I was thinking about that over that a little while ago: maybe, for whatever reason, Motorola got their stuff together a bit more and, since we have this new G5 for the pro tower crowd, maybe the other things get snappier G4s for the remainder of 2003 and early 2004...including...the...dare I say it...iBook?
The thing is, you can't really predict a company on the ropes. It's possible they'll just go out with a whimper, especially if there's a pure capitalist in charge (better, in that mindset, to simply liquidate the assets). But if the company has any sense of self-preservation, it'll start taking risks that most companies wouldn't, because no matter how far out the risk is, it's more likely to pay off than just coasting into irrelevance. So Ford sunk $3 billion into the Taurus and Sable, Apple unveiled the iMac, and Gateway is branching out into, uh, non-computer devices (this last example is just to underscore that this strategy is risky, and it can indeed fail ). I'd read enough to figure that Mot might be trying something like that, not despite the fact that they're smaller and poorer, but because they are. If you're looking at possibly going out of business, why not try to leapfrog Intel? In the worst case, you fail, and you go out of business, which you were going to do anyway.
Well, let's hope Moto can really pull an IBM on us, so that the new PowerBooks can kick serious Centrino ass! The keyword here is hope. Even though hope springs eternal, I'm not holding my breath.
Amorph: How realistic (or naive) is it to hope for a 90nm PPC 7457 from Motorola? What kind of evidence do you have that makes you believe in such a surprise?
I'm a bit confused with the clamor for the 970 in the Powerbook.
I guess because right now its the only sight we see of higher clockspeeds but the whole 64 bit aspect of the chip will never be utilized in a Powerbook.
Is it possible that IBM and Apple could be working on a different chip for portables?
Oh, I think it will, but it's gonna be more than 12 months before they do, I think it's gonna take a year and a half before we see PB G5's. And we WILL NOT see iMac or iBook G5's for a long time yet.
Oh, I think it will, but it's gonna be more than 12 months before they do, I think it's gonna take a year and a half before we see PB G5's. And we WILL NOT see iMac or iBook G5's for a long time yet.
Why so?
The 9 fans make people assume it's white hot, I think they are for quietness and dual 3GHz+ future proofing.
I know it's too keep it quiet and cool, but the numbers for the 2Ghz part are high, as they are for the 1.6, and ATM, that might be all there is, the other numbers were projections, there's the mobo and controller chips to think about, I think it's a lot more work than anyone imagines to get the whole system down to acceptable levels both in terms of heat and cost. Apple cannot afford to make their laptops any more expensive, they've just now reached a competitive level.
I'm a bit confused with the clamor for the 970 in the Powerbook.
I guess because right now its the only sight we see of higher clockspeeds but the whole 64 bit aspect of the chip will never be utilized in a Powerbook.
Not only higher clock, but better performance clock-for-clock, and much better bandwidth, FP performance and SIMD performance.
Apple has sold a boatload of laptops to people who do music and video in the field. A G5 laptop is the wet-dream for music and video on the go.
Apple has also sold the laptop as a more than capable replacement for the desktop (remember the "year of the laptop"), which now seems very slow and much less capable in comparison to the non-shipping G5. Its only marketing, but its either time to change the message or change the hardware.
Now as far as I know Skidmarks GT doesn't take advantage of the 2nd CPU when benchmarking, although Apple may have recoded it to correct this. Let's assume for now that the benchmarks above only show results for one 2 GHz G5.
If we scale down those G5 results from 2 Ghz to 1.2 GHZ (by multiplying the results by .6), we are left with the following:
---------------- 1 x 1 Ghz G4 -- 1 x 1.2 GHz G5 -- Increase
Nice, but certainly not what I'd expect. I'm sure there would be increases in performance due to improved architecture that cannot be calculated by my simple ratio, and we have no idea how much performance is to be gained by using Panther. The results are quite depressing if you calculate a 1 GHz G5 using a ratio, with the only appreciable gain to be seen in Floating Point performance.
