Apple's Processor Numbering "G" system

[wrldwzrd89]

Here's how Apple got the G numbers for the PowerPC processors:
G1 = 601
G2 = 603, 603e, 604, 604e
G3 = 750 and variants
G4 = 7447 and variants (Motorola)
G5 = 970, 970fx

 

[Zaty]

Here's how Apple got the G numbers for the PowerPC processors:
G1 = 601
G2 = 603, 603e, 604, 604e
G3 = 750 and variants
G4 = 7447 and variants (Motorola)
G5 = 970, 970fx

What happend to the 8xx(x)?

 

[wrldwzrd89]

What happend to the 8xx(x)?

Good question - I guess IBM decided not to use 8xx numbers. Motorola still might use an 8xxx number for a future processor, though.

 

[miloblithe]

What happend to the 8xx(x)?

You mean 680x0 ?

Oh, I see what you mean.

 

[idkew]

i would say that the 60X series are all Generation 2, and the 680X0 is G1.

 

[Sun Baked]

i would say that the 60X series are all Generation 2, and the 680X0 is G1.

Nope completely different processor, the PowerPC was a joint effort between Motorola and IBM -- and the 68000 series existed way before that.

The PowerPC used IBMs POWER instruction set and Motorola's 88000 bus, don't see the 68000 CISC chips in that mix.

And the very first PowerPC was the 601 chip, which still had quite a few POWER instructions.

---

On another note, today's IBM POWER event may have showed another step in the deterioration of the AIM alliance and the effort to move away from the PowerPC moniker.

Depending on how AIM moves down the road, Apple may begin using IBM's moniker -- "Power Architecture."

 

[idkew]

Nope completely different processor, the PowerPC was a joint effort between Motorola and IBM -- and the 68000 series existed way before that.

The PowerPC used IBMs POWER instruction set and Motorola's 88000 bus, don't see the 68000 CISC chips in that mix.

And the very first PowerPC was the 601 chip, which still had quite a few POWER instructions.

---

On another note, today's IBM POWER event may have showed another step in the deterioration of the AIM alliance and the effort to move away from the PowerPC moniker.

Depending on how AIM moves down the road, Apple may begin using IBM's moniker -- "Power Architecture."

but does the Generation refer the the PowerPC line, or the generations of chips apple has used? remember, the Generation moniker is apple's creation, not moto or ibm's.

 

[MisterMe]

i would say that the 60X series are all Generation 2, and the 680X0 is G1.

You would be wrong. The Generation designation refers exclusively to PPC-based Macs. It refers to the whole computer, not just the processor. There are substantial architectural changes from one "generation" to the next.

G1-NuBus, SCSI, PPC 601
G2-PCI, SCSI, PPC 603, 604, 603e, 604e
G3-PCI, ATA, PPC 74x, 75x
G4-AltiVec (seemingly a minor architectural change) PPC 74xx, 75xx
G5-PCI-X, S-ATA, PPC 970

 

[wrldwzrd89]

You would be wrong. The Generation designation refers exclusively to PPC-based Macs. It refers to the whole computer, not just the processor. There are substantial architectural changes from one "generation" to the next.

G1-NuBus, SCSI, PPC 601
G2-PCI, SCSI, PPC 603, 604, 603e, 604e
G3-PCI, ATA, PPC 74x, 75x
G4-AltiVec (seemingly a minor architectural change) PPC 74xx, 75xx
G5-PCI-X, S-ATA, PPC 970

Hmm...That gets me thinking. How is adding AltiVec enough to change chip generation designations? I wonder if we should really be on G4 chips, not G5s. Some might say that G5 PowerPCs put us ahead of Intel's "G4" Pentiums, which shouldn't really be G4 anyway - the P4 was significantly changed from the P3 with HyperThreading, but the P3 and P2 were almost the same; in that case, we're still ahead - PowerPC "G4" vs Pentium "G3".

 

[MisterMe]

Hmm...That gets me thinking. How is adding AltiVec enough to change chip generation designations? I wonder if we should really be on G4 chips, not G5s. Some might say that G5 PowerPCs put us ahead of Intel's "G4" Pentiums, which shouldn't really be G4 anyway - the P4 was significantly changed from the P3 with HyperThreading, but the P3 and P2 were almost the same; in that case, we're still ahead - PowerPC "G4" vs Pentium "G3".

