Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
more power4 hints
- Thread starter metfoo
- Start date
- Sort by reaction score
Mr. Anderson
Moderator emeritus
weird, just mentioned in passing.....hmm, I definitley like the 6.4 GB/s bandwidth - that would be one nice machine......I hope this is more than speculation
and there is no more mention of the G4 in the article, damn.
and there is no more mention of the G4 in the article, damn.
Vaporware!
No main memory RAM exists that does 6.4GB/sec. The fastest RAM is 1066 RDRAM which clocks out at 4.1 GB/sec. Next October 2003, Rambus is coming out with 1333 RDRAM which will max out at 5.2 GB/sec.
You won't see anything faster until mid 2004. You are dreaming if you think Apple will have the Power4 in the PowerMac in Jan. 2003. All you will see is a faster G4 that Motorola will call G5 clocking at best 1.6GHz, built w/ 130NM tech, with a 512 cache, and maybe at best a (real) 333 DDR front side bus. Not the overclocked crap Apple is pushing now. Motorola is too broke and losing too much money to make anything faster.
And the Power4 is incompatible with AltiVec based software. Developers won't spend millions to rewrite their software to run on the unproven Power4.
Peter
No main memory RAM exists that does 6.4GB/sec. The fastest RAM is 1066 RDRAM which clocks out at 4.1 GB/sec. Next October 2003, Rambus is coming out with 1333 RDRAM which will max out at 5.2 GB/sec.
You won't see anything faster until mid 2004. You are dreaming if you think Apple will have the Power4 in the PowerMac in Jan. 2003. All you will see is a faster G4 that Motorola will call G5 clocking at best 1.6GHz, built w/ 130NM tech, with a 512 cache, and maybe at best a (real) 333 DDR front side bus. Not the overclocked crap Apple is pushing now. Motorola is too broke and losing too much money to make anything faster.
And the Power4 is incompatible with AltiVec based software. Developers won't spend millions to rewrite their software to run on the unproven Power4.
Peter
alex_ant
macrumors 68020
Re: Vaporware!
Alex
You blasted one rumor away, but you used an equally flaky rumor to back yourself up here with regards to AltiVec. The mention that the new Power4-derived PPC will not be AltiVec-compatible is only speculation - reported by one person who talked to another person at IBM who we are supposed to believe knew what he was talking about. IBM would be mad for this new chip not to support VMX. If this new chip isn't AltiVec-compatible (but NOT called such, for AltiVec is a Motorola trademark), then bicycles don't have wheels, or something.
Alex
Yeah...
It gets confusing...
Here's what we know:
IBM will be talking about a 64-Bit PowerPC in Oct
That is all we know for sure.
What I would rate as a reliable rumor is this:
that IBM/Apple have been in discussions about it
beyond that, it's been pretty speculative.
arn
It gets confusing...
Here's what we know:
IBM will be talking about a 64-Bit PowerPC in Oct
That is all we know for sure.
What I would rate as a reliable rumor is this:
that IBM/Apple have been in discussions about it
beyond that, it's been pretty speculative.
arn
Especially when you're dealing with the Inquirer.
This is incorrect. "Altivec" is Moto's brand name for a vector processing unit. Apple calls it "velocity engine". IBM has indeed added a vector processing unit to the rumored Power4-lite-mini me.
No one knows for sure until the official announcement by IBM, but it appears no software will have to be rewritten.
MacCoaster
macrumors 6502a
While your original statement was correct, your last statement is incorrect. It's not rumored that IBM is working on a new POWER-family chip (includes Performance Optimization with Enhanced RISC and PowerPC). It's a fact. Any details beyond that is purely speculative until announced on October 15.
Re: Vaporware!
Second the MPX bus will never be a DDR bus. Not ever. Technically it's too much trouble and doing things which are technically troubling costs money, which Motorola by your own words can't afford to spend.
Motorola is already switching their G5s to on-chip memory controllers and RapidIO and assuming they were to release a next generation system for Apple they would do the same.
Third Apple does not sell overclocked systems. Anybody who says otherwise is simply ignorant.
Finally you are correct the POWER4 won't appear in Macs but you are wrong in that the POWER4 isn't an unproven technology.
Additionally IBM has already come on the record as stating they are developing a POWER4 derivative with a SIMD unit. It won't be altivec, which is Motorola's implementation of that instruction set, but Apple isn't planning on using anything in their future chips (at least for the near term) that doesn't retain altivec compatibility.
Ok just to correct a few facts. Firstly RDRAM offers less bandwidth than DDR RAM. The only reason RDRAM shows as higher is because currently you can only buy it as a dual channel offering, which effectively doubles the bandwidth. The real bandwidth of RDRAM 1066 is only ~2.1 GB/s. Dual channel DDR RAM can reach 4.2 GB/s with DDR-266, 5.4GB/s with DDR-333 and 6.4 GB/s with DDR-400. Offerings already exist for dual-channel DDR-400 memory as well.
Second the MPX bus will never be a DDR bus. Not ever. Technically it's too much trouble and doing things which are technically troubling costs money, which Motorola by your own words can't afford to spend.
Motorola is already switching their G5s to on-chip memory controllers and RapidIO and assuming they were to release a next generation system for Apple they would do the same.
Third Apple does not sell overclocked systems. Anybody who says otherwise is simply ignorant.
Finally you are correct the POWER4 won't appear in Macs but you are wrong in that the POWER4 isn't an unproven technology.
Additionally IBM has already come on the record as stating they are developing a POWER4 derivative with a SIMD unit. It won't be altivec, which is Motorola's implementation of that instruction set, but Apple isn't planning on using anything in their future chips (at least for the near term) that doesn't retain altivec compatibility.
That might be what most people know but others do know more and some people right have even gotten it right either cause they 'really know' or they were lucky. Problem is with most (not just this one) open forums is when people who do know the facts try to present them people either demand proof (understandable I guess) that the person isn't willing to provide (also very understandable).
I'm in that position myself... so rather than try and argue with those who say Alti-Vec is *totally impossible* (it isn't) or 6.4Gb/s can't happen (it can) we either post silly messages like I'm doing right now or just don't say anything and laugh.
What's best is to have much smaller (far less public) invite only forums where fools who post trash rumors (just for fun) as well as people who shout down people just because they don't like what the person is saying get booted and never get invited back. An 'insiders club' if you will.
Do those places exist? I dunno but it sure wouldn't surprise me.. 😎 Well that's the kinda place I'd like to be... I'm getting way to tired of silly photoshop mockups and tales of iWalks and the like.
Dave
Ok, I'll repeat the info I've received again. We will have power4 core macs by this time next year. Moto will have the last G4 in jan but doubt it will be called a G5 when everyone knows it's a G4. It should, however, be on a new mobo. The last bump was supposed to feature it but they had major heat issues which may explain the ports in front of the towers.
We WILL have this IBM power4 core mac nexy year according to a very creditable source.
We WILL have this IBM power4 core mac nexy year according to a very creditable source.
