This question is a result of my helping a friend select and setup a PC for dedicated Hard Drive wiping and only pertains to PCs with Parallel ATA Hard Drives. http://en.wikipedia.org/wiki/Parallel_ATA
I was under the impression that since Parallel ATA Hard Drive Controllers are standard that they would be ~ the same and the Parallel ATA Hard Drives with different rotational speeds and caches would be the significant variable. Am still sure this is correct but was surprised with the significant differences between ~ same age PCs write speeds using the same 40GB Hard Drive.
This discovery was made when we decided to check the Write Zero Speed of DBAN http://www.dban.org/ on various old PCs. These tests were all made with the Hard Drive as the Primary Master and with No Primary Slave.
Since I am a beginner concerning Audio Recording with a PC would be interested in any experience, suggestions… concerning PCs with Only Parallel ATA Controllers and Hard Drives.
Hope that someone else has also noticed significantly different write speeds with the same Hard Drive with various PCs.
Since many posters don’t know they even have hard drives, this may be more interesting than you think.
Audio (and video) editing doesn’t wait. When that B-Flat note comes down the pike, the computer has to find somewhere to put it right now, because there’s going to be another note right behind it.
That’s what kills storage. It’s not the absolute speed of the drive as much as quality of the file management and competition. Audacity will not work into a highly fragmented drive, it doesn’t matter how big or fast the drive is. You can do credible (careful) production on a very old computer with really slow drives, as long as you have lots of room and a well-behaved file system.
It regularly kills designers when they use the common data transfer numbers to spec out an entertainment system. Transfer speed isn’t nearly as important as sustained data transfer. The drive can’t stop to lick its paws and preen in the middle of a live band capture.
“Audio (and video) editing doesn’t wait.” Agree, that is why decided to test what the sequential write speed is for various PCs.
“Transfer speed isn’t nearly as important as sustained data transfer” Agree
“It’s not the absolute speed of the drive as much as quality of the file management and competition…You can do credible (careful) production on a very old computer with really slow drives, as long as you have lots of room and a well-behaved file system.” Also agree but there must be a minimum sequential write speed before the PC starts to drop data? Am sure that this is much lower for audio than video.
I don’t think the problem would be in the hardware, but mostly in the software…
Latency and speed on any recent hard drive should be more than enough to allow to record comfortably any audio stream at the required speed without dropouts.
The main issue here I think is the software… Beginning with the file system, but ultimately regarding the operating system ability to manage it efficiently.
If the operating system is accessing many files at the same time at different parts of the disk that might get you in trouble and that’s probably the most likely reason why you would have trouble writing audio data to the disk. That’s why they recommend you to close all apps before proceeding with any recording… Even then there are still dozens of things running in the background that can make the system go slow and erratic.
My opinion on this might be somewhat biased (probably a lot), but I’d say MacOS-X and GNU/Linux system will probably do a lot better than Windows.
Another thing that most people forget about is cache. Both hardware and software cache. The data is usually not written at demand to the disk it goes first through the cache in the computer’s RAM, then through the hard disk cache (often called buffer too), before hitting the physical disk’s bits… Cache can be a life saver. If the system at some point gets sluggish cached data will probably save the day. Accessing/writing to the cache is much faster than writing to the (physical) disk. Also the amount of read data still in cache may prevent the need to physically access the disk to read that file that some other app running simultaneously needs. Hard drives’ cache can vary from 2MB to 16MB. Data cached in RAM will vary depending on the available RAM and how the operating system manages it.
“My opinion on this might be somewhat biased (probably a lot), but I’d say MacOS-X and GNU/Linux system will probably do a lot better than Windows.” What OS are you using?
Will try to restate my original question. Would like to use an older PC for audio recording so when had the opportunity to use DBAN with a number of older PCs decided to run some tests using 20 to 200 GB hard drives. While noticed variation in sequential write speed using these various hard drives, also noticed significant variation with these older PCs so reran the tests using a 40GB hard drive and noticed significant variation even then.
Would be interested to find out what older PCs (and OS) others have used for successful audio recordings. Since I mainly use WindowsXP am especially interested in this but would like to expand to Linux so hope someone has this information.
I use MacOS-X and Linux. My laptops is a MacBookPro. All my other PCs run Debian GNU/Linux.