The G4 is more than Altivec, although a built-in vector processor is not to be sneezed at. That said, the G4 supports mulitprocessing whereas the G3 does not. Apple produced dual 604/604e-based computers. Apple produced dual G4s and G5s. Apple cannot produce dual G3s.

 

[wrldwzrd89]

The G4 is more than Altivec, although a built-in vector processor is not to be sneezed at. That said, the G4 supports mulitprocessing whereas the G3 does not. Apple produced dual 604/604e-based computers. Apple produced dual G4s and G5s. Apple cannot produce dual G3s.

Thanks MisterMe for clearing that up. The G4 really does deserve to be called a G4, then; but the Pentium 4 should really be a Pentium 3 (and the Pentium 3 should be the Pentium 2e, or something) so that puts us effectively TWO generations ahead. Take that, Intel!

 

[blue&whiteman]

my G4 cpu is a 7410. how is my 7410 different from the 7400?

 

[wrldwzrd89]

my G4 cpu is a 7410. how is my 7410 different from the 7400?

After looking on Motorola's web site (the makers of the 7410), there appears to be no 7400 processor. I believe that what was meant by 74xx was that multiple Motorola chips use this naming scheme, but not all possible numbers were used (like 7400).

 

[V.A.Toss]

l;hlhlkhjlkjkounm mkioh

My memory is not what it once was.

But as i recall, the 7400 was the first G4. The 7410 was the second G4, or as moto called it, the G4+. The main difference was the L2 cache, and i believe an increase in stages of the pipelines, maybe from 4 to 7, i cant remember exactly. Also a shrink in die size and it was fabbed at 0.18 microns.

The 7410 was really a stopgap from moto, after the ultra-embaressing 7400 being stuck at 500Mhz for a year, they needed something quickly. The 7410 was it, but at the time there was controversy as to whether it was actually much faster than the 7400 in "real-life", in some benchmarks it lost out i believe. But then, in some benchmarks the Pentium3 actually loses out to a 166 (i dont know whether thats as unbelievable as it first sounds).

Whilst the 7400 and 750 are very similar, generally the G4 is actually derived from the 604 rather than the 603. Though again, this isnt entirely accurate.

At the time altivec was a big thing for apple, and as apple saw it as a next generation of cpu, i guess they decided to call it a G4 rather than a G3. It still is a remarkable piece of engineering. Imagine at the time of Pentium3s and Pentium2s and MMX instruction, to hear that a G4 had a VPU unit that could process (in theory) a set of 16 instructions in one go. As opposed to a pentium, which did it in 16.

Yes the G4 has full MERSI support. There cannot be a dual G3, unless IBM come up with this full MERSI compliant 750 that the rumours have suggested.

 

[wrldwzrd89]

link

My memory is not what it once was.

But as i recall, the 7400 was the first G4. The 7410 was the second G4, or as moto called it, the G4+. The main difference was the L2 cache, and i believe an increase in stages of the pipelines, maybe from 4 to 7, i cant remember exactly. Also a shrink in die size and it was fabbed at 0.18 microns.

The 7410 was really a stopgap from moto, after the ultra-embaressing 7400 being stuck at 500Mhz for a year, they needed something quickly. The 7410 was it, but at the time there was controversy as to whether it was actually much faster than the 7400 in "real-life", in some benchmarks it lost out i believe. But then, in some benchmarks the Pentium3 actually loses out to a 166 (i dont know whether thats as unbelievable as it first sounds).

Whilst the 7400 and 750 are very similar, generally the G4 is actually derived from the 604 rather than the 603. Though again, this isnt entirely accurate.

At the time altivec was a big thing for apple, and as apple saw it as a next generation of cpu, i guess they decided to call it a G4 rather than a G3. It still is a remarkable piece of engineering. Imagine at the time of Pentium3s and Pentium2s and MMX instruction, to hear that a G4 had a VPU unit that could process (in theory) a set of 16 instructions in one go. As opposed to a pentium, which did it in 16.

Yes the G4 has full MERSI support. There cannot be a dual G3, unless IBM come up with this full MERSI compliant 750 that the rumours have suggested.

