|
Post by Johnkenn on Dec 18, 2019 14:38:26 GMT -6
I was looking in Cubase and noticed it has an option for 64 Bit Float...obviously, this will make my music twice as good, right?
|
|
|
Post by EmRR on Dec 18, 2019 14:57:04 GMT -6
Processing or file storage?
|
|
|
Post by svart on Dec 18, 2019 15:18:57 GMT -6
In layman's terms, it's a way to represent an increasingly large number that allows much smaller pieces of data to be processed. Think of a pie (mmm, pie) that's cut into 8 pieces. There's 8 pieces that represent the whole pie, but you can only eat a slice that's the size it is. Maybe you wanted a smaller slice, so now you will need to divide each piece into two pieces, giving 16 pieces that make up the whole pie..
And so on.
So you get to a point where you can't divide the pieces up lengthwise anymore, so now you want to cut them width-wise now so each piece is half as long, but now you're cutting them into fractions of each piece, which can't easily be described by a whole number. You "float" them and now you can describe a whole number, but also describe where to put the decimal place in that number to create a number less than 1. It's like a shortcut similar to scientific notation, but in a way code and processors with a set number of bits can work with.
I guess this is a pretty bad explanation, I'm sure a programmer could explain it better, maybe more simply.
|
|
|
Post by jeremygillespie on Dec 18, 2019 15:21:33 GMT -6
Hold on....
We taking a pizza pie or something like a salted caramel apple pie?
It better not be one of those Chicago style pizzas. I don’t even know what to do with those things.
|
|
|
Post by ragan on Dec 18, 2019 15:38:36 GMT -6
Computers, at their most basic level, just carry out instructions. Each instruction in a 64-bit architecture can be 64 bits. Each bit can be a 1 or a 0. For a given architecture, each place in that 64-bit "word" means something particular. A certain chunk of bits will signify what type of instruction it is, another chunk may specify a register to go grab some data from, another chunk may designate another register, another chunk may present something conditional (like 'if the first register holds a larger value than the second register, skip to _________ address and go from there...'). So, generally speaking, 64-bit processing means the data path of anything flying around in a program can be 64-bits wide, which gives value variance of 2^64.
Think of it like a town with 64-lane streets where every address is a 64-bit collection of 1s and 0s. A program drives around picking up data from addresses and taking it to other addresses and doing things with it.
Edit: Oh and the "float" part is just talking about how you deal with non-integers. There's a sort of complex way of writing and processing non-integers where one bit (furthest left) denotes the sign of the number, the next chunk of bits gives the exponent and the last chunk gives the fractional portion. It's just a way of using non-integer numbers in your calculations where instead of a decimal point you have a "binary point" which you can move around like you would in scientific notation (like Svart was saying). But it's implemented as 64-bit instructions because that's the only thing a processor can understand, ie a processor doesn't know what "multiply" means, but it knows how to "shift" one way or the other.
|
|
|
Post by Bender on Dec 18, 2019 16:03:21 GMT -6
It better not be one of those Chicago style pizzas. I don’t even know what to do with those things. Hey there guy, deep dish pizza is AMAZING so shut that mouth lol.....and a true chicago style pizza is a thin crust za TAVERN CUT into squares, don't let anyone tell you otherwise. If you wanna debate if deep dish is indeed pizza that's another bag of worms entirely, but no matter how you slice it, it's delicious; I don't see the animosity over legitimacy when it comes to the increasingly popular Detroit style pizza either, and that's a damn rectangle!
Sorry, anyone else hungry?
|
|
|
Post by ragan on Dec 18, 2019 16:11:44 GMT -6
Starving.
|
|
|
Post by popmann on Dec 18, 2019 16:24:40 GMT -6
If it factors in for you...my tests say that Cubase 10 (specifically) 32 and 64bit null....UNLESS you use the channel EQ (theoretically the channel strip stuff, but I didn't have any of that active in the test--I only noticed the EQ, because it nulled EXCEPT little bits of specific tracks where...I used channel filters.
LogicX's 64bit not only doesn't null with 32bit...it sounds noticeably better than 32. I think Apple took the opportunity to "fix" their 32bit summing code written in 1995....leaving the old one there for backwards compatibility (obviously?)...where Steinberg has insisted that 64bit float resolution is not needed forever...and they just got pressured by the market to tick the box. \
I WILL say...that I've been using their "channel strip" more lately...and my 64bit box is checked because...why would you NOT? It's beaten out and equaled some third party stuff in a way that surprised me--having tried it when they first put it in and thought "ehh--pass"...maybe I tried their strip before they did the full 64bit audio path?
Harrison maintains that Mixbus uses both 32 and 64bit float at various points in their mixer as needed...properly dithered to the output resolution. It sounds great, I'm not too into gift horse mouth inspection.
