Mac Pro: Made for Clovertown
With the Mac Pro, Apple is fully comitted to multicore processing. Previously, Apple offered dual CPU across the lineup only one time, during the dark, desperate years of the Motorola G4. Market pressure forced Apple to go dual CPU back then.
In contrast, the Mac Pro is all dual by choice, and Apple could have easily offered single CPU low and midrange systems without much criticism. Apple now competes with the entire Intel world, and with the Mac Pro has demonstrated that they intend to position themselves at the top of the performance curve. By going to a full quad core lineup, Apple makes way for bumping the high and maybe mid towers to 8 cores in January. Subsequently, Intel will release a multiprocessor version of the Clovertown, as in, >2 CPU systems. Alternatively, Intel is rumored to be releasing Harpertown in 2007, an 8 core CPU. Either way, it is likely that within the year, we will be able to buy 8 core Mac Pros, and by the end of 2007, 16 core Mac Pros. Apple's multicore express has left the train station at full speed.
So how does Apple market such a lineup? Clovertown is rumored to be clocked a bit slower than Woodcrest, and it will use a 1066 MHz FSB compared to Woodcrest's 1333 MHz FSB. On some tasks, a quad core Woodcrest Mac Pro will likely outperform an octo core Clovertown Mac Pro. Does Apple just revert to the Hydra campaign of "two heads are better than one," and just update the slogan repeatedly? "Sixteen heads are better than 8"?
As I understand it, many tasks can not be multithreaded, so if the tradeoff is MHz for cores, it's going to be a confusing time for performance.
In contrast, the Mac Pro is all dual by choice, and Apple could have easily offered single CPU low and midrange systems without much criticism. Apple now competes with the entire Intel world, and with the Mac Pro has demonstrated that they intend to position themselves at the top of the performance curve. By going to a full quad core lineup, Apple makes way for bumping the high and maybe mid towers to 8 cores in January. Subsequently, Intel will release a multiprocessor version of the Clovertown, as in, >2 CPU systems. Alternatively, Intel is rumored to be releasing Harpertown in 2007, an 8 core CPU. Either way, it is likely that within the year, we will be able to buy 8 core Mac Pros, and by the end of 2007, 16 core Mac Pros. Apple's multicore express has left the train station at full speed.
So how does Apple market such a lineup? Clovertown is rumored to be clocked a bit slower than Woodcrest, and it will use a 1066 MHz FSB compared to Woodcrest's 1333 MHz FSB. On some tasks, a quad core Woodcrest Mac Pro will likely outperform an octo core Clovertown Mac Pro. Does Apple just revert to the Hydra campaign of "two heads are better than one," and just update the slogan repeatedly? "Sixteen heads are better than 8"?
As I understand it, many tasks can not be multithreaded, so if the tradeoff is MHz for cores, it's going to be a confusing time for performance.
Comments
So how does Apple market such a lineup? Clovertown is rumored to be clocked a bit slower than Woodcrest, and it will use a 1066 MHz FSB compared to Woodcrest's 1333 MHz FSB. On some tasks, a quad core Woodcrest Mac Pro will likely outperform an octo core Clovertown Mac Pro. Does Apple just revert to the Hydra campaign of "two heads are better than one," and just update the slogan repeatedly? "Sixteen heads are better than 8"?
If Intel goes to an 8 core they better not still using fsb. Fsb does not scale well to more then 2 cpus and the upcoming quad-core are just 2 duel core cpus on the same
die.
If Intel goes to an 8 core they better not still using fsb. Fsb does not scale well to more then 2 cpus and the upcoming quad-core are just 2 duel core cpus on the same
die.
It'll scale to 4 CPUs. And remember that Woodcrest's cores communicate directly, without hitting the FSB. So only 2 CPUs. Clovertown better have a 1333MHz FSB (which I think it certainly will), or there will be some FSB saturation, with 2 CPUs on each die, and 2 dies.
im with placebo, aside from Apple apps,which one can you see taking full advantage of all those cores?
None, you can just run more and more high end Multi-CPU supporting apps. Why does everyone overlook this so frequently?
Sorry, lost myself there.
You got me. I'm of the opinion that at the very least, the poorly designed apps will each take over a core as a single process, and leave available cores for other work. Like, say, the OS? My compiles would fricking *FLY* on a quad core - it'd just start spitting the sub-builds off onto the other cores, and... gaaahhhhh...
Sorry, lost myself there.
You lost 6000 or so posts...
You lost 6000 or so posts...
I think they rolled over. 32-bit integers only go so high.
( For those who dont live in Texas, for that matter USA, in Texas everything bigger is better, the more the greater, no matter how sh..ty it tastes!
gaaahhhhh...
Sorry, lost myself there.
Kill a Black Beller and you will feel better immediately
I mean the bus is going to be saturated, higher latency, poor support for quad core, i mean all this talk about 8 cores running on a poor distributed bus, and bad app support? is this the next mhz myth? the more cores the faster the machine?
that is where AMD is way ahead of intel.
I hope amd K8L can take back the lead in the other areas.
I think they rolled over. 32-bit integers only go so high.
I don't think that he lost 4 billion posts. This isn't a roll-over, or he would have lost 4096 (12 bit unsigned, 13 bit signed) or 8192 posts (+1 bit), not 6000.
I don't think that he lost 4 billion posts. This isn't a roll-over, or he would have lost 4096 (12 bit unsigned, 13 bit signed) or 8192 posts (+1 bit), not 6000.
Yeah, I was joking
Clovertown is rumored to be clocked a bit slower than Woodcrest, and it will use a 1066 MHz FSB compared to Woodcrest's 1333 MHz FSB. On some tasks, a quad core Woodcrest Mac Pro will likely outperform an octo core Clovertown Mac Pro.
As i understand, Clovertown is the Dual Socket Capable version of Kentsfield ? which is 2 Woodrests on 1 chip, ie: total 4 cores & 8MB cache...
What possibilty is there of socket/mobo/chipset compatibility here?
ie: Could 2 x Clovertowns be dropped into a Mac Pro mobo? ...theoretically? hypothetically?
(ie: just as Merom has been dropped into the Yonah socket on the solo mini successfully...)
am i dreaming, or is this a possibility?
You lost 6000 or so posts...
6993, actually. Blame a certain mod for being puckish.
What possibilty is there of socket/mobo/chipset compatibility here? ie: Could 2 x Clovertowns be dropped into a Mac Pro mobo? ...theoretically? hypothetically?
(ie: just as Merom has been dropped into the Yonah socket on the solo mini successfully...)
Both Kentsfield and Clovertown should be drop-in upgrades to the respective Conroe and Woodcrest boards, given that the board can support the power, voltages, etc.
Unless you running a process that can take advantage of all of the cores, there's really not much point to upgrading. Then again, you shouldn't be getting a Mac Pro unless you can take advantage of all 4 cores...
6993, actually. Blame a certain mod for being puckish.
SHUUURRREEEE, NOOB!
In the past, I've suggested that any new high-level software language should be based loosely on VHDL (not Ada). We've kind of foolishly been using C and C-like languages for 30+ years. C++ came around once software projects were starting to get really big. Now things are changing again, and not C nor any of its cousins are going to cut it.
Naw....