PowerMacs updated in late January
I've been checking out Ingram Micro inventory and while they were mostly at nil for PowerMac inventory last week, they quickly restocked and are in the low hundreds on each of the models. I have heard rumors that Apple may have 970 chips in their hands by March even if a limited amount. We also are to expect the 7457 from Motorola very soon. We can all agree that PowerMacs need an update and late this month or early next is when we should expect them. I predict a staggered announcement like when the DA PowerMacs were first released: 2 models based on the G4 (7457) shipping immediately and two models based on the 970 shipping in a month for the lower speed and maybe 6 weeks for the higher end. If the 970 is not to be in the cards for this quarter then a full lineup of 7457s should hold us over with 200MHz buses, but I really hope for 970 announcements. It would convince some of my higher ups to wait on the purchase of some new Macs we need to make in the spring. My predicted specs are as follows:
All G4 line up:
1GHz SP, 1.25GHz SP, 1.3GHz DP, 1.5GHz DP
2MB L3 cache across the board
167MHz bus for the SP machines
200MHz bus for the DP machines
DDR RAM; 256MB, 512MB, 512MB, 1GB respectively
ATA133; 60GB; 80GB; 120GB, 160GB respectively
Firewire 400 & 800
USB 1.1
AGP 8X, PCI-X (!)
Bluetooth and the new Airport slots
Staggered G4/970 line up:
1.2GHz SP G4, 1.4GHz SP G4; 2MB L3 cache; 200MHz bus
1.6GHz DP 970, 1.8GHz DP 970; 800MHz and 900MHz bus
DDR RAM; 256MB, 512MB, 512MB, 1GB respectively
ATA133; 60GB; 80GB; 120GB, 160GB respectively
Firewire 400 & 800
USB 1.1
AGP 8X, PCI-X (!)
Bluetooth and the new Airport slots
All G4 line up:
1GHz SP, 1.25GHz SP, 1.3GHz DP, 1.5GHz DP
2MB L3 cache across the board
167MHz bus for the SP machines
200MHz bus for the DP machines
DDR RAM; 256MB, 512MB, 512MB, 1GB respectively
ATA133; 60GB; 80GB; 120GB, 160GB respectively
Firewire 400 & 800
USB 1.1
AGP 8X, PCI-X (!)
Bluetooth and the new Airport slots
Staggered G4/970 line up:
1.2GHz SP G4, 1.4GHz SP G4; 2MB L3 cache; 200MHz bus
1.6GHz DP 970, 1.8GHz DP 970; 800MHz and 900MHz bus
DDR RAM; 256MB, 512MB, 512MB, 1GB respectively
ATA133; 60GB; 80GB; 120GB, 160GB respectively
Firewire 400 & 800
USB 1.1
AGP 8X, PCI-X (!)
Bluetooth and the new Airport slots
Comments
i dont think the 970 will be ready that soon.
actually i dont think it will be available at MWNY
2003 either,but hopefully im wrong.
i think the 970 chip will be for the very high end.
i still think apple will used the 7457 chips and then the 7457 RM chips in mainstream macs.
just my opinion.
<strong>i agree.
i dont think the 970 will be ready that soon.
actually i dont think it will be available at MWNY
2003 either,but hopefully im wrong.
i think the 970 chip will be for the very high end.
i still think apple will used the 7457 chips and then the 7457 RM chips in mainstream macs.
just my opinion.
I'm not sure why you insist this chip will be for the very high end. They're supposedly cheap to make, they run cool, will be quick to scale...and given that the G4 lags behind its Pentium counterparts, I think Apple will get them into as many machines as they can as quickly as possible.
I see the 970 being introduced at 1.2 Ghz - not 1.6 - as the yields will be better...Even there it would mop the floor with the current offerings...Hell, perhaps even a 1 Ghz 970 might be the ticket for getting to market quickly.
I also question whether the powermacs will mix the 7457 and the 970. Wouldn't this entail considerable expense designing a compatible mobo?
