UA WebZine "Support Report" November 03 | Host Processor Load issues with UAD-1 3.2

Support Report: Host Processor Load Issues with UAD-1 3.2

PC Side
Some PC users may be experiencing a greater degree of host processing loads since 3.2. A problem introduced in 3.2 has recently been tracked down and confirmed, and a permanent fix will be included with the upcoming release of 3.3. In addition, this fix allows compatibility with Windows 2003.

Mac Side
With UA’s release of 3.2, a permanent fix was implemented for a Logic/UAD-1 specific issue that caused pops, crackles and crashes in audio for users. The fix forces all UAD-1 plug-ins into “Live Mode” or in other words, it implements the well-known “I/O Helper workaround” any time a UAD-1 plug-in is used. See the May & September Support Report for more info.

While this fix benefits UAD-1 users, it seems it may actually hinder host plug-in users by overtaxing host processors in certain situations. While we have no plans on changing this implementation under OS9, the good news is that none of this will be an issue under OSX.

Q & A Live Mode Fix

What is the “Live Mode” fix?
- The Live Mode fix is a permanent fix implemented jointly by Emagic & UA to solve a specific Logic issue with UAD-1 when using UA plugs as inserts OR sends on "live" tracks (ie.VSTi's or record-enabled audio tracks) that previously caused pops and crackles in audio due to the UAD-1 receiving two kinds of process requests "live" and "non-live" plugs with different buffer sizes. In 3.2, UAD-1 plugs ALWAYS run in "live" mode and use processor 1 of a multi-processor machine.
Which version of Logic implements this fix?
- 6.1.1
Which version of UAD software supports this fix?
- 3.2
How do you use it?
- Just instantiate a UAD-1 plug-in as normal, and it will put the track into live mode automatically.
What potential issues arise because of it?
- An appearance of higher host processing load on dual processor machines, by shifting host plug-in processing from the secondary processor to the primary processor on tracks that share inserts with UAD-1 plug-ins.

With 3.2, all plug-ins on tracks with UAD-1 plugs and ALL PLUGS ON TRACKS/BUSSES FED BY THOSE TRACKS will now be running in “Live Mode”. This means when multiprocessing is enabled, those plugs will all be running on the primary processor. Non-“Live Mode” plugs still run on the second processor.

Some users have been reverting back to 3.1 because they believe it alleviates the host processor load issue. This is true, but only if you don’t use the I/O helper. But you WILL NEED to use the I/O Helper trick if you plan to use UAD plugs on VSTi's or live audio tracks or else you open your self up to susceptibility to the clicks, pops and crashes associated with not running your UAD-1 plug-ins in Live mode. This is particularly true with users running in multiprocessor Macs.

3.2 EXAMPLE:
Normal convention in Logic is that plug-ins run exclusively on the secondary processor of a multiprocessor machine. However, in v3.2 for UAD1, if a UA plug (say an 1176LN) is used as an insert on a track with say 3 sends with 1 native plug on each separate send bus ( say a native Reverb, Tape Delay and Slapback Delay ). On a multi-processor machine, the UAD plug and the 3 native plugs will now all run on the primary processor. The UAD plug being "upstream" of the other plugs and running in "live" mode forces all the others to run on the primary processor - this ensures no buffer mismatches, but if the 3 native plugs are CPU-heavy, there is the potential for overloading the primary processor.

What this really means is that in OS9, Logic users with Multiprocessor machines have to pay a little more attention to what they are sending to from a track with UAD inserts to ensure that the main processor isn't overloaded. Obviously, if you are using a combination of UAD plugs almost exclusively, this is rarely an issue and doesn't become one unless you have some hungry native plugs that share track inserts or bus sends with UAD plugs.

Conclusion? I advise staying with 3.2 until version 3.3 is released.

--Will Shanks

Questions or comments on this article?