Jump to content

How much more speed can one get with SSE3?


wow
 Share

5 posts in this topic

Recommended Posts

Don't know about any numbers, but you'll for sure see an increase in speed, ESPECIALLY if you're running CPU intensive programs like games.

 

With SSE2 you have to emulate, which takes longer and more CPU power. SSE3 is native, so you're running the programs directly.

 

Either way, there are only a few OSx86 programs out there, the rest will still be emulated by Rosetta.

Link to comment
Share on other sites

Don't know about any numbers, but you'll for sure see an increase in speed, ESPECIALLY if you're running CPU intensive programs like games.

 

With SSE2 you have to emulate, which takes longer and more CPU power. SSE3 is native, so you're running the programs directly.

 

Either way, there are only a few OSx86 programs out there, the rest will still be emulated by Rosetta.

 

So using SSE2 is emulating SSE3 which is emulating PPC instructions?

 

I thought SSE2 Patch OSX86 didn't emulate SSE3 just took care of SSE3 instructions (is that the same as emulating?)

Link to comment
Share on other sites

You should probably read:

 

http://www.intel.com/technology/itj/2004/v...ure/p06_sse.htm

 

 

SSE3 can really cut down on some code.

 

<snip>

...

As shown below, the benefit of fisttp is two-fold: fewer instructions are needed and there is no need to modify FCW. The instruction is available in three precisions: Word (16-bit), DWord (32-bit), and QWord (64-bit).

 

Code without SSE3:

 

fstcw <old FCW>

movw ax, <old FCW>

or ax, 0xc00

movw <new FCW>, ax

fldcw <new FCW>

fistp <INT>

fldcw <old FCW>

 

Code with SSE3:

 

fisttp <INT>

 

...

<snip>

Link to comment
Share on other sites

 Share

×
×
  • Create New...