Benchmark Programs - HDD/SSD

What benchmark programs do people recommend, in particular for HDD/SSD? I have seen many screen shots but cannot determine which one the majority of people are using. I mostly want to do it as I know every once in a while a drive fails, and I would like to test that out while it is still new and returnable .

 

Crystalmark is a good one that a lot of people use

 

More on this... I was actually thinking of doing the same thing to compare the relative speeds of some external USB/Firewire drives. Do these benchmarking programs (Crystalmark) measure peripheral throughput as well? If not, does anyone know if there's some other software out there that does?

 

Instead on benchmarking programs that can only hope to imitate real life, why not use a set of your own programs/processes that you test/compare different storage subsystems with?

For example:

Time the installation of a large app (CS5 for example).

Time the copying of a standard set of files (keep them just for this purpose) from an external HD, from one folder to another and from one partition to another (if you're using partitions in your systems).

Finally, time standard tasks: like booting up, shutting down, launching programs and most importantly: creating content with those programs (whatever that content happens to be - be it PDF files, images, videos, audio, etc.).

When you have a baseline of the above, then you can confidently know that your new HDD/SSD is actually performing better (or worse) than what you had - instead of seeing a 'score' that is not reflecting any of the above real world scenarios that I outlined above.

Sure, you will not be able to directly compare to others - but that is a small price to pay when you will actually know the benefits that the new component did for you and your system.

Also, it will make it impossible for manufacturers to be able to 'cheat' on a popular benchmark program and simply tell the drive to 'report' a high score.

As to the USB/Firwire drives - Firewire is vastly superior - USB will peak out at around 30MB/s and both will kneel down to a fast drive in an eSATA setup - even USB3 will be hard pressed to outperform eSATA in certain situations that depend on as little lag as possible.

 

crystal mark and HDTune are both good. It's best to run both to verify scores. They each also have their advantages and disadvantages. Note that HDTune may glitch. I do not know why it glitches but it can on some hardrives.


This shows the gltich i am talking about
http://forum.notebookreview.com/win...hdtune-hdtune-pro-inaccurate.html#post6744790

Running an install can vary drasticly between each install. Plus other factors can apply.

Also USB3.0 is far faster than eSATA. Once the new eSATA is released that will be faster. Although nothing right now can come close to maxing that besides a RAID array or a high performance SSD

Also testing from one hardrive to another is inaccurate due to the test just shows you the speed of the slowest drive. So unless you got a high performance SSD testing a HDD or a high performance SSD testing a slower SSD that is by no means accurate. Now if you can find a program that dumps the data to RAM and sends data from ram to hardrive than that would be vary accurate.

Also to note on the installing of a program. If your installing from a disc your limit from the optical drive speed. If from installing from a file it depends on many factors from is the file on the smae drive, different drive,compressed or uncompressed, and other drives speed along with other issues

 

There's also -SSD-Copy which tries to be a bit more real world.

 

What factors will make an install faster or slower besides the physical HDD used?

This is how I can ensure all of your points above are taken out of the equation:

See:
http://forum.notebookreview.com/win...e-partition-strategy-better-than-cloning.html


As to USB3 being faster than eSATA - sure - in a purely theoretical sense.

But eSATA is using the native bus that the HDD/SSD uses and does not put any additional overhead like USB does.

So although USB3 may have a faster top speed - it's actual speed when driving a storage device is not above the native SATA interface that it is emulating - in certain situations (like when minimal lag is of paramount importance).

 

i understand the lag part but the transfer speed is still faster than eSATA no matter what.

Installing files? did you not read my post? If your installing from an optical drive..even a blue-ray drive is not faster than a hardrive. blue-rays can only transfer at 50MBps from the last thing i read and that was theoretical. DVD's are like 11MBps at best from the last time i ripped a movie. CD's???? not even going there. If you install a file from the drive itself...well thats not accurate because that's reading and writing at the same time and the latency is no where near good enough to give a good speed or a place for comparison. Installing from a hardrive to another hardrive will only give readings for the slowest drive. Again not an accurate comparison. So why did i have to repeat myself? Now i would say using boot time program to record boot times and comparing those real world numbers a good test as long as all variables are the same but still that can be easily off.

