When the sahara chip was presented officialy, IBM claimed that there will be in the future a Sahara 2 with a SIMD unit like Altivec.
It's possible that the next chip employed in the i book will be this sahara 2. Perhaps this chip will appear late this year. Afterall IBM have implemented VMX in the PPC 970, it won't be a big deal to implement VMX in sahara either.
In this case they will have a very low cost, very low watt consomption altivec chip for the i book line.
But there is something weird in the Apple offer : the old generation i mac : a 600 mhz PPC cxe (not the sahara) with a lame video card (forget the quartz acceleration) and a has-been screen. For some more buck you can have the emac. Why apple do not screw up this old thing ?
Honestly, I don't understand why there is so much venom against the "modern" G3. [puts on devil's advocate suit]
[quote]Originally posted by hmurchison:
<strong>1. No Altivec(anyone with a G4 and a G3 will tell you there's a difference even at the OS level)</strong><hr></blockquote>
"A" difference- not even a "big" difference (on OS level). Yet, lower speed G3's seem to get by just fine. Imagine, if we could have IBM's fastest G3's clocking higher than Motorola could ever get their G4's? The performance delta would probably disappear altogether on anything that isn't heavy Altivec'ed.
<strong> [quote]2. Inferior FPU compared to the G4 in particular with Double Precison math(slower)</strong><hr></blockquote>
A faster IBM-clocked G3 than a Motorola-clocked G4 may make that issue moot. I don't see why IBM couldn't put a stronger FPU in their G3 design, anyway. Ultimately, I don't see this issue as a particular obstacle.
<strong> [quote]Why go through the effort? The same effort to make a G3 support DDR could be added to the G4. </strong><hr></blockquote>
Not if Motorola has anything to say about it. The G4 situation keeps coming back to the same problem- the stigma of Motorola hangs upon its neck heavily. If anything, Apple could benefit if it is the G4 chip that would go bye-bye- anything to eliminate reliance on Motorola or what they say can and cannot be done with their chip. True, the G4 is a nice CPU. However, the larger picture asks how much do you get with what IBM can do with their G3 chip vs. what Motorola can do with their G4 chip. Surprisingly, the G3 quite possibly could come out ahead.
...and for the record, talks of a revisited G3 do not require support of "3 chips" in a product line. Logically, you could do just as well with a line that relies on the 970 and the G3, while retiring the G4. That way you have a clear performance/price demarcation- you use the 970 for maximum performance and you use the G3 for compact, thermal/battery-friendly, inexpensive solution with still good performance. With the G4, it becomes more vague why you would use one over the other, except for clockrate/ultimate performance. The downside of the 970/G3 union is that it goes back to the "eggs in one basket" premise vs. having IBM and Motorola in a friendly competition. Of course, it was Motorola that made us want to "get out" in the first place, so it isn't exactly clear how valuable having Motorola is as a secondary supplier. If it were an IBM/AMD competition, that would be alluring...
<strong>But there is something weird in the Apple offer : the old generation i mac : a 600 mhz PPC cxe (not the sahara) with a lame video card (forget the quartz acceleration) and a has-been screen. For some more buck you can have the emac. Why apple do not screw up this old thing ?</strong><hr></blockquote>
I'm guessing that it's for the low-low-end of the education market, since municipal school boards (especially the Board up here in Toronto) are usually extremely strapped for cash. I agree, it doesn't make sense, since the eMac was designed for this purpose, but I'm guessing that either Apple's trying to get rid of pesky inventory or somewhere, there is still demand for the Classic G3 iMac. They may not be able to change things like putting in a Sahara and QE-enabled gfx, but they could at least drop the price by $100-200.
It's not just us. It's many Mac users who think the G3 has a negative connotation to it. Mac users have been taught that the G4 is the chip to have and if you were to cancel the G4 and attempt to market a G3 if would be a difficult task. I guess Apple could call it a G4 and avoid this G3 stigma.
You know for all the ballyhoo over IBM and their increased clock speeds they sure haven't put their money where their mouth is. Where are the Ghz+ G3's ??
