You're missing the whole point. HT or dual core will not have much effect on speed at all. In order for this to work, the code would have to be optimized to use 2 separate cores or threads, and even then, the performance gain wouldn't be very substantial. Quite frankly, the time it would take to optimize the source for multi threading just isn't worth it. There's not enough to be gained from it to make it feasable. Even if you were to do this, you'd still only be using a CPU that has either 32 or 64 bit registers. And peak floating point operations would not increase by much, so the bottleneck is still there.DR.BETRUGER said:Yes, but i have heard that the Dual-core cpus, are more powerful than the Athlon 64 bit. Is it true? Has anyone try to run the emu with a dual-core cpu? I think that a dual core cpu will run dolphin faster than a Athlon 64, right?
No... Every person you ask tells you the exact same thing. Dolphin will not run faster on current machines right now. That's because neither Dolphin nor the current PC hardware are evolved enough. This is not going to change despite the number of times you ask.DR.BETRUGER said:Yes, but i have heard that the Dual-core cpus, are more powerful than the Athlon 64 bit. Is it true? Has anyone try to run the emu with a dual-core cpu? I think that a dual core cpu will run dolphin faster than a Athlon 64, right?
generalplot said:Athlon 64 X2 has around 5.5-6 Giga FLOPS at peak
The Athlon 64 X2 IS a dual-core processor. Besides, it sounds like you don't know very much. Just because the CPU has 2 cores doesn't mean you instantly get 2x performance. In fact in most things you wouldn't even get ANY performance boost.DR.BETRUGER said:Yes, but i have heard that the Dual-core cpus, are more powerful than the Athlon 64 bit. Is it true? Has anyone try to run the emu with a dual-core cpu? I think that a dual core cpu will run dolphin faster than a Athlon 64, right?
ector said:Although one of my plans for this spring is to build myself a dual core computer... If I find some time I might try to tear off the graphics processing into a separate thread. I'm pretty sure it can be done and would give some speed boost. No promises though![]()
Here's a note, before you go getting all excited: PCSX2 is trying to put some of the graphics rendering on a separate thread for this same reason, and you know what? Hardly a speed increase.DR.BETRUGER said:Wow!! This is the greatest idea i have ever heard!! It is true that if the graphic processing is into a separate thread, then cpu, will not have so much things to do, so this wil be increase speed a lot. Congratulations and good work!!!
TerraPhantm said:FLOPS aren't everything, there's an article on Anandtech that explains why the XBOX360 CPU, which is capable of 1 Terra FLOP, is still an inferior processor to the Athlon 64 and P4. Also the XBOX CPU IS a Celeron running at 733MHz, and yet the Xbox is more powerful then the gamecube.
I believe the coding issue is more with the PPC processors being RISC, while x86 computers are CISC. I guess theoretically it'd be easier to emulate the gamecube on a mac, because they both have PPC processors.
Also you will not see much of a performance increase for SMP apps with hyperthreading. Hyperthreading just creates a logical core with the extremely long processing pipe that P4s have. The P4D and Athlon64 X2 however have two PHYSICAL cores which are nearly as efficient as two physical processors
generalplot said:OK, I love these threads that attempt to compare emulating a console solely based on CPU clock speeds. This is when someone can come in and inform the rest of the people that don't know about what a difficult task emulating a next gen console can be. There's simply more to it than just going down the Mhz line, so to break it down, here's a general idea:
Comparisons:
PC: Generally a 32 or 64 bit environement. Athlon 64 X2 has around 5.5-6 Giga FLOPS at peak.
Gamecube: A 128 bit IBM Power PC "Gekko" 485 MHz, 13 Giga FLOPS at peak.
This simple comparison only shows a small example of the total picture. On top of the fact that the GC has a 128 bit CPU (128 bit registers which is double or quadruple that of a PC) the floating point operations (Giga FLOPS) is about double that of what one of the most powerful CPU's is capable of. If this doesn't explain enough about how difficult this can be, I can always gather more tech info, but this should give you some idea why the Mhz myth isn't enough.Combine this with inefficient code that is still a bit buggy, and low FPS shouldn't come as a surprise at all.
It doesn't have enough power to emulate, because the architectures of the two processors are completely different, it'd take more then raw power to emulate it. As bohdy said, emulation is dependent on things other then FLOPS and MHz.DR.BETRUGER said:So you say that the Athlon 64 x2 are powerful enough to emulate the XBOX 360??
TerraPhantm said:FLOPS aren't everything, there's an article on Anandtech that explains why the XBOX360 CPU, which is capable of 1 Terra FLOP, is still an inferior processor to the Athlon 64 and P4. Also the XBOX CPU IS a Celeron running at 733MHz, and yet the Xbox is more powerful then the gamecube.
I believe the coding issue is more with the PPC processors being RISC, while x86 computers are CISC. I guess theoretically it'd be easier to emulate the gamecube on a mac, because they both have PPC processors.
Also you will not see much of a performance increase for SMP apps with hyperthreading. Hyperthreading just creates a logical core with the extremely long processing pipe that P4s have. The P4D and Athlon64 X2 however have two PHYSICAL cores which are nearly as efficient as two physical processors