AMD FX6300. Leaving turbo core on or off at stock speed?

goranpaa

Member
Well yeah, 100MHz more on six cores is obviously a higher gain than 100MHz on a dual core, but you still have to deal with software not being core optimized and such.
If you were purely doing number crunching that utilized all cores, you could almost (that math is never correct) say that a 100MHz OC on a six core is a 600MHz OC.

Aha! Interesting info. Thanks!
 

Darren

Moderator
Staff member
Comparing single core to a six core or whatever the case may be is kind of pointless. You'll get performance increases if you overclock. Period. Yes mathematcally you'll be gaining more the more cores you have but programs don't fully use all cores a 100% efficiency so saying it is way better than overclocking a single core doesn't accomplish anything.

My friend has an FX 6300 like yours and I have an 8320. They're essentially the same chip (run at the same speed) but mine has 2 more cores than his. Very few games utilize all 8 cores (or 6 for that matter) so the real world differences between the two is low. If they were both clocked higher they would have a pretty similar gain in performance. Now if you're looking at raw benchmarks of how much data can you put through all the of the cores at once then yeah, the 8 core will do more than the 6, but real world won't really show that.
 

StrangleHold

Moderator
Staff member
With AMD Windows has a slight problem of knowing what threads to run on what cores. All the cores look like individual cores, not modules. If everything goes like it should, the 8 core will run slightly better because it has the capability of spreading the threads out on the cores. In other words if your running a 4 threaded program it can run 1 thread on each module instead of doubling them up on modules. But windows will screw it up and run 4 threads on 2 modules and leave 2 modules idle.

I can run a 4 threaded program and every time Windows will use different cores. Sometimes it will double them up on modules. Then run all threads on different modules. Then double 2 up on a module and run the other 2 on different modules. You can run the same benchmark 4 times and get a slightly different score each time. I even have the patch installed and it still didn't completely fix it.
 

Darren

Moderator
Staff member
To be clear an 8320 has 4 modules, each with 2 cores on it right? Does it work in any similar way to Hyper Threading that Intel has?
 

goranpaa

Member
Comparing single core to a six core or whatever the case may be is kind of pointless. You'll get performance increases if you overclock. Period. Yes mathematcally you'll be gaining more the more cores you have but programs don't fully use all cores a 100% efficiency so saying it is way better than overclocking a single core doesn't accomplish anything.

My friend has an FX 6300 like yours and I have an 8320. They're essentially the same chip (run at the same speed) but mine has 2 more cores than his. Very few games utilize all 8 cores (or 6 for that matter) so the real world differences between the two is low. If they were both clocked higher they would have a pretty similar gain in performance. Now if you're looking at raw benchmarks of how much data can you put through all the of the cores at once then yeah, the 8 core will do more than the 6, but real world won't really show that.

Ok. thank you very much for sorting this out for me.

I tried an overclock yesterday. I could reach 4.5 Ghz stable. With a max temp of 44 C doing the AIDA 64 stress test for 2 hours. The room temperature at the time where 20 C.

That is pretty neat I think. :)

But I'm satisfied with the FX 6300 at stock speed for the moment at least.
 

goranpaa

Member
With AMD Windows has a slight problem of knowing what threads to run on what cores. All the cores look like individual cores, not modules. If everything goes like it should, the 8 core will run slightly better because it has the capability of spreading the threads out on the cores. In other words if your running a 4 threaded program it can run 1 thread on each module instead of doubling them up on modules. But windows will screw it up and run 4 threads on 2 modules and leave 2 modules idle.

I can run a 4 threaded program and every time Windows will use different cores. Sometimes it will double them up on modules. Then run all threads on different modules. Then double 2 up on a module and run the other 2 on different modules. You can run the same benchmark 4 times and get a slightly different score each time. I even have the patch installed and it still didn't completely fix it.

Hi!

Did'nt know this about AMD cpu's and the treading problem in Windows. Thanks for the info. Time to get that patch then.
 
Last edited:

StrangleHold

Moderator
Staff member
To be clear an 8320 has 4 modules, each with 2 cores on it right? Does it work in any similar way to Hyper Threading that Intel has?


Right, they share some of the upper end and the L2 cache on each module. No, not really. Bulldozer is all hardware. In a virtual world the module is a great idea. But in the real word, its just to complicated for the OS to figure out what threads to run where, lol
 
Top