[quote]So in what way is the PowerPC970 still a 4th generation PowerPC? <hr></blockquote>
The PowerPC 970 comes from the IBM/POWER side of the alliance and is essentially a desktop variant of the Power4: it isn't a Power5 yet. Yes, it is not that much like the (Motorola) G4.
If you add more/better FPUs, a longer pipeline for those all important clock speeds, an SIMD unit, a RapidIO memory controller, etc.. is it still a G3?
Untill the day Apple ships a PPC970 equipped mac to you door, it is Motorola and not IBM who has made it possible to have what meager performance desktops now enjoy.
The G3 is still the best choice for laptops. I would rather have 2+ hours battery life than Altivec.
Check barefeats.org for some benchmarking. The iBook-800 is actually faster than the Powerbook-867 for most tasks, even some that use Altivec. Why? 512k cache beats 256k cache. It's also many hundreds of dollars cheaper, it runs cooler, and has a longer battery life.
The G3 is still the way to go for portable Macs. It wouldn't be out of place on the iMac/eMac line, either. Hell, if the G3 could be made compatible with the new 970 bus architecture (Altivec unit, 900mhz FSB, true DDR support), why the hell would Apple need a G4?
<strong>The G3 is still the best choice for laptops. I would rather have 2+ hours battery life than Altivec. Check barefeats.org for some benchmarking.</strong><hr></blockquote>
I wouldn't trust Bare Feats' benchmarking. I've been told to take what the guy who runs the site says with a *huge* grain of salt. However, the results aren't all that surprising, given the difference in cache sizes.
An article worth reading is Dan Knight's look at the "insignificant" difference in performance between the G3 and G4:
Timed test between systems running the same tasks are the only benches that matter. This goes for PC vs Mac tests and mac to mac tests.
Look here:
Photoshop
G4 about 3X faster
iTunes
G4 about 2.5X faster
iMovie
G4 about 2X faster
This is all for the L3-less 12" obviously the L2 advantage of the G3 doesn't translate into any advantage over the G4.
And if you compare battery life. The PB12 I've been sitting in front of has displayed very impressive battery life, even at .18u. easily up to 4+ hours of work. At .13u the G3 iBooks don't fare much better at all. At .13 a G4 should make fo a scary good laptop chip.
Any way you slice it the G3 is dead unless IBm makes enough changes to it to turn it into a G4 by default, and as others have mentioned, why would they do that for an ancient chip?
[quote] The article is quite old (it was written in 2000) but I feel it has a lot of relevance to the "G3 vs. G4" debate ongoing amongst Mac users. <hr></blockquote>
Not only is that article old but it also bases it's studies on MacBench 5.0 scores. Macbench never appromated the benefits of Altivec.
Matsu's post clearly shows that a G4 with Altivec is going to be faster. Try encoding MPEG2 for iDVD on a G3 and G4 and see your results.
Of course going from 256 kb to 512 kb of L2 cache gains tons of performance. I mean, according to AMD, that's 300 MHz of "Pentium Equivalency" right there!
Seriously, though, the 12" PowerBook is most likely a lot faster than the iBook. I haven't used it so I can't make a judgement, but even this 800 MHz iMac I'm using right now feels a lot faster than my 800 MHz iBook. If just adding a G4 makes this much difference, I can imagine how much better it would be with a 133 MHz bus, 67 MHz more processor speed, and DDR RAM.
I think we'll be living with a G4 based machines for a while longer. Moto has .13u chips in the cards. They up the L2 to 512MB, and though the announced chips still run 167Mhz as per Mot's pages, we hear (from other mot docs) that 200Mhz FSB is possible for the MPX. That's not bad for a cool runing laptop chip.
I think Mebbe the 17" PB will get a PPC970 before any other laptop, just because it suits the largess of that model. The rest will have to wait a while.
I would imagine that moto's new G4 chips will make their way into PowerBooks later this year, but if IBM can give Apple something better and sooner, maybe they would be better off getting away from Motorola ASAP.
The G3 isn't a bad chip and if it got VMX and an improved bus, it could be just as good as the G4.
I have no clue how they could name everything without creating confusion and/or misconceptions though.
