90nm 3GHz PPC 970FX by Summer

dmdimon sent in linkage to a Forbes story on the upcoming PPC chips and notes "IBM is said to be ready to deliver a new version of its PowerPC processor to Apple by the end of this year in from sizes of 130 nanometers to 90 nanometers... Apple CEO Steve Jobs has already gone on the record saying that the G5 computer will contain PowerPC chips that run at 3 GHz by the summer of 2004. A mid-step between the current systems, which top out with two chips running at 2 GHz, and systems with chips as fast as 2.6 GHz would be a logical move come January..."

Comments

Re:POWERBOOK EATS YOU.

[American AC in Paris]

I would have to wait until the second generation of that machine b/c I would like to make sure that Apple just isn't unloading G4 parts with a G5 chip.

I agree with the wisdom of letting others find the flaws in a first generation laptop--It's too easy to get burned with a brand new laptop design, pun intended.

That said, Apple puts more effort into laptop design than just about any other manufacturer I can think of. I seriously doubt they'd slap a G5 processor into a G4 design and call it done.

Re:With no Volume.

[msgmonkey]

I think you're over estimating how much apple technology went into the PPC, nearly all of the technology came from IBMs POWER risc machines (Wikipedia entry). Apple probably had a hand in specifiying the bus interface for their chipsets and other interface level stuff but I doubt anything more than that.

It takes a lot of R&D whenever you move from one feature size to a smaller one and since chip fabrication R&D costs $$$ that's why unless AMD have some kind of technology sharing agreement I doubt they would just "give away" something they've put a lot of money into.

Re:What I love about Apple

[be-fan]

That's a completely bogus post. Every chip manufacturer does die shrinks periodically. And Intel also makes a line of optimized compilers specialized for its chips. Apple and IBM are nothing special in this regard!

Spot the troll...

[Kjella]

You supply 3% of the computer market with chips, you can hand pick your chips and speed bin the rest.

Right... How does the *size* of the market relate to the yield? If a certain fraction of the chips you produce are exceptionally good, Intel/AMD can "hand pick" just as much, or as little as Apple. Their chips aren't 3% hand picked from 100%, they produce 3%, and a fraction of those again could be "hand picked".

The rest are just unsubstantiated rumors, following up a good troll. And the moderators are falling for it hook, line and sinker.

Kjella

Re:90nm Soft Error Rate

[barawn]

Some people think Failures in Time (FIT) rates will get better at 90nm than 130nm. Some think the opposite. Xilinx and Actel are arguing over it. Caches are epecially vulnerable. In a critical software application, this is unacceptable, and sometimes the cache needs to be disabled altogether.

Actually, it'd be better stated that Actel is arguing with the rest of the FPGA industry, as Actel's the only one that makes antifuse FPGAs. Xilinx is vocal, but almost everyone else would agree with them as well.

I've got a guess that Actel might be a little bit biased.

If you had to believe one or the other based on equal information, you'd tend to believe Xilinx: they can afford to give up the rad hard market, as it's not that large, so they really have very little incentive to lie. Actel, however, is completely unable to compete on price issues (god, their development kits/hardware/programmers are insane!) and so they'd have a strong incentive to lie about the reliability of the competition to get people to switch to them.

However, I also know that if I had wanted to fly a PLD on any NASA mission, I'd have to choose Actel. So someone believes them...

I wonder how many software errors will be caused by neutrons hitting the processor and upseting logic gates?

Er? I don't see many free neutrons running around in a normal environment, unless you're working near a nuclear reactor. That 11-minute half-life tends to make them go away - they're a negligible component of cosmic rays. Do you mean alphas? Alpha particle strikes on electronics are a known thing - that's why ECC is around.

Re:Disappointing progress with laptops..

[Trurl's Machine]

I am still thinking of replacing this iBook eventually with another iBook but only when they come up with the model that's at least twice as fast as the 600MHz iBook (actually CPU speed is the only reason I want to upgrade it).

Now, while I would rather recommend holding all iBook/powerbook purchases (I think that major speed progress is intevitable in 2004), you would actually notice huge difference between your iBook and a contemporary one, sometimes even surpassing the "at least twice as fast" condition. Your old iBook uses ATI RAGE 128 with mediocre 8 megabytes of video RAM, the new one is a RADEON 7500 with 32 MB VRAM. If you play games, the difference is huge. But even if you don't, MacOS X GUI heavily relies on the GPU support, so your CPU has to sweat a lot just on calculating all those pretty widgets. And finally, many applications actually take the full advantage of the G4 architecture and they also could have a ~2x boost on a new iBook (a megahertz of G4 is not the same as a megahertz of G3).