Aloha guys, (perhaps this should be in the ‘Hardware’ forum)
Till recently to do serious and consistent DAW recording there were
certain hard drive ‘techniques’ that would help.
1-one HD for the programme
2-a separate HD to read/write audio files
3-another HD to store/read samples
4-perhaps a 4th HD for storage. (always good advice)
Looking at the current read/write speeds of todays SSD’s, is this all moot?
Can we now just use one SSD for 1-2-3 and a spinning HD for number 4.
I am now doing this (with no probs) using my rig and
wondering if anyone else is also doing this?
Is the SSD big enough? Even for sample libraries? Sounds intriguing.
I had a 512G (from Crucial) installed and am about to get another 512 external connected up
Should be enough for (most of) my clients.
( and no more de-fragging and optimizing etc).
But my point is, if you get a new ‘puter with an SSD these days,
most of the ‘hocus pocus’ to get things to work (and to continue to work)
has been removed.
SSD is best for booting and applications, store sample libraries on SATA, unless you still have IDE
Most SSDs are SATA2 or SATA3.
And Steinberg actually recommends SSDs for sample libraries.
One thing to remember for SSDs though:
The more writes that are done to an SSD, the shorter its lifespan (timewise).
SSD won’t mean anything once a sample is loaded…
unless you wish to use disk streaming to lessen the load on your RAM.
ah yes, DFD native instruments style but I don’t know the Steinberg equivalent.
Good point shinta, now that there is a shift towards 64 bit it could be more or less relevant depending on your plugins.
I personally would like at least 1 ssd for booting but I’d not be too keen to store files if I could help it, ie desktop which makes me wonder if on MS windows can you store the desktop folder on another file system?
The Grand 3 has disk streaming. As does Halion 4.
A hard drive and SSD are more than likely going to have the same file system.
Windows doesn’t care what physical media it stores data on. The file system that Windows 7 is going to insist on is NTFS. This controls the disk pointer table format. The actual storage of the data bits is up to the controller of the SSD or Hard Drive.
On this computer that I’m using, my boot drive is an SSD and my Windows profile is stored on a standard hard drive.
A SSD’s high read/write speed is able to take advantage of newer interfaces’ speeds (ie SATA 3’s 6Gb/s), because the data can be read from the memory faster.
A normal hard drive has two speeds:
- The speed at which the data is read from the platter and loaded into the drive’s cache. This is dependent on rotational speed.
- The speed at which the cache transfers data to the controller. This is faster than the platter read speed.
The “bottleneck” of a standard hard drive is the platter read speed.
Having more drives share the data (ie RAID 0 style) means that one bit of data is read from one drive while the next bit is simultaneously read from another drive, thus alleviating the bottleneck somewhat.
Even with an SSD, multiple drives will give faster speeds. However, this is expensive and usually not necessary due to the high speed of a single drive.
SSD’s memory only has a specific number of read and write cycles (like all solid state memory).
(Let’s face it: an SSD is basically a USB thumb drive with a better (faster) internal controller, faster memory cells, and a SATA interface in place of USB.)
The write cycles are more limited than the read cycles (read cycles being much, much higher).
In theory, if an SSD stops being able to write data, you should still be able to read the data from it; thus allowing data to not be “lost”.
Both types of drives have their own failure methods as well.
A SSD can have its components fail (controller, memory cells).
A hard drive can have one of three things (usually) happen:
- Controller fail (can happen, but not usually)
- Platter motor failure (can’t move the platters, can’t read the data).
- Read arm motor failure (can’t move the arm containing the reading head, can’t read the data).
While electronic component failure can happen, it usually isn’t the leading cause of hard drive failures.
The motors are much more prone to failure. While the data can be recovered, it is usually costly.
Watch out for the “click of death” (when the read arm motor starts to go, it will sometimes make a “clicking” sound).
You forgot The Most Important one: Access speed (or access time, or seek time). HD always has to physically move it’s read/write head into desired track and wait for the platter to platter to spin into the sector containing the data. This time is HUGE (many milliseconds). Imagine OS/applications/audio/samples at the same drive: OS/applications are most probably at one end of the disk (they got written there first) and you audio files at another end (they were written there just few minutes/hours/days ago).
