|
Post by Johnkenn on Nov 2, 2013 12:35:20 GMT -6
Well, well...
|
|
|
Post by popmann on Nov 2, 2013 17:35:28 GMT -6
Wow...when I find myself in agreement with Ethan Winer, I feel like I should retest. But...yeah....64bit float mixers...if you still have Logic 9 loaded, you can actually run it either way. So you could literally open it as 32bit....render...open as 64bit....render....then compare those files in a third project. It's been ABLE to run 64bit for years. With LogicX, they just forced it--no 32bit and no bridging for plug ins. The latency/buffer thing IS completely unrelated. You're thinking about sample rates...which, given the same software buffer, double rate SR (88/96) will halve converter latency. 192 will halve is again. But, if you're picky like me, you simply buy a cheap analog mixer for cues and never worry about latency of the digital system. But, still record at 88.2, because it's better sounding. There...that should bring Ethan and I back to the discord I'm most comfortable with.
|
|
|
Post by jazznoise on Nov 2, 2013 18:05:34 GMT -6
Not sure why your Mac does that, does it run other things at 64bit?
Considering out destination is 16 bit 44.1Khz, to worry about whether mixing at twice the bit depth or 4 times the bit depth is a bit of a pissing contest. 24 bit goes down to -144dBFSU and no one considers it necessary. Especially when you consider we have about 120dB of a hearing range between perception of atmospheric noise and "Why did you permanently deafen me?".
I'm still working 32 bit with Reaper bridging as necessary. I'm not a perfect engineer, but if you had to list the flaws of my process it'd be way down in the noise floor.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 2, 2013 18:06:04 GMT -6
Well, Ethan, i guess Martin got puzzled a bit by our computer tech babble and meant 4 GB adressable RAM limit in 32bit against 16 GB adressable RAM limit in 64bit OS and not MB and not buffer...
Which by the way is also NOT the adressable RAM for 64bit systems! And so i think there is some need to talk about the real limits of 64bit computer systems as of today - which can be pushed up to regions where they are near meaningless..
A adressable RAM for a 64bit processor based computer system can be theoretically as large as 16 EXABYTE which is 2^64 Byte (a number with 20 digits!).
Which is virtually *endless* RAM for almost all applications. 16 GIGABYTE is the limit of many standard consumer mainboards and e.g. Win7/8 Home versions - the pro or higher versions are limited to 192 GB, which already is very huge. For home needs and common bus structures you can get intel consumer processor boards up to 128GB RAM, e.g. from MSI, since 2011 already.
If this isn't enough, Enterprise Edition Server Windows systems based on Itanium processor architecture have a practical adressable RAM limit of 1 Terabyte (the common size of your "smaller" harddisks, right?) - now. Use SSD harddisks for the rest. No bottlenecks for audio anymore, at least with nowadays audio formats.
Put as many orchestras into your RAM as you might wish. If it's full, use as many streaming samplers (e.g. KONTAKT) additionally without I/O bottleneck inside the computer. This is no science fiction. Reality, now.
So I certainly see technical benefits in using 64bit on modern computers / audio workstations because they all have 64bit processor architectures and therefore it's easier to get clean and fast assembler from the DAWs source code and more consistent data structures, making it easier to program more stable DAW versions with less overhead. And you never have to think about RAM or hard disk performance or stuff like that anymore...(at least for the next years).
Another very common misunderstanding about using 64 bit processing in digital audion workstations has to be debunked: 64-bit processing has nothing to do with the bit depth of the audio that is processed. PERIOD. The "truncation" that takes place in 32 bit processing in relation to 64 bit processing is not a truncation of integers, but of much more precise floating point audio data. Which very seriously is something completely different and not even close to 16bit vs 32bit audio FORMAT.
Sorry, had to write about this, because there seems to be some wrong impressions about the real advantages of 64bit in digital audio computing....might look like a rant...well, it is.
If i run plugins in my 64bit Sonar X3 (or any of the versions before that used 64bit internal...) i really have to say, that i often enough use the UAD-1 card still for e.g. the Pultec, and i still find it very useful. I can not tell from the sound alone, if a plugin is a 64 or 32 bit version. Or if a 32bit plugin is in the chain which should lead in the opinion of the "deeper/better/whatever sound in 64bit" opinions to a permanent degradation. And i really doubt anyone can in a double blind test. E.g. a simple plugin with the same general algorithm in both versions with taking full advantage of the 64bit processing in the 64bit one. I do not only doubt it, i am pretty sure.
What i really notice and see are faster offline/mixdown (very considerably) and more stability when dealing with modern hardware that is run with a mature 64bit environment and well programmed DAW applications in 64bit. Especially when dealing with large and complex projects, bouncing, conversions, etc... And there, time matters - also in digital audio world...