Oops! I guess I looked in the wrong place on Motorola's web site. The PowerPC 7400 does in fact exist - here's a link to Motorola's documentation about the processor. <edit> Here's another link.</edit>

 

[wrldwzrd89]

my G4 cpu is a 7410. how is my 7410 different from the 7400?

Look in this PDF for the answer. The table you seek is in section 3.10.

 

[JFreak]

the Pentium 4 should really be a Pentium 3 (and the Pentium 3 should be the Pentium 2e, or something) so that puts us effectively TWO generations ahead. Take that, Intel!

i agree - the P3 should not have been called that.

BUT...

P4 is way faster chip than G4 even if the P4 actually is third generation pentium. take that, motorola!

 

[wrldwzrd89]

i agree - the P3 should not have been called that.

BUT...

P4 is way faster chip than G4 even if the P4 actually is third generation pentium. take that, motorola!

Moto deserves it, too - for what we call "Motostagnation". However, we have the G5 now, made by IBM. All that's needed now is a wider rollout of the G5 into the rest of Apple's Mac line.

 

[Doraemon]

Here's a list of the CPUs used in Apple's G3/G4/G5 systems:

http://www.theapplemuseum.com/index.php?id=tam&page=products&subpage=cpus

 

[blue&whiteman]

my exact chip isn't even listed there. I bought it from sonnet and its a G4 7410 ver1.4. apple only lists a 7410 ver1.3. I wonder what the 7410 1.4 offered over the 1.3. must be something different or they would not have made a new version.

 

[idkew]

You would be wrong. The Generation designation refers exclusively to PPC-based Macs. It refers to the whole computer, not just the processor. There are substantial architectural changes from one "generation" to the next.

G1-NuBus, SCSI, PPC 601
G2-PCI, SCSI, PPC 603, 604, 603e, 604e
G3-PCI, ATA, PPC 74x, 75x
G4-AltiVec (seemingly a minor architectural change) PPC 74xx, 75xx
G5-PCI-X, S-ATA, PPC 970

besides PCI vs NuBus, what is different between a 601 and a 603/4 that necessitates a G change?

 

[wrldwzrd89]

What happend to the 8xx(x)?

I checked Motorola's site; it turns out that Motorola IS using 8xxx numbers, just not for chips that Apple uses. I couldn't find any evidence of 8xx series chips on IBM's site.

 

[Zaty]

Thanks MisterMe for clearing that up. The G4 really does deserve to be called a G4, then; but the Pentium 4 should really be a Pentium 3 (and the Pentium 3 should be the Pentium 2e, or something) so that puts us effectively TWO generations ahead. Take that, Intel!

As for PIII, your statement is only half true. The first generation PIII was essentially a PII + a new instructions set. That's why when the PIII was introduced; the 450 MHz PIII scored the same results as the PII 450 MHz because there was no software that could take advantage of the new instructions. But then, the original PIII was succeeded by two new revisions (Coppermine and Tulatin), which had a slightly different architecture, so those were "real" PIII.

 

[Zaty]

I checked Motorola's site; it turns out that Motorola IS using 8xxx numbers, just not for chips that Apple uses. I couldn't find any evidence of 8xx series chips on IBM's site.

Thanks for the research!

 

[V.A.Toss]

klhjklhkjhkjhkjh

As for PIII, your statement is only half true. The first generation PIII was essentially a PII + a new instructions set. That's why when the PIII was introduced; the 450 MHz PIII scored the same results as the PII 450 MHz because there was no software that could take advantage of the new instructions. But then, the original PIII was succeeded by two new revisions (Coppermine and Tulatin), which had a slightly different architecture, so those were "real" PIII.

You are right about the P3, there have been just about enough architectural changes to make it a distinct cpu. But these changes are in no way near the transformation the powerpc has gone through in recent years. I was told by a CS friend of mine that part of the reason for intel increasing the pipeline stages in the P2-P3-P4 without a great improvement in branch logic, was simply that they wanted to increase the clock speed. Without drastic changes in architecture, this was the easiest (and cheapest) way to do it.
Im not a CS expert, but it makes sense.
As for the P4 being faster than the G4, well yes it is, but it wouldnt have been if moto had invested in it, the clock speed would have improved no-end and there might have even been a true DDR fsb, and the G4 would have matched the P4. The original design of the G4 isnt bad even by todays standards.