SAND out now!

Tips & tricks, working results, technical support
Pro5
User Level V
User Level V
Posts: 55
Joined: Tue Aug 12, 2014 11:28 pm

Re: SAND out now!

Post by Pro5 » Sun Oct 09, 2016 3:28 pm

KP2009 wrote:
giancarlo wrote:We'll try to reproduce it...

Thank You Giancarlo!!
I'm on Studio one V3. I have an older system (that's almost ok for everything but Acqua sadly - I seriously need to upgrade but was holding out for skylake X but may need to bite that broadwell E bullet soon).

Anyway I tested an empty project with 32 mono tracks.

32 instances of BUS4 comp (insane mode comp A)
8 Pre4 (bus C selected)
8 EQU4 (all bands and filters active)

This was bouncing off the 100% (couldn't work this way). They took ages to load into the slots across multi selected tracks.

Saved it. Reloaded it. CPU meter in task manager was hitting 100% in spikes, ram was around 6GB.

Win 10 64 pro
16 GB RAM
SSD (Samsung Pro EVO)
I72700k o/c at 4.5ghz
88.2k sample rate
2048 buffer (on MR816x)

Am using the addon 88k pack and trial plug.

it took close to 5 minutes!!! to load this empty project with nothing but FAB 4 plugs in, was your problem you didn't wait long enough or did it just fail with an error?

G > I'm on an SSD and have decent amount of RAM, I'm not aware of it caching to HDD/Virtual Mem. Is my CPU contributing to the VERY SLOW load times? I mean these load times if you are switching around projects can kill workflow like a biatch, would a more modern CPU help here? (not sure why it would as it seems to be a drive/ram issue but aside from going to DDR4 with broadwell etc I can't see it getting much faster?).

I'd love to be able to just drag drop these plugs in Studio One as fast as I do my algos, as it's a fluid mixing experience, even so I can bear the slow load time there per plug, but when reloading a project will AA plugs with large samples to load into ram always take 5 minutes or so if you use a decent amount of them?

Is it doing something wrong with SRC maybe not picking up the 88k pack and converting the 96/44 or whatever before loading? If the project is at 88k will it always automatically look for and load the 88k samples in the AA folder? (I must delete the 44,48 and 96 to test this as I'll never use those rates at all and it'll save some SSD space)

Does anyone on the fast modern systems get lightning load times with loads of AA in a project?

Pro5
User Level V
User Level V
Posts: 55
Joined: Tue Aug 12, 2014 11:28 pm

Re: SAND out now!

Post by Pro5 » Sun Oct 09, 2016 4:00 pm

Another quick test.

Same as above (88/2048)

This time a more sensible/realistic 8+mixbus instances of

EQU4 (2 bands enabled no filters)
BUS4 (insane mode comp A)

So 9 of each.

Was showing around 50% cpu in studio one.

Closed and reloaded. This time it took 2.5 minutes to load the "empty" project (with just fab4), not as bad as before but still very long vs my usual 20-40 seconds with algo plugs and full projects.

On loading the S1 cpu meter was now hitting 100% and spiking badly. I global bypassed them all with the new S1 button, then renabled and it was better (around the same 50% as before reloading) BUT it now spikes up to 100% quite a lot.

Am aware most of my troubles come from using a hideously outdated (by AA standards) 2nd gen i7 sandybridge, so I'm not expecting anything great there, but the load times are really bad. Any advice?

Coalhouse
User Level I
User Level I
Posts: 14
Joined: Sun May 16, 2010 7:42 pm

Re: SAND out now!

Post by Coalhouse » Sun Oct 09, 2016 4:22 pm

I noticed in H******* Mixbus 32 that a 12 track project that played back without issues with CPU load at 87% suddenly hangs when i try to reopen it. It does reopen after a couple of tries but i have to disable all Sand/Fab 4 instances, then re-enable them to get it to playback without crackles and back to 87% load. It is very stable otherwise. I wonder if there could be a way to stagger-load the plugin in a project that contains several instances. That could reduce load times and instability when reopening a project. Just a thought.
Custom i7 Quad with 32GB RAM, UA Apollo Quad SF, Ensoniq Paris Pro, Antelope Orion 32 Gen 3, Lynx Hilo, Sand, Pink, Aquamarine, Amethyst, Pearl, El Ray, Emerald, Cream, Magenta, Water, Diamond, Gold, Titanium, Amber, Nebula 4, Navy, Big Ceil, Celestial, Crimson, Cola, Coffee, Coral, Ivory, Jade, Lime, Purple, Scarlet, Ultramarine, Viridian, Ebony, Erin, Taupe, Stradipad Platinum, Cubase 10.5, H******* Mixbus 32C.

