WinUAE 2.4.1 beta series

[Toni Wilen]

Can anyone duplicate following:

2.4.0 or 2.4.1b1: fastest possible CPU + low latency vsync works as expected.
2.4.1b2 and later: SND% is not stable at all and for example sysinfo shows much slower results. (even without sound enabled)

I can't see anything obvious except d3d texture size changes (to smaller size!) that shouldn't have any effect.

EDIT: I think I found the problem (This probably didn't always happen, depending on CPU and GPU speed)

[Mequa]

2.4.1 looks like an excellent update. I noticed a lot less slowdown than 2.4.0 when using new Vsync with fastest possible CPU in both windowed and fullscreen modes.

One bug I noticed in 2.4.1 beta compared with 2.4.0:
When using resolution autoswitch option with Exile AGA (possibly other games too), the screen keeps flashing black and everything slows to a crawl - presumably it is constantly trying to switch the resolution back and forth. This bug only occurs in PAL mode, NTSC works fine.

[OSH]

Toni you are one of the few people, who works with real passion, not for money...

[Mequa]

I tried the latest 2.4.1 beta with Cloanto's Amiga Forever 2012 - they still haven't officially updated from 2.3.3. Perhaps they had some issues with 2.4.0 which 2.4.1 may address?

I didn't notice any issues using the beta with Cloanto's RetroPlatform, however I noticed the horizontal centring is slightly off with Amiga Forever's setups.
I'm not sure if this is due to changed configuration format or a bug, but I'm sure Cloanto will appreciate a fix if it is the latter case.

Edit: Cloanto uses manual positioning for most setups - the A1200 one shown has horizontal position set to 260. Changing this to 292 fixes it with WinUAE 2.4.1 - not sure if this is Cloanto's job though, but it seems there has been a change, and it would be nice for them to 'officially' upgrade to 2.4.x series...

P.S. Just tested, this issue does not occur in 2.4.0 (that is positioned the same as Cloanto's "official" 2.3.3), so was introduced in 2.4.1 beta series (i.e. this is on the correct thread!)

[Toni Wilen]

Originally Posted by Mequa

2.4.1 looks like an excellent update. I noticed a lot less slowdown than 2.4.0 when using new Vsync with fastest possible CPU in both windowed and fullscreen modes.

This still needs some tweaking (probably forever..)

One bug I noticed in 2.4.1 beta compared with 2.4.0:
When using resolution autoswitch option with Exile AGA (possibly other games too), the screen keeps flashing black and everything slows to a crawl - presumably it is constantly trying to switch the resolution back and forth. This bug only occurs in PAL mode, NTSC works fine.

Not a new bug but something that wasn't as easily visible previously. (Again..)

Originally Posted by OSH

Toni you are one of the few people, who works with real passion, not for money...

Perhaps but sometimes I am not sure if it is worth the trouble...

Originally Posted by Mequa

I tried the latest 2.4.1 beta with Cloanto's Amiga Forever 2012

Don't do it. Only those that are officially released with AF are supported and I ignore everything related to AF unless it comes directly from Cloanto. (AF requires some changes, 2.4.1 won't be directly compatible, it is being worked on at the moment)

[lifeschool]

Originally Posted by Toni Wilen

Perhaps but sometimes I am not sure if it is worth the trouble...

I love the time and trouble you go to to support all us Amigans! I'm sure I'm not the only one who is eternally grateful for WinUAE since I had to sell my old A1200's a few years back. The package has certainly (almost single handedly) has kept the Amiga dream alive all these years, and without it sites like this wouldn't be anywhere near as busy.

If it was about the money I'd say why not do a Kickstarter project for V2.5 - they absolutely throw money at these projects and every Joe is running a kickstarter nowadays it seems - but I won't insult your intelligence by mentioning money again.

Perhaps what would be helpful to you would be to look forward to some kind of 'final' release. I know that sounds impossible at the moment because there are so many new features you've crammed into the package and all need fixing and sorting out. But I can't help feeling that working towards some kind of 'final' release would give you a goal - a light at the end of the tunnel - and when it's done you can sit back and relax a few months/years and reap the rewards.

Otherwise you could keep on adding features and bug fixing and you may burn yourself out and start to resent the project - and I'm sure nobody wants to see that happen - not after all these years you've put into this.

[lilalurl]

Sorry for the off-topic but:

Originally Posted by lifeschool

If it was about the money I'd say why not do a Kickstarter project for V2.5 - they absolutely throw money at these projects and every Joe is running a kickstarter nowadays it seems - but I won't insult your intelligence by mentioning money again.

