90nm G5 Supplies Improving?

[FoxyKaye]

The test models are reported to range from 2.2GHz to 2.6GHz.

What would make me really happy in a revision B machine are the following:

* A low-end tower that remains a 1.8 or 2.0 DP with a cheaper price (something akin to what Apple is currently asking for its last G4 tower systems).

* A revised form factor that actually allows for more than two internal hard drives and space for two internal CD/DVD devices.

* Bluetooth/Airport Extreme standard on all models (modem optional)

* Six USB 2.0+ (three in front, three in back), five FireWire 400 (two in front, three in back), one FireWire 800 (I don't care where they put this, no-one seems to be adopting this standard).

 

[afields]

My hope is that that the following models are announced:

2.2, 2.6, 3.0 PowerMac 1.6, 1.8, 2.0 iMac.

They could ship the lower models immediately, and then ship the 3.0 PowerMac and the 2.0 iMac by september.

 

[tortoise]

Also, POWER 5 has an on-die memory controller and I beleive uses a full implementation of hyper-transport for the FSB, so anything AMD may have on us now is just something that IBM's playing with right now or have been doing for a while already.

The real question is how much of the Power5 will actually make it into the PPC. An integrated memory controller would be huge, and a built-in ccNUMA fabric would be nice too. AMD has an advantage in this regard because they are the owners of the underlying HyperTransport technology (they bought it from someone else -- it was originally designed to be the fabric for the next generation Cray supercomputer).

The AMD cores are generally pretty slick, but comparable to PPC and Intel in basic capability. Their real advantage is the memory architecture and thorough integration of the HyperTransport fabric. If IBM manages to produce a PPC that fully exploits this fabric rather than using it as a glorified I/O bus, it would be make things interesting. As far as I know, Intel has not licensed HyperTransport, which will put them at a disadvantage until they can come up with something comparable. HT 2.0, with even greater scalability, is already well on its way, so Intel better hurry.

 

[ShnikeJSB]

Intel is on a downward spiral!

As far as I know, Intel has not licensed HyperTransport, which will put them at a disadvantage until they can come up with something comparable. HT 2.0, with even greater scalability, is already well on its way, so Intel better hurry.

Intel, as we probably all know, is abandoning the P4, since they can't keep the heat down and can't seem to scale it right, so they are going to use the Pentium M as the basis for their future processors... LOL! What was originally a NOTEBOOK processor??? HAHAHAHA!!! I hope AMD puts Intel out of its misery! Actually, I don't, because competition is always good for consumers. I just hope IBM can work out the kinks and get the new chips OUT THERE for us! -JB

 

[The Red Wolf]

64-bit G5 in 5 years

Just a thought: Anyone ever consider that when the current G5 line is out five years and whatever version of the OS Apple announces that it may actually speed up the machine? Much like putting Panther on a G3 350 MHz. Or a Keylime iBook SE 466 Mz Panther runs faster than Jaguar ever did on those machines.

How do we know that within that same 5 year time we may be seeing a 64-bit OS which will take full advantage of this hypothetical situation. Who knows having a 64-bit machine may be like not having built in USB. "OS-XII 12.1 (Komodo) will not function on any any 64-Bit machine pre Wireless Firewire" Or some such. Then again you might see "OS-XII Komodo is compatible with pre existing 32-bit machines however performance is greatly limited without at least 1 GHz bus speed".

Lastly... The "sad" 1.6 GHz G5 machines today, may see speed improvements with future OS releases. Currently, said machine out paces anything on the floor at work that I take care of. We've 80 Macs. Ranging from G3 400s to Dual 867 MMDs. The "useless" 1.6 GHz G5 is what then to our production fleet? A God? Or so much aluminum junk making all the Macs on the floor mad paperweights? Work is still done. When budget allows, with a company only goes for the low end, it's nice to know that there are cheeper versions of the high end up to date, screaming Rev B G5 PowerMac.

 

[Sun Baked]

The 130nm process looks better... hate to see the 90nm one.

Originally posted by stingerman on ARS:

It says 130NM but at the same time it says 300MM. This chart is for the Fischkill plant. It was filed today with the SEC.

Here is John Kelly's script for this slide:

Slide 8
Much has been written on our yields recently. I want to take a minute to let you know where things stand.

As our CFO, John Joyce, said when we reported our first quarter earnings, our 200 mm yields are at or above plan while our

300 mm yields, while improving, are not yet where we want them to be.

The next chart illustrates those comments. As you can see, our 130 nm, 300 mm defect densities — the number of defects in a given section of silicon — are showing rapid improvement. As you can see, we are getting much closer to where we want to be.

It's important to point out here that we are working on extremely complex logic. At a given lithography node we can provide up to 20 percent higher speeds than our competition, though in some cases these advanced chips are more difficult to yield. And while the traditional foundries currently have less than 20 percent of their volumes in 130 nm or smaller, about 50 percent of our volumes are 130 nm or below.

As John Joyce suggested, we expect to do a better job of meeting customer demand in second quarter

 

[PowerMacMan]

My hope is that that the following models are announced:

2.2, 2.6, 3.0 PowerMac 1.6, 1.8, 2.0 iMac.

