Question on the Adaption of New Processors

I was reading Wikipedia's article on the Core Duo and came across a yet to be released newer version of the Duo, codenamed Merom, the successor to Yonah, the current chip in new intel macs. Merom is supposed to outdo Yonah by 20% with respects to performance with no notable differences in power consumption. Scheduled to be released in fourth quarter of 2006, I was wondering what effects this would have on the mac line of products, especially the unreleased macbook and the other macbook pro sizes (granted they have them).

Now in the past, whenever the powepc chips were upgraded, apple always followed up with newer models...

Now that apple isn't running the powerpc line, I'm curious as to whether they will continue to update at intel's pace, or whether they will hold the current version of the duo for x years then release a mega update...

thoughts anyone?

I'd imagine they'd keep up relatively speaking though they may not adopt them straight away until the model in question is due for its scheduled refresh.

Although technically Apple did not always jump to the latest PPC processors all the time, as there are far more variants of G3 and G4 PowerPC processors than the average person cares to recall.

And then of course there was the Motorola G5 85xx series....

Vicky

Very interesting, I did not know that (learn something new everyday), well I would think that they would put this chip into the Power PC Desktop, for the professional level.

Well, the whole AIM aliance (Apple-IBM-Motorola) thing was a very curious thing indeed.

The initial PowerPC 601, 603(e) and 604(e) were just that, model numbers; some could say the 601 was G1 and the 603/4 (even though the 604 was a vastly superior chip) represented G2 series PowerPC chips.

Then came the G3, the PowerPC 750. This is where the divisions between Motorola and IBM emerged. The G3 was essentially the PowerPC 603e with the backside cache controller and memory of the PowerPC 620. It was also manufactured on a smaller die etc. allowing faster speeds.

After this came the 750CX and 750CXE from IBM; faster still variants; but manufactured by IBM rather than Motorola. These made various tweaks to the cache memory interface, the instruction and data caches etc.

At this time the PowerPC 7400 was in development. The 7400 is the first variant of the G4 as we would call it. This is where IBM and Motorola parted company, as IBM felt a vector unit in the form of AltiVec (or Velocity Engine as Apple called it) was uncecessary. The PowerPC 7400 was essentially the G3 (read 603e+620 backside cache i/f with Altivec). It was a powerful CPU.

IBM however carried on with the PowerPC 750 with the 750FX (notice the naming, it crops up somewhere else later in the line...) which was a vastly improved 750CXE with on-die L2 cache interface, faster clock speeds, more data execution units. It topped out at 1GHz, faster than any Motorola 745/750/755 G3.

Motorola hit the max with the PowerPC 7400 at 500MHz (appropriate because it was codenamed Max) and something had to be done. That was the G4e. The G4e was the G4 with a longer pipeline and some other big improvements which resulted in the PowerPC 7450 and 7410.

The 7410/50 are in fact big enough improvements for them not to even really be G4s at all. Then came the 7451 and 7411 (a low power 7450/1 without the L3 cache interface), 7455 (another big step forward) and with that followed the 7445 (a lower power 7455 without the L3 cache interface), the 7457/7447-A-B.

The 7457 was a great chip which was fast and had the L3 cache interface. Sadly it was not ramped as fast as the 7447 as it was the 7447 Apple chose for its iBook G4s in the end and its PowerBook G4, Mac Mini etc. The 7447/7447A/7447B do not have an L3 cache interface.

In essence, the 7400 and 7455/57 are very different chips and are only G4s in name.

Enter IBM.

Motorola had problems, its own G5 was late—the 85xx series and like the G4 and the G4e, the *real* G5 was the 76xx series (oddly enough) known as G5e. Through various levels of procrastination and muck ups, Motorola failed to produce the goods and Apple turned to IBM.

Unlike Motorola's G5 which is a decendent of the G4 and therefore the G3 which is in turn based on the 603+620 backside cache, IBM's G5 is the POWER-4 architecture married with the PowerPC 604. A minor difference you shout. Not quite. The 604 was 64bit in many ways and also a bloody good processor in the mid 1990s.

IBM reneged on the AltiVec decision and decided a vector unit did have benefits after all, thankfully they still had the knowledge from when they developed the 7400 with Motorola, and so the PowerPC 970 was born. The PowerPC is a mixture of the PowerPC 604 and POWER-4 architecture with the 7400s Vector Unit. (This is of course grossly simplified)

The 970 was improved again in a lower power form of the 970FX (remember IBM's G3e? the 750FX....) and then the 970MP for dual processor and finally the rumoured 970GX mobile processor. These are all IBM-G5s; not Motorola G5s.

Motorola did in the end get its 85xx series off the ground but a little too late for Apple; and it debuted in the PowerPC 8540:

http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=MPC8540&nodeId=011738

It is a G5 processor. Apple just never used it; and hence it has a bit of a stunted development, plus—it's way too powerful for its embedded applications, its give away of being the king that never was.

Of course, Motorola spun of its semiconductor unit into Freescale in 2004; hence the designation as MPC8540, MPC7448 etc.

Talking of which the 7448 is just a refined 7447.

I think this summarises why Apple wanted to use G3, G4 and G5—the PowerPC series was a lot more complicated than it appeared at face value.

Vicky