Satin goes V1.1

[sascha]

I also wanted to ask a stupid question, why does the minimum value of speed ips knob is set "1.87" ? Is there a special reason for that so specific value ?

1.875 -> 1 7/8 ips = 4.76cm/s = good old cassette speed

[Endor-8o8]

Hello U-He guys !
I recently bought my Satin license and I just wanted to stop by and express my gratitude to your team for making such great tools and instruments !
I also wanted to ask a stupid question, why does the minimum value of speed ips knob is set "1.87" ? Is there a special reason for that so specific value ?
Yes stupid question...

It probably makes more sense when you see that it misses one digit (it actually is 1.875), and that the most commonly used tape speeds are 30.0, 15.0 and 7.5 for high-end machines, 3.75 and 1.875 simply follow the series of 'powers of two' downwards to the low end.

Ah yeah you're obviously right ! Thanks mate !

[Spitfire31]

FWIW:
On a project with four instances of Satin (AU v1900) running in Reaper v4.21, I saw about a 10% CPU load increase after upgrading from v1837.
iMac 3.07gHz Core 2 Duo

So, what do you mean by "about a 10% CPU load increase"?

10 per cent of, say, a 20 percent CPU load, is an additional 2% (total of 22%).

But if you instead mean percentage units, the total from the above example would be 30%!

The difference isn't insignificant…

/Joachim

[Endor-8o8]

I also wanted to ask a stupid question, why does the minimum value of speed ips knob is set "1.87" ? Is there a special reason for that so specific value ?

1.875 -> 1 7/8 ips = 4.76cm/s = good old cassette speed

Like in my miserable walkman during my teenage years ?

[ho66it]

FWIW:
On a project with four instances of Satin (AU v1900) running in Reaper v4.21, I saw about a 10% CPU load increase after upgrading from v1837.
iMac 3.07gHz Core 2 Duo

So, what do you mean by "about a 10% CPU load increase"?

10 per cent of, say, a 20 percent CPU load, is an additional 2% (total of 22%).

But if you instead mean percentage units, the total from the above example would be 30%!

The difference isn't insignificant…

/Joachim

I repeated the observations today and made less hurried observations. I used the same project with no changes besides the version of Satin. I am using the "Activity Monitor" app built into OSX. During the monitored interval I observed:

v1837 - Minimum CPU% 47.7%; Maximum CPU% 53.3%
v1900 - Minimum CPU% 52.5%; Maximum CPU% 57.7%

So I calculate increase at minimum observed CPU% to be an additional 9.1%; increase at observed maximum CPU% to be an additional 9.2%. Not percentage units, just additional CPU usage.

If this was at 20%, not a big deal at all. Here at ~50% not a deal breaker either, but I did have to go and shut down some other running processes to avoid crackling audio. One additional thing that I noticed is that when I open the Fx GUI's the CPU load goes up markedly. If I opened three Satin GUIs I would see over 70% CPU load. (Increased to ~70% from ~50%.)

[sascha]

That looks strange. Are you talking about a single instance???

We did some further testing here at the HQ, and one of our iMacs ran more than 50 instances in Reaper. Perhaps Rob can give precise figures. It was intreaging but one has to note that, although he stuffed them all on one track, many of these ran on different cores, and it's actually hard to predict which one goes where.

I've noticed that on the PC, too. Reaper dynamically divides multiple instances up among seperate cores. In contrast, Samplitude (which I'm using mainly) does load balancing on a per-track basis, if possible. The latter gives me more predictable figures and is always my main benchmark tool (since I've known its interior myself...). On my i7 3.6Ghz, I can run 23 instances *in series*. And I keep record on this. When benchmarking, I use real large project buffer sizes (up to 16k) and the WDM driver model in order to not have any fluctuations or load spikes from the ASIO sub system. If there's an increase in CPU demand, I immediately notice, and I'm very picky about this.
Currently (v1.1, rev1900), an instance (in studio mode) uses 4.3% of the entire CPU, and v1.0 (rev 1873) was using 4.2%. A slight increase came in lately because of the more sophisticated wow & flutter which includes a change in the virtual transport model. But that's a relative increase in 2.4% only (per instance).

