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.
Quad PowerMacs and the 970MP?
- Thread starter MacRumors
- Start date
- Sort by reaction score
wdlove said:
Actually I love that design. It denotes strength and power. Every thing doesn't have to always look so artsy fartsy
renaming??
but currently rumors are spreading that GX could be a dualcore...
in IBM history GX was synonym of higher lock speed with lower power consumption (aka 750GX which never been used by Apple)Yonizzle said:
but currently rumors are spreading that GX could be a dualcore...
Hmmmmmm, all I can say is that till now Apple and IBM has been really quiet about it all... In my opinion, if Apple releases PM before WWDC I believe we will get the GX version of the G5, BUT if they will hold till WWDC I think we might see something bigger such as a possibility of a Power 5 derivative processor, I mean no one really knows if the so called 970MP is a Power4 or Power5 derivative (maybe its not a 970 afterall), anyway, I am more excited about the Power5 derivative of the Apple processor then 970whatever simply because Apple and IBM has been working on it all at the same time from scratch which would suggest to me that we might see such things as an Altivec2 or some kind of variation of Cell technology...
SMP potential....
Precisely. The MIPS ISA was fairly SMP neutral. Certainly one of the most important factors in true SMP scaling has to do with memory access conflicts and management thereof. In our particular architecture, that was a bit of a problem. However, anybody who believes an SMP will actually reach 4x of a uniprocessor in a 4 way SMP has not worked with systems in that arena. That said, and to reiterate something about SGI...in the NUMA systems they had, the access to elements like CPUs, memory etc was very interesting. They had a could of SMP systems essentially networked with the most adjacent elements performing best with the bits of the system with the least latency (e.g. "local). It's also worth noting that SGI used MIPS chips for these systems. How does this relate to a potential Mac based on dual-core 970MPs? That's speculation, but I could see a real benefit to having dual-core chips optimizing for latency by attempting to "pin" threads to processors and optimizing cache usage. As to the benefits of a quad processor, it would be huge even for non-threaded apps....such a system could have four single threaded apps running simultaneously. Mac OS X can handle that today.
ddtlm said:
Precisely. The MIPS ISA was fairly SMP neutral. Certainly one of the most important factors in true SMP scaling has to do with memory access conflicts and management thereof. In our particular architecture, that was a bit of a problem. However, anybody who believes an SMP will actually reach 4x of a uniprocessor in a 4 way SMP has not worked with systems in that arena. That said, and to reiterate something about SGI...in the NUMA systems they had, the access to elements like CPUs, memory etc was very interesting. They had a could of SMP systems essentially networked with the most adjacent elements performing best with the bits of the system with the least latency (e.g. "local). It's also worth noting that SGI used MIPS chips for these systems. How does this relate to a potential Mac based on dual-core 970MPs? That's speculation, but I could see a real benefit to having dual-core chips optimizing for latency by attempting to "pin" threads to processors and optimizing cache usage. As to the benefits of a quad processor, it would be huge even for non-threaded apps....such a system could have four single threaded apps running simultaneously. Mac OS X can handle that today.
jcrowe said:Precisely. The MIPS ISA was fairly SMP neutral. Certainly one of the most important factors in true SMP scaling has to do with memory access conflicts and management thereof. In our particular architecture, that was a bit of a problem. However, anybody who believes an SMP will actually reach 4x of a uniprocessor in a 4 way SMP has not worked with systems in that arena. That said, and to reiterate something about SGI...in the NUMA systems they had, the access to elements like CPUs, memory etc was very interesting. They had a could of SMP systems essentially networked with the most adjacent elements performing best with the bits of the system with the least latency (e.g. "local). It's also worth noting that SGI used MIPS chips for these systems. How does this relate to a potential Mac based on dual-core 970MPs? That's speculation, but I could see a real benefit to having dual-core chips optimizing for latency by attempting to "pin" threads to processors and optimizing cache usage. As to the benefits of a quad processor, it would be huge even for non-threaded apps....such a system could have four single threaded apps running simultaneously. Mac OS X can handle that today.
