First I'd like to apologize for any redundant information, possibly even including this post, and my lack of an ability to describe things.
I'm just posting this as I found some setting combinations produced interesting results on my hardware setup and hope it's helpful to someone in some way.
TLR at bottom because I'm horrible at communication and people get confused or impatient.
CPU: E5200 @ 3.6ghz (11x333fsb)
GPU: ATI 4830, running 4850 timings. Video card speed had no impact on framerate and game speed.
Game: Dragon Warrior VIII (SLUS_212.07) .iso format
OS: Vista X64
PCSX2 version(s): 0.9.6 & Beta r1888
GSdx versions: 1873, 1846, 1650, 1448, 1406, 1398, 1322 (SSSE3 builds, DX10 setting)
Never maxed out CPU usage (Windows highest=91%, pcsx2 highest=95%%, neither at the same time.)
Restarted pcsx2 and loaded from memory card each time I changed a setting, due to the random fps loss that would pop up
-----
Bugs & Glitches
Missing shadows on player and npcs (Presently intended behavior due to post-processing effects?)
Viewable: Software has it, alas broken and off center.
Also reported in another post
Cell shade outline. Like the difference between a mechanical pencil and ball-point pen line. Hardware outline can be a little to thick and too sharp at some viewing angles. I'm just assuming this issue is also somehow tied with the shadows.
Viewable: Please refer to last 4 screenshots..
Artifacts from increased internal resolution and/or filtering: well known with screenshots from many different games. Repost of many others, mabey more screenshots can help, mabey not.
Viewable A: Hardware filtering artifacts, very similar to the texture offsets Antialiasing can create with PC games. Easily fixable with a pixel offset. See moneyshot in 1st screenshot (middle).
Viewable B: Horizontal & Vertical resolution artifacts from increasing internal resolution. See moneyshot in 1st screenshot (left).
Flashing snapshot image thingy - something to do with an image buffer mabey? I'm not to sure on how ps2 gpu creates, orders and processes frames/framebuffer/FB effects.
Reproduceable: Play through the game until you can leave town (Faebury) and fight. Enter a fight and win. At the XP tally screen, as the camera moves up and around, you will get two to three flashes of the same frame snapshot of your party "entering" battle, before the battle swirl started.
The same snapshot (does not change until pcsx2 is reset) is also flashed during entering and exiting doors/zones, but less noticable and sometimes doesn't show.
*sometimes the snapshot will change after a period of gameplay; can also be a snapshot of when the "combat zone" begins/ends but this is less common from what I've encountered*
On a fresh pcsx2 process, here's my action process.
Game Initalized>Load from Memory Card>Left town>Open Menu: Yangus Casts Whistle(starts a random battle)> Battle swirl>screen flashes black for one frame>Fight screen loads>killed enemys>Left battle>Open Menu: Yangus Casts Whistle>Battle swirl>THIS SCREENSHOT FLASHES FOR ONE FRAME>Fight screen loads>ect.
-----
EMU ISSUE(?): After entering and resuming a game >6 times (6 is the lowest I've experienced this) out of nowhere there is a huge drop in fps requiring a program restart to get normal emu speed. Annoying when your going back and forth between changing settings or reading a walkthrough.
Less important settings
Interlacing: no real difference, excluding bob, which well... bobs every now and then.
microVU0 & microVU1: 48-50 fps with and without the microvu speedhacks enabled, paired or seperate. Other speed hacks in any combination also proved to have no extra benefit in relation of performance of VU#rec:fps:speed
Advanced Options: did NOT mess with any of these.
Game Fixes: None, assumed they worked only on the games mentioned in the description.
Software mode: Similar effect with hardware, but at sub-40 framerate @55-65% game speed
ZZogl & ZeroGL: Are these really needed now for a windows based report? Because they're 10-13 fps with a plethora of texture and model issues, ZeroGL being the worse off with completely missing model parts ontop of the ZZogl issues.
Other Plugins: using revisions that came with pcsx2-r1888. SPU2-x, LilyPad, Linuz Iso CDVD. All settings are default (obviously not for LilyPad) @SPU2-x: checked=Disable Time-Stretching; Enable Audio Expansion.
My Tinkering Location
Start a new game and go to the first town, Faebury. Play through until you can leave town and fight, this changes NPC positions in town, should take 5 minutes tops. Enter town during the day from the Southwest gate, turn camera to look to your left about 20-30° (tap L1 once), and there's a nice heavier-than-normal load to tweak with.
