G5 Bus Speed. WHAT!

[waynepixel]

I am thinking about buying a new G5 1.8Ghz, but when I looked at the Frontside bus on the G5 1.8 it said, it at 600Mhz that 200 less then the G5 1.6GHz.

The G5 1.6GHz has a bus speed of 800MHz. So all being said and done I really cannot see how the G5 1.8GHz can be that mush faster if any then the G5 1.6Ghz.

Or am I just getting my numbers mixed-up.

 

[jeremy.king]

According to this, the 1.8 is 900mhz (half the clock speed)

http://www.everymac.com/systems/apple/powermac_g5/stats/powermac_g5_1.8.html

Revision 2 is 600 - weird

http://www.everymac.com/systems/apple/powermac_g5/stats/powermac_g5_1.8_2.html

But the 1.6 uses slower memory.

 

[waynepixel]

I now. So why the hell would they put the first 1.8GHz at 900MHz and now spec the latest 1.8GHz at 600MHz.

More to the point. The G5 1.6GHz seems as fast as the G5 1.8GHz.

This is starting to get just a little confusing. WHAT!

 

[veedubdrew]

Perhaps it's something to do with the fact that the original single 1.8 was US$2499 when it debuted and the new (slower bus) 1.8 is $1499. I'd say a $1000 discount is a nice exhange for a 200MHz slower bus. It wasn't long ago that we were stuck at a 167MHz top end shared FSB, and for that matter my current-rev PB G4 is at a sloth-like 167, so quibbling over 600 or 800 MHz FSB makes me chuckle a bit.

Especially with the edu/gov't discount and with a combo drive, the single 1.8 is a great deal ($1299) no matter the FSB speed.

-Drew

I now. So why the hell would they put the first 1.8GHz at 900MHz and now spec the latest 1.8GHz at 600MHz.

More to the point. The G5 1.6GHz seems as fast as the G5 1.8GHz.

This is starting to get just a little confusing. WHAT!

 

[Dr. No]

Here are some Macworld tests...

http://www.macworld.com/2004/11/reviews/singlecpupowermac/index.php

It looks like the new single 1.8 is close to the old single 1.8. Both lag behind the dual 1.8, however.

 

[Makosuke]

Like veedubdrew said, it's just a cost-cutting measure. Apple has figured out ways to shave a few bucks off the cost of production, and since the 1.8 is now the low end, it gets the less-impressive components. Besides, even a 3:1 bus ratio isn't exactly bad.

And I agree that the current single 1.8 is a steal if you don't need the PCI-X slots and 8GB RAM limit.

 

[Peyote]

Like veedubdrew said, it's just a cost-cutting measure. Apple has figured out ways to shave a few bucks off the cost of production, and since the 1.8 is now the low end, it gets the less-impressive components. Besides, even a 3:1 bus ratio isn't exactly bad.

And I agree that the current single 1.8 is a steal if you don't need the PCI-X slots and 8GB RAM limit.

I seriously doubt that underclocking the bus speed saves Apple money.

 

[Sun Baked]

The new SP PM 1.8 uses the consumer chipset (from the iMac G5) which is a 3:1 multiplier.

You have to remember it's a headless iMac G5 and not really a PowerMac (just on the outside.)

The rest of the PowerMacs ran the XServe/PowerMac chipset which is a 2:1 bus multiplier.

 

[Dr. No]

The PowerMac SP 1.8 has some things the iMac doesn't have- including PCI slots, an upgradeable video card, gigabit ethernet, and FW 800.

 

[Sun Baked]

The PowerMac SP 1.8 has some things the iMac doesn't have- including PCI slots, an upgradeable video card, gigabit ethernet, and FW 800.

Yet it is a iMac G5 at it's core, no getting around it -- same chipset -- so it shares quite a few of the iMac G5's shortcomings: slower FSB, single CPU operation, slower HT bus, soldered CPU, etc.

 

[Makosuke]

I seriously doubt that underclocking the bus speed saves Apple money.

As SunBaked said, it's not an underclocked version of the same chipset in the high-end G5s, it's a different chipset with a lower bus speed.

Keep in mind that when the G5s first shipped, there was no consumer G5 machine from which to borrow an inexpensive design. The iMac G5 was developed later, and no doubt gave Apple a platform from which to borrow a low-cost motherboard architecture/chipset. You still get quite a bit more with the tower version, of course, but at its core it has more in common with the iMac than the high end towers.

Think of it like the Yikes! G4 towers--they look identical to the Sawtooth towers, but their guts are more like the B&W G3s to save a few bucks. That didn't make them bad machines (they were also a good deal at the time, and I've still got several running quite well), just cheaper at the expense of slightly less impressive components.

 

[waynepixel]

My real question is. How does the G5 1.8 600HMz Bus compeer with the G5 1.6 with the 800MHz Bus, and does the G5 1.6 use the same G5 as the New Imac.

The reason I ask this question is I could buy a G5 1.6 for about £750 UK second hand and I could save myself £250 for a machine which is as fast as the G5 1.8. Right or Wrong.

 

[Sun Baked]

The iMac G5 based machines would be a toss up compared to the SP PowerMac 1.6 -- comes down to benchmarks.

