How important are hard drive specs like RPM and cache size?

[Petros_k]

Here's my dilemma: I got a computer from a refurbisher that had a faulty hard drive in it. This is a 250 GB Seagate Barracuda ES.2 model ST3250310NS with a spindle speed of 7200 RPM and a cache size of 32 MB. They sent me a replacement for it that's 500 GB of storage, but it's a Seagate Pipeline HD ST3500312CS that has a spindle speed of 5900 RPM and a cache size of only 8 MB. I didn't need the additional storage, but should I be concerned about the potential for downgraded performance?

I see from the Seagate website that the Pipeline series hard drives are intended for things like security systems and DVR storage (in 24/7 applications like streaming media). I'm not saying these drives can't both run an operating system like Windows 7, but compared to the Barracuda the Pipeline hard drive could lead to slower performance when running certain applications such as video editing, photo editing, and digital recording, yes?

 

[homebrew2ny]

It will be a downgrade in performance, to what degree I am not to sure. I would be unhappy about it though. However, it is double the size so.......

 

[corkyg]

Definite downgrade in performance due to slower spindle speed and smaller cache. I would not like it. For my use, performance is more important than storage space.

YOu can trump the issue later on by using that HDD as a storage/data drive and get a SSD for your OS drive.

 

[sxr7171]

If you use an ssd for os it makes no difference. Probably saves you energy. If not there will be a small difference if any. This is because newer drives with higher areal density often outperform older drives even if those have higher spindle speeds.

 

[corkyg]

If you use an ssd for os it makes no difference. Probably saves you energy. If not there will be a small difference if any. This is because newer drives with higher areal density often outperform older drives even if those have higher spindle speeds.

Really? I know of no spindle drive that can outperform a SSD.

 

[sxr7171]

Really? I know of no spindle drive that can outperform a SSD.

What I mean is a newer spindle with lower rpm vs. an older spindle with higher rpm. For the case that he is not using an ssd for his os.

 

[Petros_k]

What I mean is a newer spindle with lower rpm vs. an older spindle with higher rpm. For the case that he is not using an ssd for his os.

Can you elaborate a bit more than the terms "new" and "old" ?

The hard drives in question here:

Seagate Barracuda ES.2 (7200 RPM, 32 MB cache)

Seagate Pipeline HD (5900 RPM, 8 MB cache)


Are these "old" or "new" ?

 

[smitbret]

What I mean is a newer spindle with lower rpm vs. an older spindle with higher rpm. For the case that he is not using an ssd for his os.

It's gonna depend on platter density. The fewer the platters, the faster the drive.

I.E.

320GB 7200RPM 64MB Cahce drive with 2 160GB platters

will get crushed by a

1TB single platter WD Green drive

If the 500GB is a single platter (I'll bet it's not, though) then it could be faster than the one it replaced.

 

[shabby]

The 250gb has 9msec seek times and 105MB/sec transfer speeds.
The 500gb has 14msec seek times and 120MB/sec transfer speeds.
So the smaller drive is better when reading/writing lots of small files while the bigger drive is at better at moving large files.

 

[Petros_k]

It's gonna depend on platter density. The fewer the platters, the faster the drive.

I.E.

320GB 7200RPM 64MB Cahce drive with 2 160GB platters

will get crushed by a

1TB single platter WD Green drive

If the 500GB is a single platter (I'll bet it's not, though) then it could be faster than the one it replaced.

According to Seagate specs, each of the hard drives in question (Seagate Barracuda ES.2 ST3250310NS, and Seagate Pipeline ST3500312CS) has 1 disk and 2 heads, so it's not going to matter in this case.

 

[Petros_k]

The 250gb has 9msec seek times and 105MB/sec transfer speeds.
The 500gb has 14msec seek times and 120MB/sec transfer speeds.
So the smaller drive is better when reading/writing lots of small files while the bigger drive is at better at moving large files.

But the Pipeline drive is not intended for running an OS, at least according to the Seagate website:

http://www.seagate.com/internal-hard-drives/consumer-electronics/video-3-5-hdd/?sku=ST3500312CS

Plus, what about the small 8 MB cache of the Pipeline vs. the 32 MB cache of the Barracuda? Doesn't that make the Barracuda able to handle more data faster in general?

 

[smitbret]

According to Seagate specs, each of the hard drives in question (Seagate Barracuda ES.2 ST3250310NS, and Seagate Pipeline ST3500312CS) has 1 disk and 2 heads, so it's not going to matter in this case.

That's why the 500GB drive has a faster transfer rate. Higher platter density.

 

[Petros_k]

That's why the 500GB drive has a faster transfer rate. Higher platter density.

But then you're saying the ST3500312CS Pipeline hard drive is actually FASTER than the ST3250310NS Barracuda, despite the spindle speed being only 5900 RPM (as opposed to 7200 for the Barracuda) ?

