it shouldnt take long to get performance up anyway...i mean when i joined this site they were in the rev like 2000 then i left for 9 months to a hell hole and now i get back and theyre already on rev 6000 lol so yea in a year youll have ur dolphin 3.0 thtll be able to play like anything so ya
(07-31-2010, 04:37 AM)fred52170 Wrote: [ -> ]The question is : if we have to wait 3 years because the cpu will be faster, better buy on ebay for 15€ a wii, because there will be surely a new console by nintendo at this time!
Then you're here for the wrong reason, Dolphin is not a substitute for a real wii or a simple free alternative to buying the actual thing.
This is an emulation project, not a virtual replacement console. And yes, wii is well worth the €200 or so it costs now. It's not that expensive.
Dolphin is great mainly for it's ease of use, opportunity to use very high resolution/etc... But if you want to experience the full gameplay without bugs, disappearing sounds, etc.. Play it on the wii. It's guaranteed not to crash. The emu is still a work in process.
Emulation is mostly about hardware/software preservation and is a hobby. Developers just lost sight of this because of the widespread whining of individuals due to performance issues that are associated with truly accurate/non-JIT'd emulation. Just read about all the lashback at the Mame folks because of this, cause the warazing kiddies can't play their pirated games without buying the hardware.
Consider yourself lucky that there is even an a viable alternative, had developers stuck to their original goals Dolphin and Pcsx2 would be running on the interpreter core and using software rendering with low level emulation, you would be waiting 15 - 20 years before PS2, GC, Wii emulation was possible with acceptable performance. Bsnes, an Snes emulator requires a 2 Ghz processor as the minimum for decent performance with most games. Sometimes I wish this were the case for every emulator because then more people would actually buy the hardware, instead of using an emulator to play pirated games and expecting emulation to work exactly like the actual hardware. This would also mean the forum would be filled with less garbage, and contain more useful information and intelligent discussions.
By the way, the issue mentioned in this thread was marked as invalidated on the issue tracker. Hopefully this will put a lid on the op who keeps insisting these revisions had performance regressions, which is simply untrue, read configuration issues as I've outlined in this thread.
This might sound really strange to many, but as Xtreme2damax is saying:
People enjoying playing games on the emulator = a side effect
And ironically, all the complaing is because the emulator is working too well, if the performance sucked completely, no one would hang around here to whine
(08-01-2010, 05:08 AM)KHRZ Wrote: [ -> ]And ironically, all the complaing is because the emulator is working too well, if the performance sucked completely, no one would hang around here to whine
No one says that the emulator suck.
You should accept that people can have a different opinion of yours.
Some want accuracy, others want speed... that's all.
Just be cool, it's not because we want speed that we are pirats, or something else...
I think that programmers like have a feedback from the users. If this project was just a hobby for themselves they didn't put it on internet!
I also noticed a significant speed decrease since r5982 (when the JIT optimizations were implemented)
And this time it's with a modern AMD CPU.
r5977 was the fastest build - 25.5 fps in ZTP [Faron Woods] with settings for maximum accuracy (JIT64, DualCore, Idle Skipping, cheats disabled, HLE'd IPL, Direct3D9, EFB Scale x1, no SSAA, 640x480 windowed, CPU->EFB access enabled, EFB copy to RAM, Native Mips, DirectSound Audio, no throttle).
Since the JIT optimizations in r5982, Dolphin's JIT64 was getting slower and slower. The FPS dropped from 25.5 to barely 24 fps. Repeated the test about 100 times (while resetting the emu and the PC) and still no improvement.
r5982 introduced a bigger fps drop, while 5991 introduced another smaller one.
On the other hand, JITIL64 performance dropped only minimally. 27.5fps -> 27fps, but it's still slower, not faster.
It's the optimizations that are the cause of the performance drop, not the increased accuracy.
No it isn't, that's already been confirmed and the issue was marked as invalid by the developers. Speed is fine for nearly everyone else, check your configuration. There is no possible way for those commits to cause speed issues, and the fact that this issue is limited to only a couple members while speed is fine for everyone else points to a configuration issue. There is nothing wrong with those commits, the problem lies on your end or likely another revision.
JITIL has been slower for a while, JITIL x64 has always been very slow, nothing to do with those commits.
Must I need to post comparison shots just to prove that those commits didn't cause speed issues? Other members that are also not having any issues should do the same just to prove those commits had no impact on performance, at least not a negative impact.
Users with Core iX series of CPUs aren't affected, but everyone else (with a Core2, Phenom II,Athlon II,...) is.
(08-01-2010, 05:00 PM)kirbypuff Wrote: [ -> ]I also noticed a significant speed decrease since r5982 (when the JIT optimizations were implemented)
And this time it's with a modern AMD CPU.
r5977 was the fastest build - 25.5 fps in ZTP [Faron Woods] with maximum accuracy settings (JIT64, DualCore, Idle Skipping, cheats disabled, HLE'd IPL, Direct3D9, EFB Scale x1, no SSAA, 640x480 windowed, CPU->EFB access enabled, EFB copy to RAM, Native Mips, DirectSound Audio, no throttle).
Since the JIT optimizations in r5982, Dolphin's JIT64 was getting slower and slower. The FPS dropped from 25.5 to barely 24 fps. Repeated the test about 100 times (while resetting the emu and the PC) and still no improvement.
r5982 introduced a bigger fps drop, while 5991 introduced another smaller one.
On the other hand, JITIL64 performance dropped only minimally. 27.5fps -> 27fps, but it's still slower, not faster.
It's the optimizations that are the cause of the performance drop, not the increased accuracy.
That kind of fps could fluctuate in an area even in a single revision, for me the slowest part of Arc Rise Fantasia (Diamant City) has been consistently slow from the revision you said.
I have AMD cpu too btw.
(08-01-2010, 05:47 PM)kirbypuff Wrote: [ -> ]Users with Core iX series of CPUs aren't affected, but everyone else (with a Core2, Phenom II,Athlon II,...) is.
Not true, I have a Core2Duo E8500 and many other members have AMD Phenom processors and have absolutely no performance issues with the recent builds. These optimizations are definitely not responsible for your performance issues.
FPS will fluctuate in specific areas, especially when there is more to render, because of bottlenecks in the code. FPS will never be constant, speed will vary and differ from one playthrough to the next. Sometimes you will have ok speed one time, and other times speed will go downhill, it depends on the conditions. This has nothing to do with the recent optimizations, it's just an issue with the emulator itself that everyone experiences except maybe those with overclocked Core i7 processors and high end graphics hardware.