<strong>I think we'll be living with a G4 based machines for a while longer. Moto has .13u chips in the cards. They up the L2 to 512MB, and though the announced chips still run 167Mhz as per Mot's pages, we hear (from other mot docs) that 200Mhz FSB is possible for the MPX. That's not bad for a cool runing laptop chip.
I think Mebbe the 17" PB will get a PPC970 before any other laptop, just because it suits the largess of that model. The rest will have to wait a while.</strong><hr></blockquote>
One point, Motorola does have 0.13µm in the cards, but very late to the game. It is quite possible the 970 will appear months before the MPC7457 and over a year before the MPC7457-RM.
Me personally, I can't imagine that IBM is letting the 750 development stagnate. I have absolutely no basis to say this, but it is possible that the next generation 32 bit cpu from IBM will appear before the MPC7457, let alone the MPC7457-RM(re: Multicore superscalar, Rapid I/O, 0.13µm process, integrated SIMD, n-way crossbar coreconnect).
IBM's advantages include they have been manufacturing cpu's using 0.13µm for quite some time, they have considerable experience with high speed interconnects, they have considerable experience with multicore processors. Motorola's advantages, well, errrr, uummmm, Altivec yes that's it Altivec, but well, on second thought, IBM owns several patents on VMX.
The problem with G3-based machines is often not the G3 itself. My iMac DV @ 450MHz is perfectly acceptable for home use unless it comes to Wolfenstein and the like. Get a modern graphics accelerator, faster HD and more RAM and this 'crap of G3' will easily serve two years more. One thing about my iMac @ 450MHz remains impressive and causes real envy from some of my friends ? it does not have a fan. So if there are cheap ways to further scale the process and frequency, why not?
By the way, if Apple ever rolls out an uber-tablet, G3 seems the best candidate to me.
The PowerPC 970 comes from the IBM/POWER side of the alliance and is essentially a desktop variant of the Power4: it isn't a Power5 yet. Yes, it is not that much like the (Motorola) G4.
If you add more/better FPUs, a longer pipeline for those all important clock speeds, an SIMD unit, a RapidIO memory controller, etc.. is it still a G3?</strong><hr></blockquote>
You are confusing the difference between the POWER (IBM's nomenclature, not me shouting ) line and the PowerPC line of chips. The POWER line are compatible with the PPC instruction set but actually have about 5 instruction sets they are compatible with in total (including older IBM architectures). The POWER line is at *its* 4th generation. The PPC line is also at *its* 4th generation. Bringing the 970 chip into the consumer space blurs the differences between the 2 lines but it is definitely a generation further ahead than the G4 in every way. It deserves 5th generation PPC status.
Comments
It's possible that the next chip employed in the i book will be this sahara 2. Perhaps this chip will appear late this year. Afterall IBM have implemented VMX in the PPC 970, it won't be a big deal to implement VMX in sahara either.
In this case they will have a very low cost, very low watt consomption altivec chip for the i book line.
But there is something weird in the Apple offer : the old generation i mac : a 600 mhz PPC cxe (not the sahara) with a lame video card (forget the quartz acceleration) and a has-been screen. For some more buck you can have the emac. Why apple do not screw up this old thing ?
[quote]Originally posted by hmurchison:
<strong>1. No Altivec(anyone with a G4 and a G3 will tell you there's a difference even at the OS level)</strong><hr></blockquote>
"A" difference- not even a "big" difference (on OS level). Yet, lower speed G3's seem to get by just fine. Imagine, if we could have IBM's fastest G3's clocking higher than Motorola could ever get their G4's? The performance delta would probably disappear altogether on anything that isn't heavy Altivec'ed.
<strong> [quote]2. Inferior FPU compared to the G4 in particular with Double Precison math(slower)</strong><hr></blockquote>
A faster IBM-clocked G3 than a Motorola-clocked G4 may make that issue moot. I don't see why IBM couldn't put a stronger FPU in their G3 design, anyway. Ultimately, I don't see this issue as a particular obstacle.
