SDD benchmarks = reliable impressions?

D

Dario D.

Gawd
Joined
Dec 8, 2004
Messages
582
When an SDD benchmark shows that one SSD loads programs 2x faster than another, does that mean programs will load 2x faster?

Particularly, I'm looking at the Tom's Hardware benchmark for "Application Loading Performance".

(At the end of 2009, I got burned by this, when a supposedly blazing-fast SSD I got showed little to no real-world improvement (except in my benchmark apps).)
 
Last edited:
More info:
For instance, here's the same "Application Loading Performance" benchmark for traditional hard drives... (at least it looks like the same one, but the numbers are insanely low vs. the SSD's)

Does the VelociRaptor VR200M really load stuff 2x faster than the Samsung Spinpoint F4?
 
Well, think of it this way, do you run applications
or benchmarks all day?

Real world applications vary in performance. Benchmarks
are good for problem solving, testing stability and reliability.
 
I had a WD 640 GB with a 200GB shortstroke partition for games/applications. I moved to a 120GB Vertex 2. I don't really notice a performance difference. Windows boots a little faster but I only do that once a day ;)
 
I think anandtech has the best SSD comparisons hands down, they test using a variety of benchmarks that focus on not only specific operation (e.g. sequential and random read/writes of different sizes, e.g. 4kb RW or 16kb RW) but also real-world-like tests, plus they test for how well SSDs respond to TRIM and what capacity depreciation they show. These tests often reveal crucial flaws in otherwise great SSDs.

The most informative posts to newcomers to SSD tech are from earlier this year, but for more recent examples of there reviewing structure for new SSDs you can see things like this
 
My first SSD was also the Crucial M225, in 64GB flavor (slower than the 128GB one). I didn't start fresh, but rather cloned my old windows XP install on it (taking care of alignment, which in 2009 was quite complicated). The jump in performance was not noticeable, it was incredible ! I wonder what you do with your computer ?
 
I like the way BeHardware tests SSDs -- I think they do a much better job than Anand with his proprietary record/playback-usage benchmarks. BeHardware does some benchmarks, but they also time several real-world usage examples. Here is their latest SSD roundup review:

http://www.behardware.com/articles/830-1/ssd-2011-roundup-crucial-m4-ocz-vertex-3-intel-510-320.html

Here is a recent comparison of Crucial C300 and m4 in different sizes:

http://www.behardware.com/articles/832-1/crucial-m4-vs-crucial-c300-at-64-gb-128-gb-and-256-gb.html
 
In real world use, it probably isn't going to be twice as fast (and the fact that the chart you linked uses MB/sec as the metric makes it even less likely).

Even if it was, many times you won't notice the difference because the initial load speed was so quick to begin with. If the SSD cuts the load time for an application, like Excel, from half a second to a quarter of a second, does it really matter?
 
Storage is not always the bottleneck when loading applications. It also is not the only bottleneck when booting.

Make sure where you're expecting to be storage bound is where you're actually storage bound.

I've got one box that boots in the same exact time with a 7200rpm spinner OR an SSD. It's because of a stupid Soundcard / driver that takes about 3-4 seconds to initialize. As soon as I pull out the soundcard, the SSD boots WAY faster than the spinner.
 
I like to think of it like this: It's not always the hdd or ssd that is the bottleneck. Alot of times it's the software and just the way they have it programmed to load or how it's intergrated with the OS.

Not saying that hdd's and sdd's don't help, but I find that the bottleneck isn't always the drive, memory, cpu, etc.
 
john4200 said:
I like the way BeHardware tests SSDs -- I think they do a much better job than Anand with his proprietary record/playback-usage benchmarks. BeHardware does some benchmarks, but they also time several real-world usage examples. Here is their latest SSD roundup review:

http://www.behardware.com/articles/830-1/ssd-2011-roundup-crucial-m4-ocz-vertex-3-intel-510-320.html

Here is a recent comparison of Crucial C300 and m4 in different sizes:

http://www.behardware.com/articles/832-1/crucial-m4-vs-crucial-c300-at-64-gb-128-gb-and-256-gb.html
Click to expand...

Hot damn, they also test the 120-128GB models most people will actually buy! This review didn't come up when I was searching for a scarce Intel 320 120GB review, sadly. Thanks for the link!:D
 
john4200 said:
BeHardware does some benchmarks, but they also time several real-world usage examples. Here is their latest SSD roundup review:
Click to expand...
Thanks for the links. Those real-world tests are useful indeed.

I'm having a hard time being sold on the usefulness of the non-real-world benchmarks, though. I wonder how many unsuspecting readers get the wrong ideas from them. I like how video card benchmarks are done, where it's (usually) all about measuring performance that you'll actually see.
 
I've noticed ATTO Disk Benchmark and AS SSD give completely different results. So no, I don't think they are reliable for real life performance.
 
What you really notice with SSDs isn't so much that things launch really fast. I mean with a big amount of memory Windows precaches pretty well. What it does is make the system highly immune to being bogged down. It is kinda the final piece to having a heavily multi-core and multi-threaded CPU and a lot of RAM. The system has resources to process multiple heavy requests at once without bogging down.

Programs can still take time to launch because they take time to transfer data off disk, and also process it. Sometimes you are waiting on CPU when a program is loading as it processes stuff, particularly since many programs are still single threaded for IO. However they do launch fast with SSDs, faster than HDDs, and they just don't seem to bog down.

I'm not sure of a good way to benchmark it, but it is noticeable. Having SSDs at home now I find my work computer "slow" at times. It is always when the system gets IO bound and the disk is thrashing for whatever reason. SSDs just handle that much more gracefully due to their low latency and high throughput. You can still slam them, of course, but probably not in a desktop setting.
 
bigdogchris said:
I've noticed ATTO Disk Benchmark and AS SSD give completely different results. So no, I don't think they are reliable for real life performance.
Click to expand...

Faulty logic. If they give different results, either one or both of them are unreliable.

I've seen some inexplicable results from ATTO. Also, ATTO uses easily compressible data (essentially a stream of zeros).

AS-SSD has always been reliable for me. And it provides some extra information about your SSD that ATTO does not display.
 
My Crucial M4 128GB just came in today. If I get my Win 7 Ultimate OS tomorrow I should have some feedback by the end of the weekend.
 
You must log in or register to reply here.
Back
Top