For older PCs I’d probably recommend Linux. I’ve been a fan of Debian for a very long time. Steve will probably recommend Ubuntu Studio. Might be easier to install and setup, though I have never tried it. I have issues with Ubuntu since it’s first version, but many ppl seem to be happy with it. I’ve installed Ubuntu many times in other computers for the ease and convenience of the install process. Not quite sure if I’ll do it again though.
I still say that even on an old PC the hard disk will probably not be the bottleneck… I’d bet small RAM and slow processing speed might be more troublemaking. Specially lack of RAM.
Hard disks speed tests are most often not that much accurate. Results may vary depending on a lot of uncontrolled variables. Run the tests more than once. Try to achieve some consistency in the results before jumping into conclusions.
Hard disk performance will depend on the filesystem used, free space on the disk, etc. Read/write speed is not the same at the beginning and the end of the disk. My guess is that you probably won’t notice a big difference between different disk brands and models.
If you record at 16 bits, stereo, 44100 Hz, you’ll get a stream of about 1.4 mbps or 175 KB/s. Even a very old PC with a very slow hard disk will do faster than that…
“Hard disks speed tests are most often not that much accurate. Results may vary depending on a lot of uncontrolled variables. Run the tests more than once. Try to achieve some consistency in the results before jumping into conclusions.” Agree. Maybe I should elaborate about running DBAN from a Floppy. “Darik’s Boot and Nuke (“DBAN”) is a self-contained boot disk that securely wipes the hard disks of most computers. DBAN will automatically and completely delete the contents of any hard disk that it can detect, which makes it an appropriate utility for bulk or emergency data destruction.” It is Linux not DOS OS based. http://www.dban.org/ Also since sequential write zero is the only application there is minimal OS overhead and interference.
Have tested 2 similar P4 PCs and got very significantly different sequential write speeds. After running enough tests decided to switch the 80-wire Parallel ATA cables. The difference in speed was due to a defective ATA cable.
While I have found software for testing memory, and hard drives have not found anything for ATA cables. Does anyone know of any?
Will be retesting the other PCs with known "Good ATA CAbles to see if results change?
There are learning moments here from other forums. It’s been common to recommend a relatively small 40GB to 60GB for the system drive and keep that one thoroughly error corrected and defragmented. Then use a second much larger drive for actual data. This splits computer management from show data. You can do very well with that combination since the computer doesn’t have to scurry all over the drive platters to get between the two services.
This is where a slow laptop will kill you since you can’t put in a second internal drive, you have limited opportunities for internal memory, and older machines had terrible external drive management.
There are dedicated software to benchmark disk performance…
In linux you can use hdparm to, not only tweak some of the settings (such as UDMA mode, among others), but also to make simple quick performance tests.
For more complex and accurate results you can use other utilities dedicated to filesystem and disk performance. Bonnie++ used to be a popular one on *nix systems a long time ago… I’m not sure if it’s still developed or used nowadays…
Some of those live-cd linux distro probably have some utilities for that too…
None that I know of…
That’s a good advice, though in my case I rarely do it… If I can fit two harddisk into one computer I rather use them in RAID-1. I prefer reliability to performance
I usually split the hard disk into different partitions though. It doesn’t improve performance, but it helps to keep things tidy up.
With increasing complexity in nowadays filesystems and hard disk management, things like LVM and the like things can become a lot complex and harder to say where you can win or loose in terms of performance… it’s not that linear as it used to be in the past…
I usually split the hard disk into different partitions though. It doesn’t improve performance, but it helps to keep things tidy up.
It decreases performance. We recommend against it for either video or audio production. Partition file management is not free like it is on physical drives. Getting three different files back from three different partitions takes significantly longer as the actuator buzzes back and forth than getting the same three files back from physical drives.
It also has the overhead of Drive Management, even if they are only partitions. Far better and faster to let the operating system do the work and organize things through folders.
I’m aware of that. But I highly trust the linux efficiency in managing all that… Despite the decrease in performance I believe the performance is still way above the necessary.
Yes, when I didn’t have it. I already had too many hard disk dying on my hands… I’m not taking any risks again if I can avoid them… I use raid-1 now for all system that have important data (or that recovering from a dead drive would give me too much trouble…). With raid-1 I just have to switch the faulty drive for a new one and I don’t have to worry about it again.
but getting three sequential files from one unfragmented partition is faster than getting three files that have data scattered across one partition. The pros and cons are dependent on how the partitions are managed, what software is being used and other factors. In some cases splitting disks to multiple partitions wins, in other cases it looses. With Audacity I’ve not noticed any significant performance difference of using one or multiple partitions.