As I mentioned above, older versions of Skidmarks GT only benchmarked 1 of the CPU's in a dual system. After seeing these results, I sure hope so. What I'd like to see are Skidmarks GT benchmarks for the single 1.6 and 1.8 G5's so I had a better idea of the performance gains.
Of course I'll be buying a G5 laptop as soon as they're available, but I don't see that happening for another 12 months.
Amorph: How realistic (or naive) is it to hope for a 90nm PPC 7457 from Motorola? What kind of evidence do you have that makes you believe in such a surprise?
I'll answer the second question first. See this article:
which flat out announces things that Motorola was "planning" to do. The plan appears to have worked.
Barring disaster (and remember, this is Motorola!) we are getting a 90nm PPC next year from Motorola. Whether it's a 7457 or something else is open to question (although the '57 is an obvious candidate for migration to 90nm).
As I mentioned above, older versions of Skidmarks GT only benchmarked 1 of the CPU's in a dual system. After seeing these results, I sure hope so. What I'd like to see are Skidmarks GT benchmarks for the single 1.6 and 1.8 G5's so I had a better idea of the performance gains.
Couldn't find anything that described Skidmarks GT. How is it testing performance? I really would be loath to make any speed comparison without knowing how this benchmark works. For instance does it make scheduling adjustments based on the processor profile (G3 vs G4) and was it updated for the G5? What is Skidmark GT compiled with (GCC, CodeWarrior)? Those are relevant questions no?
Maybe all it shows us is that for code that was optimally tuned for the G4, that took in account all the latencies and the number of cycles needed to complete an integer calc and etcetera optimally, that the G5 will still be able to run G4 tuned code at a decent pace. But what do I know, couldn't find any info about Skidmarks GT.
I am not sure I am clear with my point above so here is another angle: if you bother to find out what SPEC is, you know it measures not hardware performance but system performance (that is a compiler's ability to extract integer and floating point performance out of a certain piece of hardware). That makes sense to me. But if someone tells me Skidmarks is a hardware only test, I say bullocks! How can you test only hardware when you are using software on it to do so? So again, while Skidmarks might be relevant to test the performance differential of two G4s, is it relevant at all for the G5?
I am not sure I am clear with my point above so here is another angle: if you bother to find out what SPEC is, you know it measures not hardware performance but system performance (that is a compiler's ability to extract integer and floating point performance out of a certain piece of hardware). That makes sense to me. But if someone tells me Skidmarks is a hardware only test, I say bullocks! How can you test only hardware when you are using software on it to do so? So again, while Skidmarks might be relevant to test the performance differential of two G4s, is it relevant at all for the G5?
I know what SPEC is, and I'd like to see independent lab-tests that show benchmarks against various other chips using the most efficient compiler for each chip.
I'm sure we'll see more telling results in the next few weeks.
I know what SPEC is, and I'd like to see independent lab-tests that show benchmarks against various other chips using the most efficient compiler for each chip.
Didn't mean to come off so aggressive with my SPEC remark (I phrased that poorly), but since you quoted the Skidmarks bench, do you know anything about it?
Didn't mean to come off so aggressive with my SPEC remark (I phrased that poorly), but since you quoted the Skidmarks bench, do you know anything about it?
I know little about the inner workings of it, but Skidmarks GT is an benchmark which Apple bizarrely claims is purely processor based. Personally, I find that hard to believe - a point which you raised in an earlier post.
Grab the CHUD tools. Seems to have been updated to 3.0 on the 25th.
I'd download the newest version and tell you what's changed, if I weren't stuck on a PC right now. I assume they just included some G5 specific migration tools in the package and bumped the version number.
Apple seems to use promotions to clear out inventory. This promotion seems to point to a September AL PowerBook 15.4 release...
F No! Apple better not wait until Sept to update the PB G4. That would suck! That would mean the TiBook would have been languishing for 10 months w/o a refresh. Major suckage. Bring me the new Al15 now!!!
It's sad to realize that never again will a PowerBook come close to the speed of a Mac tower. I wonder how this will affect sales. Apple seems to be proud of their portable sales numbers but I expect those numbers will drop now that the portables will seem so slow when compared to the PowerMacs.
