About "Enable Dual Core" on a Single Core Processor...
|
(04-28-2009, 07:57 AM)FloW3184 Wrote: hyperthreading is not better than a real dualcore afaik.i dont see why it shouldn't be better 2 threads getting processed at the same time its the same except HT on single core uses less power o.o and i think DC cpu's are better because of the "improved" technology (aka more cache, higher speeds and other things) 04-28-2009, 09:54 PM
Well Hyperthreading just makes switching Threads faster.
You should get similar Speedups on AMD CPUs as well The problem why SC sucks is that it has to do the CPU and GPU part and keeps it in sync. So if your GPU Part runs at 30 FPS the CPU runs at half speed. It just tries to reach 60 FPS and if it cannot slows down all components equally. In DC Mode you have 2 Threads that are independent from each other. So the less expensive CPU part runs independently and gets nearly full speed. On the other hand your GPU Part runs a bit slower as it has to sync with the CPU part. So basically your "Speed Up" is just a freed CPU Thread... (04-28-2009, 09:54 PM)lenny12 Wrote: Well Hyperthreading just makes switching Threads faster.all sounds ok except Quote:The problem why SC sucks is that it has to do the CPU and GPU partwhat tell me, whats the point of gpu's if the cpu does all graphical stuff (thats what you said) EDIT: yes, onboard "gpu"'s are processed by cpu's (thats what makes them more crappy) but god, external gpu's ? 04-29-2009, 08:39 AM
(This post was last modified: 04-29-2009, 08:42 AM by thegamefreak0134.)
Quote:what I'll give you two bits of info, in reverse order then. Newsflash: "Offboard" GPUs are also processed by CPUs. All graphics cards use regular old processors to do most of their work. (Albeit, often highly specialized or optimized processors.) The catch is that a graphics card usually has a whole bunch of them. They get their speed boosts from simultaneous processing, not from raw power on one or two threads. And besides, on-board graphics cards (at least the ones I've seen) are still usually on a separate chip. Point being, a "GPU" is really just a processing chip that does graphics and nothing else, so the regular CPU doesn't have to do as much and can worry about other things. That aside, there's still a lot of work that Dolphin has to do on the CPU before the GPU can ever get that data. The Wii/GC uses a special format for all of its 3D data and textures that isn't natively understood by DirectX or OpenGL libraries. In order for Dolphin to emulate this, it has to translate that 3D data from the Wii/GC format into the DirectX/OpenGL format for the PC graphics card to display. As I understand it, this means translating all of the models for the current frame, point by point, into the correct format. Lather, rinse, repeat for each frame. They can optimize this a lot (and continually do so) but it still takes a good chunk of CPU to do that conversion in realtime. This is why Dual-Thread benefits Dolphin so much, it allows this process to happen at the same time that the CPU is processing the game (in theorey) which is very similar to how the Wii/GC would work anyway-- that is, you do things in the game in three steps. Step 1: calculate logic for frame. Step 2: throw the game's graphics at the GPU. Step 3: Wait for the GPU to finish. Naturally, every game on the market combines step 3 from the previous frame with step 1 of the next frame, so while the hardware is busy drawing frame 1, the game is busy calculating the AI and bone transformations and things for frame 2. Actually getting this all to work in sync in an emulated environment is very time consuming when you realize that you have to emulate the communication between the hardware at the same time that you're emulating the hardware itself. Edit: For some strange strange reason, Firefox on Ubuntu 9.04 is telling me that "optimized" is misspelled while "optimised" is spelled correctly. According to every other source I have, the reverse is true, "optimized" is the correct spelling. At least, in American English. Weird, seeps to happen for other words using z as well. Quite strange, has anyone else noticed this? Anyway, I appologize for any spelling errors then, since I clearly can't trust my spell checker. Further Edit: *Kicks self. Several times.* I'm sorry Daco, I should really really read the postee before making myself out to be a fool. This is the second time that I've argued with someone over something they clearly already know. >_< Regardless, I'll leave it up (again, I did this in another post) in case anyone finds the info useful, and you have all permission to slap me later. (04-28-2009, 07:43 PM)Daco Wrote: i dont see why it shouldn't be better Yeah, it's the same except in HT 2 threads have to share resources of a single core. So the threads are very limited in what they can actually do at the same time. Most of the time they just take turns to use one block or another, which is not very efficient AFAIK. In DC each thread has full resources of a core to itself and doesn't have to share. 04-29-2009, 06:39 PM
05-09-2009, 10:27 PM
05-12-2009, 01:57 PM
yeah, amd golden times, now intel owns it, but amd is always the first to bring new things....
like 64 bits and the real time video streaming. i guess amd is coming back.... but if u look straight u will see that intel cpus are faster now, but they always have brute force like agressive amounts of cache and expensive stuuf to make the latency lower... thats why its more expensive than amd. but u see once again amd brings new tech to cpus like the numa archteture, u can have a phenon II almost better than any c2q, that overclocks to 4.0 ghz for less than a half price of a core i7, so u can cut the edge with something that fits ur pocket.... now can u imagine if amd does a phenon 2 with the core i7 agressivines with an equal price tag? 05-12-2009, 02:04 PM
|
« Next Oldest | Next Newest »
|
Users browsing this thread: 2 Guest(s)