VOGONS


First post, by aleksej

User metadata
Rank Member
Rank
Member

I have read many times on this forum about "hanging notes" bug with midi routing through "fake" mpu-401 interface on creative sound blaster cards with dsp later than 4.05
Is there exist any solid test (maybe some midi playback) for 100% catching this bug and note for myself that this card is in black-list?

I ask for it because some times ago i collect various awe32 revisions for catching very similar bug with midi hanging notes. All awe32 have dsp 4.13 as i know but some models was bug-free in my opinion. I have one dumb test for this purpose.

Now i want to know - is there any ultimate strong test for this bug? Where is appear always and constantly?

Reply 5 of 55, by Great Hierophant

User metadata
Rank l33t
Rank
l33t

For those AWE32s with discrete DSPs, I believe they are all v4.12-4.13. Am I correct? In fact, I think I read somewhere that all AWE32s and 64s that do not have a discrete DSP will report a v4.13 DSP.

Reply 7 of 55, by samudra

User metadata
Rank Member
Rank
Member
c.imp wrote:

Now i want to know - is there any ultimate strong test for this bug? Where is appear always and constantly?

I can't but agree with you.

I've seen the thread where you explain the problem you had with the game Hocus Pocus. You are one of the few (only?) who actually tests instead of talking about a bug no definite test seems to be available for.

I am not saying there is no bug, but until someone tests a lot of cards (like you did in the other thread) in a lot of different configurations, instead of preferring armchair speculation, there is no solid evidence for it.

Reply 8 of 55, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie

For testing I use Doom v1.9 (the bug also appears with other Doom versions). In the music of the first level (E1M1), the hanging notes are very apparent and cannot be overheard.

These are the cards I tested (with both SCB-55 and WaveBlaster I):

CT1750, DSP 4.11 - hanging notes (scarce)
CT1759, DSP 4.05 - OK
CT2910, DSP 4.13 - hanging notes
CT2960, DSP 4.13 - hanging notes
CT3980, DSP 4.13 - OK

I got the last card (AWE32) today and I didn't expect it to work without hanging notes, since it also has the v4.13 DSP, but apparently it does.

Reply 10 of 55, by Great Hierophant

User metadata
Rank l33t
Rank
l33t

I think if there is a consensus, then it is this: v 4.05 DSPs work fine, but v4.11-4.13 DSPs are suspect. However, DSPs 4.13 seem to have a better chance of working fine than 4.11 or 4.12. The card is also a factor here as well. c.imp also said in his other post on the subject:

CT2760: midi - corrupted, fm after emu8k wt- mute CT3900: midi - ok, fm after emu8k wt - mute CT3980: midi - ok, fm after emu8 […]
Show full quote

CT2760: midi - corrupted, fm after emu8k wt- mute
CT3900: midi - ok, fm after emu8k wt - mute
CT3980: midi - ok, fm after emu8k wt - mute, Qsound 3D on/off - always on
CT3990: midi - corrupted, fm after emu8k wt - ok, Qsound 3D on/off - ok

I think that if you want an AWE that doesn't suffer from this problem, you must go with the 3900 or 3980.

I read somewhere that the SB16 had problems with 8-bit digital sound, leading to cracks and pops. Maybe the later DSP revisions fixed that at the expense of the midi. Finally, to respond to the complaints, Creative did something with the board to fix the midi.

Reply 11 of 55, by aleksej

User metadata
Rank Member
Rank
Member

But....
Unfortunatelly even with these two bug-free AWE32 revs. similar "hang note" bug exist but now it have phantom behavior and appears occasionally.
In example - it may appear after long playing (20-30m) same midi song during play game.
Maybe it bacause of extremal overclocking in my rig (83mhz FSB/ 20.7mhz 16bit ISA) and cards just heat up?

Reply 12 of 55, by Great Hierophant

User metadata
Rank l33t
Rank
l33t

Maybe it bacause of extremal overclocking in my rig (83mhz FSB/ 20.7mhz 16bit ISA) and cards just heat up?

I'm surprised that the ISA cards work at all! The 16-bit ISA bus is supposed to be run from 6MHz to 8.33Mhz. The bus may be getting ahead of the card. Try playing without the computer overclocked and you may see that hanging note bug disappear. What kind of processor are you using anyway?

