How can we count quad core Processor speed ?
- Thread starter krzysiek.uk
- Start date
name='llwyd' said:
in anything.
I thought the point of a dual/quad core cpu was basically fitting multiple cpu's into one?
so there shouldnt be much difference between 4 seperate processors and a quad core (same frequency, cache etc).
I used to run a dual celeron setup, and certain encoding programs benefitted 100% from having another CPU to work with, which you could then compare to 2xMHz single CPU.
Obviously not all apps will use multiple cpus, although the mains apps I use which are hard on the cpu (photoshop for instance) benefit from 2 cores alot.
Rastalovich
New member
What it boils down to is the 1xCPU being able to do a task u give it, whether it be multiple cores or just the 1.
IF the modern cpus behaved in a way that the OS would see it as 1 cpu, but internally the cpu split the commands between it`s cores to get the job done, u could effectively rate it as a single entity of it`s combination of cores, even tho it`s speed is still the clock rating.
Unfortunately, or fortunately depending on how u favor core theory, the OS (any) isn`t as advanced as that, and programmers (not all of them, but unfortunately those in key positions in the computer world) are lazy and/or ignorant of how to use these cores effectively to a required level.
It is wholey possible that u could give a job for a single core... a theoretical S6600 a job and a quad core Q6600 the same, and the quad could do it quicker - not 4 times quicker, but for arguements sake 4x(0.66) times quicker - eg fold a work unit.
Whichever way u looked at it, both CPUs are running at the same speed, say stock clock 2.4g.
However, if u ran the S6600 @ 4x2.4g (seeing as we`re using our imagination) it would complete whatever task u gave it quicker than a Q6600 @ 2.4g. A SuperPI score would probably be in excess of 4x quicker, depending on the architecture of this single core cpu.
To compare the clocks in terms of the Quad being a multiple of the Single just doesn`t work tho. For example, a Tricore Phenom ? (rated at 2.4ghz? dunno) and a Phenom quad that ran 2.4ghz - the quad Phenom is not 4x(1/3) faster.
It`s all down to the architectures of the cpus really, and the dumbass software houses that haven`t been issued a manual on how to use them.
IF the modern cpus behaved in a way that the OS would see it as 1 cpu, but internally the cpu split the commands between it`s cores to get the job done, u could effectively rate it as a single entity of it`s combination of cores, even tho it`s speed is still the clock rating.
Unfortunately, or fortunately depending on how u favor core theory, the OS (any) isn`t as advanced as that, and programmers (not all of them, but unfortunately those in key positions in the computer world) are lazy and/or ignorant of how to use these cores effectively to a required level.
It is wholey possible that u could give a job for a single core... a theoretical S6600 a job and a quad core Q6600 the same, and the quad could do it quicker - not 4 times quicker, but for arguements sake 4x(0.66) times quicker - eg fold a work unit.
Whichever way u looked at it, both CPUs are running at the same speed, say stock clock 2.4g.
However, if u ran the S6600 @ 4x2.4g (seeing as we`re using our imagination) it would complete whatever task u gave it quicker than a Q6600 @ 2.4g. A SuperPI score would probably be in excess of 4x quicker, depending on the architecture of this single core cpu.
To compare the clocks in terms of the Quad being a multiple of the Single just doesn`t work tho. For example, a Tricore Phenom ? (rated at 2.4ghz? dunno) and a Phenom quad that ran 2.4ghz - the quad Phenom is not 4x(1/3) faster.
It`s all down to the architectures of the cpus really, and the dumbass software houses that haven`t been issued a manual on how to use them.
name='AdamR79' said:
No not really. As Rast has just hinted at, GHz does not = processing power. Thats how the athlon rating system worked, with 2.3 GHz athlons out performing 4 GHz P4s. It comes down to architecture. If you think of it as 4 1/4 CPUs, individually they are nowhere near as effective as a whole single core CPU equal to the speed of the whole processor. But combined they effectively increase the processing bandwidth (NOT speed), allowing 4 different processes to run simultaneously without cutting each other up on the way through.
