Samsung SpinPoint F4 320GB 16MB comparable to SSD

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

sub.mesa

Senior member
Feb 16, 2010
611
0
0
tweakboy, your higher than 'possible' results with HDtune must be from caching since you use Intel drivers with 'write caching' enabled; which uses your RAM as write back. This also means you can't do simple HDTune benchmarks and you can get burst rates as high as your RAM. I recommend both AS SSD and CrystalDiskMark to test storage, particularly more complex storage such as with the Intel driver. No other onboard RAID does writeback like Intel is capable of; on ICHxR southbridge chipsets.

I still suspect alot of software is running in the background on your PC; anti-virus for example can significantly slowdown your I/O performance. But i have no reason to suspect anything is wrong with your setup; the caching that Intel does should help quite a bit in most circumstances. Do mind that write-back can corrupt your filesystem; so keep good backups.
 

tweakboy

Diamond Member
Jan 3, 2010
9,517
2
81
www.hammiestudios.com
Wow thank you Sub.Mesa for the amazing technical response. I understand now, so the drive is ok its doing its thing. I cant trust benchmarks basically. Also I just did what you said with passmark I exited all processess except for ones to keep windows running. I ran test same I get like even 600 points at times,, my max is 756 points once I got that, other then that avg is like 650 to 700. Soo the sreenie of the techreview benches shows mine as 934, if I cant even get 800 points, what can be the problem there. Everything is exited out btw, all processes like spybot steam etc etc,, Should I still believe the drive is ok and Im getting most performance out of drive. Thank you soo much

Armen Abcarians
 

SickBeast

Lifer
Jul 21, 2000
14,377
19
81
I was excited at first about this Samsung drive, but now I'm not. It has poor access times and basically only scores really well in sequential transfers.

With 120gb SSDs starting to go for $160, it's starting to look like the ship has sailed for mechanical hard drives, particularly drives like this one or the Raptors.

I still like mechanical drives for mass storage. You can get 1tb drives for only $10 more than this 320gb drive if you look hard enough.

You're not going to notice fast sequential transfer speed in a vast majority of situations.
 

tweakboy

Diamond Member
Jan 3, 2010
9,517
2
81
www.hammiestudios.com
SickBeast I dont need 1TB nore 500GB, 320GB is perfect for my primary disk,,, a true fast SSD 120GB is 250 350 dollars.... I am not rich for a SSD, this is the next best thing, until I can afford one in 2013 maybe...
 

SickBeast

Lifer
Jul 21, 2000
14,377
19
81
A true fast 120gb SSD like the Patriot sandforce-based drive I just saw costs $160.

IMO the thread title should be changed. There is no way that this hard drive can compete with an SSD. I would have to see some real-world benchmarks; nothing synthetic.
 

SickBeast

Lifer
Jul 21, 2000
14,377
19
81
Also Tweakboy, even if you don't need 1TB, I'll bet a 1TB drive would be faster than the 320gb drive if you just short stroke it to 320gb.
 

computer

Platinum Member
Nov 5, 2000
2,735
2
0

FishAk

Senior member
Jun 13, 2010
987
0
0
Also Tweakboy, even if you don't need 1TB, I'll bet a 1TB drive would be faster than the 320gb drive if you just short stroke it to 320gb.

This

Look at a HD Tune graph of the 320Gb F4, and a 1Tb F3 Spinpoint.

The 320 starts @ 150Mb/s, and drops to 75, with the access time starting @ about 7ms, and climbing to about 23ms or so.

The F3 starts @ 145Mb/s, and drops to about 125Mb/s @ 320Gb, with access time starting @ about 8ms and climbing to only about 13ms @ the 320Gb point.

The F3 is less than half the price of the F4, with 3 times the room, and much better performance in the first 320Gb. You can partition the first 320Gb for files that need better performance, and still have 660Gb left over to use as auxiliary storage, movies, or whatever.

A true fast 120gb SSD like the Patriot sandforce-based drive I just saw costs $160.