Coalhouse
User Level I
User Level I
Posts: 14
Joined: Sun May 16, 2010 7:42 pm

Re: SAND out now!

Post by Coalhouse » Sun Oct 09, 2016 4:27 pm

giancarlo wrote:Never uninstall
Install always over the previous one
Giancarlo,
What if you have unistalled already. can one just install the latest version ? would this cause any performance related issues?
Custom i7 Quad with 32GB RAM, UA Apollo Quad SF, Ensoniq Paris Pro, Antelope Orion 32 Gen 3, Lynx Hilo, Sand, Pink, Aquamarine, Amethyst, Pearl, El Ray, Emerald, Cream, Magenta, Water, Diamond, Gold, Titanium, Amber, Nebula 4, Navy, Big Ceil, Celestial, Crimson, Cola, Coffee, Coral, Ivory, Jade, Lime, Purple, Scarlet, Ultramarine, Viridian, Ebony, Erin, Taupe, Stradipad Platinum, Cubase 10.5, H******* Mixbus 32C.

Pro5
User Level V
User Level V
Posts: 55
Joined: Tue Aug 12, 2014 11:28 pm

Re: SAND out now!

Post by Pro5 » Sun Oct 09, 2016 4:54 pm

Coalhouse wrote:I noticed in H******* Mixbus 32 that a 12 track project that played back without issues with CPU load at 87% suddenly hangs when i try to reopen it. It does reopen after a couple of tries but i have to disable all Sand/Fab 4 instances, then re-enable them to get it to playback without crackles and back to 87% load. It is very stable otherwise. I wonder if there could be a way to stagger-load the plugin in a project that contains several instances. That could reduce load times and instability when reopening a project. Just a thought.
For now, for bus comp at least I've gone back to Duende, in my own A/B tests under heavy compression on drums there's very little in the difference of 'punch' and snap etc. The Sand EQ is very nice though and I doubt that's so easily done in ALGO but as it was the bus comp I was mostly interested in it's a bit too much of a workflow/cpu sacrifice to make for now...(and I REALLY want to wait for skylake x with new platform features if I'm spending that much on a new rig, which I use for video, rendering, development/programming and VR too - mid 2017 is fine for me) when I upgrade my PC I'll be back to try again, but right now - music calls and bus comp wise Duende is still doing a great job (better than slate's grey imo)

SlayermanGR
User Level II
User Level II
Posts: 23
Joined: Sat Dec 13, 2014 10:21 am

Re: SAND out now!

Post by SlayermanGR » Sun Oct 09, 2016 5:12 pm

Coalhouse wrote:
giancarlo wrote:Never uninstall
Install always over the previous one
Giancarlo,
What if you have unistalled already. can one just install the latest version ? would this cause any performance related issues?
Personally, I just uninstalled SAND (noticed that .aut & .ser files remain at the acustica64 folder), installed latest version (first 44.1-96, then 48-88 and finally the fab4) and everything's working like a charm here...
Amethyst, Azure, Cobalt, Coral, Crimson, Diamond, Ebony, Gold, Ivory, Lemon, Lift, Lime, N4, Pearl, Pink 2, Pink, Pink Compressors, Ruby, Sand

User avatar
giancarlo
Founder
Founder
Posts: 11215
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

Re: SAND out now!

Post by giancarlo » Sun Oct 09, 2016 5:18 pm

Pro5 wrote:Another quick test.

Same as above (88/2048)

This time a more sensible/realistic 8+mixbus instances of

EQU4 (2 bands enabled no filters)
BUS4 (insane mode comp A)

So 9 of each.

Was showing around 50% cpu in studio one.

Closed and reloaded. This time it took 2.5 minutes to load the "empty" project (with just fab4), not as bad as before but still very long vs my usual 20-40 seconds with algo plugs and full projects.