Best regards, Martin
|
|
|
Post by Johnkenn on Nov 2, 2013 20:44:29 GMT -6
Lets move on...
|
|
|
Post by Martin John Butler on Nov 2, 2013 21:50:00 GMT -6
Yep, thanks guys, I really am out of my league here when it comes to deeper computer tech, so thanks for helping me to know a bit more about it.
Ethan, thanks for your kind offer, I bet it would be fun. Unfortunately at this time, my taking time for such interesting, but recreational endeavors would be like a guy stopping off to catch a movie while his house is burning. There are some difficult personal things going on right now. So if you don't mind, perhaps I can get a rain check on an audiophile afternoon for sometime further down the road.
I can't say exactly why it is, but I notice Logic X feels a little smoother, and I find I'm a bit more comfortable listening, able to mix a little longer (I had zero expectation of there being the slightest change in audio quality) It might have nothing to do with it being 64 bits now, but if given a choice, I'd jump in the 64 bit thing, just in case.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 3, 2013 10:28:58 GMT -6
if you still have Logic 9 loaded, you can actually run it either way. So you could literally open it as 32bit....render...open as 64bit....render....then compare those files in a third project. Exactly. SONAR does this too. Even more easily, with no need to reload the project. One audio option is to use either 32-bit or 64-bit math. So you render with one setting, switch, then render again. As Martin explained, this has nothing to do with the bit-depth of the source files. Again, DAWs mix and process audio using math calculations, and each calculation adds a tiny amount of noise and distortion. 64-bit math adds less, but it's all so far below the noise floor of the music that the difference between 32 and 64 is totally irrelevant. --Ethan
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 3, 2013 11:07:14 GMT -6
Ethan, can you come to my studio and hang out/help me hang some panels? :-D
|
|
|
Post by wreck on Nov 4, 2013 9:24:36 GMT -6
I tried to make the jump to 64 this weekend, but I bit off way too much. I tried to upgrade my motherboard and processor, OS (Win8) and upgrade to PT11. I made it through the motherboard, processor and OS over the weekend and have PT11 installed, but not working properly. It seems like I go through this every 3 years or so. I spend an entire weekend trying upgrade and never make it. Despite all of the above about less processing, I still needed more processing power even though my ultimate goal is to process less. Right now my PT output sounds like it's 2bit. Even playing a cd in the computer through my 003 sounds like 2bit. Off to the forums to see what I have done wrong.
|
|
|
Post by Martin John Butler on Nov 4, 2013 10:02:25 GMT -6
Ethan, I believe you, and can understand why you would say the difference between 32 and 64 is totally irrelevant, but that's what I'm talking about. I have a very subtle feeling using 64 bits, something I'm sure I couldn't prove or notice in a test, but I just feel a little more relaxed when using 64 bits in Logic, so I went with it. There may be things happening on Apple's end I'm not aware of. I tend to expect the opposite of what many people do, and I expected to hear or feel no difference whatsoever. So, if I had a bias, it would be that 32 or 64 makes no difference to what I hear. But there it is, I feel something, very subtle, can't explain it, but since using 64 is basically no different than 32 in practical applications, I went for 64.
It is interesting, but I don't have the time to spend researching such things. Perhaps it the math of hundreds of plug-ins and calculations adding some distortions, I don't know, but I don't really care that much about it anyway.
|
|
|
Post by tonycamphd on Nov 4, 2013 11:31:35 GMT -6
IMO, The simple answer is....64bit is better, unless of course you're never going to buy any new software or plugs, and you're going to stick with late 90's OS technology for the rest of your days mate...
|
|
|
Post by Martin John Butler on Nov 4, 2013 14:32:30 GMT -6
That's the facts Jack !
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 4, 2013 14:44:56 GMT -6
Ethan, I believe you, and can understand why you would say the difference between 32 and 64 is totally irrelevant, but that's what I'm talking about. I have a very subtle feeling using 64 bits, something I'm sure I couldn't prove or notice in a test, but I just feel a little more relaxed when using 64 bits in Logic, so I went with it. I'll put $1,000 down - really! - on the side of flawed perception, expectation bias, and the placebo effect. This is the whole point of scientific investigation: to separate belief from fact. I understand that knowing you have your DAW set to 64-bit math makes you feel better, and I'm sure that feeling is real! But this is no different than preferring to drink wine from a wine glass versus a coffee mug, even though the taste is the same. I think someone else mentioned a wine analogy earlier? In this case there is zero cost to using 64 bits, so there's no reason not to do that! But in some cases there is a price, such as using sample rates and bit depths greater than necessary, which wastes disk space and reduces track and plug-in count. --Ethan
|
|
|
Post by popmann on Nov 4, 2013 16:33:25 GMT -6
No, coffee DOES taste different from different mugs. The more accurate analogy would be to explain that an unrelated attribute as the reason. Like coffee tastes better from a cup that's red...because that's the color of the mug that tastes better.
