Welcome to the MacNN Forums.

Crusoe · Mar 6, 2006, 01:35 PM

In lieu of any 30th anniversary rumors, what's the chance they'll give us a Quad Processor Dual Core G5 as the swan song of the the PowerMac?

Steve played down the 30th saying he didn't know what they're gonna do. Anything as special as I'm eluding to is is highly unlikely but what else could or do you think they'll do for their 30th? Discounts? Sweepstakes? A Party with free Schwag at the stores? Bikini Clad Female Wrestlers in Apple Sauce?

Surely they won't only release iBooks and call it an event.

F*ckDell · Mar 6, 2006, 01:48 PM

You've got a great point there, but after reading other info on the possible release of Intel's processors - I don't think it's happening.

Don't get me wrong... I can't wait to see the new Intel Towers. My only worry is, will the perform as well as the quads. I'm looking to buy the quad, but don't know if I should wait until the Intel Towers.

Job's got something planned, we'll just have to wait and see!

Crusoe · Mar 6, 2006, 02:41 PM

I thought it was reletively certain we won't see desktop intels till then end of '06 at best. My PowerMac would be a G5 not Intel.
I'm in the market for a new system and am going to wait till the 30th Anniversary before deciding since I'm in no rush. Leaning toward a G5 PM DC refurb atm.

Some off the wall ideas I thought of are:
the aforemention 4x2 G5.
20" MacBook.
Bluetooth iPod with Stereo Headset
Sweepstakes with a Mini Cooper S JCW GP with iPod and built in Mac Mini with LCD.
Something with green or orange shag carpet and panelling (circa 1976)
%30 discounts for everything!
Willy Wonkaish Sweepstakes with Cupertino Tour Prize.

hickey · Mar 6, 2006, 02:42 PM

Sauce

02gtstang · Mar 6, 2006, 03:02 PM

I almost bought an open box demo QUAD from the apple store this weekend. it was
2799. I tried like hell to get a couple hundred more off, but no dice, so no buy.

Todd Madson · Mar 6, 2006, 03:35 PM

I suspect it will be more of a higher visibility product or something iPod related.

rhashem · Mar 6, 2006, 03:41 PM

Originally Posted by Crusoe

In lieu of any 30th anniversary rumors, what's the chance they'll give us a Quad Processor Dual Core G5 as the swan song of the the PowerMac?

What'd be the point? Most programs wouldn't be able to use the extra four core (the utilization of the two extra cores in the Quad is already pretty low in many programs), and even for the programs that did, the extra four cores would be hampered by the shared memory bus. 8.4 GB/sec is barely enough for a quad-core system, much less an 8-core one!

mduell · Mar 6, 2006, 07:13 PM

No, the first 8 core machines from Apple will be Clovertown based (which won't be out until later this year).

driven · Mar 7, 2006, 12:52 PM

I'm fairly certain Intel doesn't have anything comperable with the current top of the line G5 machines, do they? (Certainly in the 64 bit category ... where AMD shines.)

Crusoe · Mar 7, 2006, 01:14 PM

Ok Ok, so no 4x2s but what do you think we might see for their 30th Anniversary?

hookem2oo7 · Mar 7, 2006, 01:14 PM

intel's current desktop dual core chips (900 series) have to be clocked something on the order of 1-2ghz higher to achieve similar performance to their amd counterparts...I'd say the powermacs won't be switched until the Conroe chips come out

driven · Mar 7, 2006, 02:13 PM

Isn't the 800 series also Dual Core?

1-2 Ghz? That's a HUGE difference. Wow.

F*ckDell · Mar 7, 2006, 05:26 PM

Originally Posted by Crusoe

Ok Ok, so no 4x2s but what do you think we might see for their 30th Anniversary?

I don't remember where... but I read something that they may drop the wheel all together, making the new iPod with a much larger screen. The new larger screen would be touch sensitive. I also remember reading that we'll be able to download full length movies from iTunes. With the new bigger screens, it would compare to the PSP for watching HD movies.

I could have just dreamed this all up, but I'm kind of sure I read it somewhere.

rhashem · Mar 7, 2006, 10:20 PM

Originally Posted by driven

Isn't the 800 series also Dual Core?

1-2 Ghz? That's a HUGE difference. Wow.

It's supposed to be a huge difference. The P4's architecture was an engineering decision. Engineering is all about trade-offs. One of those trade-offs is circuit complexity versus circuit delay. Simple circuits can be clocked higher, but can't be as sophisticated, which limits the performance per clock achievable by using them.

The P4 was the product of a set of simulations that showed that the IPC versus clockrate tradeoff reached an optimal region with a chip that had 20-50 pipeline stages. In such a chip, IPC was lower than in other designs, but overall performance was higher because the higher clockspeed more than made up for it.

This particular architecture was an entirely sensical one given the information Intel had in the late 1990s when the chip was designed. Intel's historical clockrate improvement per process generation had been about 30%. The P4 hit 2 GHz at 180nm. Intel was entirely justified in expecting it to be in the 4.5-5 GHz region at today's state of the art 65nm process. A 4.5 GHz Northwood P4, with the improvements that have since been added to Prescott (bigger caches, etc), would've been an extremely impressive chip.

Unfortunately, the assumptions Intel had at the time the P4 was designed did not hold. Process technology hit a bottleneck that none of the major CPU manufacturers predicted. Everyone was affected, but Intel was affected more than most. Because of power dissipation limitations, the P4 was left with the inherently lower IPC of its design, without the clockrate advantage that was supposed to make up for it.