Another location in the same town during the day is to enter and leave the Church and take a step or two foward. This location seems to have a slightly higher workload.
Gameplay speed vs FPS *Context: Limit fps setting (no skipping);enter town from SW gate, tap L1 once, 52fps without speedhacks*
While trying to replicate the game speed one would get on a ps2 on pcsx2, I've come to notice three things...
A) Framerate, without speedhacks, is the speed of the game outside of any slowdowns you'd encounter on actual ps2 hardware.
B) Speedhacks don't really make the game faster, despite what the fps counter says. It does, however, help keep the audio from skipping as its timing is tied to the framerate with Time-streching disabled.
C) Last two ticks of "VU Cycle Stealing" simulate point A's actual gameplay speed at whatever fps it would get, with an added 10-25% speed decrease. (this may not be true for everyone else on this game in the same location) *See below for a better description*
Now allow me to elaborate a bit.
I tried a combination of all gpu, cpu and speedhack settings in hope of finding one that worked better than the rest. I repeated this for various gpu plugin revisions, and repeated both of those trials with pcsx2 0.9.6.
My results were the same, plus or minus 1-2 fps between gpu plugin versions. Excluding 1448, all mentioned GSdx revisions had 51-53 fps range, most recent averaging out to 53 constant and dipping low every now and then.
Besides checking for framerate gains or loss, I also moved about 5-10 steps in various directions numerous times to check for animation speed fluctuations (i.e. checking for emulated ps2 frameskipping or slowdowns).
I would then enable, one by one, combination by combination, the speedhacks checking again for framerate gain/loss and overall game speed improvement.
Noticing that the framerate was improving with some speedhacks I thought I was making progress, as music and sounds were no longer skipping, until I returned to town only to see myself move about at the same speed as without the hacks. Clearly the game hasnt increased in speed, just framerate (allowing better syncing music & sounds).
Boggled as I was, I tried a even more combinations, with the "VU Cycle Stealing" changes as the last of the bunch I tinkered with.
"VU Cycle Stealing" Moderate with 1.5x or 2x EE sync, -AND- Large & Very Large (all EE sync settings) gave the game a "slow motion" effect even though the framerate was still 60.
To better explain this, lets use the points I made above as the example.
With no hacks involved fps=53 & game speed ~90%. With hacks, fps is 60 & speed is still ~90% (but without skipping music & sounds). With the more advanced VU steals, fps is still 60, but speed is ~55-65% now (music/sounds still a-ok).
Any place in the game where I would get 60 fps without speed hacks its fine, but if I got just even 59 fps w/o speed hacks it would be susceptible to the wierd slow motion effect.
So instead of 59 fps at 98% speed, I would wind up with 60 fps at %60-70% speed. Sorry if that dosen't make any sence.
Note of interest: my machine is usually running at 3.8ghz (11.5x333fsb) which is Folding@home, OCCT, & Orthos (Small/Large FFT's and Blend) stable @42h each. Pcsx2 crashes after a while, but it was long enough to let me test well enough.
@3.8ghz, no speedhacks, I get 60fps at the town gate, and 57-59 at the church, which leads me to rule out any emulated frameskips from the locations I chose to test in as game speed was as it should be given framerate difference.
Screens@imageshack
[Image: dqviiiscreens.th.jpg]
Sorry about the bottom text, dunno how it ended up like that (it was placed on the same layer as the rest of the inverted text). If you can't read it, it says "Funky Shadows".
[Image: hwuf.th.jpg][Image: hwf.th.jpg][Image: hwn.th.jpg][Image: 52833450.th.jpg]
Left to Right: Unfiltered Hardware, Filtered Hardware, Hardware Native box checked, Software
512x512 internal, scaled to HDTV resolution
TLR:
Bugs as stated above.
Wierd behavior with framerate and its relationship with game speed in action and not on paper.
Speed hacks, while being what they are and basically unsupported(?) have basically zero benefit, and only at extreme settings do they have a negative effect.
Screenshots and Wall o' Text in hope that it can be of use.
Sorry for the wall o' text post, tried to include all details that might be of some use.
Thank you for your work!
I'm just posting this as I found some setting combinations produced interesting results on my hardware setup and hope it's helpful to someone in some way.
TLR at bottom because I'm horrible at communication and people get confused or impatient.
