PDA

View Full Version : We can thank Moto for crippling the new PowerBooks


illumin8
Sep 17, 2003, 10:14 AM
Like almost everyone on this board, I've been waiting for months now for Powerbook updates. Now that they are finally here, I think it's a shame that Apple had to cripple them because Motorola couldn't deliver the proper chips. Hear me out for just a second here:

1. Apple plans on releasing new PowerBooks based on the Moto 7457 processor. This proc has 512KB L2 cache and 1MB L3 cache, and runs at up to 1.33 ghz. Moto estimates this should give you a 40% increase in performance over the 7455 which is currently used in the G4 TiBooks.

2. Motorola fails to deliver the 7457 in sufficient quantities for a PowerBook release. Apple is forced to postpone their product release and miss the critical back to school buying season.

3. In order to release a new PowerBook at all, Apple has to retrofit the Motorola 7447 processors into the new PowerBooks, which have no L3 cache. This is most definitely a step backwards for Apple. Even though the L2 cache has increased from 256K to 512K, the lack of an L3 cache is going to cripple these new PowerBooks.

4. Apple releases them anyway, because they have to release *something*, and they also have to appear faster, which they do, due to the increased Mhz. rating.

So, from someone who was about to buy a AlBook as soon as they were released, now I am wondering if I should just skip this product release entirely and wait for the G5 Powerbooks. The lack of an L3 cache is pretty embarrassing, even though they did go with the right video card with the ATI 9600.

I also think it's pretty obvious that Motorola caused the delays in the new PowerBooks, as well as the lack of an L3 cache.

Did I get everything right here? What are your thoughts on this.

Daveman Deluxe
Sep 17, 2003, 10:29 AM
1. WRONG. The 7457 is in the PowerBooks, but it only SUPPORTS up to 2 MB of Level 3 cache. It's called "Level 3" because it's off-chip. Apple decided that 512k of L2 offset any benefit gained by adding the L3 (L3 has high latency compared to L2 and takes eight clock cycles to access). Furthermore, the L3 cache memory is VERY expensive and, from what I understand, adds quite a bit to the heat dissipation of the computer--which is probably why it wasn't in the original 12".

2. Possibly correct. Much has been made of Motorola's embarrassingly low chip yields for the 7457. Apple should have had new models out two weeks ago, five weeks would have been much better. Steve reportedly got really P.O.'ed at Moto execs, saying the G4 is a "Tonka Truck".

3. Wrong but also correct. The 7457 is the chip in the PowerBooks, but it is correct that the 7447 doesn't support L3 cache.

4. Even keeping the same clock frequencies, the 7457 is supposed to be much faster than the 7455. I haven't read much about the 7457's architecture, so I can't confirm this.

So to all who want to know: Yes, they are 7457s, and NO they don't have L3 cache. L3 cache isn't a given just because a chip supports it since the L3 cache is off-chip.

chazmox
Sep 17, 2003, 10:44 AM
I think Dave's got it right. When I heard of the lack of L3 cache, I thought that the 7447 might have been used, but now that I've looked at the specs, it appears to be the 7457.

I still agree with the original assumption that Motorola screwed up, just not as bad as not being able to produce ( in acceptable volumes ) the 7447.

ftaok
Sep 17, 2003, 10:46 AM
Originally posted by Daveman Deluxe
2. Possibly correct. Much has been made of Motorola's embarrassingly low chip yields for the 7457. Apple should have had new models out two weeks ago, five weeks would have been much better. Steve reportedly got really P.O.'ed at Moto execs, saying the G4 is a "Tonka Truck".
I really have a hard time believing this one. It's all predicated on a rumor that Macbidoulle posted and it's snowballed into this.

The only facts that we know is that Moto had samples of 7457s available to "select" customers earlier this spring. Originally, Moto planned to release the 7457 in Q4 2003 (which begins in about 2 weeks). Sometime in the summer, Moto revised their shipping date for the 7457 to Q3.

All of this talk about Moto not shipping Apple chips in May is just conjecture. The chips were supposed to be out in Q3 and they made their date.

bousozoku
Sep 17, 2003, 11:12 AM
Sure, Motorola didn't have enough 7457s for Apple to introduce the new PowerBooks earlier. Motorola released information that they would have 1.5-1.8 GHz 7457s available this autumn anyway.