SGI is abandoning MIPS in favor of Itaniums. I expect SGI will drop the Origin line within 3 years. They have implemented NUMA LINK/Flex on all of their new Itanium Systems with outstanding results.
We have 68 Origin 3400s that we would like to replace. Wish we had the cash for the Altix
daveL said:
I wouldn't call it a bad decision the Itanium is much more efficient at floating point operations than the Power4 and Power5. It is also much less expensive.
Take a look at these efficiencies achieved with Kazushige Goto's BLAS implementation.
Itanium dgemm 99%
Power 4 dgemm 74.6%
PPC970 running OS X dgemm 83.3%
PPC970 running Linux dgemm 88.2%
http://www.cs.utexas.edu/users/kgoto/signup_first.html
ddtlm said:
Exactly.
Just look at the top 500 list. SGI has the number 2 system. It took IBM 3 times as many processors to beat it.
The efficiency of SGI 85%
IBM Bluegene 77%
Yes, I do. I don't see it as a contender, and it may very well die, at some point. It's great that it has superior floating point, but that quality, in and of itself, doesn't make it a viable processor in the marketplace. Even if it does survive, it may lag in terms of the amount of attention it gets from HP and Intel. IBM has withdrawn from it altogether, which tells you they see no customer demand for it. Even HP has backed off a bit and put more energy into x86/64 offerings. I guess only time will tell, although I'm sure I'll now hear how stupid I am from a dozen different folks.ddtlm said:
This is great, but the artist forgot to add 2 more HD bays, 2nd optical drive
and at least one PCI-e expansion slot.
...well sorrrrry...
daveL:
http://www.theregister.co.uk/2004/12/06/intel_updates_pricelist/
It's gona come down to price vs performance, and then the market will decide.
Since when does superior performance mean nothing? Intel has plenty of money to see this through, and once Itanium and Xeon share the same socket (2007 I think) anyone with Linux servers will find it an easy switch to make. You can already buy an Itanium processor for well under $1k, the problem is getting something to put it in.
http://www.theregister.co.uk/2004/12/06/intel_updates_pricelist/
It's gona come down to price vs performance, and then the market will decide.
daveL said:
Well, I'm not going to call you stupid, I am however, going to point out that Itanium server sales more than tripled in 2004 while purchases of Power and Sparc both fell.
"So far, Itanium machines have made a modest dent in the server market. Server customers purchased $1.6 billion worth of Itanium machines in 2004, up from $448 million in 2003, Reynolds said."
"By comparison, x86 server purchases grew from $21.1 billion to $24.6 billion between 2003 and 2004. Sparc system purchases shrank slightly from $6.9 billion to $6.7 billion, while Power purchases shrank from $7.2 billion to $7.1 billion, Reynolds said."
Doesn't that give us a good idea of where the demand is heading?
daveL said:
Maybe they read the rumor sites and said..... Oh know these guys found the new stuff in our code.
On no our competitors will have an advantage against us now... LOL
~loserman~:
Yeah sometimes you have to wonder how this works in Apple's collective head. What are they afraid of in a case like this? People who see this rumor and delay a purchase hoping to get a 4-processor Mac where either going to buy something expensive anyway (no loss), or are simply being drawn upmarket (Apple's benefit).
Yeah sometimes you have to wonder how this works in Apple's collective head. What are they afraid of in a case like this? People who see this rumor and delay a purchase hoping to get a 4-processor Mac where either going to buy something expensive anyway (no loss), or are simply being drawn upmarket (Apple's benefit).
I, on the other hand, am going to call you stupid. No, no it doesn't. It says Itanium is growing, certainly, but a 1.5% contraction is well within any sort of normal variance.JCheng said:
~J
Very curious they pulled the software....
Although it doesn't really help or harm the rumor mill - its just showing that Apple knows they slipped up somehow.
Now if they put up another new version that has just 2 cpu options that might give reason to think that we won't see the quads anytime soon.
D
Although it doesn't really help or harm the rumor mill - its just showing that Apple knows they slipped up somehow.
Now if they put up another new version that has just 2 cpu options that might give reason to think that we won't see the quads anytime soon.
D