CPU: E5200 @ 3.6ghz (11x333fsb)
GPU: ATI 4830, running 4850 timings. Video card speed had no impact on framerate and game speed.
Game: Dragon Warrior VIII (SLUS_212.07) .iso format
OS: Vista X64
PCSX2 version(s): 0.9.6 & Beta r1888
GSdx versions: 1873, 1846, 1650, 1448, 1406, 1398, 1322 (SSSE3 builds, DX10 setting)
Never maxed out CPU usage (Windows highest=91%, pcsx2 highest=95%%, neither at the same time.)
Restarted pcsx2 and loaded from memory card each time I changed a setting, due to the random fps loss that would pop up
-----
Bugs & Glitches
Missing shadows on player and npcs (Presently intended behavior due to post-processing effects?)
Viewable: Software has it, alas broken and off center.
Also reported in another post
Cell shade outline. Like the difference between a mechanical pencil and ball-point pen line. Hardware outline can be a little to thick and too sharp at some viewing angles. I'm just assuming this issue is also somehow tied with the shadows.
Viewable: Please refer to last 4 screenshots..
Artifacts from increased internal resolution and/or filtering: well known with screenshots from many different games. Repost of many others, mabey more screenshots can help, mabey not.
Viewable A: Hardware filtering artifacts, very similar to the texture offsets Antialiasing can create with PC games. Easily fixable with a pixel offset. See moneyshot in 1st screenshot (middle).
Viewable B: Horizontal & Vertical resolution artifacts from increasing internal resolution. See moneyshot in 1st screenshot (left).
Flashing snapshot image thingy - something to do with an image buffer mabey? I'm not to sure on how ps2 gpu creates, orders and processes frames/framebuffer/FB effects.
Reproduceable: Play through the game until you can leave town (Faebury) and fight. Enter a fight and win. At the XP tally screen, as the camera moves up and around, you will get two to three flashes of the same frame snapshot of your party "entering" battle, before the battle swirl started.
The same snapshot (does not change until pcsx2 is reset) is also flashed during entering and exiting doors/zones, but less noticable and sometimes doesn't show.
*sometimes the snapshot will change after a period of gameplay; can also be a snapshot of when the "combat zone" begins/ends but this is less common from what I've encountered*
On a fresh pcsx2 process, here's my action process.
Game Initalized>Load from Memory Card>Left town>Open Menu: Yangus Casts Whistle(starts a random battle)> Battle swirl>screen flashes black for one frame>Fight screen loads>killed enemys>Left battle>Open Menu: Yangus Casts Whistle>Battle swirl>THIS SCREENSHOT FLASHES FOR ONE FRAME>Fight screen loads>ect.
-----
EMU ISSUE(?): After entering and resuming a game >6 times (6 is the lowest I've experienced this) out of nowhere there is a huge drop in fps requiring a program restart to get normal emu speed. Annoying when your going back and forth between changing settings or reading a walkthrough.
Less important settings
Interlacing: no real difference, excluding bob, which well... bobs every now and then.
microVU0 & microVU1: 48-50 fps with and without the microvu speedhacks enabled, paired or seperate. Other speed hacks in any combination also proved to have no extra benefit in relation of performance of VU#rec:fps:speed
Advanced Options: did NOT mess with any of these.
Game Fixes: None, assumed they worked only on the games mentioned in the description.
Software mode: Similar effect with hardware, but at sub-40 framerate @55-65% game speed
ZZogl & ZeroGL: Are these really needed now for a windows based report? Because they're 10-13 fps with a plethora of texture and model issues, ZeroGL being the worse off with completely missing model parts ontop of the ZZogl issues.
Other Plugins: using revisions that came with pcsx2-r1888. SPU2-x, LilyPad, Linuz Iso CDVD. All settings are default (obviously not for LilyPad) @SPU2-x: checked=Disable Time-Stretching; Enable Audio Expansion.
My Tinkering Location
Start a new game and go to the first town, Faebury. Play through until you can leave town and fight, this changes NPC positions in town, should take 5 minutes tops. Enter town during the day from the Southwest gate, turn camera to look to your left about 20-30° (tap L1 once), and there's a nice heavier-than-normal load to tweak with.
Another location in the same town during the day is to enter and leave the Church and take a step or two foward. This location seems to have a slightly higher workload.
