. All it is is a G4+ architecture fabbed on a 0.18u SOI process. It's uses are purely up to the manufacturer.
<strong>To compete in non-vector performance computing applications with high-end Athlon / P4 machines Apple will need the G5. The latest Athlon/P4 machines will also eclipse the 7455's vector performance.</strong>
In integer, they are fine with a 20 to 30% clock rate gap with the Athlon and 30 to 40% gap with the P4. In floating point, they'll need a second FPU unit or higher MHz. So a G5 maybe needed there. But in vector performance, it is very debatable if the G4 needs to close the gap at all.
As long as Moto fabs a G4 on HiP 7 (0.13u), there will be faster G4 chips in 2002. If Moto ships a 0.13u G4 (7460) with support for a DDR processor bus, Apple will be able to use DDR DRAM. So there can be a lot of life in it left.
I view any all G5 rumors as fantasy right now. There has zero solid information on it. It's almost if it doesn't exist.
Agreed... while you were posting I appended my post. The roadmap on which the 7460 first showed up indicated it was intended for consumer and notebook applications, but you're right that this doesn't preclude use in higher end machines.
Eh? The 8540 would show some sign of life to the G5, would it not?</strong>
I hope not. The 8540 won't ship until Q4 02 (if Moto is on time), will only ship from 600 to 1000 MHz on a 0.13u process, and doesn't perform any better than the current 1 GHz 7455 shipping in the current Power Macs.
The only thing I hope the next gen PPC processor (that Apple uses) has in common with the 8540 is the non-blocking on-die switched fabric and perhaps the on-die memory controller. Otherwise, we better hope the G5 Apple uses is totally different from the 8540.
"The only thing I hope the next gen PPC processor (that Apple uses) has in common with the 8540 is the non-blocking on-die switched fabric"<hr></blockquote>
What about Rapid I/O?
Apple - Hypertransport
Motorola - Rapid I/O
Is it at all possible that the next generation processor from Motorola for Apple is not currently on Motorola's Roadmap?
G4 w/ 10 stage pipeline, on die controlller, ala System on a Chip?, and Hypertransport, 0.13µ process?
Eh? The 8540 would show some sign of life to the G5, would it not?</strong>
I hope not. The 8540 won't ship until Q4 02 (if Moto is on time), will only ship from 600 to 1000 MHz on a 0.13u process, and doesn't perform any better than the current 1 GHz 7455 shipping in the current Power Macs.
The only thing I hope the next gen PPC processor (that Apple uses) has in common with the 8540 is the non-blocking on-die switched fabric and perhaps the on-die memory controller. Otherwise, we better hope the G5 Apple uses is totally different from the 8540.[/QB]<hr></blockquote>
Err...major edit...
I thought the 8540 was shipping now? Doesnt the fact that it's book-e compliant, and (IIRC) on a 13µ process indicate the direction that moto is heading in the 85xx line?
<strong>Well more MHz can be sqeezed out of the G4 when it moves to 130nm. And 1GHz may not be the top speed for the 7455. When the 7450 came out, top speed was 733. Then they bumped it to 866 later on. Add SOi and you can bump it to 1GHz. As Motorola gets better at fabbing these 7455's then I don't see why 1.13, 1.2 and 1.26GHz versions are not in the pipeline on 180nm. That's if the process change to 130nm doesn't come sooner rather than later, then we may be looking at 1.3-1.5GHz G4's (by then, though, I think g5's will be out).</strong><hr></blockquote>
I'm fairly certain that even if MOT is fabing chips that will run faster than 1GHz right now, the current apple MBs only have multipliers up to 7.5 and as we can all figure out, 133MHz x 7.5 = 1000MHz (1GHz). It's going to take a new motherboard (likely with faster bus) in oder to use chips faster than 1GHz.
Comments
<strong>
No, no. The original G4 does 917 MIPS @ 500MHz which theoretically is 1834 MIPS @ 1GHZ thus making the MPC7455 24.3% faster.</strong><hr></blockquote>
My mistake, you are right.. I must have skipped a row.
<strong>I have a feeling that a 7455 is Apollo on a 0.18 SOI process. I think Apollo 7460 will be on 0.13 SOI, but basically they're the same chip. Yes we have Apollo!</strong><hr></blockquote>
I don't what's to be happy about if this is apollo. if it is then that means apple must wait to the G5 for anything faster than 133Mhz bus
<strong>
again - look over at <a href="http://e-www.motorola.com/brdata/PDFDB/docs/MPC7455FACT.pdf"; target="_blank">http://e-www.motorola.com/brdata/PDFDB/docs/MPC7455FACT.pdf</a>;
there is a low power version of the 7455 running at 10,3 Watts @ 733 Mhz and is available up to 800 Mhz
greets,
krassy</strong><hr></blockquote>
convinced.