Anyone can donate through the winuae website (link just above News):
http://www.winuae.net/

Not sure how much Paypal takes for fees, but it will probably be less than Kickstarter:

What fees does Kickstarter charge?

Kickstarter collects a 5% fee from the project’s funding total if and only if a project is successfully funded. Amazon (our payments processor) also charges credit card processing fees that generally work out to 3-5%.

Also, unless there is some other clause for non-US citizens:

Am I eligible to start a Kickstarter project?

To be eligible to start a Kickstarter project, you need to satisfy the requirements of Amazon Payments:
Be a US resident and at least 18 years of age with a social security number (or EIN), a US bank account, US address, US state-issued ID (driver’s license), and major US credit or debit card.
Please note that anyone, anywhere (with a major credit card) can pledge to Kickstarter projects.
We’re working hard to open up to more countries. If you’ve been waiting, we really appreciate your patience.

http://www.kickstarter.com/start

[Toni Wilen]

offtopic: I'll move above posts (and this) to some other thread if people seem interested enough.

Seriously, I'll do this as long as it is interesting, money won't buy real interest anyway.. I think the only thing that requires some kind of bounty or something is: CD32 FMV cartridge emulation. It is mostly useless and it isn't trivial at all. But it would be fun.

[FrodeSolheim]

Originally Posted by Toni Wilen

Seriously, I'll do this as long as it is interesting, money won't buy real interest anyway..

No, I don't think money will help in that regard, but I made a donation anyway, as a token of appreciation for your stewardship of UAE for the last few years

[lifeschool]

^ I just donated too. Great idea Frode. Just wanted to share my support. Sorry to use up blog space. Please move these comments to 'WinUAE Appeciation Society' or something.

[Toni Wilen]

http://www.winuae.net/files/b/winuae_2410b7.zip

Beta 7:

- DirectDraw full-window mode on non-primary monitor was blank.
- Adjusted (again) fastest possible CPU mode extra CPU time scheduling, trying to find balance between better timing syncronization and not causing unnecessary slowdowns on some systems..
- OCS Denise bug emulation update, PAL STREQU counts are 8 (short field) or 9 (long field) lines, not static 9 lines. (Probably no program cares)
- F11 in gameports/input test mode shows next page (if list is longer than visible area).
- Manual display positioning X coordinate offset fixed (Filter panel or gfx_center_horizontal_position) Origin should be hardware position 0 (same as sprites), not first possible visible position.
- Fastest possible low latency vsync should work better again (b2), plus some other adjustments done, CPU throttling supported.
- Autoresolution worked badly in some situations.
- Removed USB_DEVICE_ID_RETRO_ADAPTER from USB "quirks" list, apparently it is working correctly according to firmware sources. (Perhaps older version was wrong, not exactly sure why it was included in Linux HID quirks list)
- Save also selected monitor name (gfx_display_name) to config file because order of displays can change when removing or adding monitors or replacing display cards.

(and thanks to 2 posters above)

using d3d filtershaders with this new beta 7, cause crash

[Mequa]

Originally Posted by gicgioc

using d3d filtershaders with this new beta 7, cause crash

I get the same issue.

[rsn8887]

It is a bit hard to describe, but b7 broke at least the game Hyperdrive for me.

In b6, A1200 fastest possible low latency vsync, the game runs perfectly.

In b7 the player space ship sprites flickers and moves much more jerkily and flickery. Happens both in Direct3D and Directdraw modes.

Config and log files attached.

Since the graphical effect is hard to describe, I uploaded movies of the flickery b7 gameplay here:
www.coolatoms.org/output.avi (~6 MB)
And a movie of the 100% correct b6 gameplay here:
www.coolatoms.org/output_b6.avi (~6 MB)

I think I have also seen other occurences of flickering in beta 7 during regular WB usage and screen flipping etc but that seems harder to reproduce.

[Toni Wilen]

Originally Posted by gicgioc

using d3d filtershaders with this new beta 7, cause crash

It does..

Originally Posted by rsn8887

It is a bit hard to describe, but b7 broke at least the game Hyperdrive for me.

In b6, A1200 fastest possible low latency vsync, the game runs perfectly.

Unfortunately this is yet another victim of fastest possible CPU mode updates. It was never meant to used with games that depend on timing too much but unfortunately old one was quite bad which made it too "compatible" and here is the result

Now even tiny adjustment can change behavior of some games..