Now, in the situation you are streaming your audio from the disk AND at the same time your OS or application wants some data from the disk. Your read/write head has to move into end of the disk containing OS/applications. At the same time your DAW application desperately needs it’s audio stream. The poor read/write head is banging around the disk platter to fullfil all the requests and spending more time moving around than actually reading the data. And soon (as your disk cache is exhausted) you’ll be getting dropouts in your audio stream.
Enter SSD: Access time? Hey … it’s virtually ZERO compared to HDDs. And what’s most important: it’s always the same. So, in the case your interface speed is enough, you won’t get surprises.
Based on your setup, I can tell you how you’re going to see improvement. It’s then up to you to decide what is considered “marginal”
Faster boot times and faster program loads (for data into RAM, don’t know how this will affect Cubase’s load times).
Could be marginal improvement.
Faster saving times and load times. SSD will also have less data loss worries.
Again, could be marginal.
If your VST supports disk streaming (like you’re asking in the other thread) then a SSD will shine. This will also free up RAM for other plugins and processing.
If you have a large range of notes and are using sample-based content, this could be a huge RAM save.
It really depends on what vexes you the most.
A slow computer boot and program load?
A slow project load/save time?
Too many samples clogging the RAM?
I’ve always kept my samples on one drive, audio on another, and OS on a third. All regular fast hard disks and have never had a problem. Is there really a problem to solve here?
I mean, faster is always better to some degree, but what “no brainer” are we talking about? This has been a non-issue for years.
You CAN stream from a fast hard drive, but even the fastest hard drives out there (comparison drive: WD Velocoraptor @ 200 MB/s sustained) pale in comparison to a SATA3 SSD’s speeds. A SATA3 SSD nearly maxes out the 6Gb/s that the controller offers (comparison drive: Crucial M4 @ 500 MB/s sustained).
You’re more likely to get dropouts and missed samples with a regular HD.
I’ve seen upwards of 1TB SSDs…for about $2750.00 CDN.
I’d say that when SSDs come down in price, we’re going to be seeing less and less of regular HDs. They may even disappear entirely.
Probably. But that day isn’t tomorrow. For a long time there will be applications (not talking about DAWs) which will be more cost-efficient with HDDs.
IMO for a regular DAW a single SSD should be enough already (if you can afford to buy a large enough one). It’s just a matter of your storage space requirement. If everything doesn’t fit into single SSD, buy a HDD for your audio, if you still run out of space buy another for samples. If you’re now out of money and can’t afford a SSD fo´r as your OS/application disk, buy a HDD for that too.
Of course, there are special cases which break my rule:
- Limited audio streaming requirements and heavy sample use with huge sample library: put OS/applications and audio on one SSD and samples at HDD.
- Limited sample use, but requirement of having huge audio projects: put OS/applications and samples on one SSD and audio at HDD.
Samples on HDD is fine once they are in memory, which is what this struggle around 64 bit is all about so we don’t need DFD.
That’s why using SSD for system/applications drive is advantageous.
A SSD in this instance would only provide extra speed (other than at startup) if the application was accessing the disk for files located in its install directory.
To my knowledge, most programs have their major disk accessing at startup and then not much once they are in RAM.
I don’t know how often Cubase accesses the HD for files within its install directory once loaded.
Even so, unless those files are huge, not much of a noticeable performance gain there.
I’ve used a three HDD system for years and it has performed flawlessly, speed never being an issue (except pproject loading times) but then again, I don’t use hundreds of tracks.
I recently bought (just 'cause I can;)) a 128GB Crucial m4 SSD, partitioned for my two W7 OSs and that is fine too, startup time is really nice and quick.
I probably will soon buy another larger SSD for my sample libraries and would expect project loading times to speed up. A bit slow at the moment as I’m using Trilian, Omnisphere, Superior2 and Kontakt a lot on several tracks and some of the sample sets are quite big.