Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

brhmac

macrumors regular
Original poster
Apr 21, 2003
175
0
Planet Earth
Can someone explain the significance of bus speed when comparing computers?

For example, on consumer machines from Apple and Dell.

Apple iMac/new eMac
Processor: 1.25 GHz G4
Bus speed: 167 MHz

Dell Dimension 2400
Processor: 2.4 GHz Celeron
Bus speed: 400 MHz

We've all been told by Apple not to pay attention to megahertz because of other factors. OK, what else besides megahertz and bus speed is there to consider? :confused:
 
brhmac said:
Can someone explain the significance of bus speed when comparing computers?

For example, on consumer machines from Apple and Dell.

Apple iMac/new eMac
Processor: 1.25 GHz G4
Bus speed: 167 MHz

Dell Dimension 2400
Processor: 2.4 GHz Celeron
Bus speed: 400 MHz

We've all been told by Apple not to pay attention to megahertz because of other factors. OK, what else besides megahertz and bus speed is there to consider? :confused:

The bus is how the computer moves data between the processor and memory. It can be a big bottleneck and in all seriousness is the one thing really hampering the G4. The G5 is fantastic the bus runs as fast as the processor. I don't know how much it will make a difference when dealing with a Celeron, or when using applications that aren't overly demanding.
 
Um, ignore the Celeron. It doesn't matter how much bandwidth it has, it is crippled compared to your run-of-the-mill Pentium 4. In that case (with the P4), then yes the bandwidth matters--and it matters with memory performance.

Most notably, games and graphic applications (like Photoshop) depend on memory performance, and thus will benefit from increased bandwidth.

This is the reason the P4C's (800fsb) are considerably stronger than the P4B's (533fsb). This is ALSO the same reason the Opteron and Athlon 64's do so well in Photoshop.

http://www.gamepc.com/labs/view_content.asp?id=athlon64&page=8

The Athlon 64 (ClawHammer) 3200+ @ 2.0ghz clockspeed [on-die memory controller, memory limited by HT @ 1.6ghz], manages 90.6seconds.

The older Athlon XP (Barton) 2800+ @ 2.083ghz clockspeed [slower dual-pumped 200fsb or 400fsb] manges 116.1seconds

While the Barton seems to have nearly a 100mhz advantage, the much faster bus of the Athlon 64 allows it to overcome its clock disadvantage, while also being 10% faster in the same task.
 
bradz_id said:
They only call celeron's 400FSB but they only actually have a 100MHz FSB "Quad Pumped" which is a load of crap

All of Intel's busses are Quad Pumped. It isn't as bad as your saying, but it still isn't as good as the double pumped bus on the PowerMac G5 (400/450/500 double pumped to 800/900/1000).
 
Rincewind42 said:
All of Intel's busses are Quad Pumped. It isn't as bad as your saying, but it still isn't as good as the double pumped bus on the PowerMac G5 (400/450/500 double pumped to 800/900/1000).
In case any of you are wondering just how a "double-pumped" bus transfers data at the speed it does, I have an explanation for you. Double-pumping a bus works just like DDR memory does - it reads and writes data on both the rising and falling edge of each cycle. Non-pumped buses and single data rate memory only uses the rising edge of each cycle to send/receive data. Quad-pumping is similar to double-pumping; an electronic device called a pullup resistor is used to double the effective clock rate, then data is read/written on both edges to achieve a quadruple data rate. However, a double-pumped 200 MHz bus (effective rate 400 MHz) is not as fast as a single-pumped/non-pumped 400 MHz bus (effective rate 400 MHz) because the pumping process has some overhead associated with it.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.