• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

What are the standart write/read speeds for sata HDD?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Artas1984

New Member
Joined
Sep 22, 2009
I want to know the standarts of write/read speeds for standart sata hard drives in Megabytes per second with different RPM rates. I can't find a source - lot's of BS talk all over the internet, but nothing serious.


There are standarts of transfer speeds for sata interfaces, but it does not mention what kind of transfer speed is that. For example sata2 is set to be working at 300 MB/s rate.

Question 1: is that 300 MB/s write or read speed?

Question 2: does a 7200 rpm sata2 HDD have the same 300 MB/s rate as a 5400 rpm sata2 HDD?

Question 3: how does memory cache affect transfer rate?
 
There is a short answer and a long answer.

The short answer is that there is no standard read/write speed for hard drives. No hard drive does 300MB/s.



The long answer is considerably longer. 300MB/s is the SATAII interface speed. This is significantly faster than any current hard drive can read or write. The only things pushing this limit are SSDs. For the SATA interface the 300MB/s is for both read and write, at least in theory.

A 7200rpm hard drive will likely have the same 300MB/s SATA interface as a 5400rpm hard drive. This does not mean they are as fast. All else being equal the 7200 rpm drive will be faster.

Then again, all else is never equal. The actual read/write speed of a hard drive depends on multiple factors, primarily amongst them is the rotational speed and platter density. The read/write of modern hard drives are in the 50-150MB/s range. Every model will be different. There is no standard read/write speed.

The cache help a little bit in allowing the drive to do read-ahead which helps with read speed, and in write caching which can help with writes. Data that already is in the cache can often be read much faster, even up to the limit of the SATA interface. The number of heads and platters in the drive also has a minor effect.

Additionally the read/write speed of a hard drive is not constant over the hard drive. With a constant rotational speed the outer edge of the platter is moving faster then the inner edge, which results in higher read/write speeds the further out on the hard drive you are. Here is a picture of the Samsung drive in my HTPC:
SAMSUNG-HD501LJ-690G-MrAlpha.png
As you can see the read speed starts out at around 90MB/s at the outer edge of the platters and sinks down to 50MB/s at the inner edge.


All of the previous only applies to sequential read and write. Since a hard drive is a mechanical machine it has to get the head to the correct position before it can read or write data. If you don't read or write sequentially but need pieces of data that spread out over the drive the hard drive has to spend a lot of time seeking out the data, in which case the read/write speeds drop to around 1MB/s.

This is also a case where the cache helps a lot, and can push the random write speed up to 2-3MB/s by caching the writes.
 
Additionally the read/write speed of a hard drive is not constant over the hard drive. With a constant rotational speed the outer edge of the platter is moving faster then the inner edge, which results in higher read/write speeds the further out on the hard drive you are. Here is a picture of the Samsung drive in my HTPC:

Hey! That was so obvious, and yet i did not think of that. Yesterday i was doing the same test and was wondering why my HDD is slowing down - obvious why now! But anyway - it is the other way around - i mean that the outer plate of any rotational disc is moving slower than the inner, so the data is being read from "inside to outside" - slowing down while doing it. And now comparing my WD Caviar Green speeds to the pointed ones, i see it's ok.

The actual read/write speed of a hard drive depends on multiple factors, primarily amongst them is the rotational speed and platter density.

And what about the density of the platter? You mean how many layers are there for info to store?
 
Hey! That was so obvious, and yet i did not think of that. Yesterday i was doing the same test and was wondering why my HDD is slowing down - obvious why now! But anyway - it is the other way around - i mean that the outer plate of any rotational disc is moving slower than the inner, so the data is being read from "inside to outside" - slowing down while doing it. And now comparing my WD Caviar Green speeds to the pointed ones, i see it's ok.
No, the drive is read from the outer edge to the inner and the outer edge is faster.

Think about it. The diameter of the platter of a 3.5" drives is 9.5 cm. So the outer circumference is 29.8 cm. If the platter is spinning at 7200 rpm, that is 120 rotations per second. Since the head doing the reading on the outer edge stays in place and lets the platter move under it, the platter will move 120 * 29.8 = 3039.6 cm/s under the head. This determines the read speed of the outer edge.

The inner diameter of a platter is 2.5 cm. This gives a circumference of 7.9 cm. Since the rotational speed is the same the inner edge of the platter will move 120 * 7.9 = 948 cm/s under the head.

Now these numbers are a bit rough. The innermost and outermost tracks aren't quite at the circumference of the platter and the density of the inner track is a bit higher than the outer track. This means difference is not as big as these numbers would suggest, but the idea holds. The outer track is faster.

And what about the density of the platter? You mean how many layers are there for info to store?
The data is only stored on one layer on the platter (technically two layers since it is stored on both sides of the platter). Density means how many physical bits there are on a square centimeter. Or, expressed in a different way, how small each bit is. Since the track moves at a specific speed under the head (depending on the rpm and how far out the track is on the platter) the more bits you have per cm of track the faster you will read and write.
 
Yes, nice of you to prove it. I was confused at the case, where in some other rotational situations the objects near the axis have to spin faster than the objects near the perimeter in order to travel the same length. That's another case of course...
 
Back