Gameplay speed vs FPS *Context: Limit fps setting (no skipping);enter town from SW gate, tap L1 once, 52fps without speedhacks*
While trying to replicate the game speed one would get on a ps2 on pcsx2, I've come to notice three things...
- [A] Framerate, without speedhacks, is the speed of the game outside of any slowdowns you'd encounter on actual ps2 hardware.
[B] Speedhacks don't really make the game faster, despite what the fps counter says. It does, however, help keep the audio from skipping as its timing is tied to the framerate with Time-streching disabled.
[C] Last two ticks of "VU Cycle Stealing" simulate point A's actual gameplay speed at whatever fps it would get, with an added 10-25% speed decrease. (this may not be true for everyone else on this game in the same location) *See below for a better description*
A) Framerate, without speedhacks, is the speed of the game outside of any slowdowns you'd encounter on actual ps2 hardware.
B) Speedhacks don't really make the game faster, despite what the fps counter says. It does, however, help keep the audio from skipping as its timing is tied to the framerate with Time-streching disabled.
C) Last two ticks of "VU Cycle Stealing" simulate point A's actual gameplay speed at whatever fps it would get, with an added 10-25% speed decrease. (this may not be true for everyone else on this game in the same location) *See below for a better description*
Now allow me to elaborate a bit.
I tried a combination of all gpu, cpu and speedhack settings in hope of finding one that worked better than the rest. I repeated this for various gpu plugin revisions, and repeated both of those trials with pcsx2 0.9.6.
My results were the same, plus or minus 1-2 fps between gpu plugin versions. Excluding 1448, all mentioned GSdx revisions had 51-53 fps range, most recent averaging out to 53 constant and dipping low every now and then.
Besides checking for framerate gains or loss, I also moved about 5-10 steps in various directions numerous times to check for animation speed fluctuations (i.e. checking for emulated ps2 frameskipping or slowdowns).
I would then enable, one by one, combination by combination, the speedhacks checking again for framerate gain/loss and overall game speed improvement.
Noticing that the framerate was improving with some speedhacks I thought I was making progress, as music and sounds were no longer skipping, until I returned to town only to see myself move about at the same speed as without the hacks. Clearly the game hasnt increased in speed, just framerate (allowing better syncing music & sounds).
Boggled as I was, I tried a even more combinations, with the "VU Cycle Stealing" changes as the last of the bunch I tinkered with.
"VU Cycle Stealing" Moderate with 1.5x or 2x EE sync, -AND- Large & Very Large (all EE sync settings) gave the game a "slow motion" effect even though the framerate was still 60.
To better explain this, lets use the points I made above as the example.
With no hacks involved fps=53 & game speed ~90%. With hacks, fps is 60 & speed is still ~90% (but without skipping music & sounds). With the more advanced VU steals, fps is still 60, but speed is ~55-65% now (music/sounds still a-ok).
Any place in the game where I would get 60 fps without speed hacks its fine, but if I got just even 59 fps w/o speed hacks it would be susceptible to the wierd slow motion effect.
So instead of 59 fps at 98% speed, I would wind up with 60 fps at %60-70% speed. Sorry if that dosen't make any sence.
Note of interest: my machine is usually running at 3.8ghz (11.5x333fsb) which is Folding@home, OCCT, & Orthos (Small/Large FFT's and Blend) stable @42h each. Pcsx2 crashes after a while, but it was long enough to let me test well enough.
@3.8ghz, no speedhacks, I get 60fps at the town gate, and 57-59 at the church, which leads me to rule out any emulated frameskips from the locations I chose to test in as game speed was as it should be given framerate difference.
Screens@imageshack
[Image: dqviiiscreens.th.jpg]
Sorry about the bottom text, dunno how it ended up like that (it was placed on the same layer as the rest of the inverted text). If you can't read it, it says "Funky Shadows".
[Image: hwuf.th.jpg][Image: hwf.th.jpg][Image: hwn.th.jpg][Image: 52833450.th.jpg]
Left to Right: Unfiltered Hardware, Filtered Hardware, Hardware Native box checked, Software
512x512 internal, scaled to HDTV resolution
TLR:
Bugs as stated above.
Wierd behavior with framerate and its relationship with game speed in action and not on paper.
Speed hacks, while being what they are and basically unsupported(?) have basically zero benefit, and only at extreme settings do they have a negative effect.
Screenshots and Wall o' Text in hope that it can be of use.
Sorry for the wall o' text post, tried to include all details that might be of some use.
Thank you for your work!