Sign in with
Sign up | Sign in
Your question

Most reliable benchmark test?

Last response: in Storage
Share
January 27, 2013 10:36:53 PM

hi. what are some of the most reliable and accurate SSD read/write benchmark tests?

thanks!
a c 168 G Storage
January 27, 2013 10:59:08 PM

That would be a controlled comparison test of YOUR workload.

In general, I would discount most of the synthetic benchmarks. They use apps that drive the SSD's to their maximum rates at very high queue rates.
That might be appropriate for a server.

We usually do things one or two at a time, so the response time for individual reads and writes are more relevant. This number is not usually reported, and for the most part, all modern SSD's perform about the same in normal desktop work.

That is about 50x faster than a hard drive in random I/O which is what the os mostly does, and 2-3 times faster in sequential.

Perhaps one I might pick out would be the anandtech ssd bench selecting light workload disk busy time.
Here is the current report. Only the samsung 840 pro is as good as 10% faster.
http://www.anandtech.com/bench/SSD/269
m
0
l
January 27, 2013 11:06:21 PM

im picking up my SSD tommorow and i just want to see if it comes close to the advertised speeds. i want to know if its a defected so i can return it within the 30 day time frame.
m
0
l
Related resources
a c 168 G Storage
January 28, 2013 12:47:07 AM

computernewb said:
im picking up my SSD tommorow and i just want to see if it comes close to the advertised speeds. i want to know if its a defected so i can return it within the 30 day time frame.


I might suggest as ssd.
Not so much for the speeds which you can compare to many benchmarks in reviews, but to note that your ssd capacity will be in green which means that your drive is aligned on a 4k boundary, or in red if it is not.

On a clean install, you will be aligned. on a clone, perhaps not.
Also, verify that your sata mode is set to AHCI, not raid or IDE.
m
0
l
a b G Storage
January 28, 2013 12:58:56 AM

ATTO
m
0
l
a c 87 G Storage
January 28, 2013 2:55:48 AM

*Find a review for YOUR exact SSD then run the SAME benchmark that they use in the review.

Different applications vary significantly in the numbers they generate. You can't run ATTO then look at an AS-SSD benchark to see if things are okay.

It will also depend on your SATA controller. Make sure to use the BEST one for the job. Possibly the INTEL one.

Notes:
- if numbers seem about HALF of what you expect you likely are on a SATA2 controller for a SATA3 SSD, or are on a SATA3 controller but stuck at SATA2 speed for some reason (it happens).

- If the numbers of your benchmark are within about 15% (controller variance) you are fine.

- SSD's are so fast that a SATA3 drive may not feel much different than a SATA1 drive for basic Windows usage, especially if your RAM is at least 3GB.
m
0
l
a b G Storage
January 28, 2013 5:07:27 AM

Some factor for choosing SSD

- Controller Chip (samsung, sandfoce, etc....) same type (generation) of controller even in different brands usually had same performance....

- size, bigger is better but the price is higher but the bigger they are usually the price/size ratio is better...

- Brands, each brand charge differently for same (spec) SSD.... choose what u trust most

m
0
l
a c 87 G Storage
January 28, 2013 1:47:29 PM

He didn't actually ask, but if I had to choose an SSD it would be the:

SAMSUNG 840 PRO

*Note that the Samsung 840 (non-Pro) uses TLC which has a lower lifespan. That doesn't mean it's a bad drive if the price is right though. My current issue is simply that this technology is quite new so the issues are unknown.

TLC will very likely dominate the market in another year or two and most likely will end up being my recommend for value until something better comes along. Even the TLC-based Samsung 840 will likely last for five years easily and by then a replacement will be a fraction of the price.

**No matter what SSD drive you get, make sure to flash the Firmware. The (beta?) firmware for the Samsung 840 and 840 Pro's could brick the device completely but the issue has been solved.
m
0
l
a c 283 G Storage
January 28, 2013 4:20:56 PM

geofelt said:
That would be a controlled comparison test of YOUR workload.



I absolutely, positively agree with geofelt.
m
0
l
!