We can be pretty certain from the official and semi-official announcements that the 970 will start production in 6 months. But what about the 7457? Motorola hasn't said crap. If they surprise us by turn out a 1.6 plus ghz part in the next 2 months it might really change Apple's time table. I'm not saying that Apple won't quickly adopt the 970... I'm saying that what Motorola actually produces might change HOW quickly and in what manner Apple moves to the 970. For example, if the 7457 can hold down the fort on the low end, then Apple might wait an extra couple of months for yields of faster 970s to come up. On the other hand, the 7457 might be a boondoggle, and we could see the 970 introduced as quickly as possible. In that case, Apple might get more mileage out of low yields by moving, for one generation, back to single processors.
[ 01-08-2003: Message edited by: Bacon ]</p>
<strong>I doubt that we will see a 200 mhz front bus, PC 3200 memory are not officially supported in the PC world due to their lack of stability. Otherwise the difference of performance is not huge comparing to PC 2700 memory, especially if you take law latency PC 2700 (CL2 one).</strong><hr></blockquote>
Unless Apple uses an asynchronous bus. 200Mhz bus with 333MHz DDR RAM. It's unknown if the present chipset supports that feature.
<strong>A few things:
I see the 970 being introduced at 1.2 Ghz - not 1.6. First off... yields will be better...Even there it would mop the floor with the current offerings...Hell, perhaps even a 1 Ghz 970 might be the ticket for getting to market quickly.
I also question whether the powermacs will mix the 7457 and the 970. Wouldn't this entail considerable expense maintaining?
We can be pretty certain from the official and semi-official announcements that the 970 will start production in 6 months. But what about the 7457? Motorola hasn't said crap. If they surprise us by turn out a 1.6 plus ghz part in the next 2 months it might change really change Apple's time table. I'm not saying that Apple won't quickly adopt the 970... I'm saying that what Motorola actually produces might change HOW quickly and in what manner Apple moves to the 970. For example, if the 7457 can hold down the fort on the low end, then Apple might wait an extra couple of months for yields of faster 970s to come up. On the other hand, the 7457 might be a boondoggle, and we could see the 970 introduced as quickly as possible. In that case, Apple might get more mileage out of low yields by moving, for one generation, back to single processors.</strong><hr></blockquote>
Really? I would imagine a 16 stage pipeline 130nm processor like the 970 would at least start at 1.6GHz and if it can't reach 1.8 on it's first run, then well lets just say I don't think IBM's fabs are that pathetic. 1.2GHz was quoted only to show the low power version, its ability to be used in a portable.
<strong>
Unless Apple uses an asynchronous bus. 200Mhz bus with 333MHz DDR RAM. It's unknown if the present chipset supports that feature.</strong><hr></blockquote>
What's DDR-II cloecked at actually? Doesn't it support 400 mhz as well (low end)?
My guess is that we wil see one more speed bump of the G4 towers. Getting single CPUs back is very very unlikely. The new lowend shoud be at least as fast as the current low end. So to replace a dual 867. A 1.5 GHz would be needed. Way faster in nonSMP like most games and stuff and about on par with the old one for ripping. A dual 1.5 as a high end is OK but then what about the midrange? A dual 1.25? Why pay 1000 dollars premium for a box that is slower than the budget one at many application
<img src="graemlins/hmmm.gif" border="0" alt="[Hmmm]" />
My hope is a dual ranging from a resonably1.2-1.35-1.5 GHz to wildy optimistic 1.4-1.6-1.8 GHz. This followed by 970 in the second trimester of 2003. <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
<strong>
Really? I would imagine a 16 stage pipeline...</strong><hr></blockquote>
I thought that depending on the operation the number of stages in the 970 could vary from 9 to 20 stages??
I think all we will see from Apple in the PowerMac lineup this year is speed hikes and maybe some new motherboard gear (Firewire 800 USB 2.0 etc...) I think this is strongly indicated by Steve's 'This is the year of the laptop' blurb during his keynote. Also the fact that IBM's Power chips aren't going to be production ready in volume until Q3 at the earliest. I predict that Macworld next January will be PowerMac rebirth time...
Whad'ya think..? :cool:
Apple know these will sell like hot cakes, and therefore they cant let customers wait several months as with the 17" iMac, when it was released. Furtheron, they can NOT get through a disaster as when the 500 Mhz G4 was first introduced.
<strong>
I thought that depending on the operation the number of stages in the 970 could vary from 9 to 20 stages??</strong><hr></blockquote>
Well depends on the unit and how complex the intruction is. I think 16 stages is the maximum steps in the integer pipeline. It may take as little as 9 or max out at 16 depending on the instruction.
