Am I wrong or right?

[i_build_nanosuits]

Is there some sort of reference to that?

?! ...its how this work its the way AMD has gone with designing the bulldozer cmt architecture that has been improved with piledriver and has remained the same since...if you cant take my words for it just try to google your way into this mate...

[Godlygamer23]

?! ...its how this work its the way AMD has gone with designing the bulldozer cmt architecture that has been improved with piledriver and has remained the same since...if you cant take my words for it just try to google your way into this mate...

I looked at Anandtech's article on Bulldozer. No mention of it at all.

[Bit8xPr0]

Well im not even going to bother to talk about the CMT architecture and how it exaly work.

Because i think that people should learn and do their own research.

I see so manny people make the same mistake over and over again wenn they talk about the FX processors.

 

But yeah thanks for adding this lol

 

Back to the main question, "OP" you are completely wrong

Am I wrong or are you wrong?

[Bit8xPr0]

Maybe @LinusTech or @Slick knows DUN DUN DUN

[Bit8xPr0]

Maybe @AMD knows

[Bit8xPr0]

I looked at Anandtech's article on Bulldozer. No mention of it at all.

It's quite confusing how the answer of this is unclear even after being out for so long. 

[i_build_nanosuits]

Am I wrong or are you wrong?

you are mostly wrong...the fx 8 core (4 modules thus 4 FPu) is likely to be faster at anything that use 3 threads or more assuming there are some integer based instructions and some fpu instructions as its pretty much always the case...the 3 module fx will never be faster than the fx8 if both are clocked at the same frequency.

[Bit8xPr0]

you are mostly wrong...the fx 8 core (4 modules thus 4 FPu) is likely to be faster at anything that use 3 threads or more assuming there are some integer based instructions and some fpu instructions as its pretty much always the case...the 3 module fx will never be faster than the fx8 if both are clocked at the same frequency.

So why do the FX 6300 and the FX 8320 both score the same score in Single performance? Same frequency

[Bit8xPr0]

you are mostly wrong...the fx 8 core (4 modules thus 4 FPu) is likely to be faster at anything that use 3 threads or more assuming there are some integer based instructions and some fpu instructions as its pretty much always the case...the 3 module fx will never be faster than the fx8 if both are clocked at the same frequency.

Does this also mean that the Single Core Performance isn't really "Single Core Performance" in a benchmark? Because they are forced to help each other.

[Faa]

You understand how the CMT architecture works?

Well there's not much or no performance difference between a 128 bit FPU & 256bit FPU in most tasks except ones with support for AVX (256bit requirement). Also it's a FlexFPU 256 bit that can split in two 128 bit FPU's for when you pump a 2nd thread in the core.

You're only covering the back-end (execution resources like FPU's & ALU's) but that's not the reason why a 8350 will do better in 4 threaded tasks than a 6300, the front-end is whats steering the back-end and if the front-end is crap the back-end will be heavily bottlenecked. On a 8350 you have 4 front-ends and 3 front-ends on a 6300. The front-end is the cause of low IPC and becomes quickly the bottleneck so having an additional front-end to pump the 4th thread will soften it and result in a performance gain.

Theoretical difference is 33% (4 vs 3) when you're purely looking at the front-end count but it doesn't remain that simple, it will be much lower than that. 

[D2ultima]

So why do the FX 6300 and the FX 8320 both score the same score in Single performance? Same frequency

Single performance? Assuming you mean single-threaded performance, then obviously it will. He said if any program uses three or more threads AND involves integer calculations (which use the floating point unit) that the CPU will use all the FPUs on itself to calculate, and thus be better than the hexacore. If you go single-thread versus single-thread, even an i3 will beat an i5 if the i3 is clocked higher. It's a moot point. Single-threaded performance only helps a few programs which are still stuck in single-thread performance.

 

Anyway... as far as AMD goes, it doesn't make much sense for YOU personally to upgrade from a 4.3GHz 6300 to a 8350... unless you plan to OC that more. If you do, it's quite easy to get it to hit 4.8GHz. If that's unattainable easily with your 6300, then the 8350 will be an "upgrade". Of course, I've no idea what you're using your CPU for.

[CDHoward]

No way at all.

I have the 8350 Black Edition. The 8350 obviously has more cores, but it also newer architecture over the 6300.

Those who say "All the FX chips are the same chips, just with cores turned off" are wrong. They have slightly different architectures and protocols.

[Godlygamer23]

No way at all.

I have the 8350 Black Edition. The 8350 obviously has more cores, but it also newer architecture over the 6300.

Those who say "All the FX chips are the same chips, just with cores turned off" are wrong. They have slightly different architectures and protocols.

The 6300 and 8350 both use Piledriver.

[harrynowl]

For the same reason if an application uses 2 cores a 5GHz G3258 will crush a 3GHz 5960X

 

If they don't get it don't waste your time

[CDHoward]

The 6300 and 8350 both use Piledriver.

Oh. I thought the 6300 was Bulldozer, and the 6350 Piledriver.

[Godlygamer23]

Oh. I thought the 6300 was Bulldozer, and the 6350 Piledriver.

The x100 series is Bulldozer. The x300 series is Piledriver. The 6350 is just clocked higher. That's all.

[Bit8xPr0]

Does this also mean that the Single Core Performance isn't really "Single Core Performance" in a benchmark? Because they are forced to help each other.

Anyone highlight on this?

[Guest]

Link them this thread.

[Bit8xPr0]

Link them this thread.

Link who?