SSD worth for Libraries? DAW?

[kingozrecords]

It's easier on the cpu. It depends on the platter and headers, 4 is better. The 8g/b uses two lanes of in and out. It's not much diff because usually 2 and 2 lanes are used for in/out anyways, coming from a technical aspect. Sata and m.2 have the same amount of lanes, their in/out lanes will always be limited to 2/2 if both reading and writing; I'd think honestly gaining more is impossible. If pcie 2.0 of course even worse.

The strength of ssd is the accessibility of the file and the lack of buffering. but, like I said with lanes, depending on the files you're accessing you still have to buffer the same way. I imagine that would be in the sum of about 8-12gb/s. So that would be the max per the 2 lanes you'd likely have available and then you would divide that power by the number of active connections. SSD is only better by a few gb when you consider everything; but of course those gb are great to have.

I made this song and video thinking of this conversation. It was done live. The audio though; after re-processing in videopad has a few static pops that I couldn't track down in the wav file I used to remaster it.
Tells Me I should use openshot for the same next time so as to replace the audio.

[Guenon]

It's easier on the cpu. It depends on the platter and headers, 4 is better. The 8g/b uses two lanes of in and out. It's not much diff because usually 2 and 2 lanes are used for in/out anyways, coming from a technical aspect. Sata and m.2 have the same amount of lanes, their in/out lanes will always be limited to 2/2 if both reading and writing; I'd think honestly gaining more is impossible. If pcie 2.0 of course even worse.

The strength of ssd is the accessibility of the file and the lack of buffering. but, like I said with lanes, depending on the files you're accessing you still have to buffer the same way. I imagine that would be in the sum of about 8-12gb/s. So that would be the max per the 2 lanes you'd likely have available and then you would divide that power by the number of active connections. SSD is only better by a few gb when you consider everything; but of course those gb are great to have.

You do not know what you are talking about.

It's tricky commenting on your posts/threads, and one should always think twice, because - especially for the casual observer - it just makes the one commenting look like a meanie, for repeatedly correcting and talking back , and it just keeps going on, if one expects a constructive response from you acknowledging that something indeed is wrong. However, you are repeatedly presenting yourself as an expert here, so at least occasionally someone should comment, in my opinion. Otherwise it's just... anything goes, man.

In other words, yes, it's pretty inevitable that it only becomes a drawn-out conversation where people who actually know their stuff are just shaking their heads and feeling a kind of frustration and even sympathy, and chances are that the conversation will not lead to you correcting your points or realizing that you are talking about things you don't understand, anyway.

"Someone being wrong on the internet" is the name of the game here, and so be it, haha. You just do that a lot, while at the same time saying things like "in sound quality I can beat any company that makes audio software", "I can prove to you that I make the very best software in the world", and so on (direct quotes from your youtube video stuff).

You have claimed that using vector graphics in the GUI of a plugin, versus static bitmaps, causes aliasing in the actual audio signal, because it "needs more math." You describe numerous things that don't really work the way you say they do. And when someone points these things out, it's a no go. But I repeat, at least occasionally someone should do it, as otherwise it's just... anything goes, man, whatever.

So, with this Kontakt and SSDs stuff, random technical details are one thing, but knowing the depths that await if correcting you on matters like these, I'm trying to strictly keep it relevant to Kontakt performance here: it has already been explained previously what is important in a storage device for Kontakt library use, and how the massively better random read performance of an SSD, reading from a large sample pool, makes a difference in practice when using disk streaming in Kontakt.

But instead of even addressing actual Kontakt performance, you just say these things. Like in your last post, quoted above. You often point out throughput values - and disparate things contributing to theoretical maximum throughput - and so on. There is no real idea behind it, how it all comes together, and what the actual discussed use case needs.