I don't think a 970 mac will be ready until some late announcement perhaps in december or november.
<strong>In my opinion the 970 will be part of a completely new product - the chip is far different than the G4, much faster and 64 bits. Any anouncement of the 970 will be coupled with an announcement of the 64 bit version of OS X, which takes more work than just a simple recompile.</strong><hr></blockquote>
Specific mention is made in online discussion of the 970, that it is able to run 32-bit OS's with only minor modifications, in 32-bit mode. It seems to me that if Apple wanted to get to market faster, they could well ship with a non-64-bit version of OS X very quickly. It may well be that 64-bit Darwin and OS X are actually far along - but if they are not, there is a logical fallback plan.
G-news
I've always wondered what Apple's plans were with regards to 64-bit.
When IBM and Motorola were the first companies to "break off" from the MIPS
standardization efforts of the late '80s, and created the new PowerPC, they made
it compatible with their Power ISA (instruction set architecture). A few years
later, they attempted to create the first 64-bit PowerPC, only to suffer a total
design failure. The result is that IBM created a 64-bit Power4 series chip with
Power ISA only, which wasn't compatible with 32-bit PowerPC.
But now it looks like IBM is ready to announce a new Power4 chip that introduces
full 32-bit PowerPC compatibility at Microprocessor Forum next week, according
to this EETimes article. And contrary to previous rumors that such a chip would
be compromised of two separate cores, the design is supposed to be unified. In
essence, the chip should be completely compatible enough to boot 32-bit PowerPC
OSes with no issue -- opening the way for Apple to adopt it in quantity and keep
costs commodity.
The move is a good one for IBM too, because Apple commodity equates to RS/6000
commodity, and improves the economies of scale by a couple orders of magnitude.
About the only "analysis" in the article I differ with is this one:
"The chips could be used in Apple's new line of Xserve servers even
in a 32-bit mode. However, Apple would have to heavily rework its
Mac OS, which has just gone through a major release cycle, to
support 64-bit addressing."
MacOS X is "Darwin," the API compatible BSD release based on the FreeBSD kernel
and core. FreeBSD runs on the Alpha, and is fairly "64-bit clean" already --
even when Apple original began their proprietary creation. You only run into
such porting issues when the codebase wasn't designed for a larger bit width --
which is clearly not the case with FreeBSD, an codebase that only increases the
ease of moving to 64-bit. That's just poor application of "Microsoft logic" by
the EETimes author -- especially with regards to a company that _has_ already
ported its OS from one architecture (68000) to another (PowerPC) before.
It would sure be a heck of a lot easier than moving to the rumored AMD's x86-64,
one of the most complex CISC designs, and far more ABI compatible with existing
software. Besides, PowerPC is RISC and far more simple. Yes, the x86-64 core
itself is actually RISC inside (NexGen RISC86 ISA with Intel CISC x86
microcoding), but it still has to support all the microcode for x86 ISA backward
compatibility**.
EETimes Article: "IBM processor hints at Apple's 64-bit future"
<a href="http://www.eet.com/semi/news/OEG20021009S0020"; target="_blank">http://www.eet.com/semi/news/OEG20021009S0020</a>;
[ **NOTE: Even more humorous is the fact, there are many articles on the issues
with the Intel P4 being less 386/486 ISA compatible than the Athlon. It seems
even Intel is not making its own 32-bit chips 100% backward compatible because
of all the every-increasing number of instructions to support in doing so. ]
--
Bryan J. Smith, E.I. Contact Info: <a href="http://thebs.org"; target="_blank">http://thebs.org</a>;
A+/i-Net+/Linux+/Network+/Server+ CCNA CIWA CNA SCSA/SCWSE/SCNA
I think this sums it up best. 64-bit is within reach. I think hardware is a bigger block right now that the OS or applications.
I think if we go here:
<a href="http://www.apple.com/contact/feedback.html"; target="_blank">http://www.apple.com/contact/feedback.html</a>;
and constantly complain that our 3000 is not getting us the highest end speeds and give them links to how we get blown out in benchmark test, we will eventually get what we need. namely a chip that can scale and satisfy our needs. or atleast a announcement