New performance test result between r5985 and r5995.

[finalmars]

I have to announce that there is NO other task which was using CPU while the testing was runing again and the test was runing under the same configuration and condition.

game for testing: resident evil 4 (gamecube version)
method of test:
Just run the game and start a specific stage, then wait until the fps counter becomes steady, take the screenshot for result finally.

below is the screenshot of the result:

First is r5985:
    -- fps:41
    -- fps:25
    -- fps:53

---

second is r5995:

    -- fps:37

    -- fps:24

    -- fps:49

---

compare to the Dolphin Released 2.0

    -- fps:73

    -- fps:29

    -- fps:129

I don't know how it happened. but I just want to say, the released version is still the best one to play this game...could someone tell me why the performance has reduced so much since the released version to the newest svn on this game?

---

finally post my configuration details of dolphin



           

SSAA:4x
EFB scaled copy: disabled

[Xtreme2damax]

There must be something else wrong then, specific to your system or configuration that affects no one else. No one else is reporting any speed issues except for yourself, speed has remained the same or improved for everyone else. I can even take some snapshots or videos and prove to you that none of those revisions had any negative impact on speed, sorry to say but you need to look everything over on your end carefully since no one else is reporting reduced speed. It is highly unlikely that the commits you mentioned would have a negative impact on speed, they were just optimizations.

The developers will likely not look into it unless there is widespread reports and other members reporting the same thing, all I can suggest is that you take snapshots of your current configuration and post them. Also post a list or snapshot of the plugins you are using as well.

I also noticed the snapshots you posted are inconsistent, you need to take comparison snapshots in the same spot/location in both the old revision and new revisions for a proper valid comparison. The reason for this is that some areas of the game run slower than other areas due bottlenecks in Dolphin, and rendering more on screen in specific areas. Idle skipping doesn't work correctly with this game, which could explain the decrease on the title screen, you'll have the same issue with the ending credits with idle skipping enabled.

[finalmars]

(07-30-2010, 03:35 AM)Xtreme2damax Wrote: There must be something else wrong then, specific to your system or configuration that affects no one else. No one else is reporting any speed issues except for yourself, speed has remained the same or improved for everyone else. I can even take some snapshots or videos and prove to you that none of those revisions had any negative impact on speed, sorry to say but you need to look everything over on your end carefully since no one else is reporting reduced speed. It is highly unlikely that the commits you mentioned would have a negative impact on speed, they were just optimizations.

The developers will likely not look into it unless there is widespread reports and other members reporting the same thing, all I can suggest is that you take snapshots of your current configuration and post them. Also post a list or snapshot of the plugins you are using as well.

I also noticed the snapshots you posted are inconsistent, you need to take comparison snapshots in the same spot/location in both the old revision and new revisions for a proper valid comparison. The reason for this is that some areas of the game run slower than other areas due bottlenecks in Dolphin, and rendering more on screen in specific areas. Idle skipping doesn't work correctly with this game, which could explain the decrease on the title screen, you'll have the same issue with the ending credits with idle skipping enabled.

actually, I can't understand your whole meaning.

As the snapshots which you could see, I have taken the test on 3 stages in the game for each of version respectively. And the older one just compares to the newer one with the same scene, the same setting, the same condition and even the same game's timing of snapshot.

I really can't understand why you still said that the snapshots you posted were inconsistent .

I have also posted the snapshots of my current configuration and plugings I am using.

anyway, thanks for your attention.

---

pluging details:

   

[Jotokun]

I havent been tracking the revisions, but I've been experiencing speed loss as well as new revisions have come out. I used to be able to play SSBB at 640x480 in full speed at all times, now it's hovering between half and 75% in-game, and full speed in the menus. I'll have to try the 2.0 release, I've always tried the revisions.

[finalmars]

(07-30-2010, 04:04 AM)Jotokun Wrote: I havent been tracking the revisions, but I've been experiencing speed loss as well as new revisions have come out. I used to be able to play SSBB at 640x480 in full speed at all times, now it's hovering between half and 75% in-game, and full speed in the menus. I'll have to try the 2.0 release, I've always tried the revisions.

could you provide any more detail? such as which revision you are using ? and the configuration details.