Where do you see a Patriot Inferno in the $160 range? The cheapest 120Gb drive I see is $100 more than that.
 

computer

Platinum Member
Nov 5, 2000
2,735
2
0
This

Look at a HD Tune graph of the 320Gb F4, and a 1Tb F3 Spinpoint.

The 320 starts @ 150Mb/s, and drops to 75, with the access time starting @ about 7ms, and climbing to about 23ms or so.

The F3 starts @ 145Mb/s, and drops to about 125Mb/s @ 320Gb, with access time starting @ about 8ms and climbing to only about 13ms @ the 320Gb point.
Tests like HDTune, HDTach, Crystal, AIDA32, et al, while they can tell you a HD's potential and capability, aren't exactly accurate for real-world use. The only way to really compare HD's is with a more "real word" benchmark like PCmark or WinBench because they use actual app algorithms (such as XP Startup, App Loading, Virus scan, File Write, General HD usage; and in WinBench High-End Disk & Business Disk WinMark, Frontpage, Photoshop, Visual C++, etc.). I've seen a HD get great results in HDTune & HDTach, yet falter in PCmark & WinBench.

For example, a WD6401AALS I tested didn't do so great in HDTach, HDTune, AIDA32 & Crystal (it was beat by a lot with the HD103SJ), yet it beat the HD103SJ by a good bit in PCMark04, PCMark05 and WinBench. That tells me the WD would be better for an OS drive, and the HD103SJ would be better for a drive onto which you'd read/write large files (audio/video creation/editing, and similar).

There's another great incredibly simple program called FCTest which could be the tell-all in benchmarks. It can take a long time to run all tests (I run them repeatedly with various options and file sizes), but it's worth it in the end if you really want to know which HD may be fastest for not only an OS drive--but also overall--because some of its tests can be written/read/copied repeatedly for as many GB as you want to see maximum throughput.


The F3 is less than half the price of the F4, with 3 times the room, and much better performance in the first 320Gb. You can partition the first 320Gb for files that need better performance, and still have 660Gb left over to use as auxiliary storage, movies, or whatever.
:confused: The F4 (322GJ) is only 43 bucks. I believe I paid about $80 for my 1tb F3 (HD103SJ) but that was about a year ago, they're about ~$68-75 now, sometimes cheaper AC.
 

FishAk

Senior member
Jun 13, 2010
987
0
0
:confused: The F4 (322GJ) is only 43 bucks. I believe I paid about $80 for my 1tb F3 (HD103SJ) but that was about a year ago, they're about ~$68-75 now, sometimes cheaper AC.

HaHa! Good catch! I meant to say the F3 is less than half again the price of the F4. (less than 2x$ the F4)

Compared to the WD, one would be hard pressed to actually perceive much of a difference between the drives. For the OS, one should spring for a small SSD, but for programs that use many 1 to 5Mb files, the SSD doesn't have much of an advantage yet. In this case, it still economical to partition a chunk of a 1Tb drive to glean some of it's higher performance.
 

tweakboy

Diamond Member
Jan 3, 2010
9,517
2
81
www.hammiestudios.com
I agree, those apps dont mean squat. Its real world performance.

I can tell a difference, Thunderbird used to come up after boot up and 15 minutes idle come up in 3 seconds. Now it takes 1 second and half for first launch after boot. My pics dealing with larger files it zooms in faster,, used to take 1 second delay now its 0 seconds. Also I can see real world benefit with my DAW Sonar 8.5 Producer Edition @ 2ms latency. 24bit 192khz recording. thanks and gg
 

SickBeast

Lifer
Jul 21, 2000
14,377
19
81
I'm not saying that you're not noticing a difference, tweakboy. What I'm taking issue with is the fact that you're comparing the drive's performance to that of an SSD, when in reality that is only the case when performing large sequential transfers, which are rare.

Other people on the internet (namely the newegg website) have said that these drives perform like an SSD, so I guess I can't really blame you. If you look at the Crystal benchmarks of the drive, yes, it comes close to a crappy SSD if all you look at is the transfer rates.

