Is GHz not uniformly scaled across CPUs? 5 year old CPU vs. new

[ninaholic37]

Different architectures can execute more instructions at the same clock:

I was curious what this was actually measuring, so I found the original article:

http://www.tomshardware.com/reviews/processor-architecture-benchmark,2974.html

Looks like the Pentium 4 does twice as bad at everything there at 3Ghz (which I guess makes sense since it's also the oldest). It would have been interesting to see how well a Pentium M at 2.0Ghz does in comparison to the Pentium 4 at 3.0Ghz.

 

[DrMrLordX]

AM2 / AM3 had a setting for the dual-channel memory controller, where both 64-bit channels could act together as a single 128-bit channel (ganged), or as two independent 64-bit channels (unganged). I had always heard that unganged was best for quad-cores.

Ah yes, ganged vs. unganged. I believe this setting is also available on FM2+ systems (I'd have to double-check, but I think it's in there). It really depends on the app, but yeah, lots of people ran unganged back in the day. If you have single-threaded apps then ganged is supposed to be faster.

I think APUs need ganged mode for the benefit of the iGPU.

 

[Denithor]

Here's the article you want to go read.

http://www.anandtech.com/show/2045

This is the launch for the Core 2 Duo/Quad lineup that marked the drastic shift away from Pentium 4 architecture back to a focus on performance/watt. After Pentium 3, the paths diverged into desktop chips (moar gigahertz!) and mobile chips (lower power, higher IPC). Intel switched gears to the more efficient pathway to fight back against AMD's A64/X2 chips that were crushing P4 outright.

Go have a read...