EDIT:and synthetic tests are still real world tests....just a more in-depth accurate account of raw speed. It's easy to compare 2 synthetic tests and see which one would be faster in various "real world" situations. If one has a faster read and write than it'll be better in a sequential data transfer. Now if it has a slower transfer rat but a far lower latency than that will be faster in non-sequential writing and reading...simple.

EDIT: id didn't read your whole thing right now becuase its too in-depth for me right now....i am feeling lazy ^^

 

The (in)efficiency of the protocol can easily make the theoretical throughput advantage disappear.

Where can I find a benchmark that shows the USB 3.0 throughput ?

 

Never seen anything even come close to passing USB3.0 settings...let alone it capping.

 

What do you mean ?

 

I haev never even seen a benchmark test of anything making USB 3.0 cap out at its max speed. You need an extremely fast SSD or even 2 SSDs in a RAID. I never seen anyone try that yet.

Side note Wikipedia said eSATA was limited to 120MBps or something and this was tested using a raid array lol. Which is stupid and wrong so i fixed it and posted screen shots of my single 7200rpm hardrive pushing 136MBps

 

Notebookreview.com reviewed the WD MyBook with USB 3.0 back in March. The USB 3.0 benchmarks came out slower than the same drive removed from the MyBook casing and hooked up to a SATA port. You can read the review here. So while the USB 3.0 specifications might say that it's faster, current implementation seems to be coming up short, whether due to driver issues or something else.

 

first did they run multiply tests? Crystal mark is notorious for having benchmarks all over the place. I can test my eSATA 7200rpm 1TB hitachi 10 times and get tests from 80MBps to 140MBps. Reason for this being crystalmark tests all over the drive and you have no control over where it tests. my drive ranges from 130-140MBps on the outer platter and 70-80MBps on inner platter. So if they just did a single test than that doesn't say anything. Thats why HDTune pro is more accurate except on some drives it glitches. My eSATA drive on HDtune pro says its capped at 90-100MBps but sisoft sandra's and crystal mark prove that to be inaccurate. Sisoft Sandra's test show my drive from 136MBps on outer platter and 70 something MBps on inner. So i would have to say those benchmarks could be in accurate if they didn't test them like 10 or more times. Or used HDTune or Sisoft Sandra's benchmark to verify.

Check this thread. I talk about my findings with my eSATA external. I say Sisoft Sandra's test has been the most accurate out of all of them. Than HDTune pro...i would like HDTune pro more except it glitches on some drives.

http://forum.notebookreview.com/win...hdtune-hdtune-pro-inaccurate.html#post6744790

 

I don't know, you could try PMing Kevin O'Brien, who wrote the original review to ask. From what we were talking about at the time, it seemed the fault may have laid with the NEC USB3 daughter card. I don't recall any other reviews of USB 3.0 peripherals recently, and I also seem to remember something about USB 3.0 in notebooks being limited to 2.5 Gbit/s due to the the single PCI-E interface, although that may have just been the Expresscard 1.0 implementation.

 

you are correct on the Expresscard ones but the full blown usb3.0 should be full speed.

Did you find my findings on crystal mark and HDTune Pro interesting? Have you noticed your scores varying a lot on crystal mark? Also the varying scores on crystal mark are easy to spot on 1TB-2TB drives because 1-4gig scans can be anywhere. On small drives a 4 gig scan take up a good portion and not be in just a single section of the drive. With a 1-2TB drive 4gig is only .2-.4% of the drive so who knows where that scan took place. On a 320 gig that's 1.25% of the drive. see the difference?

 

I've never personally bothered to run a benchmarking utility on my hard drive, although most of that is because mine is pretty ancient (100 GB Hitachi Travelstar 7K100). I can understand what you're trying to say, but I don't know enough about the benchmarking utilities themselves to know if the tests are conducted randomly, or if the utility is supposed to "know" where the outer portion of the platter is and check on that fastest part of the disk. To that end, I could see, for example, extra partitions confusing it or other similar effects. After all, numerous defragmentation utilities can identify the outer, and thus the theoretically fastest portions of platters, so I don't see why benchmarking utilities couldn't do the same. Whether or not they do, of course, is an entirely separate issue.

Also, for the "full-blown" integrated USB 3.0, I think the problem becomes how many slots are left in the mobile chipsets. I'm not sure there are enough PCI-E slots left in many chipsets after graphics cards and what-not to support full speed USB 3.0. That probably comes down to an individual model issue, though.