Author
| Symbian emulator?
|
GuyveR800 msx guru Posts: 3048 | Posted: January 27 2004, 17:26   |
Kinda like NLMSX... Haven't seen the code, but it does seem like the only fMSX code left in these programs is some minor glue-code! |
|
Bart msx professional Posts: 654 | Posted: January 27 2004, 20:31   |
JR rules  I'm actually considering buying a 6600 just for JR's emu
If JR starts to rewrite core stuff, shouldn't we start calling it JRMSX/60 instead of fMSX/60?? I guess we should! |
|
jr msx addict Posts: 328 | Posted: January 28 2004, 11:00   |
The idea about improving the frame skipping was nice; I implemented it and now the speed of the emulation keeps approximately the same and only display update is slower. I tested the speed (with the new Z80 core) quickly on my 6600 with v0.99b and two setups:
a) MSX1 + F1 Spirit, sound enabled
b) MSX2 + Metal Gear 1, sound enabled
Both games use PSG and SCC. I didn't play the games for hours so I don't know if there would be a glitch in the sound sometime later but anyway I got the following results when I tried to find a frame skip value that provided crackle-free sound with the specified rendering mode:
Raw unzoomed: a)0 b)1
Fast fit: a)0 b)1
Smart fit: a)1 b)4
Resample fit: a)3 b)-
In reality the frame skip is always one bigger than what you define (i.e. if you set frame skip to zero the actual frame skipping would be one).
I guess it depends on the game which frame skip you need to get good sound. Perhaps I need to increase the maximum value (now five) to enable resample fit with good sound on the b) case too... actual game play starts to get a bit difficult with a very high frame skip though...
|
|
jr msx addict Posts: 328 | Posted: January 28 2004, 12:33   |
I did some further processing and I think that case b) with smart fit in the above test runs fine with a frame skip of three now on a 6600. Also, I raised the maximum frame skip to 10 and got a result for the resample fit case b) on a 6600, it was 7. So, final results for 6600:
Raw unzoomed: a)0 b)1
Fast fit: a)0 b)1
Smart fit: a)1 b)3
Resample fit: a)3 b)7
I did the same test on a 7650 with v0.99a, no surprises there, it's a bit faster:
Raw unzoomed: a)0 b)0
Fast fit: a)0 b)0
Smart fit: a)0 b)2
Resample fit: a)3 b)6
|
|
karloch
 msx addict Posts: 476 | Posted: January 28 2004, 13:21   |
Quote:
| I did the same test on a 7650 with v0.99a, no surprises there, it's a bit faster:
|
Curious, Symbian 7 was suposed to be faster with sound because of the new sound arquitecture, right? Anyway I have no complains since I use the Symbian 6.1 from my 3650. ^^ |
|
Bart msx professional Posts: 654 | Posted: January 28 2004, 13:38   |
Hmzz. I opened up another thread for this one. But it seems some peeps link to this topic...
|
|
snout
 msx legend Posts: 5011 | Posted: January 28 2004, 15:38   |
I'm quite surprised to see the 6600 perform worse than the 7650 indeed. I thought there was a workaround for that already.
|
|
jr msx addict Posts: 328 | Posted: January 28 2004, 17:14   |
Aye, architecturewise the sound streaming in the new OS should be faster for applications. And it is in practice also.
But somehow the overall performance of a 6600 is lower than a 7650. This is quite interesting, I don't know what is causing it, SW? HW? Anyway, the emulator runs faster on a 7650/3650 than 6600 -- even if I did manage to narrow the gap with the B version of the emulator the A version is still running faster on a 7650/3650. Sigh. Well, I guess we have to wait for more Series 60 based phones to appear and then we'll see if this slowness is just an issue with the 6600 SW/HW or the OS SW in general.
|
|
jr msx addict Posts: 328 | Posted: January 28 2004, 17:30   |
I just uploaded v0.99.
About the opcodes and other non-architectural things I changed in the Z80 emulation, if someone is interested:
- I added lots of undocumented EDxx and DDCBxx/FDCBxx opcodes that were not implemented in the original fMSX Z80 emulation.
- IMHO the original version did not handle properly DDEDxx and FDEDxx cases (both prefixes were ignored - I assume this should be handled as EDxx and just run a NOP as the DD/FD).
- The original version had zero timings for some opcodes causing wrong(?) interrupt timing.
|
|
snout
 msx legend Posts: 5011 | Posted: January 28 2004, 18:16   |
G-r-e-a-t!!!! I'm going to test it right now!!
And already another feature request...
the raw/unzoomed mode (although I almost never use it)... maybe an option to focus on other parts of the screen (center/top right/ top left/bottom left/bottom right) would be useful? Especially since in rotated mode with focus on center a lot of games are quite playable (I guess)... |
|
jr msx addict Posts: 328 | Posted: January 29 2004, 13:39   |
Yes, someone already asked for it a while back and it is on my to-do list. Of course it will also introduce a marginal perfomance hit in the raw rendering mode but I guess it doesn't matter. I'll try to include it in version 1.00 and perhaps speed up the rotated raw&fast modes as well.
|
|
snout
 msx legend Posts: 5011 | Posted: January 29 2004, 13:42   |
... which reminds me... when rotating the screen, could the left/right/up/down controls auto-rotate as well?  |
|
jr msx addict Posts: 328 | Posted: January 29 2004, 15:03   |
Yep, also on the to-do list. In fact, a "slot" for it has already been reserved in the config file for one or two releases but I haven't got around to implementing it  But I'll definetely try to put it in 1.00  |
|
karloch
 msx addict Posts: 476 | Posted: January 29 2004, 15:38   |
I just tested the new version (0.99a). No bugs found, and increased performace. Excellent work Jr!
|
|
snout
 msx legend Posts: 5011 | Posted: January 29 2004, 15:40   |
I just played Space Manbow on my 6600 with perfect sound emulation. WHAAAAA!!!  |
|
|
|
|