Separate names with a comma.
Discussion in 'Mac Basics and Help' started by ppc_michael, May 15, 2005.
Stupid question: is "dual core" the same as "dual processor" or am I sadly mistaken?
Dual Core is two processors on one physical chip.
Dual Processor is two individual processor chips, each chip has its own cache memory and controllers
Theoretically, you ccould have a dual-core-dual-processor machine which would have 4 processors in all
Okay. Thank you for clearing that up.
So does the Powermac G5 have four total cores (two per processor chip) or are the processors both "single" core?
The PowerMac Dual G5 has two single core chips. Apple has not released a dualcore Macintosh yet. Intel and AMD have demonstrated them but I haven;t kept up with whether they have shipped any commercially yet.
Normally dual-core chips also have separate caches for each chip. And most chips (eg. the G5) don't have integrated memory controllers.
Correct. Only AMD's dual core chips have independent memory controllers. Almost all dual core chips retain individual caches. What also varies is how the two processors communicate. Intel is considered by some to be "cheating" and using the front side bus for communication, which is noticeably slower than AMD's method which interlinks the two CPUs directly on a much faster connection than an FSB solution.
True, but each processor in the Dual G5 has a dedicated buss and buss controller (just not on the chip). So a dual processor machine is generally more expensive, runs hotter, and is potentially faster than an imagined dual-core machine of the same specification - depending how much is shared between the cores and whether that presents a bottleneck.
How much of the support circuitry (other than the CPU core) of the dual-core chip is shared between the processors, or duplicated, is an architectural decision made by the chip designers. AMD's design has 2 L2 caches onboard, but one shared memory controller. IBM's has two L2 cache onboard and no onboard L3 cache controller.
Each processor has its own FSB link, but no dedicated controller. They also connect directly into the same system controller, and the system bus is shared. Nothing else in the computer can feed the huge bandwidth offered by the independent FSBs, though it allows for ultrafast movement of data already in the processor(s). Getting the data from RAM into the CPU moves across a single bus, through the single system controller, and then at that point speeds up. A dual-core arrangement would probably keep this setup.