<strong> [quote]Why go through the effort? The same effort to make a G3 support DDR could be added to the G4. </strong><hr></blockquote>
Not if Motorola has anything to say about it. The G4 situation keeps coming back to the same problem- the stigma of Motorola hangs upon its neck heavily. If anything, Apple could benefit if it is the G4 chip that would go bye-bye- anything to eliminate reliance on Motorola or what they say can and cannot be done with their chip. True, the G4 is a nice CPU. However, the larger picture asks how much do you get with what IBM can do with their G3 chip vs. what Motorola can do with their G4 chip. Surprisingly, the G3 quite possibly could come out ahead.
...and for the record, talks of a revisited G3 do not require support of "3 chips" in a product line. Logically, you could do just as well with a line that relies on the 970 and the G3, while retiring the G4. That way you have a clear performance/price demarcation- you use the 970 for maximum performance and you use the G3 for compact, thermal/battery-friendly, inexpensive solution with still good performance. With the G4, it becomes more vague why you would use one over the other, except for clockrate/ultimate performance. The downside of the 970/G3 union is that it goes back to the "eggs in one basket" premise vs. having IBM and Motorola in a friendly competition. Of course, it was Motorola that made us want to "get out" in the first place, so it isn't exactly clear how valuable having Motorola is as a secondary supplier. If it were an IBM/AMD competition, that would be alluring...
<strong>But there is something weird in the Apple offer : the old generation i mac : a 600 mhz PPC cxe (not the sahara) with a lame video card (forget the quartz acceleration) and a has-been screen. For some more buck you can have the emac. Why apple do not screw up this old thing ?</strong><hr></blockquote>
I'm guessing that it's for the low-low-end of the education market, since municipal school boards (especially the Board up here in Toronto) are usually extremely strapped for cash. I agree, it doesn't make sense, since the eMac was designed for this purpose, but I'm guessing that either Apple's trying to get rid of pesky inventory or somewhere, there is still demand for the Classic G3 iMac. They may not be able to change things like putting in a Sahara and QE-enabled gfx, but they could at least drop the price by $100-200.
<strong> The downside of the 970/G3 union is that it goes back to the "eggs in one basket" premise</strong><hr></blockquote>
Which is an interesting irony, since we got Moto into the AIM alliance so this wouldn't happen.
It's not just us. It's many Mac users who think the G3 has a negative connotation to it. Mac users have been taught that the G4 is the chip to have and if you were to cancel the G4 and attempt to market a G3 if would be a difficult task. I guess Apple could call it a G4 and avoid this G3 stigma.
You know for all the ballyhoo over IBM and their increased clock speeds they sure haven't put their money where their mouth is. Where are the Ghz+ G3's ??
The PowerPC 970 comes from the IBM/POWER side of the alliance and is essentially a desktop variant of the Power4: it isn't a Power5 yet. Yes, it is not that much like the (Motorola) G4.
If you add more/better FPUs, a longer pipeline for those all important clock speeds, an SIMD unit, a RapidIO memory controller, etc.. is it still a G3?
Check barefeats.org for some benchmarking. The iBook-800 is actually faster than the Powerbook-867 for most tasks, even some that use Altivec. Why? 512k cache beats 256k cache. It's also many hundreds of dollars cheaper, it runs cooler, and has a longer battery life.
The G3 is still the way to go for portable Macs. It wouldn't be out of place on the iMac/eMac line, either. Hell, if the G3 could be made compatible with the new 970 bus architecture (Altivec unit, 900mhz FSB, true DDR support), why the hell would Apple need a G4?
<strong>The G3 is still the best choice for laptops. I would rather have 2+ hours battery life than Altivec. Check barefeats.org for some benchmarking.</strong><hr></blockquote>
I wouldn't trust Bare Feats' benchmarking. I've been told to take what the guy who runs the site says with a *huge* grain of salt.
An article worth reading is Dan Knight's look at the "insignificant" difference in performance between the G3 and G4:
<a href="http://www.lowendmac.com/tech/insignificant.html" target="_blank">G4 insignificantly superior to G3</a>
The article is quite old (it was written in 2000) but I feel it has a lot of relevance to the "G3 vs. G4" debate ongoing amongst Mac users.
<strong>Hell ..just shelve the G3. Once the 970 hits it should be 970=Pro G4= Consumer.