[Toni Wilen]

http://www.winuae.net/files/b/winuae_2410b8.zip

Beta 8:

- 68EC020/68020 state files didn't load properly if current CPU config didn't match.
- Adjusted JIT timing a bit, possible fix for slowdowns on some systems.
- Made sound sync a bit more aggressive. (2.4.0 relaxed it a bit too much, I think)
- D3D crash fixed (b7)
- CIA related rare statefile CPU TRACE error fixed.
- Crash fixed when CDFS CD drives enabled without any harddrives.
- Disable USB HID RAWINPUT (Fallback to Direct Input) if "blacklisted" devices found, added RM-203 USB adapter to list because it requires calibration which WinUAE can't support.

[Toni Wilen]

http://www.winuae.net/files/b/winuae_2410b9.zip

Beta 9: (quick fix because this can be very annoying)

- Fixed D3D texture reallocation ordering. Caused blank screen or other strange side-effects (b7)

[rsn8887]

I think I can make a proper statement about mouse pointer motion now:

For me, in A1200 fastest possible modes, mouse pointer motion is smooth in b6, and jerky in b7+ (b7-b9).

Using fastest possible A1200 low latency vsync mode. For log files and config, see my last post (making the hyperdrive tests led me to notice the effect).

I tried to record a avi video of it, but in the video mouse pointer always moves smooth and very fast?!?

The jerkyness goes away completely when switching from fastest possible to approximate A500/A1200 speed.

Did something that was introduced in b7 mess with sprite movement smoothness? I do notice similar jerkyness in the hyperdrive space ship motion. That jerkyness didn't show up in the videos I posted though (only the flickering shows up in the avi, not the jerky motion for some reason).

EDIT: In order to test I simply booted into a cli window. Turning sound on/off makes no difference. Direct 3D/Directdraw, no difference. CPU idle, no difference. Scanlines/Double, no difference. It maybe seems better with neither doubling nor scanlines, but the screen is very squished making it hard to tell. USB mode on/off, no difference.

Quick way to tell: In b7+, the mouse pointer also jumps up and down vigorously when dragging the workbench screen up and down. In b6 it moves smoothly. It doesn't quite stay on the screen title bar, buit doesn't jump. When I turn off low latency vsync, the motion in b7+ is much smoother, maybe even as smooth as in b6 with low latency vsync on.

EDIT 2: Line/Double etc. does NOT make a difference. However, I can make the mouse pointer in b6 similarly jerky when I set CPU speed to -10% or lower. It becomes very jerky then. Only CPU speed 0% works perfectly in b6 (the default). In b7 even cpu speed 0% is jittery. In both b6 and b7, setting CPU speed to anything else then 0% also messes with the fps counter more or less (jumps to 68 or so fps and constantly changes, even though it should be flat 50).

[Toni Wilen]

Originally Posted by rsn8887

For me, in A1200 fastest possible modes, mouse pointer motion is smooth in b6, and jerky in b7-b9.

Test using some demo with horizontal scrolling, the bigger the scroller the better. It is much better vsync test. (Any jump is very easily noticeable and easier to duplicate than manual mouse movement)

Also always test all buffer modes (no, double, triple)

EDIT REPLY: I meant display panel display buffering mode (Display panel, top), not line doubling modes.

[rsn8887]

Horizontal scrolling is perfectly smooth in any case. I see this jerky motion effect only with mouse pointer (and space ship in Hyperdrive but might not be related).

Selected buffer modes (no double triple) make no difference. Switching low latency vsync off makes mouse pointer smooth again.

Horizontal scrolling does become jerky however when setting CPU speed to anything less than 0% (logs for setting of -10% attached)

[rsn8887]

Interesting side note:
In "low latency vsync" and "no vsync", the mouse pointer does not stay on same point in title bar when dragging the wb screen. In first case it is jumpy, in second case it is smooth but still moves a bit different than screen.

However, when using "legacy vsync" the pointer DOES stay on same pixel in title bar and drags perfectly together with the screen when moving up down. AFAIK that is how the real amiga behaved.

[Toni Wilen]

Originally Posted by rsn8887

Horizontal scrolling is perfectly smooth in any case. I see this jerky motion effect only with mouse pointer (and space ship in Hyperdrive but might not be related).

Selected buffer modes (no double triple) make no difference. Switching low latency vsync off makes mouse pointer smooth again.

Hmm... This is really strange. Does it move smoothly if you use keyboard to move mouse (Gameports panel, select some keyboard layout, switch type to "mouse" from "default").