Personally, after using a PowerBook for over a year, I'll never switch back to a tower. So, I hope there is a solution to the speed, heat, power consumption problem before long. I hope by the time I'm ready to buy a new PowerBook, they'll be wicked-fast, cool and quiet compared to the 667Mhz I have now.
F No! Apple better not wait until Sept to update the PB G4. That would suck! That would mean the TiBook would have been languishing for 10 months w/o a refresh. Major suckage. Bring me the new Al15 now!!!
Apple has set the deadlines for their promotions a good month or two after the eventual product upgrade before. If you read the fine print, they say they can change the date around more or less at will. They can choose a pessimistic date partly to cover their own butts in case the product hits a snag close to its intended release date, partly to throw off rumormongers, and partly to get people to actually use the promotion to buy computers instead of using it as an indicator for how long they should wait before buying.
That said, September is certainly possible, if it takes Apple that long to stockpile an adequate number of low-voltage 970s, or 7457s, or whatever ends up powering the next generation PowerBooks.
iDave: If it's any consolation, you can think of it as happening because the towers just jumped up into a previously unheard of product category for Apple (the high-end UNIX workstation), not because of any failure of the PowerBooks. Or, you can think of the PowerBook's longtime parity as a side effect of Apple's reliance on an embedded processor for their high-end hardware.
EIther way, rest assured that the PowerBooks will continue to be high-performance machines, even if they can't muster the raw power of the PowerMac anymore. (And who knows? The 970 might end up in there yet. I'm not going to bet on it, but you know Apple wants it to happen...).
Comments
Originally posted by pscates
I was thinking about that over that a little while ago: maybe, for whatever reason, Motorola got their stuff together a bit more and, since we have this new G5 for the pro tower crowd, maybe the other things get snappier G4s for the remainder of 2003 and early 2004...including...the...dare I say it...iBook?
The thing is, you can't really predict a company on the ropes. It's possible they'll just go out with a whimper, especially if there's a pure capitalist in charge (better, in that mindset, to simply liquidate the assets). But if the company has any sense of self-preservation, it'll start taking risks that most companies wouldn't, because no matter how far out the risk is, it's more likely to pay off than just coasting into irrelevance. So Ford sunk $3 billion into the Taurus and Sable, Apple unveiled the iMac, and Gateway is branching out into, uh, non-computer devices (this last example is just to underscore that this strategy is risky, and it can indeed fail
Amorph: How realistic (or naive) is it to hope for a 90nm PPC 7457 from Motorola? What kind of evidence do you have that makes you believe in such a surprise?
Escher
I guess because right now its the only sight we see of higher clockspeeds but the whole 64 bit aspect of the chip will never be utilized in a Powerbook.
Is it possible that IBM and Apple could be working on a different chip for portables?
Originally posted by Matsu
Oh, I think it will, but it's gonna be more than 12 months before they do, I think it's gonna take a year and a half before we see PB G5's. And we WILL NOT see iMac or iBook G5's for a long time yet.
Why so?
The 9 fans make people assume it's white hot, I think they are for quietness and dual 3GHz+ future proofing.
I'm a bit confused with the clamor for the 970 in the Powerbook.
I guess because right now its the only sight we see of higher clockspeeds but the whole 64 bit aspect of the chip will never be utilized in a Powerbook.
Not only higher clock, but better performance clock-for-clock, and much better bandwidth, FP performance and SIMD performance.
Apple has sold a boatload of laptops to people who do music and video in the field. A G5 laptop is the wet-dream for music and video on the go.
Apple has also sold the laptop as a more than capable replacement for the desktop (remember the "year of the laptop"), which now seems very slow and much less capable in comparison to the non-shipping G5. Its only marketing, but its either time to change the message or change the hardware.
Originally posted by Blackcat
Why so?
The 9 fans make people assume it's white hot, I think they are for quietness and dual 3GHz+ future proofing.
97 Watts of heat output per 2 GHz G5 processor ... that's why there are so many fans.
Compare that to approximately 22 Watts per 1 GHz G4.
