Hello,
does anybody have experience with AMDs new X2 CPUs?
Are they better (or faster) with Blender than the FX CPUs?
By the way, there a 2x512 and 2x1024 Cache CPUs (Manchester & Toledo). Is there a noticable difference between them?
Thanx,
Mic
For one, Blender 3d doesn’t support 64b yet, so the performance gain that would be seen would be little, with a two core chip the threads may work like the Hyperthreading does on a P4 chip, giving a little bit more speed improvements. I have a 64b AMD 3200 and only saw a small performance gain, when combined with a 64b OS and compiling Blender 3d for yourself. If you use a stock Blender and OS their will be no gain, except for the faster chip.
On the cache size, more cache usually means faster performance. But sometimes the smaller cache chips can be overclocked higher giving better performance all around.
The tests that I have done with a P4 3.2g vs an AMD 64 3200 where just about the same, on the same PC with the same OS. The P4 was slightly faster on the internal renderer due to the hyperthreading support. Yafray on the other hand was almost 2x faster on the AMD 64, due to the fact that it does support 64b OSes.
Mic, I saw that you asked a lot of hardware questions in several threads, it may have been better to ask them all at once. Here is a response I gave to someone a couple weeks ago who was asking about a good blender hardware setup:
If Blender is multithreaded, the X2 CPU is better.
But I don’t know if blender is multithreaded. Is there someone who can tell me that?
Just look this render test with Maya and 3dsm. These two app are multithreaded and X2 CPU are the fastest to render the scene.
I did a test of Intel Hyperthreading and speeding up Blender. Someone commented on the times their 64bit AMD (there’s a 64bit Linux build of Blender) processor acheived. May be of some use.
http://www.digital-air.co.uk/2005/07/intel-hyper-threading.html
Ricky Dee
Current Blender does take advantage of 2 cpus for rendering. I don’t know of exact numbers, but I’m sure it’s a pretty substantial gain if you use an X2 over an FX.
The different cache sizes don’t matter in the majority of programs from the benchmarks I’ve seen, though I can’t comment on blender specifically.