When you look at these numbers again:

The 250gb has 9msec seek times and 105MB/sec transfer speeds.
The 500gb has 14msec seek times and 120MB/sec transfer speeds.

How is 14msec faster then 9msec?

 

[Petros_k]

I thought I'd put these specs out there if others want a quick comparison to comment:

Pipeline HD.2 ST3500312CS

Spindle Speed: 5900 RPM
average seek time, read: 14msec
Sustained data transfer rate: 120 Mbytes/sec. max
cache buffer: 8 Mbytes


Barracuda ES.2 ST3250310NS:

Spindle Speed: 7200 RPM
average seek time, read: 8.5msec
Sustained data transfer rate: 105 Mbytes/sec max
cache buffer: 32 Mbytes


I thought not only for OS files but for applications that are more hard drive intensive like video editing, photo editing, and digital recording that a higher spindle speed and larger cache buffer was desirable. That's not true in this case and the Pipeline will still be faster?

If you go to the Seagate website, you can see the Pipeline hard drive is intended for security and DVR storage applications (which might explain the low 8 MB cache).




The reason I expect the drive with the larger cache size (the Barracuda in this case) to outperform the other (explained at a website on digital recording):

"A larger buffer{cache memory} is essentially a fast buffer between the drive platters and the computer. It is faster than physically accessing data from the drive platters when reading or writing. Complex algorithms determine the practical use of cache memory.

As specific data is requested from the hard drive, the drive logic may anticipate that more accesses will ask for the same data. To increase performance and to avoid having to keep reading the same data from the drive several times, the data is put into the cache memory, so that the drive will be able to handle future requests immediately. Often, additional data that is usually requested afterwards is cached as well.

Reading data from the cache memory is much faster than reading directly from the drive platters themselves. The more data that is cached, the better the chance for cache access over physical platter access. Thus, as drive cache size increases, so does overall drive performance." http://homerecording.com/bbs/general-discussions/digital-recording-computers/24-bit-recording-82798/

 

[Petros_k]

Just wanted to share this from the PassMark Software website:
http://www.harddrivebenchmark.net/hdd_list.php

Seagate Barracuda ES.2
model: ST3250310NS
Disk Rating (higher is better) 604
Rank (lower is better) 1376


Seagate Pipeline HD.2
model: ST3500312CS
Disk Rating (higher is better) 620
Rank (lower is better) 1332

For what it's worth, these rating seems to be congruent with comments here suggesting the Pipeline is a slightly better hard drive.

I was misunderstanding the data transfer rate and realize now the Pipeline HD.2 actually beats the Barracuda by 15 Mbytes per second, despite having only a 5900 RPM spindle speed.

 

[shabby]

But then you're saying the ST3500312CS Pipeline hard drive is actually FASTER than the ST3250310NS Barracuda, despite the spindle speed being only 5900 RPM (as opposed to 7200 for the Barracuda) ?

When you look at these numbers again:

The 250gb has 9msec seek times and 105MB/sec transfer speeds.
The 500gb has 14msec seek times and 120MB/sec transfer speeds.

How is 14msec faster then 9msec?

Why are you looking at the seek times, look at the transfer speeds. Anyway, both drives will be slow in the end. I can't even use a pc that uses a mechanical drive, when it starts looking up files and windows freezes i feel like punching the monitor.

 

[Petros_k]

Why are you looking at the seek times, look at the transfer speeds.

I was misunderstanding the data transfer rate and realize now the Pipeline HD.2 actually beats the Barracuda by 15 Mbytes per second, despite having only a 5900 RPM spindle speed.

Anyway, both drives will be slow in the end...

Actually, "slow" is quite relative. I was only looking for a comparison of these two hard drives and trying to determine if the replacement hard drive with a spindle speed of 5900 was going to degrade performance compared to the hard drive with 7200 RPM. I can see now it's not so simple.

 

[smitbret]

The reality is that they are pretty even.

The Pipeline will certainly be faster with large file transfers and more suited to file storage.

The Barracuda with the faster spindle speed and faster seek times will probably be better as an OS drive where loading multiple, smaller files will be faster.

 

[jolancer]

just my opinion

the question is moot. ur getting an old used drive to replace an old used drive, which may or maynot be having issues. there real world performance is probably similar and moot compaired to SSDs. All in all, again IMO no matter what kinda drive you recieved as replacement(within reason) from the seller, I'd consider yourself lucky unless your purchase came with some type of warranty, most sellers i assume would probably blow you off without hard evidence of defect and even then thy may not give a dam.

