i don't know any of that stuff, i actually don't know anything about how HDD is built, i just know that the higher RPM/larcher cache, newer tech make the thing go faster and bench marks can show how a HDD can perform in real life situation, on paper 5400RPM speed is not far away from the 7200RPM but in real life when it actually go to work, it's much more "sluggist". Manufacture are leaving the 5400RPM drives so new tech NCQ ect... aren't gonna get into those drives.
NCQ is basically a gimmick. We called it Tagged Command Queueing back in the day, and when you had a SCSI bus that ran in the single digits, it made some difference (though not huge even then). These days, a SATA bus is so many orders of magnitude faster than anything but the burst rate for the mechanism that the real-world impact of NCQ is near zero on the average. Yes, there are pathological loads where it makes a big difference, so on a server, it makes sense to do it, but we're talking about a
laptop here. In fact, for audio, NCQ actually will reduce performance, as audio is a mostly sequential load (for long strides anyway).
The heat/power consumption issue of a "recording" computer is not there, i don't think any one is gonna go in a library or any other place that have no access to an outlet and do his recording, where's he gonna plug the mic/amp in? obviously you know your tech and spec, but you have to think about the real life situation where it's getting the application.
Who is talking about power consumption? I'm talking about heat. Faster drives get hotter. Hotter drives in a small enclosure fail sooner. It's really pretty easy to understand.
I don't think any of the current 5400RPM can match up to the current 7200RPM if i remember correctly.
Of course not. They're at the same level of tech.
Next year's 5400 RPM drives will be roughly comparable to today's 7200 RPM drives, which are, in turn, roughly comparable to last year's 10,000 RPM drives. Yes, I'm exaggerating slightly---it's probably closer to two years---but you get the idea.
well the thing is, the HDD @ 7200RPM we can actually use it to it's full potential, the HDD is the only moving components in a computer to make it run, it is the slowest part of them all, you want as much speed out of it as possible.
With the exception of booting, it is unlikely that you will ever find anything that is I/O bound (limited by the performance of your hard drive). While the hard drive is the slowest part of the system, there are very few real-world system loads that actually are hindered by that bottleneck. Audio definitely is not one of them.
I seriously don't get your argument about platter and "increasing platter" maybe i don't have a good understanding of HDD, but i would like to hear an example of how to increase platter size to make my 5400RPM go as fast as a 10000 RPM raptor.
The math is a little horrible, but I'll take you all through it just because I enjoy causing people mental anguish.
Performance in hard drives is bounded by several things:
- rotational latency---the amount of time it takes to spin the disk around to where the head needs to read. On the average, this is half of a revolution, so on a 5400 RPM drive, it is on average 1/10800th of a minute, or about 5 milliseconds. In random access loads (which audio is not), this is a major performance bottleneck. Thus, for random access loads (like booting), the faster the drive, the better performance will be. Again, though, audio isn't randomly scattered across the disk. Audio software normally reads a few seconds of audio per track at a time.
- seek/settle time---this is the time it takes the head to move over the track and adjust its position until it is precisely in the right spot. It is the largest delay involved (often several rotations of the platter), but again, hurts random access performance a lot more than continuous reads. There is also usually a "fast seek" mode on most modern drives for shifting one or two tracks at a time. This is rarely described in detail, but is much more relevant to audio workloads, and is usually on the order of a quarter of a rotation of the disk (very small)
- raw data throughput---this is the speed at which bits fly off the disk. For mostly-continuous reads like audio software uses, this is the most critical factor. You'll usually find this listed as continuous read performance.
So the most important thing for audio is the third one, continuous read performance. This is determined by two things: how dense the data is on the platter and how fast the disk is moving.
Pretend we're talking about a tape instead of a disk for a minute. That way we don't have to worry about RPM and can just talk about the speed the magnetic material moves past the head in inches per second. It will make the example clearer. If you have ten bytes per inch (ludicrously low, just giving a simple example) and the tape is moving under a head at 10 inches per second, it reads 100 bytes per second. If there's twice as many bytes per inch in that track, it reads 200 bytes per second. If the density is the same, but the disk is moving twice as fast, it also reads 200 bytes per second. Thus, the raw throughput is the number of bytes per inch times the speed in inches per second.
On a spinning platter, this gets more complex, as the speed under the head in inches per second varies. Because the platter is spinning at a constant speed, the outer tracks past the heads move much faster than the inner tracks, and thus, assuming the data density is constant (it usually is), the read/write speed on the outer tracks is faster. We generally ignore this, however, and talk about it as an average across the entire disk. It usually isn't worth breaking it down further.
The biggest thing to remember is that disks are getting more dense. Ten years ago, we were at about 5 gigs per platter. The current record as far as I can tell is held by Samsung, which has a whopping 333 gigs per platter. Seagate comes in second at 250 gigs.
A platter of a 3.5" hard drive has a diameter of about 3.74 inches (no, really). The radius is thus 1.87 inches, and the usable portion of that is typically about 1.2 inches because of the spindle in the center. Thus, the area is about 10.98 - 1.41 = 9.57 square inches, give or take. I'm going to be sloppy and just use 10 because it makes the math easier.
If you have an older drive with an 80 gig platter (say something from 2002-2004 or so), the density is 8 gigs per square inch. A 333 gig platter would have 33 gigs per square inch. Now the question is how this translates to a single pass around the platter.
Let's say that the 80 gig drive has 100,000 tracks per inch, and there's about 1.2 inches of usable space, for a total of about 120,000 tracks. That means that each track has on average 666 kilobytes per track. Assume that the track at 1.5 inches out from the middle of the spindle is about that average value. (This is a ballpark guess.) That means the circumference of that track is pi * 3, or about 9.4 inches. That comes out to about 71kB per track-inch.
Now to get the actual read performance of this drive, you would take that 666 kilobytes and realize that it reads it 90 times per second (for a 5400 RPM drive). Multiply. It's about 60 megabytes per second! Of course, you'll have some seek and settle time reducing that, but that's a pretty fast drive.
Now, take the higher density drive. The number of tracks will be greater because the density is greater in both directions. To approximate the number of tracks, take the ratio of total data (333/80), divide by pi. Take the square root of this value. This is now proportional to the radius instead of to the area. Best guess is that the disk should have about 1.15 times as many tracks, or on the order of 174,000 tracks. Each track contains 1,913 kilobytes, or just shy of three times as much as the 80 gig platter on average, but the capacity is more than four times as much.
So that's what I meant when I said that the performance doesn't scale linearly with the total capacity of a platter. What does this mean for the data per second, though? About 1,913 kB/rev * 90 revs/sec. = 172.19 megs per second. That's still at the exact same RPM as the original drive. If we took the original 80 gig platter and spun it at 10,000 RPM, it would come in at 111 megs per second. Thus this demonstrates that, this drive from about three or four years ago running at 10k RPM would still be slower than a modern drive running at 5400 RPM.
All of these numbers are somewhat higher than real-world numbers for drives because I pulled several of the numbers out of my backside. My guess is they're all off by about a factor of 1.5-2, but they should be good enough to give you an idea of the issues involved.