EDIT
Also..."double precision" processing has been done for a long time. In hardware and software...I feel lie the Waves L2 and stuff is running 64bit float internally. You could always do that, even when say Apple only gives you a 32bit AU bus. Just like a lot of DSP hungry plug ins are running 96khz and higher internally despite you sticking with 48khz sessions. The setting in Cubase is for the mixer's resolution--so it's faders, pan law, bussing, EQ, and THEORETICALLY (though I don't know this) the "channel strip" DSP. Not sure about the VST buss of plug ins (since I do know Apple's AU is 32bit at least prior to Catalina).
|
|
|
Post by kcatthedog on Dec 18, 2019 16:25:22 GMT -6
I followed the pie and neighborhood ideas, so went and bought 64 types of ice cream and 64 types of coffee, did I get it right ? What time you guys showing up ?
|
|
|
Post by Johnkenn on Dec 18, 2019 16:32:22 GMT -6
So...I shouldn’t worry about changing my bit rate from 24 to 64?
|
|
|
Post by jeremygillespie on Dec 18, 2019 16:56:43 GMT -6
It better not be one of those Chicago style pizzas. I don’t even know what to do with those things. Hey there guy, deep dish pizza is AMAZING so shut that mouth lol.....and a true chicago style pizza is a thin crust za TAVERN CUT into squares, don't let anyone tell you otherwise. If you wanna debate if deep dish is indeed pizza that's another bag of worms entirely, but no matter how you slice it, it's delicious; I don't see the animosity over legitimacy when it comes to the increasingly popular Detroit style pizza either, and that's a damn rectangle!
Sorry, anyone else hungry? Hey man any style - I’m game. I literally just don’t know what to do with Chicago pizza. Pick it up and slop it over my beard? Fork and knife? Burn the roof of my mouth? I had great pizza last time I was in Louisville too. You gotta give it to the hipsters - they do a pretty good pizza game and it gets people away from Dominos at least!
|
|
|
Post by popmann on Dec 18, 2019 17:13:06 GMT -6
So...I shouldn’t worry about changing my bit rate from 24 to 64? That's the bit depth of the FILE format. Cubase has never been a 24bit mixer that I'm aware of--certainly not in this century. There's no reason to TRACK with it set to anything but 24bit integer. Once you get into editing and rendering, you can change that in Cubase--it will only use the floating point moving forward from that point for edits/freeze/renders.
|
|
|
Post by Blackdawg on Dec 18, 2019 18:17:39 GMT -6
This has more to do with the DAWs processing than anything. 32 and 64 bit processing has been around for ages but only recently has 64 bit architecture really started to be the main way to do things though many many apps have been using it for years.
From our stand point it really just helps how our DAWs process audio/plugins/ect.
|
|
|
Post by Johnkenn on Dec 18, 2019 18:35:01 GMT -6
So...I shouldn’t worry about changing my bit rate from 24 to 64? That's the bit depth of the FILE format. Cubase has never been a 24bit mixer that I'm aware of--certainly not in this century. There's no reason to TRACK with it set to anything but 24bit integer. Once you get into editing and rendering, you can change that in Cubase--it will only use the floating point moving forward from that point for edits/freeze/renders. So it could possibly speed up renders and freezing? That sounds good to me.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 18, 2019 18:41:28 GMT -6
Changing the DAW to 64-bit, won't automatically change plugins to 64-bit. It's true that 64-busses and mix-gain may have some advantages, but that's more likely to turn up in huge sessions. The majority of plugs will probably stay largely 32-bit for quite a while. 64-bit floats require a lot more processing and will often require that many operations make calls to a library rather that being native to the processor. So there's a non-neglible chance that performance would take a real hit.
Still, there are places where 64-bit processing is worth that hit. That's especially true in some types of EQ. But plugin developers have been able to selectively do 64-bit processing for years. For example, most Exponential Audio plugins have long done EQ as 64-bit, even when the rest of the plug is 32-bit.
|
|
|
Post by popmann on Dec 18, 2019 19:43:24 GMT -6
That's the bit depth of the FILE format. Cubase has never been a 24bit mixer that I'm aware of--certainly not in this century. There's no reason to TRACK with it set to anything but 24bit integer. Once you get into editing and rendering, you can change that in Cubase--it will only use the floating point moving forward from that point for edits/freeze/renders. So it could possibly speed up renders and freezing? That sounds good to me. If anything, it will slow it down. twice as many numbers to crunch and write. Steinberg has been very open about the performance penalty of setting the mixer to 64bit. I've not really seen that, but I might just have a lot of CPU headroom. The advantage is academically a sonic one. Basically--rather than either adding dither noise OR quantization errors to every single thing you render/crossfade/tune/bounce....it leaves it in the format that the DAW's MIXER is always running at...so that you keep it in floating point format until it's time to master and reduce to whatever.
|
|
|
Post by swafford on Dec 19, 2019 2:03:53 GMT -6
So...I shouldn’t worry about changing my bit rate from 24 to 64? No man, you should worry if you have enough pie to satisfy your guests. I think.
|
|
|
Post by svart on Dec 19, 2019 7:12:03 GMT -6
So...I shouldn’t worry about changing my bit rate from 24 to 64? No man, you should worry if you have enough pie to satisfy your guests. I think. All you can eat? Do I look like a man who's had all he can eat?
|
|
|
Post by popmann on Dec 19, 2019 12:33:49 GMT -6
Define "satisfy". I am suddenly very hungry for pie.
|
|
|
Post by Johnkenn on Dec 19, 2019 19:02:38 GMT -6
|
|
ericn
Temp
Balance Engineer
Posts: 15,022
|
Post by ericn on Dec 20, 2019 10:44:47 GMT -6
That's the bit depth of the FILE format. Cubase has never been a 24bit mixer that I'm aware of--certainly not in this century. There's no reason to TRACK with it set to anything but 24bit integer. Once you get into editing and rendering, you can change that in Cubase--it will only use the floating point moving forward from that point for edits/freeze/renders. So it could possibly speed up renders and freezing? That sounds good to me. To Quote the great Barney Stinson author of the Bro Code “ I have one rule new is always better”. Now at some point it probably will be better, once developers all start righting for it, but for now it’s an easy selling point that really doesn’t mean much other than it goes to 11!
|
|