On loading the S1 cpu meter was now hitting 100% and spiking badly. I global bypassed them all with the new S1 button, then renabled and it was better (around the same 50% as before reloading) BUT it now spikes up to 100% quite a lot.

Am aware most of my troubles come from using a hideously outdated (by AA standards) 2nd gen i7 sandybridge, so I'm not expecting anything great there, but the load times are really bad. Any advice?
compared with algo timings could be worse, especially if you don't use ssd hdd or similar things
Also cpu makes things a lot faster, for example the normal amount of time I need for loading bus4 on a mac book pro is below the second for instance.

About the sound it is subjective. If you hear a small difference I can tell you the difference in such case would be huge when you compare many tracks. Normally in the hardware realm we pay a lot of money for subtle improvements. So if you hear a difference that improvement is not subtle at all for a single track, go figure for a whole mix.

A thing: I suggest to compare "not" compression alone but the sense of air, of transparency you get on the track.
It is not a matter of puch alone, but also a matter of damages you could get (and they are not easy to spot from a quick glance).
Normally processors with a lot of aliasing are compressing perfectly and you cannot spot a real difference just from a simple quick A/B test.
You hear issues after an amount of processed tracks and you cannot go back any more. If you don't care about aliasing then you could avoid insane mode at all and in such case you could open an awful lot of instances even in bus4. I guess bus4 not in insane mode has still less aliasing of many processors (for the exception of very good ones).


What I think
- bus4: yes you could use something else. But in this case loading time should be really fast, and the plugin pretty light on most of systems. I see people are running 40 insane instances on outdated computers. Loading time is not bad even on old systems. So why someone should use something else? In case of bad performances I would suggest to upgrade the system, because normally people find bus4 very light (over 100 instances on new systems, or even more). The sound is imho very good, one of best bus compressors around. It almost never breaks, and it is pretty transparent on audio, it doesn't damage things. Aliasing is very low.

- equ4: this is heavier, but here you appreciate the technology we are using. Loading time is slower than normal plugins on outdated systems. On new ones performances are better. I see many users are using 60-90 equ4, which is amazing. I guess this one is the most difficult to replace in the digital realm, cause our good performances around nyquist, the complete lacking of oversampling, the perfect aliasing and the fidelity to the sampled gear (we have a PROPER PHASE!)

- pre4: I see performances are very good here. Loading time could be a little slow on outdated systems. But we have a really good aliasing, a proper phase representation both in the fundamentale and in harmonics!

User avatar
giancarlo
Founder
Founder
Posts: 11215
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

Re: SAND out now!

Post by giancarlo » Sun Oct 09, 2016 5:25 pm

SlayermanGR wrote:
Coalhouse wrote:
giancarlo wrote:Never uninstall
Install always over the previous one
Giancarlo,
What if you have unistalled already. can one just install the latest version ? would this cause any performance related issues?
Personally, I just uninstalled SAND (noticed that .aut & .ser files remain at the acustica64 folder), installed latest version (first 44.1-96, then 48-88 and finally the fab4) and everything's working like a charm here...

perfect :D

Coalhouse
User Level I
User Level I
Posts: 14
Joined: Sun May 16, 2010 7:42 pm

Re: SAND out now!

Post by Coalhouse » Sun Oct 09, 2016 5:49 pm

Good to know. That was my thought initially. However, with AA and a tech that is constantly improving you just never know. Having said that, in Mixbus 32, i have a mix that takes me back to my analog days. :)

Good job Gian and AA!
Custom i7 Quad with 32GB RAM, UA Apollo Quad SF, Ensoniq Paris Pro, Antelope Orion 32 Gen 3, Lynx Hilo, Sand, Pink, Aquamarine, Amethyst, Pearl, El Ray, Emerald, Cream, Magenta, Water, Diamond, Gold, Titanium, Amber, Nebula 4, Navy, Big Ceil, Celestial, Crimson, Cola, Coffee, Coral, Ivory, Jade, Lime, Purple, Scarlet, Ultramarine, Viridian, Ebony, Erin, Taupe, Stradipad Platinum, Cubase 10.5, H******* Mixbus 32C.

KP2009
User Level X
User Level X
Posts: 136
Joined: Wed Mar 31, 2010 8:46 am

Re: SAND out now!

Post by KP2009 » Sun Oct 09, 2016 6:54 pm

Pro5 wrote:
KP2009 wrote:
giancarlo wrote:We'll try to reproduce it...