Consider that, for example, a single one-layer sampled voice at 48 kHz 24 bit, when played back at its natural sampled speed, is 288 kilobytes per second of sample data. Even this is assuming uncompressed data, and for mono recorded sources it's half of that, but just some hundreds of kilobytes per second gives you a ballpark idea. What do you think are the actual limiting factors of a mass storage device these days for real-world Kontakt voice counts, when one voice sits at those figures for the required raw sample data being played back? Is it just the maximum throughput of the device, or does something else factor in, limiting the operation in this use case well before that? What might those other factors be, why do they come about, and might SSDs give you significantly better performance because they happen to nail a good combination of those factors?

To refresh your memory, in an earlier video you say:

If you were to be running many different SSHDs, in my opinion, as long as they were the two terabyte model of the FireCuda, with the four heads on a platter, then I think that they would be so much faster than an SSD that the difference would be appalling. And the reason is because, a hard drive based on a file allocation table, cluster size, which is usually NTFS, is naturally gonna be able to load them faster. The amount of connections a hard drive has, is... is... is... is quadruple, cos' it's got four arms, reading at the same time, and then providing that data onto that eight gigabytes of SSD memory. This is Robert from DSPplug, I hope I've given you some really great ideas for your Kontakt, for your sampler, for your SFZ and your audio libraries in general, like DirectWave for instance, and that maybe you can save a little money as well. I'll try to find a two terabyte model of the Seagate FireCuda so that you can hopefully save a little bit of money and have a faster, more reliable experience.

...

Again, you just say things. I have no wish to use much time on this again, and I am absolutely sure anyone who is actually looking for info on how Kontakt libraries perform can make their own call on whether to take your word for it or perhaps ask the official support of the tools in question and see if they agree.

But all this is how you usually operate, just throwing things out there like that. A random example from another recent video I checked out:

That's the difference between using stereo and mid/side. Mid/side has a lot of additive aliasing, and it's just, it happens cos' the math is naturally gonna do that, and there's no way to get around that.

Would you wanna record mid/side? Lord, no. [Laughs.] Maybe the real deal, but I mean, that's uh, it's very rare, if you would, like, find a classic mid/side mic, like they used to use back in the day. I'm telling you, that would be a very rare affair indeed.

Just saying things. Heaven forbid if someone started a discussion on the technical claims and implications of stuff like that. But your stuff is just littered with things like this.

Anyway, last year you even recommended SSDs for sampler use and said "The speed on hdd just is not enough for a sampler imo", and now it's different. The technical stuff coming from your direction is all over the place and random and even goes a full 180 degrees sometimes (like above), and yeah, I know it's just... not worth it to get involved, really. You take zero criticism and keep doing stuff like this over and over, regardless of the subject. Sometimes, though, someone needs to comment, you know, just once in a while, as otherwise anything goes.

[fairlyclose]

That's how you operate, just throwing things out there like that. A random example from another

Would you wanna record mid/side? Lord, no. [Laughs.] Maybe the real deal, but I mean, that's uh, it's very rare, if you would, like, find a classic mid/side mic, like they used to use back in the day. I'm telling you, that would be a very rare affair indeed.

Just saying things. Heaven forbid if someone started a discussion on the technical claims and implications of stuff like that. But your stuff is just littered with things like this.

Anyway, last year you even recommended SSDs for sampler use and said "The speed on hdd just is not enough for a sampler imo", and now it's different. The technical stuff coming from your direction is all over the place and random and even goes a full 180 degrees sometimes (like above), and yeah, I know it's just... not worth it to get involved, really. You take zero criticism and keep doing stuff like this over and over, regardless of the subject.

that is pretty weird about mid/side, which is an incredibly standard technique where the maths has no impact on aliasing at all - cant see how it possibly could, makes no sense at all. Almost always two mics are used for mid-side so the mic comment is also strange

[starflakeprj]

Everything (but backups and temporary storage) should be at least SSD, it is definitely worth it.

[Krixu]

I keep my samples on a hdd. I personally don't feel that it's necessary to drop an extra bag for speed.

[fairlyclose]

I keep my samples on a hdd. I personally don't feel that it's necessary to drop an extra bag for speed.

depends on the software to some extent - for large libraries Falcon benefits enormously from a fast drive, Kontakt not as much.