Processor usage oddity in PTLE 6.9 and now PTLE 7

[Mark Seyler]

Weird this:
PC is an Athlon 3700, ASUS Mboard with 2Gb RAM, SATA drives etc and seems to work OK. I am using Digi 002 work surface (which is a lovely bit of kit!). Was using LE6.9, now on to LE7 and the problem is the same.

The problem: take a small session (say 4 tracks with EQ and Dverb on each) and in PLAY the processor usage is around 15-20%. STOP the playback and the processor usage climbs (Task Manager tells me this is Protools only, no other processes working) and can quickly reach 80% and error message time!! Not every time but often enough to cause real grief! If caught in time, returning to PLAY instantly restores the 15-20% usage until STOP .....


I have tried deleting all preferences, databases etc, done all the optimisations that I can find but to no avail. At times this means closing the session and starting again as CPU overload won't allow me to get back into play again!!! And obviously gets worse the bigger the track / DSP list is as the starting point is higher.

Any ideas?

Mark Seyler

[BigMetalDave]

Right Click My Computer and choose Manage.
Click on Disk Management under Storage.
Right Click the Partition you are saving audio to and choose properties.
On the general tab at the bottom make sure that "Allowing indexing service to index this disk for fast file searching" is checked.

My PT did the same thing when I created a new audio partition and checked that off.

[MorganPost]

Right Click My Computer and choose Manage.
Click on Disk Management under Storage.
Right Click the Partition you are saving audio to and choose properties.
On the general tab at the bottom make sure that "Allowing indexing service to index this disk for fast file searching" is NOT checked.

My PT did the same thing when I created a new audio partition and checked that off.

Just out of curiosity, where did you come up with this piece of info to help smooth out Pro Tools? Was this a Digi-tip or did this come from another source, and if so, what source?

[Ducky]

You are experiencing the denormalization problem with Dverb.
Put a 16bit dither above the dverb, and things should work fine.
This has been a sanctioned work around for this issue.

[BigMetalDave]

Just out of curiosity, where did you come up with this piece of info to help smooth out Pro Tools? Was this a Digi-tip or did this come from another source, and if so, what source?

I'm glad you called me out on this... i actually have it backwards. The checkmark does need to be there. that's the last time i post from work w/o checking my rig at home.

anyways, it was process of elimination. i had just partitioned a new drive before a session and that was the only one i was experiencing the symptoms with. that was the only setting i had different.

[Mark Seyler]

Well thanks for these rapid replies!

Morganpost and BigMetalDave: this is one to bear in mind: not the problem here it seems as the box was ticked already.

BUT: Ducky! What a star! Solved immediately! The only drawback being the increase in RTAS activity (although the Dither activity seems to have a very low cpu requirement) and the fuss of having to insert an extra plug-in on each DVerbed track. Is there a (reasonably) simple explanation for this problem?

I am afraid this also begs the question: if this is a "known getaround", why has it not been sorted for PT7LE7??? Any answers Digidesign?

Mark Seyler

[ejwells]

That's certainly a good question

The denormalization issue has been around since DVerbs incept, and here we are...years later...with a complete overhaul of PTLE...and we still need to put a dither plug before DVerb?

I know NOTHING about plugin development, so I have to assume that if it could be fixed it would be fixed. Makes sense to me. So I have to assume it either can't be fixed or won't be fixed

Those seem to be the only two scenarios possible

Taken in turn, if it's "it won't be fixed," that makes no sense to me. So much so that I won't accept this as an answer, so I'll have to boil it down to the only other possibility...

It can't be fixed. Huh? Well..that's just too taxing on my little brain.

I mean, TrueVerb and RVerb don't have denormalization issues
IR-1 doesn't have denormalization issues
Altiverb? No. AnyOtherVerb? Not that I'm aware of.

Why not either slap a hidden dither into it so it doesn't SEEM have the issue, or...
Why not just toss DVerb into the trash heap, design a new free reverb that DOESN'T have the denormalization issue, and call it a day?

I mean, for the love of Pete, if you have a VST2RTAS wrapper, FreeVerb is, well...FREE, it's as good a verb as DVerb...and IT has no denormalization issue

[lemix]

I find this thread interesting because...
As far as I can remember all denormalization issues w/Dverb were pointed in the direction of Intel Pentium systems. And you have experiencing it on an AMD.
Anyone else care to chime in on this ?

cheers,