[rsn8887]

Originally Posted by Toni Wilen

Hmm... This is really strange. Does it move smoothly if you use keyboard to move mouse (Gameports panel, select some keyboard layout, switch type to "mouse" from "default").

Same when using keyboard (but easier to see this way). B6 CPU 0% perfectly smooth. b6 CPU -10% jerky. b9 jerky regardless of CPU %

Please also see above (I edited post to include logs).

[Toni Wilen]

Does it also happen if you start from scratch? Start from (for example) A1200 quickstart, set fastest possible and vsync.

There must be some hidden variable..

[rsn8887]

Originally Posted by Toni Wilen

Does it also happen if you start from scratch? Start from (for example) A1200 quickstart, set fastest possible and vsync.

There must be some hidden variable..

It also happens when I start from scratch:

Started with A1200 quickstart, chose Workbench 3.0 disk, pressed "set configuration". Then Display - Fullscreen, 100 Hz PAL, no buffering (but doesn't matter), low latency vsync, and changed game port one to keyboard with mouse type.

Did the above both for B6 and B9.

Same behaviour. B6 not stuttering, B9 pointer stuttering. Resulting configs (I saved them afterwards) and logs attached.

Actually buffering does seem to change stutter frequency (more stutter at no buffer than triple buffer can be seen only using keyboard mouse), but doesn't make it go away.

EDIT: Tried 60 Hz too, same behaviour.

[Toni Wilen]

Originally Posted by rsn8887

Started with A1200 quickstart, chose Workbench 3.0 disk, pressed "set configuration". Then Display - Fullscreen, 100 Hz PAL, no buffering (but doesn't matter), low latency vsync, and changed game port one to keyboard with mouse type.

Same also in 50Hz and 60Hz modes? (Still can't see anything wrong in 60Hz or 100Hz low latency vsync modes and it does make no sense have perfectly smooth scrolling but mouse sprite working badly..)

Same also in JIT mode? There has to be some option that changes the behavior. There just has to be..

EDIT: what about sound panel automatic switching? Does it work if you untick it?

EDIT2: I can duplicate it (FPS jumping wildly) if sound is disabled. Finally something..

[rsn8887]

Yes same in JIT mode. I forgot I was using portaudio, not directsound. No change when switching sound system or automatic switching. Same problem in 60 Hz. My monitor doesn't support 50 Hz

Interesting: In b9, if I go to "approximate A1200 speed" where mouse pointer is smooth, I can then adjust the CPU speed anywhere, even up to the max of +500%and mouse pointer stays smooth. But as soon as I select "Fastest possible" however, mouse pointer starts to stutter and doesn't stop even if I go down in CPU % then...

[Toni Wilen]

Does http://www.winuae.net/files/b/winuae.zip work better? At least it fixes my no-sound + non-0% CPU FPS issue.

[rsn8887]

FPS issue for CPU <0% fixed, but mouse pointer still jerky. Now I was able to test <0% CPU speed.

Result: Mouse pointer becomes smooth if I reduce CPU speed from max (0%) to -70% or lower (this corresponds to approx 34% CPU LED number in emulation). This is in fastest possible mode. In "approximate A1200 speed" it is still always smooth (and the fastest setting of 500% corresponds to approx 40% CPU LED number). Tested with Quickstart A1200 as before.

EDIT: It still stutters every couple seconds at -70%, but at -80% it is perfectly smooth (~30% CPU LED).

[msayed1977]

Your beta files became at smaller sizes, as the non-beta fixes.
At least be sure your final releases is CPU-optimized.

[Toni Wilen]

Originally Posted by rsn8887

FPS issue for CPU <0% fixed, but mouse pointer still jerky. Now I was able to test <0% CPU speed.

winuae.zip updated again, fastest possible low latency vsync vsync detection rewritten partially, hopefully it is a bit better now..

Originally Posted by msayed1977

Your beta files became at smaller sizes, as the non-beta fixes.
At least be sure your final releases is CPU-optimized.

Official betas and finals have full optimization (=much longer compilation time). Non-official betas only have partial optimization enabled.

[rsn8887]

Hi Toni, the mouse pointer is still stuttering in latest winuae.zip.

I finally managed to get the effect onto video. It was a question of using a good, fast codec. The video "smooth_approximate_speed.avi" shows the smooth behaviour when "approx. speed" is selected. The video "stuttery_fastest_possible.avi" shows the jerky pointer when "fastest possible" is selected.