However if the machine was one of the 8 DIMM slot SP PowerMac 1.6 machine -- then it would be worth it, since the G5 seems to be very memory sensitive.

The SP PowerMac 1.6 also contains a PCI-X capable chipset, if you decide to figure out which resistors to add/remove -- to reverse Apple's feature crippling.

 

[oingoboingo]

The iMac G5 based machines would be a toss up compared to the SP PowerMac 1.6 -- comes down to benchmarks.

However if the machine was one of the 8 DIMM slot SP PowerMac 1.6 machine -- then it would be worth it, since the G5 seems to be very memory sensitive.

The SP PowerMac 1.6 also contains a PCI-X capable chipset, if you decide to figure out which resistors to add/remove -- to reverse Apple's feature crippling.

Are the PCI slots in the SP PowerMac 1.6 physically any different to the machines which officially support PCI-X though?

IMHO the PCI-X slot issue is a distraction...I'd be willing to bet that no more than a handful of users on this site have expanded their G5 PowerMacs with a true PCI-X expansion card.

 

[Sun Baked]

Are the PCI slots in the SP PowerMac 1.6 physically any different to the machines which officially support PCI-X though?

IMHO the PCI-X slot issue is a distraction...I'd be willing to bet that no more than a handful of users on this site have expanded their G5 PowerMacs with a true PCI-X expansion card.

No physical difference in the slot, the SP PM 1.6 even has a PCI-X HT Tunnel, it just operates at PCI speed locking out the 100/133MHz PCI-X speeds.

Look up the PCI-X HT Tunnel Apple uses you'll probably find something like this...

If the systemboard supports PCI-X mode operation for a bridge, then a pullup resistor to VDD33 must be placed on the bridge’s PCIXCAP pin. To limit the frequency of a PCI-X-capable bridge to 66 MHz on a systemboard, the systemboard must also include a pulldown resistor from the bridge’s PCIXCAP pin to ground. The strapping options on GNT[4:3]# are used to distinguish between systems that support 100MHz and 133 MHz; in either of these two cases, the system board should include no pulldown resistors on PCIXCAP.

 

[Mechcozmo]

Think of it like the Yikes! G4 towers--they look identical to the Sawtooth towers, but their guts are more like the B&W G3s to save a few bucks. That didn't make them bad machines (they were also a good deal at the time, and I've still got several running quite well), just cheaper at the expense of slightly less impressive components.

They even had the ZIF slot for processors. It was, literally, a B&W G3 but they dropped in a G4 processor into that slot. Some people have dropped G3s in them and they've worked. The Yikes! series were extremely cost-cut-ed. (New word for you all: cost-cut-ed!)

 

[Daveman Deluxe]

FWIW, the Yikes! G4s weren't built with the PMG3 chipset to save a few bucks. They were built that way because at the last second, Apple found a flaw in their chipset design they were planning on using. Since Apple didn't have time to fix it before the ship date, they just put a G4 into a G3 chipset. They called it "Yikes!" as in, "Yikes! This chipset design catches fire on contact with oxygen!"

I don't know if that was the flaw at all though. It just didn't work.

 

[Makosuke]

FWIW, the Yikes! G4s weren't built with the PMG3 chipset to save a few bucks. They were built that way because at the last second, Apple found a flaw in their chipset design they were planning on using.

Are you sure about that? I don't remember hearing rumors about that anywhere at the time, Google didn't give me any reminders, and I thought the Yikes! and Sawtooth (other than the delayed 500MHz version) shipped at the same time.

I'm honestly curious, though, since I follow Apple trivia and the only problems I remember from that time were Motorola's inability to produce fast-enough G4s for the lineup initially announced (though I will note my lab bought 5 of the 400MHz Yikes! machines before they got dropped to 350MHz--that was a heck of a deal).

 

[livingfortoday]

Well, when I asked at the Apple store about it, I was told that the bus speed was cut down because the G5 overheated in such a small enclosure, without the cooling that a standard tower G5 had.

Same reason I was told not to expect a G5 laptop anytime soon.

 

[Sun Baked]

Well, when I asked at the Apple store about it, I was told that the bus speed was cut down because the G5 overheated in such a small enclosure, without the cooling that a standard tower G5 had.

Same reason I was told not to expect a G5 laptop anytime soon.

Reducing the speed of the FSB reduces the power used and heat generated by the Memory Controller, which is a hot little chip itself.

 

[Mechcozmo]

And it makes you want to buy the next level up...

 

[Daveman Deluxe]

Are you sure about that? I don't remember hearing rumors about that anywhere at the time, Google didn't give me any reminders, and I thought the Yikes! and Sawtooth (other than the delayed 500MHz version) shipped at the same time.

I guess I'm only half-right. It wasn't a last-minute flaw, it was just that they didn't have enough time to get the kinks worked out of the AGP chipset design.

Originally, the MPC 7400 chip had been planned to debut in the G4 AGP model, but Apple was not able to get the new machine ready in time. The G4 PCI had been in the works in case of such an event, and allowed Apple to ship 7400-equipt machines while they worked out the final bugs of the G4 AGP.