Thank You Giancarlo!!
I'm on Studio one V3. I have an older system (that's almost ok for everything but Acqua sadly - I seriously need to upgrade but was holding out for skylake X but may need to bite that broadwell E bullet soon).

Anyway I tested an empty project with 32 mono tracks.

32 instances of BUS4 comp (insane mode comp A)
8 Pre4 (bus C selected)
8 EQU4 (all bands and filters active)

This was bouncing off the 100% (couldn't work this way). They took ages to load into the slots across multi selected tracks.

Saved it. Reloaded it. CPU meter in task manager was hitting 100% in spikes, ram was around 6GB.

Win 10 64 pro
16 GB RAM
SSD (Samsung Pro EVO)
I72700k o/c at 4.5ghz
88.2k sample rate
2048 buffer (on MR816x)

Am using the addon 88k pack and trial plug.

it took close to 5 minutes!!! to load this empty project with nothing but FAB 4 plugs in, was your problem you didn't wait long enough or did it just fail with an error?

G > I'm on an SSD and have decent amount of RAM, I'm not aware of it caching to HDD/Virtual Mem. Is my CPU contributing to the VERY SLOW load times? I mean these load times if you are switching around projects can kill workflow like a biatch, would a more modern CPU help here? (not sure why it would as it seems to be a drive/ram issue but aside from going to DDR4 with broadwell etc I can't see it getting much faster?).

I'd love to be able to just drag drop these plugs in Studio One as fast as I do my algos, as it's a fluid mixing experience, even so I can bear the slow load time there per plug, but when reloading a project will AA plugs with large samples to load into ram always take 5 minutes or so if you use a decent amount of them?

Is it doing something wrong with SRC maybe not picking up the 88k pack and converting the 96/44 or whatever before loading? If the project is at 88k will it always automatically look for and load the 88k samples in the AA folder? (I must delete the 44,48 and 96 to test this as I'll never use those rates at all and it'll save some SSD space)

Does anyone on the fast modern systems get lightning load times with loads of AA in a project?
I'm not sure I had a session with other third party plugins. Never taken long to load with other Nebula/Aquas..it just looked like it was hanging while loading the SAND EQ4..I let it keep loading for probably 5 min..Will try 1 more time I only had 3 EQ4..a few PRE4..and two SAND Comps along with other third Party Slate, Waves Fabfilter etcetc..
Will let it load over 5 min to see..The other SAND Comp/Eq/Pre-p loads fine when SAND EQ4 is taken out the vast Folder.

Edit:
Loading session.. So far waiting if it will load for the SAND EQ4 to load. Disabled the other sand Pre/COMP as well just to make sure but still waiting 16 minuites so far for session to load.. I see the ram in the Task Manager Win & the studio One app is slowly using ram up over time will wait another 5 min.. SAND EQ4 may not be usable right now till a fix comes can't wait for long load time on a current system.

Love SAND So much Hope to get EQ4 working..

Update: Waited 25 minutes for session to load have to cancel out of it.......... SAND EQ4 is hanging the session from loading. Didn't get a Error. But Studio One Load GUI was frozen..

As far as SAND Instances Only had.. 3 SAND EQ on Busses,5 SAND Pre4 and 2 SAND Buss4.. All the rest were other Third Party Plugs, Slate Waves, UAD, Eventide, Hofa...
Dual 8 Core Xeon 2.6 (Turbo) 32gig ram

User avatar
giancarlo
Founder
Founder
Posts: 11215
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

Re: SAND out now!

Post by giancarlo » Sun Oct 09, 2016 7:50 pm

best way: open a ticket
Without further details I cannot say anything more, I need someone in the office successfully reproduces it.


From our experience it is very very stable, so when there is an issue on a specific system we need to recreate the perfect conditions

KP2009
User Level X
User Level X
Posts: 136
Joined: Wed Mar 31, 2010 8:46 am

Re: SAND out now!

Post by KP2009 » Sun Oct 09, 2016 8:32 pm

giancarlo wrote:best way: open a ticket
Without further details I cannot say anything more, I need someone in the office successfully reproduces it.


From our experience it is very very stable, so when there is an issue on a specific system we need to recreate the perfect conditions
Thanks Giancarlo. Do you have any Xeon setups in your labs?
Ticket was open the other day..