They could ship the lower models immediately, and then ship the 3.0 PowerMac and the 2.0 iMac by september.

Seems very realistic... Shipping times should be earlier though... IMO

 

[afields]

very true...ideally everything would be available immediately. I have a feeling that the top model will ship later. Hopefully july or something.

 

[ingenious]

That can change, that was just one of the possibilities. The 3GHz could use a 3:1, so it would still have a 1 GHz bus but be 3GHz. If the low-end was a 2.4, then it could have a 800MHz bus, the same as it is now. The 2.7 could be 900Mhz, the same as the 1.8. Just because the CPU speed increases doesn't mean that the bus will. Intel has used 400 and 533 MHz for how long?

and we want to be like intel, why?

 

[Mav451]

As far as I know, Intel has not licensed HyperTransport, which will put them at a disadvantage until they can come up with something comparable. HT 2.0, with even greater scalability, is already well on its way, so Intel better hurry.

Intel, as we probably all know, is abandoning the P4, since they can't keep the heat down and can't seem to scale it right, so they are going to use the Pentium M as the basis for their future processors... LOL! What was originally a NOTEBOOK processor??? HAHAHAHA!!! I hope AMD puts Intel out of its misery! Actually, I don't, because competition is always good for consumers. I just hope IBM can work out the kinks and get the new chips OUT THERE for us! -JB

Laptop cpus aren't necessarily bad--its just that the Pentium-M's aren't running the same 800fsb for memory bandwidth as the P4's.

The reason laptop CPUs are particularly interesting, especially for desktop environments, is that they MUST, absolutely MUST run at a very low voltage. With this in mind, if it is run in a desktop environment, while adhering to the laptop standards, it would run EXTREMELY cool in a standard tower (probably <20-25C's compared to a more normal 32-35C's).

What does this mean? It means it can be "overclocked" much much higher for the desktop environment. Say an default desktop version runs @ 1.83ghz @ 1.65vcore. Pretty standard. Now, the laptop or "mobile" version, can ALSO run @ 1.83ghz, but @ only 1.45vcore. Anyone familiar with AMD overclocking knows how huge even .05 vcore is, but .2 vcore? That's ridiculous. This is easily 400-500mhz that can be squeezed through such an increase. Hence the reason many gamers hit 2400-2500 "out of the box" - literally set-it-and-forget-it RIGHT after setting up the computer.

These mobile procs from AMD are no joke: A Mobile 2500+ @ 2.7ghz (224fsb) is roughly equivalent (slightly less) to a 3.8ghz (254fsb) P4 in gaming performance--3dmark 2001se scores of 21,1k and 22,3k respectively. Obviously the P4 has a much higher memory output @ > 1ghz true cpu-memory fsb

THIS is why I am interested in seeing how Intel will work with this laptop proc in a desktop environment...with AMD's mobile proc being undoubtedly the most powerful, <100 dollar investment for serious computing power...we shall see how Intel uses its laptop technology...

 

[SiliconAddict]

Intel, as we probably all know, is abandoning the P4, since they can't keep the heat down and can't seem to scale it right, so they are going to use the Pentium M as the basis for their future processors... LOL! What was originally a NOTEBOOK processor??? HAHAHAHA!!! I hope AMD puts Intel out of its misery! Actually, I don't, because competition is always good for consumers. I just hope IBM can work out the kinks and get the new chips OUT THERE for us! -JB

Dude you ARE aware of how damn good the Pentium M is right? It actually runs rings around the G4 at = clock cycles. The CPU may have been made for laptops but the thing is efficient enough that Intel realized that this CPU can easily scale to a desktop without even breaking a sweat. AFAIK a 1.7Ghz Pentium M actually is neck and heck with a 3Ghz Pentium 4 which is pathetic as hell and now that Dothan, Intel's 2MB cache Pentium M, is shipping..... As they say don't knock it till ya try it. I know exactly where Intel is going with this. Its their typical brute force approach to computing. The Pentium M is great at only 1.7Ghz but in laptops you have to worry about batt performance. Not so with desktops. I will bet that we are going to see 3Ghz (prob higher since its using a 90nm process.) Pentium M based systems in the not to distant future that are equipped with multi-cores. Intel has been a big go getter in SMP lately most likely because they are struggling with CPU speeds. Do NOT discount the M. Its quite possibly the best CPU Intel has ever released. Quite frankly the folks on the Desktop development team should be scared pissless that Israel's Pentium M team may end up getting their jobs.

 

[gekko513]

AFAIK a 1.7Ghz Pentium M actually is neck and heck with a 3Ghz Pentium 4

Well .. the Pentium M is not always that great. In some test runs that I've done, a Pentium M @ 1.6GHz was slightly slower than a Xeon @ 2GHz.

 

[Travl]

I have a fairly simple question I hope someone can answer. Im just curious as to how hard it would be to upgrade from say a dual G5 2.0 ghz to one of the newer chipsets once they are released or if this is even possible. I know this is a little off topic but kinda has to do with the release of the rev b. Thanks