Yamaha DS2416/SW1000XG

[VintageSynth]

Has anyone out there succeeded in getting ASIO4ALL vers 2 working with the good old Yamaha DS2416/SW1000XG cards? It seems that these otherwise wonderful devices have been binned in their thousands for want of some support from Yamaha. Their only shortcoming is their latency that is far too great for soft synth use, at least when using the last drivers provided by Yamaha back in '02.
I discovered ASIO4ALL when it was at vers 1.8 and, using Cubase running virtual instruments like the B4 and Halion found that I could get the 2416 latency down to (allegedly) 3mS, but very acceptable anyway. This was great, but could only configure Cubase to output to one of the 8 channels on the 2416 at a time. So when I saw version 2 of ASIO4ALL with its multi-channel capability, I thought my prayers has been answered. Unfortunately though, life is never that easy, as all channels on both cards are now :-[shown as "beyond logic".
Short of trying to hack the Yamaha drivers myself to try to reduce the buffer size further from its current minimum of 512, I'm a bit stuck, although still reluctant to take a trip down to the local dump.

[Ash]

Hi,

I've the same "Beyond Logic" problem with my dear SW1k... (see the previous topic)
Michael Tippach (a great idol for us ;D) told me the next version of ASIO4ALL (ver 2.3) will works with the SW1k as well...
So we must wait for now, and works with the old versions of ASIO4ALL.

Cheers.

[VintageSynth]

Thanks for that, Ash, I'll await version 2.3 with great interest.

[Michael Tippach]

Actually, it's supposed to already work with the pre-release:

mtippach.proboards40.com/index.cgi?board=FYI&action=display&thread=1103593538

[VintageSynth]

Thanks, Michael; the pre-release version does indeed work with both cards. I can do everything that was possible with the Yamaha drivers, but with a third of the latency (Yamaha reported value: 28mS, ASIO4ALL: 9mS).
This was obtained with 2 wave out channels enabled on the SW1000 and 6 wave outs and 2 wave ins enabled on the DS2416 (2.8GHz P4, Gigabyte M/B, 520MB RAM).
The best results seem to be obtained with the ASIO buffer size set to the minimum of 64 and the hardware buffer enabled with offset=8mS, on both cards. Strangely, on the DS2416, there seems to be better immunity against crackles with the ASIO buffer set to 64 than there is at any other point up to 128, where the delay is obviously greater. One major advantage of this arrangement is that you can just enable whichever I/Os are needed at the time, optimising performance.
You have given these much-loved sound cards a new lease of life!
Thanks again, John C

[Ash]

VintageSynth said:

Thanks, Michael; the pre-release version does indeed work with both cards.

I wasn't able to work with this new version... It crash regularly, and when works I get no sound again... But this time the icon 'Beyond Logic' has not appeared.

[VintageSynth]

I haven't had any crashes using this pre-release version, but have occasionally lost sound when playing around with the buffer settings. One problem I have found on further use is that any audio tracks recorded in Cubase using the new ASIO4ALL are substantially out of sync with the MIDI tracks when replayed. I can compensate for the delay using the audio-to-MIDI delay setting in the Cubase Audio System Setup dialog box, but it needs a value of -14500, (about 0.33 sec advance @ 44.1kHz).
This seems to be happening during the record process, as recordings made using the original Yamaha drivers are still playing in sync. I can take some of this out by adjusting the Latency Compensation slider in the ASIO4ALL configuration to its max value, but as this only goes up to 1024, the required delay compensation setting then only reduces to about -13500.
Not a killer problem, but thought the feedback might be useful.

[Michael Tippach]

An offset of 14k samples certainly is a bit excessive.

Do you have the hardware buffer option enabled?

[VintageSynth]

Yes, the hardware buffer is enabled. In fact, I can't record anything unless it is. When disabled, any attempt to record results in a track with no audio and, after hitting the stop button, the old favourite from Cubase: "Last recording too short to use" , which normally means "Processor too slow". (I'm using 2.8GHz P4).
I've tried adjusting the ASIO buffer size and number of Kernel Buffers in non-hardware-buffer mode, but always get this result. I've also played around with the ASIO buffer settings and the hardware buffer offset value when using the hardware buffer to see if the audio delay on replay suddenly drops (in case of some buffer wrap-around condition), but the additional delay just adds to the required 13.5K sample value (more accurate measurement of precise delay than original estimate of 14.5K).
Although I can offset the delay by inserting this -13500 number into the MIDI-to-Audio delay box in the Cubase Audio Setup, I've found that I have to be very careful not to save the song with this parameter set to such a high value. If I do, when I reload the song, I get the worst crash in history with audio being played at some phenomenal speed and the kind of lock-up where you can't even access the OS to shut down (I'm running W2K). On the last occasion, another PC on the network even thought it was being attacked! So a workaround would be nice, although I think I can get by as long as I remember to reset the delay value to zero before shutting down.