It's probably the same problem I reported a few years ago: viewtopic.php?p=100793#p100793TTTTTsd wrote:This emulator is fantastic, although I can only report one oddity and that may be in the sound emulation (although I lack the real hardware on me t the moment to test if this is true.) It happens (most prominently) in Castlevania 3 and Akumajou Densetsu. The drums (I believe the Noise channel, as they're called in the emulator) don't sound....I guess proper is the word. This is the only issue with puNES that I can report. If this gets fixed at any point in time it will be literally perfect.
Speaking of, well, 'problems' reported a long time ago: Perhaps you, FHorse, might still be interested in a realistic sprite delimiter enhancement option that's not as potentially destructive concerning accuracy?thefox wrote:It's probably the same problem I reported a few years ago: viewtopic.php?p=100793#p100793TTTTTsd wrote:This emulator is fantastic, although I can only report one oddity and that may be in the sound emulation (although I lack the real hardware on me t the moment to test if this is true.) It happens (most prominently) in Castlevania 3 and Akumajou Densetsu. The drums (I believe the Noise channel, as they're called in the emulator) don't sound....I guess proper is the word. This is the only issue with puNES that I can report. If this gets fixed at any point in time it will be literally perfect.
Other than that, I can only second the "literally perfect" comment. Thank you for your hard work! I'm enjoying the emulator immensely.
Also, the APU controls appear to be in linear space, which is inconvenient UX as the perceptible differences for volume are non-linear. Any of (slider percentage^2), a dB -> percentage power (1 - 10 ^ (dB/10)) or dB -> percentage voltage (1 - 10 ^ (dB/20)) would work better.
Also, sorry to say, but I'm having some significant audio lag, regardless of the settings I use. It's very noticeable - several miliseconds at least. I'm running Manjaro Linux, a derivative of Arch. At first I thought it may have been input lag, but after thorough testing with Contra, as I am acutely familiar with the timing nuances in that game, I can confirm that input is not the issue. I am running the ALSA backend, and I do not experience audio lag with other emulators (I used to when I had JACK enabled, but not since I removed it a while ago).
Other than that, man....this emulator is an impressive piece of work. Keep it up.
At power-on the $F000 bank should be set to the high bank, but at reset it should not be changed.
I made some test ROMs. Look at "STARTUP MODE" in the test display. It should correctly say "POWER-ON" or "RESET" depending on how the ROM is started.
Both Nestopia and puNES implement¹ the CHR banking registers as:
Code: Select all
$B000 : [...D EFG.] $B002 : [.... ABC.]
Code: Select all
$B000 : [.... EFG.] $B002 : [.... ABCD]
Code: Select all
STA $B000 LSR LSR LSR LSR STA $B002
Attachment includes a validation ROM for CHR banking: it uses cc65 and knes, and I was too lazy for a Makefile so there's just a script that runs cl65 three times instead. The source even includes a (commented out) "officially wrong" implementation that produces the desired results in Nestopia/puNES but the wrong results in FCEU/Nintendulator. In the center of the screen, it should appear to count from 0 to 127, without skipping any, and the color emphasis bits should change twice as fast (indicating that the LSB is ignored). ¹ I think I can blame Nestopia's original author for being too clever here.
This error doesn't occur when vol=100%. But if you try to reduce volume, then:
1) On squares/noise slider work like threshhold filter, causing cut-off when channel volume goes to fade-out (lower slider volume = stronger cut-off)
2) On triangle slider work like distortion, causing distort triangle timbre.
Here is test ROM i've made for show this bug.
- (32.02 KiB) Downloaded 687 times
FHorse can confirm that this bug was real. He fixed it just now.
Also, i tried Nestopia and BizHawk, it doesn't have this bug.
Is this statement entirely accurate? In this video taken from the real NES pixels on the right half of the scanline seem to flicker for about a second at 1:15:Great Hierophant wrote: Nintendulator, nemulator and real hardware only show the scanline flicker on the left side of the screen.
- The resampler reviewed and optimized. Now the sound quality is dramatically better and the audio output more accurate and correct.
- Added a russian translation (many thx to EXL and Eugene.S).
- Added an option ho hide the mouse cursor.
- Added a new mouse cursor when using the Zapper.
- Fixed APU volume sliders that did not work properly.
- Correct identification of submapper for MMC1 (viewtopic.php?f=3&t=9350&start=90#p153298) and for the mapper 34 (viewtopic.php?f=3&t=9350&start=90#p153334).
- Fixed a bug in the CHR banking registers of mapper VRC2A (viewtopic.php?f=3&t=6928&start=495#p153176).
- Tons of others bugfixes.