Originally posted by audiopollution
97 Watts of heat output per 2 GHz G5 processor ... that's why there are so many fans.
Compare that to approximately 22 Watts per 1 GHz G4.
Yup, but iMacs would be 1.6GHz or less, with no hot 7200rpm SATA drive. I thinks it's doable - 1.2GHz 970s are 19W.
Originally posted by Blackcat
Yup, but iMacs would be 1.6GHz or less, with no hot 7200rpm SATA drive. I thinks it's doable - 1.2GHz 970s are 19W.
Yes, maybe, but consider the following:
Think Secret have posted benchmark results for the Dual 2 GHz G5 versus a single 1 GHz G4.
---------------- 1 x 1 Ghz G4 --- 2 x 2 GHz G5 --- Increase
Integer ------------ 100 ------------ 172 ----------- 72%
Floating Point ----- 100 ------------ 270 ---------- 170%
Vector ------------- 100 ------------ 208 ---------- 108%
Now as far as I know Skidmarks GT doesn't take advantage of the 2nd CPU when benchmarking, although Apple may have recoded it to correct this. Let's assume for now that the benchmarks above only show results for one 2 GHz G5.
If we scale down those G5 results from 2 Ghz to 1.2 GHZ (by multiplying the results by .6), we are left with the following:
---------------- 1 x 1 Ghz G4 -- 1 x 1.2 GHz G5 -- Increase
Integer ------------ 100 ----------- 103.2 --------- 3.2%
Floating Point ----- 100 ------------ 162 ----------- 62%
Vector ------------- 100 ----------- 124.6 -------- 24.6%
Nice, but certainly not what I'd expect. I'm sure there would be increases in performance due to improved architecture that cannot be calculated by my simple ratio, and we have no idea how much performance is to be gained by using Panther. The results are quite depressing if you calculate a 1 GHz G5 using a ratio, with the only appreciable gain to be seen in Floating Point performance.
As I mentioned above, older versions of Skidmarks GT only benchmarked 1 of the CPU's in a dual system. After seeing these results, I sure hope so. What I'd like to see are Skidmarks GT benchmarks for the single 1.6 and 1.8 G5's so I had a better idea of the performance gains.
Of course I'll be buying a G5 laptop as soon as they're available, but I don't see that happening for another 12 months.
Originally posted by Escher
Amorph: How realistic (or naive) is it to hope for a 90nm PPC 7457 from Motorola? What kind of evidence do you have that makes you believe in such a surprise?
I'll answer the second question first. See this article:
http://www.electronicstimes.com/tech...G20030623S0089
which flat out announces things that Motorola was "planning" to do. The plan appears to have worked.
Barring disaster (and remember, this is Motorola!) we are getting a 90nm PPC next year from Motorola. Whether it's a 7457 or something else is open to question (although the '57 is an obvious candidate for migration to 90nm).
Originally posted by audiopollution
As I mentioned above, older versions of Skidmarks GT only benchmarked 1 of the CPU's in a dual system. After seeing these results, I sure hope so. What I'd like to see are Skidmarks GT benchmarks for the single 1.6 and 1.8 G5's so I had a better idea of the performance gains.
Couldn't find anything that described Skidmarks GT. How is it testing performance? I really would be loath to make any speed comparison without knowing how this benchmark works. For instance does it make scheduling adjustments based on the processor profile (G3 vs G4) and was it updated for the G5? What is Skidmark GT compiled with (GCC, CodeWarrior)? Those are relevant questions no?
Maybe all it shows us is that for code that was optimally tuned for the G4, that took in account all the latencies and the number of cycles needed to complete an integer calc and etcetera optimally, that the G5 will still be able to run G4 tuned code at a decent pace. But what do I know, couldn't find any info about Skidmarks GT.
I am not sure I am clear with my point above so here is another angle: if you bother to find out what SPEC is, you know it measures not hardware performance but system performance (that is a compiler's ability to extract integer and floating point performance out of a certain piece of hardware). That makes sense to me. But if someone tells me Skidmarks is a hardware only test, I say bullocks! How can you test only hardware when you are using software on it to do so? So again, while Skidmarks might be relevant to test the performance differential of two G4s, is it relevant at all for the G5?