[Xtreme2damax]

(07-30-2010, 04:04 AM)Jotokun Wrote: I havent been tracking the revisions, but I've been experiencing speed loss as well as new revisions have come out. I used to be able to play SSBB at 640x480 in full speed at all times, now it's hovering between half and 75% in-game, and full speed in the menus. I'll have to try the 2.0 release, I've always tried the revisions.

Some speed has been lost due to accuracy over the past 100 - 500 revisions or so, that was bound to happen and is actually a good thing, just wait for the developers to optimize.

What is being claimed in this thread and the other thread is that the past 10 - 15 revisions have caused reduced speed which simply isn't possible as these revisions were optimizations and had virtually no effect on performance. Not only that, so far it has only been limited to one person reporting reduced speed while everyone else is reporting better speed, not worse.

The op needs to test multiple times under the same conditions, try to spread the tests apart and not immediately run the emulator again after stopping it.

http://forums.dolphin-emu.org/thread-11113-post-104937.html#pid104937

---

First of all, some things I see wrong..

You have 4x SSAA enabled, depending on what resolution you are running at this may be stressing your graphics hardware. Your GTX 260 is just a re-branded 9xxx series, running at higher resolutions especially with SSAA is taxing on the hardware, and may be negatively affecting your performance. I recommend setting EFB scale to 1x with no SSAA or with 4x SSAA for best results. I also recommend running at 1024 x 768, any higher than that and it's going to start impacting performance. Window mode is also faster than stretching the window or using full screen.

You are using safe texture cache which is not needed for Resident Evil 4, native mips is also not needed and these options may be producing this undesired effect on performance. Disable idle skipping, while it will generally improve speed this option will also cause some issues which could affect speed, especially with the intro movie and ending credits. Resident Evil 4 is also coded to be locked at 30 FPS ingame and 60 FPS for cutscenes and movies, 30 FPS NTSC/25 FPS Pal is full speed for this game and you should be using the frame limiter. I also wouldn't recommend enabling both audio throttle and the frame limiter, enabling one or the other should be sufficient since they both accomplish the same task.

I also recommend an actual aspect ratio instead of stretch to window, depending on your monitor you may want to try 4:3 (Standard) or 16:9 (Widescreen).

Keep in mind that some areas will have slowdowns because there is more to render in the scene and due to bottlenecks with Dolphin itself. Even my Core2Duo E8500, Geforce 9800 GT, 4GB of Ram has trouble keeping up with this game. I'll have framerates dip down into the teens or mid 20's and the framerate will generally fluctuate, this has been an issue with Dolphin for a while now and has no relation to the past 10 - 15 commits, it's very difficult although not impossible to maintain full speed with this game and that's it.

And as it has already been stated, test multiple times under the same conditions while spreading the benchmarks apart and use a proper configuration. I've already pointed out issues with your configuration that are likely affecting performance with your hardware.

[qlum]

I have the same with ark rise fantasia the 2.0 version gave about 8fps more than the latest reversion and changing the graphics didn't help a single fps I use a hd 5770 , a phenom2 x4 2.6 ghz and 6gb ddr3 ram on windows 7 x64 yes I use the x64build opencl didn't change much nether did switching from dx9 to 11.

[hyperspeed]

I agree with finalmars.

But I think the loss of fps is because they corrected the gfx error.

   
   

[Xtreme2damax]

(07-30-2010, 09:16 AM)qlum Wrote: I have the same with ark rise fantasia the 2.0 version gave about 8fps more than the latest reversion and changing the graphics didn't help a single fps I use a hd 5770 , a phenom2 x4 2.6 ghz and 6gb ddr3 ram on windows 7 x64 yes I use the x64build opencl didn't change much nether did switching from dx9 to 11.

Again as I mentioned above, this thread and the other thread is referring to the last 10 - 15 revisions only, which were optimizations. It's highly unlikely these commits would have decreased speed that much. So far he is the only member reporting performance issues with the past 10 - 15 revisions.

The speed decrease you are referring to is due to improved accuracy and issues being fixed, this is a good thing even if it decreases speed temporarily. All you can do is wait for the developers to optimize the emulator, it will be well worth it in the end.

[skid]

(07-30-2010, 09:34 AM)hyperspeed Wrote: But I think the loss of fps is because they corrected the gfx error.

Yes, if I remember correctly, it was the mario's eyes fix in SMG that slowed the graphics down since 2.0. I remember Tino saying that the video cards in PC's are not efficient in making the calculations needed for the fix. However, I think we'd prefer graphics accuracy over speed.