You need to look at access times to see which drive will be "faster" for 95% of the things you'll do on a computer. This Samsung drive might have a 12ms seek time, compared to a 0.001ms seek time on an SSD.
 

SickBeast

Lifer
Jul 21, 2000
14,377
19
81
Tests like HDTune, HDTach, Crystal, AIDA32, et al, while they can tell you a HD's potential and capability, aren't exactly accurate for real-world use. The only way to really compare HD's is with a more "real word" benchmark like PCmark or WinBench because they use actual app algorithms (such as XP Startup, App Loading, Virus scan, File Write, General HD usage; and in WinBench High-End Disk & Business Disk WinMark, Frontpage, Photoshop, Visual C++, etc.). I've seen a HD get great results in HDTune & HDTach, yet falter in PCmark & WinBench.

For example, a WD6401AALS I tested didn't do so great in HDTach, HDTune, AIDA32 & Crystal (it was beat by a lot with the HD103SJ), yet it beat the HD103SJ by a good bit in PCMark04, PCMark05 and WinBench. That tells me the WD would be better for an OS drive, and the HD103SJ would be better for a drive onto which you'd read/write large files (audio/video creation/editing, and similar).

There's another great incredibly simple program called FCTest which could be the tell-all in benchmarks. It can take a long time to run all tests (I run them repeatedly with various options and file sizes), but it's worth it in the end if you really want to know which HD may be fastest for not only an OS drive--but also overall--because some of its tests can be written/read/copied repeatedly for as many GB as you want to see maximum throughput.



:confused: The F4 (322GJ) is only 43 bucks. I believe I paid about $80 for my 1tb F3 (HD103SJ) but that was about a year ago, they're about ~$68-75 now, sometimes cheaper AC.

In this case, when you short stroke a drive, it the head has to travel only 25% as far to perform the seeks. If the sequential transfer speed matches up in the benchmarks, then theoretically, the large short stroked drive will destroy the smaller non short stroked drive in real world applications.

I would be willing to wager that a short stroked 7200rpm 1TB drive would beat this Samsung drive in pretty much all benchmarks. Then you also get the benefit of a large slow partition at the end of the drive to store all of your archival stuff like photos and videos.
 

sub.mesa

Senior member
Feb 16, 2010
611
0
0
HDDs are slow for any random I/O access; they can only be fast with sequential transfers. So use them for that, and use SSDs when possible for your system disk.

Short stroking can sometimes be wise. But it can also lower performance if you access both partitions at the same time. So it depends per user what is the best configuration. If you want more seeking performance, try adjusting AAM (Acoustic Management) settings of the drive. These settings do not affect sequential throughput, only Random I/O.

HDDs remain excellent for mass data storage; movies in excess of 1GB shouldn't be stored on SSDs as it provides zero benefit and costs alot more per GB. HDDs still do excellent in this area so splitting your data in 'active' (OS+apps) and 'passive' (large data files) would be preferable, especially when looking at the near future, where both HDDs and SSDs will be used commonly because they each have their own advantages.
 

Makaveli

Diamond Member
Feb 8, 2002
4,801
1,265
136
Those HDD Tune numbers don't make any sense. Am I missing something here? Why are your values ridiculously high even compared to top of the line SSDs?

I was wondering this myself since I saw the first screenshot and shocked that it took 4 pages in for someone to notice.

Those numbers look like they are from a RAID 0 setup not a single drive.

Edit *Just read Sub.mesa's Post at the top of this page its due to Write caching!
 
Last edited:

tweakboy

Diamond Member
Jan 3, 2010
9,517
2
81
www.hammiestudios.com
Thank you all for contributing to this thread. Ill get a SSD when 320GB version comes out for like 100 dollars in 2013. Until then I am very happy with my setup; as long as I notice a speed difference Im happy. Thanks
 

FishAk

Senior member
Jun 13, 2010
987
0
0
If you buy a good 40Gb drive when they are $50 next year, and just put your OS/programs on it, you will see a huge speed boost.
 

tweakboy

Diamond Member
Jan 3, 2010
9,517
2
81
www.hammiestudios.com
I just went with my gut got a F4 320GB. I read it came out in June 2010 and its built for performance and its single platter. Soo I bought it for 47 dollars shipped to the door step. I have 127GB free space out of it right now, after putting on image on it.

