swilczak :
8350rocks :
swilczak :
rishiswaz :
swilczak :
I'm sure that's how AMD wanted it to work
AMD made a new architecture based on more ALUs and less FPUs to deliver more real-world performance where it counts while still having floating point calculations available to those who need it. When launching the FX processors they clearly said that they were re-inventing what a core should be, so not it is not Hyperthreading at all. Some programs will report each module as a core instead of each core. 1 module contains two cores so it may seem that way but they are all physical cores with no special thread handling.
The Tom's Hardware website commented that the lower-than-expected performance in multi-threaded workloads may be because of the way Windows 7 currently schedules threads to the cores. They point out that "if Windows were able to utilize an FX-8150's four modules first, and then backfill each module's second core, it'd maximize performance with up to four threads running concurrently." This is similar to what happens on Intel CPUs with HyperThreading – Windows 7 "schedules to physical cores before utilizing logical (HyperThreaded) cores
What it is and how it actually works in windows are two different things. It works like hyperthreading just like I said.
http://en.wikipedia.org/wiki/Bulldozer_(microarchitecture)
You do realize the only shared cache is L3 right? Each core otherwise has separate L1 and L2 cache, which are far more crucial...
Additionally...for scheduling purposes it would work better that way...however, windows scheduler is typically poor compared to other things. If you look at the PD performance in Linux you can see the difference between a "efficient" and "less efficient" scheduler.
Yes I realize that, but since only 5% of computer users use Linux.....need I say more? I'm tired of arguing about this, I showed proof from tomshardware quoted on wiki, so I'm done. You can sit in your moms basement all day and try to prove me wrong.
So, let me get this straight...
You're basing your argument on windows scheduler being a poor scheduler. Because it's a poor scheduler, if it handled the scheduling more efficiently...like it does for a HTT enabled Intel CPU...it may improve performance. Since that may improve performance, you are insinuating that the modular design of AMD's CPU architecture is similar to HTT...? All this, even though HTT has no physical resources outside of an extra register stack, and AMD architecture has a physical core with 2 ALUs and dedicated L1 and L2 cache??
Hmm...then you justify this by quoting a misinformed writer, trying to make an off the cuff comparison, that was quoted by Wikipedia?
That argument holds as much water as a minnow bucket...(if you're not a fisherman, they have holes in them from top to bottom)