Someone else suggests they could send the coffee from both mugs to a lab and prove its the same liquid...therefore, since the liquid enters and leaves the cup the same, it tastes the same.
|
|
|
Post by Johnkenn on Nov 4, 2013 16:52:04 GMT -6
I was told there would be no math
|
|
|
Post by Martin John Butler on Nov 4, 2013 18:08:35 GMT -6
I use Reidel glassware for drinking wine. They're designed to guide wine to particular areas of the tongue and palette. I didn't believe it could make any difference, until I tried them. I've asked a dozen people or more to taste the same wine from two different glasses and see if they noticed any difference, and they did, all of them. Could I blindfold them and would they be able to say that's from glass A or B? I doubt it, but I say it's the test method at fault, and there are many factors that can skew test results, including the effect of a second taste or third taste on the palate, the factors are endless.
So, we could argue the scientific validity of the glass makers claim, or simply enjoy the wine in the glass we like it in, while someone else is too busy telling me there's no difference to fully appreciate how good it is, ;-)
|
|
|
Post by tonycamphd on Nov 4, 2013 19:37:25 GMT -6
I could put a piece of straw on your head, and you could walk about all day without noticing anything but a couple folks looking at you funny, then i could put a bale of hay on your head, and you'd notice it crushed you for a second before you died. Either way there was a piece of straw on your head? This has no relevance to this thread at all, i just didn't want to erase it after i wrote it. The art of recording music, is the sum of many small things, daw performance and workflow included, you could have 2 dozen pieces of gear that are marginally better or more stable than a slightly lesser piece, differences hardly noticeable at all on their own, but without a doubt, you'll hear the sum of those marginal improvements summed in the end, otherwise you need to acquire more betterer..r..r gear so you can hear that difference....mmmkay All kidding aside, it's very true my tired analogy, the single fly in the room can be unnoticed, 1000 fly's in the room, you notice nothing else. One track with a marginally high noise floor, and a bad word clock= no big deal, 32 tracks with a marginally high noise floor, and a bad word clock= dog doo snow cone! I never understood how people could argue a small difference means nothing, and never look at the bigger picture?
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 4, 2013 20:40:56 GMT -6
Cough.... ehem, please, tony, would you be so kind to tell me if you sum up all flies into one couple of flies, and if you notice them then... Or... if you sum tracks up in the box... what do you expect happens to the noisefloor of 2, 8, 32, 64 or 128 tracks, when they are summed to a 2-track? Do you have the volume of all tracks added? Really? How much voltage do you have from a single track, and how much from the master tracks? Line level and at the master Line Level x128? Does noise behave differently for you and me at summing? Just asking. Because the "noise adds up at summing" argument is one of those evergreens of audio myth...
Sorry tony, i just couldn't resist :-) Best regards, Martin
|
|
|
Post by tonycamphd on Nov 4, 2013 22:54:17 GMT -6
Cough.... ehem, please, tony, would you be so kind to tell me if you sum up all flies into one couple of flies, and if you notice them then... Or... if you sum tracks up in the box... what do you expect happens to the noisefloor of 2, 8, 32, 64 or 128 tracks, when they are summed to a 2-track? Do you have the volume of all tracks added? Really? How much voltage do you have from a single track, and how much from the master tracks? Line level and at the master Line Level x128? Does noise behave differently for you and me at summing? Just asking. Because the "noise adds up at summing" argument is one of those evergreens of audio myth... Sorry tony, i just couldn't resist :-) Best regards, Martin No worries martin, i'm not sure what your getting at, but i think you misunderstood my premise, i was veering slightly off topic, i was referring to small improvements over all of your equipment mostly OTB, even though when you track, your noise floor ITB is only as good as the outboard gear that put it ITB, so...? And yes, your noise floor gets summed at the 2buss like everything else right? just like every piece of gear that sounds harsh sums up, just like every piece of gear that sounds great sums up, my point was all the small things, digital or analog, make the big picture, the better the small things, the better the end result. you know smallbutfine
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 5, 2013 1:18:34 GMT -6
Well. i mean, your flies analogy is not valid. At least not for summing. It would be valid if you put a signal thru noisy channels in series, not in parallel, you understand? ;-) best regards, Martin
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 5, 2013 4:07:22 GMT -6
OK, i try to be precise now and give facts to proof my claim that the efffect of summing up noise and distortion is overrated. A number of tracks N with a noise floor (in-coherent random phasing signal sources) of Y dB add up with a resulting noise like: Y + 10 x log N
Example: 24 tracks with a noise floor showing -95dB in your DAW add up to a signal with a noisefloor of -95dB + 10 x log 24 x dB = -95dB + 13.8dB = -81.2dB. I.e. a noise floor of 10 tracks is always raised by 10dB, 24 tracks always by ~14dB, 32 tracks by ~15dB, 48 tracks by 17dB....128 tracks by ~21dB. Very easy to remember and good for estimation of resulting audio qualities... -95dB is e.g. the real-life measured noise floor of an old $30 list price Behringer USB interface ... Resulting noise floor is not bad at all with no gating/noise suppression involved....