My final review of drive is.

The noise level when thrashing occurs in W7 is very audible and loud. But when normal usage you cant hear the drive too much when its working. Its normal I would say.

Speed. Thunderbird after 15 minute after boot up not touching anything opening it took 2 and a half to 3 seconds. Now it takes one and a half second on the F4 , Opening a large image zooming it,, some pics took 1 second when you click to zoom, this does it in 0 seconds. Everything is snappier.

I really dont care about benchmarks cuz it says my drive is a 934 score on passmark hd mark test,, but I got only 757 points after many retries. I dont care all I know is my drive is defragged soo no fragmentation,, and its speedy,, I can tell in my DAW with Sonar 8.5 Producer Edition ,, loading of samples is faster I can tell difference.

Soo bottom line Im happy with my single platter mechanical drive, Ill ride it until you show me a SSD that is 320GB for 100 dollars. Thank you ,, from Intel of course... maybe in 2012
 

sub.mesa

Senior member
Feb 16, 2010
611
0
0
@tweakboy

The Passmark isn't even an I/O test; it tests all aspects and the HDD is only small part of it. A faster CPU will directly give you higher passmark score; so don't use this to determine the speed of your current drive. Also, you can't compare this score with anything but your own system, as your system has different CPU, memory, chipset etc than other systems. So you can't compare Passmark numbers at all

Like PCMark, i suggest to avoid these benchmarks since it is totally unclear what the number you get actually represents.
 

computer

Platinum Member
Nov 5, 2000
2,735
2
0
Like PCMark, i suggest to avoid these benchmarks since it is totally unclear what the number you get actually represents.
Have you ever used PassMark or PCMark? Both PPT and PCMark give a detailed breakdown of the score, more so with PCMark.....like I have mentioned. So with them it is clear what the #'s represent.

PassMark:
Sequential Read
Sequential Write
Random Seek + R/W

And the overall HD score (Disk Mark), (but the PassMark site doesn't list the details of the overall #'s).

PCMark04:
XP Startup
Application Loading
File Copying
General HDD Usage

Then an overall HDD Suite score

PCMark05:
XP Startup
Application Loading
General Usage
Virus Scan
File Write

Then an overall HDD Suite score

That's why I said it (and WinBench & esp FCTest) are better for determining how a HD will act in actual use rather than tests such as HDTune, HDTach, CrystalDiskMark and the like.
 

sub.mesa

Senior member
Feb 16, 2010
611
0
0
I'm not sure we are in agreement. However, my point was that PassMark depends more on things like CPUs so you can't say person X has 900 score with the same HDD i have, so my HDD is not performing like it should. Basically the scores do not mean anything unless you compare them with the exact same setup except one thing changed (such as different HDD). I looked on that website a bit at the test scores they publish, and it just looks alot like sequential I/O; the SSDs are only twice as high scores as HDDs which means they don't rely on random I/O alot.

The PCmark benchmark uses 'captured I/O' using traces like 'XP startup'; the problem is that you can't really reproduce them the same way and you leave out of the equasion the CPU utilization necessary between the I/O requests. Also multiqueue is not handled properly; though it does appear to simulate multiple queue depth.

My opinion is that benchmarks are only useful if you can explain exactly what the number you get means. The more complex benchmarks you use, the more potential for wrong conclusions and the more factors that play a role in determining that score. I would argue that if you don't exactly understand the meaning of the score then it's exact value is meaningless; it doesn't answer the question how the HDD would perform for you in your situation.

