-
Join the MyBroadband community
South Africa’s biggest forum. Discuss, discover, and connect with thousands of members.
-
Question of the Day: What's the most interesting small town you've visited in South Africa?
HDD: Does size affect speed
- Thread starter HANDsolo
- Start date
Platter density has a significant impact on speed, particularly sustained transfer rates. Other than that there is no yes or no answer to every circumstance.
I say this reservedly - Usually larger capacity drives will exhibit "marginally" better performance given that all other factors are equal i.e. Same family of drives from the same manufacturer.
Last edited:
LazyLion
King of de Jungle
no, only rpm and cache size would make the difference.
If anything a bigger drive will be slower... bigger partitions and longer seek time.
If anything a bigger drive will be slower... bigger partitions and longer seek time.
No, it's as tb says. Higher platter density generally makes for faster sustainable transfer rates. The Raptors with their high RPMs and low densities have quicker seek and bursts transfers, but a high density 1TB drive will be faster when it comes to big long transfers.
LazyLion
King of de Jungle
but then they are not really equal, are they? the OP said... "all things being equal"
but you cant have equal densities on equal platters and then end up with different capacities!? or are you harry potter?
if all things are equal, besides capacity, then that means the only difference is platter density. more data fits onto the platter, hence it is more densely packed hence in that one zillisecond drive A (less dense) will present less data to the read/write head assembly than drive B (more dense)
but there is a heck of a lot of variables, so you cant make a blanket statement, except to say that in myadsl-world, where we control all variables, the bigger capacity drive will be quicker.
quovadis
Honorary Master
There are quite a few things that can influence performance of larger and smaller disks and it all depends on what drives you're comparing - the main factors would include number of platters, number of heads, cache size, interface, spindle speed and in general any technology that may be specifically on one drive or another.
You cannot possible say a smaller drive of the same family is faster than a bigger drive of the same family as one may have one platter (80GB drive with 1 80GB platter) and the other may have two (146GB drive with two 73GB platters). The drives would need to be assessed based on each drive's physical construct and as such you'd have to evaluate each drive on a case by case basis.
Partitions rarely have anything to do with it as you can't say a drive that has 3 partitions will be faster than a drive with 2 partitions or 1 - It varies completely on how much data is being accessed and where it is on the disk. Alot of people partition a drive into two - 1 for OS and 1 for Program files/Data - This is done primarily so that the OS can be more readily reinstalled without having to reformat the entire drive (involving time consuming backups etc) and not as a major performance increase.
You cannot possible say a smaller drive of the same family is faster than a bigger drive of the same family as one may have one platter (80GB drive with 1 80GB platter) and the other may have two (146GB drive with two 73GB platters). The drives would need to be assessed based on each drive's physical construct and as such you'd have to evaluate each drive on a case by case basis.
Partitions rarely have anything to do with it as you can't say a drive that has 3 partitions will be faster than a drive with 2 partitions or 1 - It varies completely on how much data is being accessed and where it is on the disk. Alot of people partition a drive into two - 1 for OS and 1 for Program files/Data - This is done primarily so that the OS can be more readily reinstalled without having to reformat the entire drive (involving time consuming backups etc) and not as a major performance increase.
as far as I understand.. drive are rated on averate seek time... with fastest seek time being at the rim of the platter and slowest at the core.. because of the speed at which the plater head is higher at the out side... I have read its can be 40% faster at the rim... this is one of the reasons the first partioning should be for Os and page file.. forceing this data to the outside of the disk... where seek time is maximised
Uhh. No!
Faster transfer rates occur on the outer cylinders.
Seek times vary depending on the distance (no. of cylinders) from the current head position.
Yes the linear velocity of the platters under the head is faster on the outer edges and is what gives you your disk to head "transfer rate".
The traversal of the heads between cylinders (movement perpendicular to the discs rotation) is know as "seek time". It is this distinction that I was trying to correct in your first post.
Yes the linear velocity of the platters under the head is faster on the outer edges and is what gives you your disk to head "transfer rate".
The traversal of the heads between cylinders (movement perpendicular to the discs rotation) is know as "seek time". It is this distinction that I was trying to correct in your first post.
Quick quessy.. do all cylinders contain the same amount of data at the rim vs the core.. with the cylinder be physically longer ?
No the number of sectors reduces as you move inwards, so the number of sectors per track (cylinder )reduces. This results in less data (bytes) being stored as you move inwards. The data density however remains constant.
Last edited:
Ahh, Ok I see where you are going. In essence what you are saying is correct. This is why a slower seeking drive with higher density platters "can" under certain conditions outperform a faster seeking drive.
It was the context in which you used the term "average seek" in your original post that was incorrect. The change of bytes stored per track as you move inwards has no direct bearing on "seek "times which is a very specific measure of an attribute of a drive.
It is the statistical reduction of the requirement to perform a seek due to the greater amount of data stored per cylinder that allows higher densities to improve performance under certain usage patterns.
Hope that all makes semse.
Dan C
Honorary Master
How much does the cache size matter in performance ?
Once again I'm going to precede anything I say by saying that there are no black and white answers in hard disc performance. No single attribute of a hard drive is going to make that hard drive the best performer under all circumstances. Performance is a very general term - "Performance doing what?"
Now to try and answer your question I will on this one defer to what two of the drive manufacturers are doing and saying regarding cache sizes.
Seagate - uses 32MB caches on their higher capacity drives
WD - is using 16MB and stated that "there was little performance gain beyond this size"
You'll have to google that for yourself but IIRC it came from the WD site.
Most of the tests now are showing the WD drives to be faster than the seagates. This is the situation now and will probably change when next one of these manufacturers releases a newer model.
These cache sizes are really too small that you are likely to have a large number of read requests satisfied from cache. Having said that a certain amount is necessary for the buffering of data between the disc and host.
So I will answer your question by saying that without a cache you will likely see a huge performance hit but increasing cache beyond what the current high end drives will likely show little benefit.
Check out this site for a wealth of independent information on discs in general.
http://www.storagereview.com/
I have made a lot of generalised statements but the drives I am referring to when comparing the WD's and Seagates are the high capacity desktop models from their respective ranges.
Seems that WD is using a 32MB cache on their 1GB Black Edition (WD's Top line). They are however still stating that there is little to be gained beyond 16MB.
Last edited:
Dan C
Honorary Master
thanks kb !
For those that have the time and inclination.
Here is a direct link to the HDD performance section of the website I often refer to in HDD performance discussions. It goes into far more detail and probably explains the subject far better than I will ever be able to.
http://www.storagereview.com/guide/index_perf_perf.html
Please read you will probably find most of your answers here!
Here is a direct link to the HDD performance section of the website I often refer to in HDD performance discussions. It goes into far more detail and probably explains the subject far better than I will ever be able to.
http://www.storagereview.com/guide/index_perf_perf.html
Please read you will probably find most of your answers here!