Reply 13 of 55, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie
Great Hierophant wrote:

I think if there is a consensus, then it is this: v 4.05 DSPs work fine, but v4.11-4.13 DSPs are suspect. However, DSPs 4.13 seem to have a better chance of working fine than 4.11 or 4.12. The card is also a factor here as well.

What if the bug was unrelated to the DSP version? What about the other chips? Could the problems with v4.1x just be a coincidence?
It seems strange that the bug vanishes on the CT3980, it has exactly the same DSP as the CT2910 (I can't tell with the 2960, since it's an all-integrated Vibra16C).

When I look at all the chips, it seems to boil down to the following:

CT1746B + CT1741v405 = OK
CT1746B + CT1741v411 = Bug
CT1746B + CT1741v413 = Bug
CT1705B + CT1741v413 = OK

Can someone verify this?
Here I assume that CT1746B and CT1705 basically do the same thing, judging from their position on the board.

c.imp wrote:

Maybe it bacause of extremal overclocking in my rig (83mhz FSB/ 20.7mhz 16bit ISA) and cards just heat up?

Although Intel 430xx chipsets do not have selectable clock dividers for the PCI bus, the ISA bus is decoupled from FSB. (I also ran HX and TX chipsets at 83 MHz 😉)
It's extremely unlikely that an ISA SoundBlaster still works at 20+ MHz. Most of them already have problems at 12-16 MHz. I destroyed two AWE64 Gold cards at 16 MHz until I found that out 🙄

Reply 14 of 55, by aleksej

User metadata
Rank Member
Rank
Member

Ofcourse is my mistake - default ISA bus clock is 8mhz not 16mhz, then if my FSB 83mhz i have ISA clock 10.3mhz 😉

Anyway this is enough for make many old ISA cards non-worked. In example:
My Matrox Millenium II work solid only with 75mhz FSB / 37,5 PCI. With 41mhz PCI he lock up with cyrillic font driver loaded when with 33/37.5mhz all ok.
Roland SCC-1 just lock up with any FSB/ISA clock higher than default - very old card 😉
Now i set all my cards wich i have in my rig that work just fine on FSB 83mhz.

Reply 15 of 55, by samudra

User metadata
Rank Member
Rank
Member

In the readme files of games incompatibilities are always listed.

One finds lists documenting the various videocards and soundcards, tested and not-tested, and their quirks when using the software

So, I wonder, are there any games out there that list this problem in the readme?

Or even better, has anyone ever seen a PC magazine in the day documenting this problem, or receiving letters about it and placing it in the mail section?

There must have been numerous people experiencing this problem in the day.
Or was the amount of people using the waveblaster connector rather slim?

Or perhaps there was a massive conspiracy on the part of the PC magazine editors funded by Creative, heh?

Reply 16 of 55, by rubinio

User metadata
Rank Newbie
Rank
Newbie

excusme people.
i will have a roland scb-55 in my house next week, but i have read in your forum the problem for the DSP of the sb16.
i have two sb16:
ct2960 dsp 4.17
ct2940 dsp 4.13
this cards works ok?

Reply 17 of 55, by aleksej

User metadata
Rank Member
Rank
Member

rubinio, keep in mind - this is not ultimate test. I just found a midi melody where this bug always exist. It may be used as a dumb test for first time.
Do you have hexen?
If not you may download demo version here.
launch it with -warp 02 parameter.
"hexen.exe -warp 02". Right in begining when game starts (not in loading screen) you must hear (or not hear fortunatelly) hanging note.

Reply 18 of 55, by StickByDos

User metadata
Rank Member
Rank
Member

I tried Hexen -warp 02 with my db50xg
CT3990 4.13 -> bug
CT2890 4.13 -> bug
Maxi Sound 16 -> no bug (not a Creative card)

The bug exist with 83MHz and 66MHz fsb
with or without cache
with K6-2 and with PentiumMMX

Type win to loose the power of your computer !

Reply 19 of 55, by aleksej

User metadata
Rank Member
Rank
Member

huh. thnx for such detailed testing! 😉

Maybe you try just play this mid in any player?
Try this on your buggy CT cards.
Just for knowing reason - necessary a game itself or not?
Maybe enough only midi file for testing.

Attachments

  • Filename
    Hxmidtst.zip
    File size
    17.08 KiB
    Downloads
    490 downloads
    File comment
    Hexen, map02 test midi. Reuploaded with more accurate converted midi and original mus.
    File license
    Fair use/fair dealing exception
Last edited by aleksej on 2006-10-23, 15:09. Edited 1 time in total.