AMD Core Per Module

[csgoSparroW]

For the average game, would an AMD CPU like the AMD 8150, perform better for most games using 2 cores per module, or the 1 core per module method.

Also, what are some good ways to benchmark this type of test? Any free benchmarking software would be appreciated.

 

Thanks,

SparroW

[Sintezza]

Short straight to the point answer NO.

 

Those cores like you allready mentioned your self, are integer cores inside a module that share re-sources like the fpu, cache and decoder.

There for they wont perform better then standalone true cores.

For example if you take the FX6300 which has 6 "cores"

And you compair that to lets say a 5820K with HT disabled.

Then you will see that the 5820K performs ways better.

Because that are true 6 cores.

And not 3 modules with each 2 integer cores as on the FX6300.

Of course you cannot realy compair it that way, because we talk about totaly diffrent architectual designs and price ranges.

BUt i think you will get the idea.

 

[Aytex]

I think you can stress per core with prime95 im not sure.

[csgoSparroW]

@Sintezza Thanks for that, I haven't been able to test it myself so that is exactly what I needed. I agree with the fact you can't compare due to architecture and price ranges, as that is the hardest thing to find on a video for benchmarks, is comparison due to price. Thanks again!

[SpaceGhostC2C]

On 11/6/2016 at 4:48 AM, csgoSparroW said:

For the average game, would an AMD CPU like the AMD 8150, perform better for most games using 2 cores per module, or the 1 core per module method.

Also, what are some good ways to benchmark this type of test? Any free benchmarking software would be appreciated.

 

Thanks,

SparroW

It would perform better with all cores enabled. The reason is that (after some Windows patching for older versions) cores are used "in the right order", i.e., one core per module until you run out of modules, then you use the second core in each module. Hence, a game running in 4 cores will  run almost equally well with 1 core per module or 2 (almost because there's always some background process, what would make 2 better). But if the game can use more than 4, then the 2 core per module option will be better, since you are not shutting down any cores.

The shared resources within a module imply that the scaling from 1 to 4 is better than from 4 to 8, but you won't make net losses. Hence, disabling cores you lose something without really gaining anything.

 

Disabling cores could be helpful to reach higher overclocks. In that sense, if you are using games that take advantage of 1 or 2 cores only, shutting down 4 cores and OCing the remaining 4 more heavily could translate into a better performance for those games.

[Sintezza]

4 hours ago, csgoSparroW said:

@Sintezza Thanks for that, I haven't been able to test it myself so that is exactly what I needed. I agree with the fact you can't compare due to architecture and price ranges, as that is the hardest thing to find on a video for benchmarks, is comparison due to price. Thanks again!

Well i did some testing and played arround with disabling cores and what not.

But the main problem is that most motherboards wont realy allow you to shut down individual cores, only modules like Asus boards.

And in that case you not only lose cores but also the fpu´s.

So that will result in a performance decrease.

If you have a specific motherboard that could allow you disable one core per module.

Then still it isnt realy going to be an increase of performance.

Because basicly both integer cores have their own FP pipeline.