IMO if they were both new, id prefer the 500gb, %50 more space and higher probable reliability since its lower rpm, also maybe quieter. however there both used, and since your research turned up the 500gb may of been pulled from a DVR, that drive maybe just as beat or more so than the one your replacing, tho perhaps still quieter because lower rpm.

beggers cant be choosers but if your going the ultra cheap route, at least you have another drive now for backup, even if it sounds odd, if its not physically damaged, you can keep it powered off and it will always be there till you kill it.

re someones comment about higher density platters... I would disagree with ur example unless your ruling out other variables. Im not a HDD engineer so if im wrong please correct... but this would be my assumption in your previous example, a higher density platter will be read faster yes, but the disk with lower desnsity and more platters will be able to read/write to more platters simultaniously. so the difference in that instance is probably moot unless other variables come into affect, or the density is more than twice as much in equal platter size and rotation speed. re different density but SAME platter count, then i would completely agree, which would seem to apply to the OPs HDDs as he stated.

 

[Concillian]

If you search Seagate's site, you'll see that "Pipeline" drives show up in their 'Video' section. This type of drive tries to sustain some minimum throughput for large file sizes. They try to keep above some threshold throughput, as momentary drops below threshold will cause pauses in video.

ES.2 is a nearline storage drive, This will focus a lot more on random fetches used in datacenters.

So in addition to the spindle speeds, you have to account for the firmware differences that will tweak performance towards the intended applications. ES.2 will be better at random performance, as will high spindle speed, since that will lower seek latency. Pipeline will be better with consistency in maintaining some minimum throughput. Surely the pipeline is overall slower, but it probably offers more consistency in large transfers, with a focus on never dipping below whatever it's throughput threshold is.

 

[VirtualLarry]

re someones comment about higher density platters... I would disagree with ur example unless your ruling out other variables. Im not a HDD engineer so if im wrong please correct... but this would be my assumption in your previous example, a higher density platter will be read faster yes, but the disk with lower desnsity and more platters will be able to read/write to more platters simultaniously.

Sorry to burst your bubble, but HDDs simply don't work that way.

 

[jolancer]

Sorry to burst your bubble, but HDDs simply don't work that way.

no offence taken, not actually bursting anyones bubble i never looked into it, i just assumed. yeah i know assuming is bad, but when its not mission critical saves a lot of time

tho *IF what your saying is true, IMO that would be Amazingly stupid of hdd manufacturers. unless there just handicapping themselves to make SSDs more marketable LOL

 

[smitbret]

no offence taken, not actually bursting anyones bubble i never looked into it, i just assumed. yeah i know assuming is bad, but when its not mission critical saves a lot of time

tho *IF what your saying is true, IMO that would be Amazingly stupid of hdd manufacturers. unless there just handicapping themselves to make SSDs more marketable LOL

They are usually pretty quick to introduce and convert to the higher density platters when they come out.

I think that also probably explains why external drives are the same price or cheaper than a same size internal. If you pop one open, it's usually the older drive with more platters and performs a little worse. The older stock with more platters goes into the externals and onto the shelves for customers that just care about how much it will hold.

 

[Concillian]

no offence taken, not actually bursting anyones bubble i never looked into it, i just assumed. yeah i know assuming is bad, but when its not mission critical saves a lot of time

tho *IF what your saying is true, IMO that would be Amazingly stupid of hdd manufacturers. unless there just handicapping themselves to make SSDs more marketable LOL

Amazingly stupid? Really?

Care to explain the logistics of multiple heads reading at once?

Keep in mind a track is somewhere around 0.000005 inches wide and to read multiple platters at once. So your tolerance is probably somewhere around 1/10th of that? Call it ±0.0000005 inches cumulative across 2 to 8 surfaces.

Oh, and it has to be able to align all those tracks repeatably, a couple dozen times a second...

I eagerly await your explanation on how you can keep heads aligned with a track on one surface without messing up the alignment on another.

 

[jolancer]

@Concillian -

theres nothing to align, the disks are affixed to the same armature, and the heads are affixed to the same arm mount.

assuming your not confusing yourself, only thing i can assume your trying to say is that the gaps between tracks on the disk are not a mirror image to each other because of the manufacturing proccess, and this gap is sizable enough to have a relevant interference between disks with a fixed head position?, otherwise even IF tracks are Not synced and the gap is not relevant between tracks, who cares witch track it reads from when as long as the heads are synced wouldn't matter.

IF for whatever reason the gap would cause it to be out of tollorance, and there to lazy to have some kinda process in production to fine tune alignments afterword, one may also imagine they could simply test the tolorances of each disk and pair ones that reach spec together then use those only for drives with synced I/O and sell'm at a premium, and continue the rest as normal.

IF what your tying to say is the physical gap between tracks is Not oriented circular around the disk, well that makes it easyer IMO since one may imagine you could compensate for the gap then through software calibration.

..any which way no matter how wrong my assumptions maybe, its probably would of been more than doable.