Relationship between clock speed and performance (Same CPU)

[Jebonia]

Is the relationship between clock speed and performance linear when you are talking about the same cpu? i.e. Will an underclocked 4670k running at 2ghz take twice as long to render the same video if it was being rendered with a 4 ghz 4670k ?

[Cacao]

Yes, it's proportional with programs that can take advantage of the extra speed. I have tested and confirmed this in Sony Vegas.

[Guest]

No - yes, higher clock speeds will increase performance however it isn't a linear relationship. Also, clock speeds between CPU's do not mean equal performance. A prime example of this is the new AMD CPU clocked at 5GHZ getting the same FPS in Battlefield 4 as a 3960X which is clocked at 3.5Ghz. 

[nishank93]

Interesting question. I have no evidence for it, but at least theoretically, I'd think this to be true. Also, I think it would depend heavily on the testing methodology you use to compare computational power.

 

EDIT: @Windspeed36, could you elaborate or link to why it isn't linear?

[Cacao]

No - yes, higher clock speeds will increase performance however it isn't a linear relationship. Also, clock speeds between CPU's do not mean equal performance. A prime example of this is the new AMD CPU clocked at 5GHZ getting the same FPS in Battlefield 4 as a 3960X which is clocked at 3.5Ghz. 

From what I understand 2x clock speed = 2x performance for programs that are 100% CPU dependent. Obviously this doesn't apply to gaming and doesn't apply when you're talking about 2 different CPUs. But I think a CPU at 2ghz will perform 2x as well as the same 1ghz CPU in an optimized (CPU only) program. We should test this with cinebench or something.

[Guest]

Interesting question. I have no evidence for it, but at least theoretically, I'd think this to be true. Also, I think it would depend heavily on the testing methodology you use to compare computational power.

 

EDIT: @Windspeed36, could you elaborate or link to why it isn't linear?

There are a number of factors that affect it, most noticeably heat. This might begin to point you in the right direction. http://arstechnica.com/gadgets/2011/04/ask-ars-whats-the-relationship-between-cpu-clockspeed-and-performance/

 

When I get my new rig up and running, I'll add this sort of thing to the list of CPU based benchmarks I want to try. 

[Cacao]

I'm testing this, 1st run:

 

3.3ghz - 8.84

 

 

Overclocked test coming in a few mins.

[Jebonia]

No - yes, higher clock speeds will increase performance however it isn't a linear relationship. Also, clock speeds between CPU's do not mean equal performance. A prime example of this is the new AMD CPU clocked at 5GHZ getting the same FPS in Battlefield 4 as a 3960X which is clocked at 3.5Ghz. 

That's why i said for the same cpu. I know that you can't compare clock speeds with different cpus

[Cacao]

test #2:

 

4.7ghz - 12.7

 

 

So we have a about a 42% increase in clock speed which gives us a about a 49% increase in our cinebench score. Well that went better than expected

[MeltyMoon]

There is no straight equation for clockspeed gains on one single architecture. The law of diminishing returns still apply, so after about 1 Ghz or so you pretty much wasting your time when you overclock IMO.

[Spannerhands]

Is the relationship between clock speed and performance linear when you are talking about the same cpu? i.e. Will an underclocked 4670k running at 2ghz take twice as long to render the same video if it was being rendered with a 4 ghz ?

 

Think of it like a car engine...

 

AMD is a petrol Engine and Intel is a diesel engine...and GHZ is RPM

 

A 200BHP Petrol engine is high revving and gets ok fuel economy, and makes it peak power at 6500RPM

 

A 200BHP Diesel engine is lower revving but gets better fuel economy and makes it peak power at 4500RPM

 

So you cant compare both engines @ 4500 RPM as the petrol has not reached peak power at that RPM, but you can compare there total power output and fuel economy...

 

Like wise if an engine revs to 6500RPM and make peck power of  200BHP then @ 3250RPM its not going to make 100BHP so its not linear... 

[Jebonia]

Think of it like a car engine...

 

AMD is a petrol Engine and Intel is a diesel engine...and GHZ is RPM

 

A 200BHP Petrol engine is high revving and gets ok fuel economy, and makes it peak power at 6500RPM

 

A 200BHP Diesel engine is lower revving but gets better fuel economy and makes it peak power at 4500RPM

 

So you cant compare both engines @ 4500 RPM as the petrol has not reached peak power at that RPM, but you can compare there total power output and fuel economy...

 

Like wise if an engine revs to 6500RPM and make peck power of  200BHP then @ 3250RPM its not going to make 100BHP so its not linear... 

Apologies if i may have mislead people on my question. My question was is it linear if it was the same cpu. I know that you can't compare clock speeds between and amd and intel CPU at this day and age

[Cacao]

Apologies if i may have mislead people on my question. My question was is it linear if it was the same cpu. I know that you can't compare clock speeds between and amd and intel CPU at this day and age

Look at my tests.. It is linear if you are using a 100% CPU dependent program.