Toni, it would be awesome if you could take a look at the attached avis!
The filesize of the videos is only ~400 kb. The videos are tinted green for some reason but show the effect. They play with the freeware player program "mediaplayer classic" http://mpc-hc.sourceforge.net/ (I am not sure if they play on all machines with the official windows mediaplayer due to codec used).

I just turned on JIT and off again and with JIT on, the stuttering is much less (one stutter every 10 seconds or so). The effect is very clear when using keyboard mouse.

I think since scrolling etc. is all smooth, tested with Turrican 2, it must be something in the timing between b6 to b7 that affects only the mouse pointer (maybe other things that I haven't found yet), but not the overall vsync in general.

EDIT: JIT is NOT affecting it in 60 Hz mode, only in 100 Hz mode.

[Toni Wilen]

Did you try both no buffer, double and triple buffered modes? This can affect vsync behavior a lot, especially no vs double.

EDIT: I can see something similar if 100Hz and dragging screen. But only if 100Hz or 120Hz.

[pmc]

I can confirm what you say there Toni.

I had big screen and sound slowdowns since WinUAE 2.4.0

I realised that if I changed triple buffering down to no buffering everything then worked fine - now, on that setting, I have no issues with slowdowns at all.

[rsn8887]

Originally Posted by Toni Wilen

Did you try both no buffer, double and triple buffered modes? This can affect vsync behavior a lot, especially no vs double.

EDIT: I can see something similar if 100Hz and dragging screen. But only if 100Hz or 120Hz.

It happens in no buffer, double buffer and triple buffer in b9. I tested again with 60 Hz. It happens in 60 Hz NTSC for me, too. It is actually more visible here in 60 Hz. It looks like, during mouse pointer motion, every once in a while a ghost image of the pointer from the past flickers up behind the real pointer. I would say it is somewhat of a flickering effect in 60 Hz.

The "pointer jumping while screen dragging effect" might be related, but seems a bit different. These are myh test results using 60 Hz low latency Vsync (whether I use 50/60 vsync option or not seems to make no differenct)

BETA 6

Fastest possible:

no buffering - motion smooth, pointer lags behind when dragging screen but no jumping

double buffering - motion slightly jerky/flickery (see above discussion), screen dragging pointer position seems perfect

triple buffering - motion jerky/flickery, screen dragging pointer position seems perfect

Approx 1200 speed:

regardless of buffering - motion smooth, pointer jumps when dragging screen


BETA 9 (latest winuae.zip):

Fastest possible:

regardless of buffering - motion jerky/flickery, pointer jumps when dragging screen.

Approx 1200 speed
regardless of buffering - motion smooth, pointer jumps when dragging screen

[Toni Wilen]

Another winuae.zip update..

I found more corner cases that weren't handled correctly, no idea if it helps with your problem.

60Hz/120Hz skipped randomly in fastest possible no-buffer mode. (50Hz worked fine), wait period was calculated from hardcoded value that was too small for >50Hz rates.

NOTE: This version can randomly seem to stop responding or get really slow ((in low latency vsync only), just enter and exit the GUI to fix it. No time to fix it today.

[rsn8887]

Still the same problem here.

Btw, don't know if it helps, but in cycle exact mode the mouse pointer is perfectly smooth.

EDIT: But pointer position jumps when dragging screen still.

EDIT8: I said a bunch of nonsense here before about scrolling in Turrican 2 etc. not being smooth with latest winuae.zip, but never mind it still is fine. The sprite movement is perfectly smooth in Turrican 2. The problem so far is isolated to the mouse pointer as far as I can tell.

[rsn8887]

Toni, I just edited above post: Mouse pointer is still jerky/flickery, scrolling and sprite movements in Turrican 2 are smooth. Scrolling in Hyperdrive also smooth.

[rsn8887]

Here is another mousepointer related problem. This is very minor, and I have no clue how to reproduce it as of yet:

Sometimes during boot up of my wb (using latest winuae.zip), I get a "squished" mousepointer. See attached picture. This happened to me with an earlier 2.4.1 beta too once, but I don't remember which one it was.

I am guessing it has something to do with interlace/non-interlace switching, because an interlace picture is displayed during boot up time. But that's just a guess. It is hard to figure out since it almost never happens.

[Toni Wilen]

Originally Posted by rsn8887

Here is another mousepointer related problem. This is very minor, and I have no clue how to reproduce it as of yet:

It is normal when using OCS/ECS configuration and hires mouse cursor is enabled (prefs/pointer). AGA supports hires sprites, OCS/ECS does not and AOS automatically removes every other line to make it ugly "hires-like"..