In your case, I can either assume there's something wrong with your setup (driver/buffering mismatch, background processes, PCI bus latency, hybernation/cpu stalls, temperature etc.), or the cpu is switching to denormal mode on occasion, for whatever reason. Hard to diagnose remotely, but perhaps you can provide us with as much info on your system as possible, probably in conjunction with a system diagnosis/health tool. Does something exist on the Mac?

[Urs]

our iMacs

pssst - we have MacMinis, not iMacs.

[ho66it]

That looks strange. Are you talking about a single instance???

We did some further testing here at the HQ, and one of our iMacs ran more than 50 instances in Reaper. Perhaps Rob can give precise figures. It was intreaging but one has to note that, although he stuffed them all on one track, many of these ran on different cores, and it's actually hard to predict which one goes where.

I've noticed that on the PC, too. Reaper dynamically divides multiple instances up among seperate cores. In contrast, Samplitude (which I'm using mainly) does load balancing on a per-track basis, if possible. The latter gives me more predictable figures and was my main benchmark tool (since I've known its interior myself). On my i7 3.6Ghz, I can run 23 instances *in series*. And I keep record on this. When benchmarking, I use real large project buffer sizes (up to 16k) and the WDM driver model in order to not have any fluctuations or load spikes from the ASIO sub system.
Currently (v1.1, rev1900), an instance (in studio mode) uses 4.3% of the entire CPU, and v1.0 (rev 1873) was using 4.2%. A slight increase came in lately because of the more sophisticated wow & flutter which includes a change in the virtual transport model. But that's a relative increase in 2.4% only (per instance).

In your case, I can either assume there's something wrong with your setup (driver/buffering mismatch, background processes, PCI bus latency, hybernation/cpu stalls, temperature etc.), or the cpu is switching to denormal mode on occasion, for whatever reason. Hard to diagnose remotely, but perhaps you can provide us with as much info on your system as possible, probably in conjunction with a system diagnosis/health tool. Does something exist on the Mac?

Sascha, Thank you for your response. I had four instances running, two on one track and one each on two other tracks. I will put together as much info as I can and attach to an email for you. I would be very interested to know if it's the way I have things set up.
Regards.

[sascha]

our iMacs

pssst - we have MacMinis, not iMacs.

Argh. This outa-space aluminum technology is alien to me

[sascha]

I will put together as much info as I can and attach to an email for you. I would be very interested to know if it's the way I have things set up.

Ok, thanks. I'm out-of-office for the rest of the week now, but I guess Rob can sort it out with you to a greater detail.

[wnick]

in delay mode the high cut knob needs a little attention - the characteristic isn't on par with that of the bass cut.
Nothing serious

I forgot
Satin 1.1 win7/64 Reaper 4.52 64

[Basspartout]

Hey
just tested the new update on Presonus Studio One 2.6 on my WinXP 32bit home PC - With V1.1 I now get a nasty audible Full CPU Peak every time I press the Bypass Button in the PlugIn.(both VST3 and VST2) That hasnt happened before. Also the GUI flickers more wildly all over the screen now when opening Satin. ( Was just an almost unnoticeable flicker before.) Now it looks as if the Window opens at three different places in short succession.

Will test it tomorrow in the Studio on WIN7 64bit.
Grüße
Sascha

[Tp3]

Any chance of making the two last dots of the "peak bar" in the color red ? and let them fade only after 3 seconds or so (if they aren't surpassed by a new, higher, value)?

Kinda like the meters on The Glue or SixtyFive...

[sascha]

Any chance of making the two last dots of the "peak bar" in the color red ?

Erm... the region from 0 to 3db IS red, or did I miss something here... ?

[Garycactus]

Yes, Satin 1.1 uses more CPU now.
Logic can't play some of my songs anymore.....

I kept version 1.0 of Satin, will be downgrading today!