(07-09-2009, 06:07 AM)jasong Wrote: [ -> ]I can post screenies later.
I don't think we need screenshots. Just give us your FPS, we'll believe you!
I did a little testing myself:
Super Mario Galaxy 32 bit: 27 FPS
Super Mario Galaxy 64 bit: 32 FPS
Mario Kart Wii 32 bit: 45 FPS
Mario Kart Wii 64 bit: 56 FPS
Mario Kart Double Dash 32 bit: 27 FPS
Mario Kart Double Dash 64 bit: 30 FPS
Too bad, I could use the additional frames. If I'd only knew what makes the 64 bit version crash from time to time...
The only difference I noticed is, that the 32 bit version sometimes during gameplay says "Cleared cache". It appears top left, where all the information appears when you start a game. I never saw that in the 64 bit version. Could that be a clue?
also 64 bit support 4 gb DDR Ram
I always see the "Cleared code cache" thing in x64, unless you enable Unlimited JIT cache in debugger.
those numbers are pretty much in keeping with my findings as well. Paper Mario was 49/60, SMS was 40/50, Zelda TP was 30/28. 64 bit is faster, that's pretty definitive.
(07-09-2009, 02:35 AM)SantaC Wrote: [ -> ]you also need a 64 bit OS right
Yes, you do. And a 64 bit processor, of course.
(07-09-2009, 03:57 AM)CacoFFF Wrote: [ -> ]Considering my 64 bit OS can't even run a x86 version of Dolphin.
Why? Your signature says XP 64. 64 bit Windows can handle 32 bit applications very well. Or are you referring to another operating system?
(07-09-2009, 07:55 AM)ESwormL Wrote: [ -> ]also 64 bit support 4 gb DDR Ram
Yes, 64 bit operating systems support more than 4 GB RAM. Since Dolphin never needs that much, it's no argument... well, as far as I understand. Apart from that, x64 Dolphin is still faster!
(07-09-2009, 08:40 AM)CacoFFF Wrote: [ -> ]I always see the "Cleared code cache" thing in x64, unless you enable Unlimited JIT cache in debugger.
I didn't enable anything, but I'm really sure I never encountered it in x64.
How are people comparing the differences between running x86-32 or x86-64 versions of Dolphin?
Running both versions on a 64bit OS?
Running the x86-64 on Vista 64, and the x86-32 on Vista 32? Obviously with the same underlying hardware?
Both versions are compared on a 64 bit OS. Yes, I guess that's not very fair, because 32 bit processors tend to clock faster than 64 bit processors (is that true?). So the 32 bit versions could be faster.
Semi off topic: can someone explain to me what the "Lock threads to cores" option means?
If it's enabled, I've got two of my four cores working at near 100%. So the temperature of these cores go up to about 75°C.
If it's disabled, all cores are working with about 50% and the temperature stays at 65°C in all cores. (Measured with CoreTemp)
Considering the temperature and the fact that 100% usage make me nervous, I'd like to leave it disabled. Is there a serious drawback with that?
Interesting does it change the speed at all?
btw 65C seems very excessive and you aren't overclocking - you sure everything is plugged in as normal?
(07-09-2009, 09:15 PM)jasong Wrote: [ -> ]you sure everything is plugged in as normal?
Well, no!
Idle temperature is 40°C, according to CoreTemp or RealTemp. I'm not overclocking and the cooler is running fine. I've got no idea what could be wrong (if) and thought that's just fine for a CPU, even if it's Quad.
Besides, I think that those sensors only report the distance to the max temperature TJMax. This distance is around 60°C in idle and 25°C in Dolphin and 14°C with crunching numbers. My TJMax is 100°C.
The problem with testing 32bit dolphin on a 64bit Windows OS is the extra overhead from wow64.
I've no idea how much it is, but running an emulator on an emulator could explain some of the performance difference?