The L3 cache is never a certainty and, if Apple believe it necessary, they can add it later without throwing everything away because they're already using the 7457. This motherboard is most likely a stopgap design until the PPC970 motherboard and cooling system are ready. It's unlikely that Apple want to keep a G4 in the PowerBooks any longer than necessary due to the divergence from the PowerMac line.

ftaok
Sep 17, 2003, 11:20 AM
Originally posted by bousozoku
Sure, Motorola didn't have enough 7457s for Apple to introduce the new PowerBooks earlier. Motorola released information that they would have 1.5-1.8 GHz 7457s available this autumn anyway.
Moto never stated that the 7457s would be shipping at 1.5-1.8 ghz. They stated that initially, they would be available at speeds up to 1.3ghz.

cubist
Sep 17, 2003, 11:21 AM
L3 cache is only a stopgap technique because of low bus bandwidth. These machines are using faster RAM, and have a bigger L2 cache.

I don't think you need to worry about the lack of L3 cache. You wouldn't insist on a catalytic converter on an electric car, would you?

Flynnstone
Sep 17, 2003, 11:35 AM
I think we need some verification of the use of a 7457 in the new Powerbooks.
The 7457 is a 483-pin CBGA package. The 7447 is a 360-pin CBGA package. The 7457 will be more expensive than the 7447. So if Apple is trying to save money by eliminating L3 cache, why put the added expense of a 7457 in? Perhaps Motorola has messed up and the L3 cache interface doesn't work properly. If this is the case, then perhaps rev B will have L3 cache back.
If there is a 7447 in the PowerBook, then there will not likely be L3 cache in the near future.
So if there is no plans for L3 cache, then I'd be very surprised to see a 7457 in a PowerBook !

I'm not surprised by Motorola's problems with the 7447/7457, it is a technology shift from the 7445/7455 tech. The former are 0.13 um and the latter is 0.18 um.

illumin8
Sep 18, 2003, 09:27 AM
After reading the front page today and seeing confirmation that Apple is indeed using the 7457 and not the 7447 as I had originally suspected, I guess Arn can kill this thread if he wants to.

I think all signs point to Apple offering a speed-bump in the next 6 months or so which will add L3 cache (hopefully 2MB L3 cache) across the line, but it seems that these new PowerBooks don't even need the L3 cache. The benchmarks look really good.

Flynnstone
Sep 20, 2003, 02:09 PM
I guess we have our confirmation ... a 7447

http://forums.macrumors.com/showthread.php?s=&threadid=38918

Perhaps L3 cache has diminishing returns when L2 is bigger.
Perhaps Motorola couldn't deliver.
Perhaps the truth is stranger

Daveman Deluxe
Sep 20, 2003, 02:25 PM
Reading Motorola's technical documents, it appears the only real difference between the 7447 and the 7457 is the lack of L3 cache support on the 7447 and the fact that the 7447 is only 360 pins.

The 7457 is pin-compatible with the 7455 and 7445, the 7447 is pin-compatible only with the 7445.

Both processors run at up to 1.3 GHz and use the same amount of power.

The 7447 is billed as a processor for space-constrained applications.

I don't see how we're getting screwed if it is indeed a 7447 since the only thing missing is L3 cache support. With double the L2 cache, I don't think there's much to miss.

panphage
Sep 20, 2003, 02:34 PM
Originally posted by Flynnstone
I guess we have our confirmation ... a 7447

http://forums.macrumors.com/showthread.php?s=&threadid=38918

Perhaps L3 cache has diminishing returns when L2 is bigger.
Perhaps Motorola couldn't deliver.
Perhaps the truth is stranger

With 512MB L2 cache, the L3 cache would have had to jump up to 2MB to really add anything to performance. Imagine you are trying to fill a trough with water. You have two buckets. One is one gallon, the other is either 2 or 4 gallons. Someone, who is 4x slower than you are, is walking to the river to fill the larger bucket. He stops 20 feet away, maybe there's a fence. The river is 200 feet away. If you are filling your trough by dipping your bucket in his bucket, which would you rather he have, the 2 or 4 gallon? Or would you rather just go to the river yourself since you're so damn fast?

I am also of the opinion that the 4 gallon bucket would cost a lot and generate too much heat. It's better just to let the new L2 do the work, it's 50% faster than the old L2 and 2x the size.

ZildjianKX
Sep 20, 2003, 08:29 PM
I waited for the PB for over 4 months, and I think the most dissapointing aspects are their poor performance, weight, and battery life compared to a Centrino. G5s are top notch... but the PBs still aren't a good value.

QCassidy352
Sep 20, 2003, 08:38 PM
who cares if it's a 7447 or a 7457... the only difference between them is that the latter *supports* L3 cache, but given that in reality the powerbooks do not *have* L3 cache, what does it matter if it is theoretically supported or not?

The bottom line is that these new powerbooks seem to have good benchmarks compared to the old ones.