And as for distortion: If you use the same kind of preamp/channel/interface, distortion does not add up at all. The distortion of the sum is the same as the distortion of every single channel. The characteristics of the sum is the characteristics of the channel. That was essentially the idea behind the first Dangerous Music passive summing box that utilized a pair of your favourite tracking preamps for gain makeup for getting the sonic fingerprint of an analog console without having one.....
I mean...it is not like a room full of flies really, isn't it...? ;-) The impact of summing up noise and distortion over many tracks is strongly overrated, as long as the gear is half way decent... that's what i meant and try to say...
Best regards, Martin
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 5, 2013 8:28:35 GMT -6
And as for distortion: If you use the same kind of preamp/channel/interface, distortion does not add up at all. The distortion of the sum is the same as the distortion of every single channel. This doesn't make sense. are you saying 1+1 = 1? Could you go a little bit more in depth on this, like you did with the noise floor thing?
|
|
|
Post by tonycamphd on Nov 5, 2013 10:42:15 GMT -6
I think there is a misunderstanding going on here?...^ i agree, 1+1=2, if you take 2 identical tracks, side by side at the same amplitude, the volume increases over 1, and so on for every duplicate. everyone has seen this when they initially duplicate a track for parallel compressing. I just did this on my mini delta a couple weeks ago, i was having a crackling problem on 1 channel. With a single channel mic pre blasted, the rest set reasonably, the console makes a little noise, engage the rest of the channels in the same fashion, and it sounds like a white noise generator. If your noise floor is marginal, one channel is ok, if you sum 40 of the same marginal noise floors up, you get a lot of noise=fly's analogy, this is my experience.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 5, 2013 13:47:55 GMT -6
Hi, mkatmusic. The distortion e.g. a preamp has does not add up at all, regardless how many tracks you sum. In this case it really is 1+1=1. Let's take a preamp with 0.001% distortion. Record a track. This one has 0.001% distortion. Record a second Track thru the same pre. Also 0.001% distortion. Add them up. Actually NOT 0.002% distortion, still 0.001% distortion! OK, for linear distortion, it is pretty easy to explain. First, linear distortion acts like an equalizer, a constant change of the frequency response. This never adds up. Assume you add up 24 tracks of material from the same preamp. Assume your pre makes a bump at 10khz. Assume, you want to compensate for that with an EQ. Then it is exactly the same if you equalize every single track the same or use the exactly same equalization once on the sum...you obviously do not need 24x as much EQ in the sum, right? I hope, linear distortion does not need more explanation at this point. Since im not a native speaker, it is pretty hard for me to explain...also, i do not think about things like this for years or decades, once understood then i take most of it just for granted...:-( But well, i try hard....: Additionally to linear distortion, there is the relative non-linear distortion, that is the relation of the level of the added spectral content in relation to the frequency content of the signal to the level of the signal itself. (Wow....i do not understand it myself anymore after writing, but am pretty much sure it is right...somehow...) ASSUMING you sum up different signals that then are not harmonically related to each other, which is the very common case in audio summing of different sources, the added distortion in the different tracks takes place in different locations of the frequency spectrum and ends up staying relatively the same level-wise. This is the same for harmonic (Klirr) destortion and intermodulation distortion. Again, this is all for the same preamp for all tracks (like in a console...). If you have different preamps and therefore different kinds of distortion, this acts differently. Hm. I am not sure if i brought the thing over, but i am kind of tired today, so i just stop right here and just want to add a link with alot of valuable information and handy calculators for background information and audio math to get deeper into it... Tontechnik Dipl.-Ing. Eberhardt Sengpiel englishBest regards, Martin
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 5, 2013 14:34:06 GMT -6
No, coffee DOES taste different from different mugs. The more accurate analogy would be to explain that an unrelated attribute as the reason. Like coffee tastes better from a cup that's red...because that's the color of the mug that tastes better. Okay, fair enough. But still, any perceived difference between using 32-bit versus 64-bit math within a DAW is only perception and not real. --Ethan
|
|