With basic sequential and IOps performance, you got the basic performance characteristics of the HDD, which would play a significant role in all the I/O related performance characteristics. If you insist on 'real world tests' then do tests with a stopwatch instead, such as extracting a file with many small files inside it. But benchmarking properly can be very hard and it is easy to 'contaminate' your benchmark scores; for example by not preventing OS filecaching from making it a RAM-test instead of a disk-test.

Especially many Windows benchmarks look very nice, but the numbers they give you are not authoritative at all. The best benchmark i know is Intel's iPeak suite with Rankdisk; but it is not easy to use at all.
 

computer

Platinum Member
Nov 5, 2000
2,735
2
0
I'm not sure we are in agreement. However, my point was that PassMark depends more on things like CPUs so you can't say person X has 900 score with the same HDD i have, so my HDD is not performing like it should. Basically the scores do not mean anything unless you compare them with the exact same setup except one thing changed (such as different HDD). ......
No, I agree, all I'm saying is that tests like HDTach, HDTune, Crystal (and similar) are not as accurate as the others I mentioned on telling you how a HD will act as your Windows OS drive.


My opinion is that benchmarks are only useful if you can explain exactly what the number you get means.
And PCMark, WinBench & similar do that.


it doesn't answer the question how the HDD would perform for you in your situation.
Of course it does. ;) If you run them on your own PC.


If you insist on 'real world tests' then do tests with a stopwatch instead, such as extracting a file with many small files inside it.
That's exactly what FCTest does. I also do that myself. (I have unchanged test folders consisting of dozens of files types, and drag and copy them several times, I zip them, double zip them, unzip, also do the same with rar, etc., etc., and time the results). And what I'm saying is that HD's that do better on those types of aforementioned tests also do better on PCMark & WinBench. Where as HD's that may do better on HDTune, HDTach and similar do not do well on PCMark and WinBench. That's because those two (and similar) are better at showing you which HD will be faster with your own typical Windows use.

Throughput tests are nice to look at, nice to know, and I do indeed use them, however they are not and should never be the tell-all for deciding which HD you use as your OS drive. For that, more credence and precedence is place on the results for those like PCMark & WinBench.

I agree you can't compare results at some website with your own, but it's all one can do, it is at least a starting point to indicate a general figure that you should see with the same HD tested.

What I do, is get some of the fastest HD's of the day and proceed to test them. Generally you can conclude that while a testing website's numbers may not be the same that you will see, their fastest HD's will also generally be the fastest HD's on your own setup. In other words, if you get one of the HD's at the bottom of their lists, it's probably also going to be slow on your own PC. If you get one of the HD's at the top of their lists, it's probably also going to be fast on your own PC.

I've been testing HD's for a long time, and I have seen some oddball (high) results on HDTach, HDTune, and similar, (and older SANDRA versions), and those particular HD's may do poorly on more realistic tests such as WinBench, PCMark, et al. Yet when it comes to testing those HD's with FCTest, and my own timed tests I described above, those HD's that do well on the more accurate tests (WinBench, PCMark, et al), also excel at the FCTest and timed tests.

Therefore, if one has X number of HD's at their disposal and cannot make a decision on which to use as their OS drive, they should turn to tests such as PCMark, WinBench, FCTest, etc. On the other hand, if they want a storage HD onto which to write/read/edit large A/V files for example, in that case then something like HDTune, HDTach may be a more accurate representation (and the timed tests with large archived single files such as .zip, .rar, .avi, .wmv, .exe, etc.), than those results of PCMark & WinBench.
 

tweakboy

Diamond Member
Jan 3, 2010
9,517
2
81
www.hammiestudios.com
Very nice post computer my friend. Totally true what you say. good stuff.


When I used to zoom in on a large pic, it used to take 1 second.
Now when I zoom in it takes 0 seconds.

Thunderbird used to launch on a clean reboot and wait 15 minutes in 3 seconds.
Now it launches in 1 second and a half on first launch.

DAW is snappier with samples loading them etc.

Soo at the end Im happy with my F4 and am glad I made this decision. I thank you all for posting to this thread and special thanks to FishAk for his contributions to this thread and good job computer your techy saavy very cool. thx and gg and gb