<strong>Well more MHz can be sqeezed out of the G4 when it moves to 130nm. And 1GHz may not be the top speed for the 7455. When the 7450 came out, top speed was 733. Then they bumped it to 866 later on. Add SOi and you can bump it to 1GHz. As Motorola gets better at fabbing these 7455's then I don't see why 1.13, 1.2 and 1.26GHz versions are not in the pipeline on 180nm. That's if the process change to 130nm doesn't come sooner rather than later, then we may be looking at 1.3-1.5GHz G4's (by then, though, I think g5's will be out).</strong><hr></blockquote>
motorola TODAY is able to produce 1.08Ghz G4's given the fact that SOI adds 25% to the former 867 Mhz. -> therefor: enough 1Ghz G4s for the dualies...
if you add 30% for the change to 130nm you get 1.4 Ghz - that's all...
how long will the change to 130 nm take?
if i think about it. if they put all those new technologys to the G5 and add the goodies of the BookE etc the G5 will be a mind blowing chip....
Also, everyone, read <a href="http://news.com.com/2100-1040-824621.html"; target="_blank">this</a>. In particular, from the article:
[quote]
Motorola is the first chipmaker to bring SOI to the desktop.
[...]
The 1GHz PowerPC runs at 1.6 volts and will consume 15 watts to 17 watts of power, Motorola executives said. Its predecessor, the 533MHz PowerPC 7450, ran at 1.8 volts and consumed 14 watts of power.
[...]
The new 733MHz PowerPC 7445, an even lower-power version of the 7455, runs at 1.3 volts and consumes about 10 watts.
<hr></blockquote>
(Emphasis mine.)
You have to admit, that's pretty damned impressive.
<strong>seems to me, thats the Apollo..</strong><hr></blockquote>
Except that Apollo is supposed to use the 0.13 micron process. This uses the 0.18 micron process.
The Apollo was always promised to be a low-power / higher speed version of the G4+ intended for the consumer machines. Clock rates should continue to climb, but I'd expect them to top out at 1.5 - 1.8 GHz by the end chip's end of life. To compete in non-vector performance computing applications with high-end Athlon / P4 machines Apple will need the G5. The latest Athlon/P4 machines will also eclipse the 7455's vector performance.
Addendum: on further reflection perhaps the expected nomenclature of 7460 is accurate, and this is just a minor update to the 7541. Add SOI to the current process in order to get speeds higher and power consumption lower, maximizing what can be obtained from the current process. 7460 may be a bigger jump, including changes to the memory subsystem (DDR) and process (-> 1.5 GHz) to gain substantial performance improvements.
[ 01-29-2002: Message edited by: Programmer ]</p>
<strong>This is not the (original) Apollo. The technically specifications officially announced by Moto. (some times ago) were:
- Name: 7460
- L2 cache size 512 KB
- low-power/heat-dissipation
So this is a tweaked 7451 on a new process (SOI). Maybe they simply lowered their goals in order to deliver. Or G5 is faster progressing than expected while the Apollo slipped behind schedule? But then: What about new PowerBooks? 550 and 667 MHz is not competative.</strong><hr></blockquote>
I don't think they ever gave a name, and I know they never said that the L2 would be 512k. The Apollo was always supposed to be a 7450 on a new process (+ maybe a little tweaking). I was under the impression that the new process would be .13u + SOI, but looking back over the press releases and news clips over the years, I think it might have been .18u all along.
Having said that, it makes a lot more sense for MOT to release a .18u SOI rather than jumping to .13u SOI. The former is only a single process change whereas the latter is 2 process changes. It's always wiser to change things one at a time.
On another note, AMD has been working with MOT on SOI and 0.13u. AMD started fabbing .0.18u SOI in Dec 2001, and MOT must have
[ 01-28-2002: Message edited by: msp ]</p>
The low power version is impressive and should show up in the portables in the next few months, but I'd bet that the 7455 has production priority.</strong>
Apple can use the 800 MHz 7455 fine. I do believe they use the 7451 in the current Powerbooks, not the 7441. A 667 and 800 MHz Powerbook by the end of March seems like a reasonable upgrade to me.
<strong>The Apollo was always promised to be a low-power / higher speed version of the G4+ intended for the consumer machines.</strong>
The Apollo was never promised for anything. All it is is a G4+ architecture fabbed on a 0.18u SOI process. It's uses are purely up to the manufacturer.
<strong>To compete in non-vector performance computing applications with high-end Athlon / P4 machines Apple will need the G5. The latest Athlon/P4 machines will also eclipse the 7455's vector performance.</strong>
In integer, they are fine with a 20 to 30% clock rate gap with the Athlon and 30 to 40% gap with the P4. In floating point, they'll need a second FPU unit or higher MHz. So a G5 maybe needed there. But in vector performance, it is very debatable if the G4 needs to close the gap at all.