User avatar
giancarlo
Founder
Founder
Posts: 11215
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

Re: SAND out now!

Post by giancarlo » Sun Oct 09, 2016 8:52 pm

During weekends generally we are not in office!!!
Reproducing a bug requires time

Pro5
User Level V
User Level V
Posts: 55
Joined: Tue Aug 12, 2014 11:28 pm

Re: SAND out now!

Post by Pro5 » Sun Oct 09, 2016 8:56 pm

Hey G thanks for the reply, I do agree somewhat. It's not that I can't hear how Sand is 'better' and seems to hold up more in more cases, it's a balancing act with my system/workflow etc as usual. Sound is ultimately the deciding factor. For the BUS COMP yes I can get others to sound similar, but you are correct, even the best algo bus comp I've tried either sounds faked (slate, ik) or good but not as open/analog/airy (Duende). With Sand it's like you get all the pros and none of the cons (at least sound wise)

I did some more tests with combinations on drum bus of Sand + sand EQ (low and hi boost), then sand comp with duende eq, sand comp with softube EQF 100 eq.

Then I did tests with duende bus comp + sand eq, duende eq, eqf eq.

These were all bounced down to blind test 6 files and I sorted them by overal preference and feel. And sure enough when finished, the "all sand" one was the best overall, it does have some mojo and air, punch, depth that adding an algo eq or algo comp seems to damage.

Sand comp + sand EQ = 1st
Sand comp + duende EQ = 2nd
Sand comp + EQF eq = 3rd (of course the 'british' style EQ was better for this task I just used EQF as a wildcard it didn't really suit the low boost got too muddy nowhere near as tight as SAND).

Then the last 3 were
Duende comp + Sand EQ
D comp + D eq
D comp + eqf eq

which seems to check out. I guess, as I'd be using this kind of setup more on busses than tracks anyway (bus comp and shaping eq) I'd usually only need 2-4 bus comps (drums, mix maybe synths/gtrs bus) so it's not such a problem in the short term to use them just that way, and only use Sand EQ where algos fail me - important tasks on busses, maybe 2-3 instances. That is workable even on my old system if I don't overload it with other plugs, it's just those loading times that put me off, however it's not the end of the world and if in time when I upgrade (not so far away) loading times may improve and I can start using more PRE4s too then it's probably worth it.

I try to justify NOT buying it as I already own lots of plugs and I'm always careful with money but I can't deny it does seem to be giving me results and mixing 'security' I'm not getting elsewhere (I've demoed pretty much every algo plug known to man).

tl;dr? I'm not giving up yet on SAND! I'm going to move to another mix/song next week and try it out there, it's a very punchy upbeat rock/pop song with synth/dance elements and could really benefit from SAND's quality I think.And as you say, the more instances you use over a mix builds up to the higher total quality (I know this from using too many waves plugs in the past which did the opposite - made it sound worse and worse - harsher, brittle, smeared, thin..)

back to music!

User avatar
giancarlo
Founder
Founder
Posts: 11215
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

Re: SAND out now!

Post by giancarlo » Sun Oct 09, 2016 9:38 pm

Agreed!
A thing maybe I never explained enough: when we release a tool we are not interested in convincing all people to buy; we have an idea about something interesting for someone - it could be a console family, a compressor type, a complex effect used in real world from real engineers - and we deliver the best thing possible using our approach. We know our approach is always bringing something new on the table, so there is a need for it. If we fail we cancel the product. If we succeed (we check carefully feedbacks from betatesters, friends and ourselves) we release it. At this point we accept the tool is not for everyone but it will please at least someone, which is enough.
In hardware realm you will find people loving N**e and not A*I, and viceversa. It means they prefer different tools. Even if you love N**e, maybe you love 1073 and not 1081. You love 8016 and not 8068 for the same exact reason. You love first N**e and not recent ams 1081. This is life. We are just happy recreating the different shadows and the more we produce successful emulations, the more they split the audience. Normally a "general purpose clone" leads to more positive reviews than a better matched one.
And here the main point: we are not interested in selling all our tools, but to produce at least a tool which is good enough for you. When I go on a forum and I read user xxxx loves honey and hates purple, and I read the opposite comment from an other user I know I'm doing correctly my job.

Post Reply