</strong><hr></blockquote>
I think Apple needs to get away from Motorola and the G4 name as soon as possible.
970's in the PowerMac to start. Die shrink...then to PowerBooks/ Towers get speed bump/ and iMac's get the lowest speed 970.
iBooks use the 750Fx with 200MHz bus and VMX.
I wouldn't worry about chip names either. Just call them the PowerMac/PowerBook 970 and the iBook 750...or iBook Fx
drop the G(x) name completely.
Look here:
Photoshop
G4 about 3X faster
iTunes
G4 about 2.5X faster
iMovie
G4 about 2X faster
This is all for the L3-less 12" obviously the L2 advantage of the G3 doesn't translate into any advantage over the G4.
And if you compare battery life. The PB12 I've been sitting in front of has displayed very impressive battery life, even at .18u. easily up to 4+ hours of work. At .13u the G3 iBooks don't fare much better at all. At .13 a G4 should make fo a scary good laptop chip.
Any way you slice it the G3 is dead unless IBm makes enough changes to it to turn it into a G4 by default, and as others have mentioned, why would they do that for an ancient chip?
Not only is that article old but it also bases it's studies on MacBench 5.0 scores. Macbench never appromated the benefits of Altivec.
Matsu's post clearly shows that a G4 with Altivec is going to be faster. Try encoding MPEG2 for iDVD on a G3 and G4 and see your results.
G3= old relic
G4= barely passable by todays standards
970= future pony to ride as far as we can.
Seriously, though, the 12" PowerBook is most likely a lot faster than the iBook. I haven't used it so I can't make a judgement, but even this 800 MHz iMac I'm using right now feels a lot faster than my 800 MHz iBook. If just adding a G4 makes this much difference, I can imagine how much better it would be with a 133 MHz bus, 67 MHz more processor speed, and DDR RAM.
I think Mebbe the 17" PB will get a PPC970 before any other laptop, just because it suits the largess of that model. The rest will have to wait a while.
The G3 isn't a bad chip and if it got VMX and an improved bus, it could be just as good as the G4.
I have no clue how they could name everything without creating confusion and/or misconceptions though.
<strong>I think we'll be living with a G4 based machines for a while longer. Moto has .13u chips in the cards. They up the L2 to 512MB, and though the announced chips still run 167Mhz as per Mot's pages, we hear (from other mot docs) that 200Mhz FSB is possible for the MPX. That's not bad for a cool runing laptop chip.
I think Mebbe the 17" PB will get a PPC970 before any other laptop, just because it suits the largess of that model. The rest will have to wait a while.</strong><hr></blockquote>
One point, Motorola does have 0.13µm in the cards, but very late to the game. It is quite possible the 970 will appear months before the MPC7457 and over a year before the MPC7457-RM.
Me personally, I can't imagine that IBM is letting the 750 development stagnate. I have absolutely no basis to say this, but it is possible that the next generation 32 bit cpu from IBM will appear before the MPC7457, let alone the MPC7457-RM(re: Multicore superscalar, Rapid I/O, 0.13µm process, integrated SIMD, n-way crossbar coreconnect).
IBM's advantages include they have been manufacturing cpu's using 0.13µm for quite some time, they have considerable experience with high speed interconnects, they have considerable experience with multicore processors. Motorola's advantages, well, errrr, uummmm, Altivec yes that's it Altivec, but well, on second thought, IBM owns several patents on VMX.
kind..
I recall quite clearly an article from an Australian microprocessor conference discussing
the addition of a vector unit to the 750fx. I can't recall any details save the focus..adding
the vector unit..
By the way, if Apple ever rolls out an uber-tablet, G3 seems the best candidate to me.
<strong>
The PowerPC 970 comes from the IBM/POWER side of the alliance and is essentially a desktop variant of the Power4: it isn't a Power5 yet. Yes, it is not that much like the (Motorola) G4.
If you add more/better FPUs, a longer pipeline for those all important clock speeds, an SIMD unit, a RapidIO memory controller, etc.. is it still a G3?</strong><hr></blockquote>
You are confusing the difference between the POWER (IBM's nomenclature, not me shouting
MM