As long as Moto fabs a G4 on HiP 7 (0.13u), there will be faster G4 chips in 2002. If Moto ships a 0.13u G4 (7460) with support for a DDR processor bus, Apple will be able to use DDR DRAM. So there can be a lot of life in it left.
I view any all G5 rumors as fantasy right now. There has zero solid information on it. It's almost if it doesn't exist.
[ 01-29-2002: Message edited by: THT ]</p>
<strong>[qb]Originally posted by Programmer:
The low power version is impressive and should show up in the portables in the next few months, but I'd bet that the 7455 has production priority.</strong>
Apple can use the 800 MHz 7455 fine. I do believe they use the 7451 in the current Powerbooks, not the 7441. A 667 and 800 MHz Powerbook by the end of March seems like a reasonable upgrade to me.
<strong>The Apollo was always promised to be a low-power / higher speed version of the G4+ intended for the consumer machines.</strong>
The Apollo was never promised for anything. All it is is a G4+ architecture fabbed on a 0.18u SOI process. It's uses are purely up to the manufacturer.
<strong>To compete in non-vector performance computing applications with high-end Athlon / P4 machines Apple will need the G5. The latest Athlon/P4 machines will also eclipse the 7455's vector performance.</strong>
In integer, they are fine with a 20 to 30% clock rate gap with the Athlon and 30 to 40% gap with the P4. In floating point, they'll need a second FPU unit or higher MHz. So a G5 maybe needed there. But in vector performance, it is very debatable if the G4 needs to close the gap at all.
As long as Moto fabs a G4 on HiP 7 (0.13u), there will be faster G4 chips in 2002. If Moto ships a 0.13u G4 (7460) with support for a DDR processor bus, Apple will be able to use DDR DRAM. So there can be a lot of life in it left.
I view any all G5 rumors as fantasy right now. There has zero solid information on it. It's almost if it doesn't exist.
[ 01-29-2002: Message edited by: THT ][/QB]<hr></blockquote>
Agreed... while you were posting I appended my post.
Eh? The 8540 would show some sign of life to the G5, would it not? It's book-e compliant, etc..
Eh? The 8540 would show some sign of life to the G5, would it not?</strong>
I hope not. The 8540 won't ship until Q4 02 (if Moto is on time), will only ship from 600 to 1000 MHz on a 0.13u process, and doesn't perform any better than the current 1 GHz 7455 shipping in the current Power Macs.
The only thing I hope the next gen PPC processor (that Apple uses) has in common with the 8540 is the non-blocking on-die switched fabric and perhaps the on-die memory controller. Otherwise, we better hope the G5 Apple uses is totally different from the 8540.
"The only thing I hope the next gen PPC processor (that Apple uses) has in common with the 8540 is the non-blocking on-die switched fabric"<hr></blockquote>
What about Rapid I/O?
Apple - Hypertransport
Motorola - Rapid I/O
Is it at all possible that the next generation processor from Motorola for Apple is not currently on Motorola's Roadmap?
G4 w/ 10 stage pipeline, on die controlller, ala System on a Chip?, and Hypertransport, 0.13µ process?
<strong>[qb]Originally posted by harp:
Eh? The 8540 would show some sign of life to the G5, would it not?</strong>
I hope not. The 8540 won't ship until Q4 02 (if Moto is on time), will only ship from 600 to 1000 MHz on a 0.13u process, and doesn't perform any better than the current 1 GHz 7455 shipping in the current Power Macs.
The only thing I hope the next gen PPC processor (that Apple uses) has in common with the 8540 is the non-blocking on-die switched fabric and perhaps the on-die memory controller. Otherwise, we better hope the G5 Apple uses is totally different from the 8540.[/QB]<hr></blockquote>
Err...major edit...
I thought the 8540 was shipping now? Doesnt the fact that it's book-e compliant, and (IIRC) on a 13µ process indicate the direction that moto is heading in the 85xx line?
[ 01-31-2002: Message edited by: harp ]</p>
<strong>Well more MHz can be sqeezed out of the G4 when it moves to 130nm. And 1GHz may not be the top speed for the 7455. When the 7450 came out, top speed was 733. Then they bumped it to 866 later on. Add SOi and you can bump it to 1GHz. As Motorola gets better at fabbing these 7455's then I don't see why 1.13, 1.2 and 1.26GHz versions are not in the pipeline on 180nm. That's if the process change to 130nm doesn't come sooner rather than later, then we may be looking at 1.3-1.5GHz G4's (by then, though, I think g5's will be out).</strong><hr></blockquote>
I'm fairly certain that even if MOT is fabing chips that will run faster than 1GHz right now, the current apple MBs only have multipliers up to 7.5 and as we can all figure out, 133MHz x 7.5 = 1000MHz (1GHz). It's going to take a new motherboard (likely with faster bus) in oder to use chips faster than 1GHz.