In summery --- quit ragging on Intel. They made an entirely sound engineering decision, and decided to gamble in making a very innovative processor. That gamble simply didn't work out for them. I should point out that IBM is making the same gamble with POWER6. By all accounts, the POWER6 will be a narrower design then POWER5, but will use aggressive circuit design to achieve a high clock-rate that makes up for any IPC losses.

driven · Mar 7, 2006, 11:13 PM

Thanks for the explaination. It was educational.

One question regarding your last paragraph? How was pointing out that a 1-2 Ghz difference is huge (which you agreed with) qualify as "ragging on Intel" ?

mduell · Mar 7, 2006, 11:43 PM

Originally Posted by driven

I'm fairly certain Intel doesn't have anything comperable with the current top of the line G5 machines, do they? (Certainly in the 64 bit category ... where AMD shines.)

Conroe (out later this year) should be as fast/faster than the dual G5s for most everything and supports 64-bit. Kentsfield and Clovertown (expected later this year) allow for 4 and 8 (16 with Clovertown MP) CPUs in 2 sockets.

driven · Mar 7, 2006, 11:48 PM

Originally Posted by mduell

Conroe (out later this year) should be as fast/faster than the dual G5s for most everything and supports 64-bit. Kentsfield and Clovertown (expected later this year) allow for 4 and 8 (16 with Clovertown MP) CPUs in 2 sockets.

Sounds exciting!!!

Catfish_Man · Mar 8, 2006, 03:38 AM

Originally Posted by rhashem

In summery --- quit ragging on Intel. They made an entirely sound engineering decision, and decided to gamble in making a very innovative processor.

Of the three major Pentium 4 core revisions, I would say that only one of them was an entirely sound engineering decision. Willamette was prematurely released* in response to the Athlon, Northwood competed well for most or all of its lifespan, and Prescott was (as far as I can tell) was simply a demonstration of willful ignorance of engineering realities.

Was Intel expecting the 90nm process to bomb so spectacularly? No, of course not, but the writing was still on the wall. There was no way the Pentium 4 core could scale to the levels its design would have allowed had thermal limits not existed. Increasing the length of the pipeline only exacerbated the problem by further reducing IPC and increasing power overhead.

*it also had to deal with the optimization transitions required by the new architecture (non-free fxch, slow as hell rotate/shift, etc...), AND the rdram fiasco, so the core itself may not have been quite so much to blame. Hard to say, really.

Ken_F2 · 04:32 AM

It will still be another nine months before we see an 8-core PowerMac.

Intel's next-generation "Conroe" processor is slated for release between July 31 and August 31 ("early in 3Q 2006" according to Intel today), but the quad-core versions aren't coming until January, 2007. The "Kentsfield" quad-core will only be for single processor desktop systems, but there will be two server versions. "Clovertown" is the server version of the quad-core processor, while "Clovertown MP" is the server version for systems with up to four processsors (16 cores). All coming January, 2007.

Anandtech ran benchmarks on the new dual-core "Conroe," which you can see here. Note those benchmarks are only for the mid-range 2.66GHz version. It will be available in 2.93Ghz and 3.33GHz versions at release. One of the more interesting comments from the Intel demos:

We saw a very quick demo of Battlefield 2 being played on Intel's Conroe chip, today.

We watched a quick demo of two gamers loading up a Battlefield 2 map. Anyone who's played Battlefield 2 knows that it takes an absolute age to go from 'Enter map', through 'Verifying data' all the way to frag-o-matic.

With Robson working, and what was called 'Intel's special sauce' (leading to quips amongst the audience of 'Intel wants to feed gamers its special sauce'), the load times for a Battlefield 2 were cut, roughly in half.

This led to an amusing demo where the chap playing on the Conroe system was in the map, had grabbed a plane and was able to frag his opponent with a hail of missile fire just as the poor chap was spawning.

mduell · Mar 8, 2006, 04:00 AM

The "special sauce" is NAND flash being used as a disk cache.

I saw a link on it earlier today showing the performance difference. Opening Office files and running macros in 3s instead of 15s. Booting the OS faster. All sorts of good stuff.

Coming with Santa Rosa in Q1 2007.

rhashem · Mar 8, 2006, 02:07 PM

Originally Posted by Catfish_Man

Of the three major Pentium 4 core revisions, I would say that only one of them was an entirely sound engineering decision. Willamette was prematurely released* in response to the Athlon, Northwood competed well for most or all of its lifespan, and Prescott was (as far as I can tell) was simply a demonstration of willful ignorance of engineering realities.

Prescott was a last ditch effort to plug the leaks in the ship while waiting for Merom to ship. Everything points to Prescott being a rush job, from the small but high-latency L1 to the 30+ stage pipeline just to reach 3.8 GHz.

Was Intel expecting the 90nm process to bomb so spectacularly? No, of course not, but the writing was still on the wall.

I think you're overestimating the information available in hindsight. As I said, IBM didn't see this "writing on the wall" either, otherwise the G5 would've been a high-IPC design. Perhaps AMD saw the writing on the wall, or perhaps the fact that AMD can't keep a processor design team together for more than one job prevented them from releasing a brand new core in response to the Pentium 4.

There was no way the Pentium 4 core could scale to the levels its design would have allowed had thermal limits not existed.

Intel engineers aren't stupid. They used projections of the power dissipation of future processors to determine how high they could clock the design while staying within a reasonable power envelope. Their projections were just hugely wrong. Of course, so were everyone else's --- it was just Intel's design that took the biggest hit from that mistake.