Originally posted by razor
I am not sure I am clear with my point above so here is another angle: if you bother to find out what SPEC is, you know it measures not hardware performance but system performance (that is a compiler's ability to extract integer and floating point performance out of a certain piece of hardware). That makes sense to me. But if someone tells me Skidmarks is a hardware only test, I say bullocks! How can you test only hardware when you are using software on it to do so? So again, while Skidmarks might be relevant to test the performance differential of two G4s, is it relevant at all for the G5?
I know what SPEC is, and I'd like to see independent lab-tests that show benchmarks against various other chips using the most efficient compiler for each chip.
I'm sure we'll see more telling results in the next few weeks.
Originally posted by audiopollution
I know what SPEC is, and I'd like to see independent lab-tests that show benchmarks against various other chips using the most efficient compiler for each chip.
Didn't mean to come off so aggressive with my SPEC remark (I phrased that poorly), but since you quoted the Skidmarks bench, do you know anything about it?
Originally posted by razor
Didn't mean to come off so aggressive with my SPEC remark (I phrased that poorly), but since you quoted the Skidmarks bench, do you know anything about it?
I know little about the inner workings of it, but Skidmarks GT is an benchmark which Apple bizarrely claims is purely processor based. Personally, I find that hard to believe - a point which you raised in an earlier post.
Here's a link to Apple's page on it.
http://developer.apple.com/tools/performance/
Grab the CHUD tools. Seems to have been updated to 3.0 on the 25th.
I'd download the newest version and tell you what's changed, if I weren't stuck on a PC right now.
Apple seems to use promotions to clear out inventory. This promotion seems to point to a September AL PowerBook 15.4 release...
http://store.apple.com/Catalog/US/Im...dle_terms.html
This bundle features an iPod plus PowerBook combo and it runs through September.
PowerBook G5 in 12 - 18 months --> MWSF 2005
Originally posted by backcheck
Hello everyone, first post...
Apple seems to use promotions to clear out inventory. This promotion seems to point to a September AL PowerBook 15.4 release...
F No! Apple better not wait until Sept to update the PB G4. That would suck! That would mean the TiBook would have been languishing for 10 months w/o a refresh. Major suckage. Bring me the new Al15 now!!!
A comrade in arms...
Personally, after using a PowerBook for over a year, I'll never switch back to a tower. So, I hope there is a solution to the speed, heat, power consumption problem before long. I hope by the time I'm ready to buy a new PowerBook, they'll be wicked-fast, cool and quiet compared to the 667Mhz I have now.
Originally posted by DHagan4755
F No! Apple better not wait until Sept to update the PB G4. That would suck! That would mean the TiBook would have been languishing for 10 months w/o a refresh. Major suckage. Bring me the new Al15 now!!!
Apple has set the deadlines for their promotions a good month or two after the eventual product upgrade before. If you read the fine print, they say they can change the date around more or less at will. They can choose a pessimistic date partly to cover their own butts in case the product hits a snag close to its intended release date, partly to throw off rumormongers, and partly to get people to actually use the promotion to buy computers instead of using it as an indicator for how long they should wait before buying.
That said, September is certainly possible, if it takes Apple that long to stockpile an adequate number of low-voltage 970s, or 7457s, or whatever ends up powering the next generation PowerBooks.
iDave: If it's any consolation, you can think of it as happening because the towers just jumped up into a previously unheard of product category for Apple (the high-end UNIX workstation), not because of any failure of the PowerBooks. Or, you can think of the PowerBook's longtime parity as a side effect of Apple's reliance on an embedded processor for their high-end hardware.
EIther way, rest assured that the PowerBooks will continue to be high-performance machines, even if they can't muster the raw power of the PowerMac anymore. (And who knows? The 970 might end up in there yet. I'm not going to bet on it, but you know Apple wants it to happen...).