Is intel 520 240GB still one of the fastest real world drives?

[MaxT]

Hi,
Trying to decide on which SSD to get for a new build.
I am mostly interested in the real world performance, such as windows boot time, file copying, application load times, and more importantly game loading times.

There are a few reviews that have mentioned that Intel 520 was outperformed by the likes of Vertex 4 and Plextor M3 Pro. From the theoretical synthetic benchmarks I can see where that is true, however, from the real world scenarios listed above Intel 520 seems to actually dominate both Vertex 4 and Plextor M3 Pro.


Am I missing something? Since I have picked up a 240GB 520 for 209.99 from newegg, should I keep it or look for something newer/better?

Reviews with the real world scenario benchamrks:

Look at the bottom of the page for the real world usage section
Page 8 and check out page 9
Another one

And there were some others I can't really find right now.

So assuming they are around the same price, is Intel 520 really faster (240GB/256GB real world scenarios) than Crucial M4, Samsung 830, Plextor M3 Pro, Mushkin Chronos Deluxe, OCZ Vertex 4, etc...?

Thanks.

[johnw]

A lot of the so-called "real-world" tests you will see on review sites are not truly real-world. Many of them, such as the StorageReview one you linked to, are actually playing back "recordings" of IO-activity. There are at least two big problems with such recordings -- few (if any) of them actually record real-world data. Instead, they write some repeating pattern of canned data that they have, or some of them just write streams of zeros. Depending on exactly what they write, it can give a big advantage to Sandforce SSDs, since the write speeds are much higher for some types of synthetic, easily-compressible data. The other problem is that most of those recording play back hours of real-world activity in a matter of minutes. So it exercises the SSDs at high queue-depths, but that is not necessarily indicative of real-world usage.

If you want to see real-world usage, then time real-world tasks. For example, some of your links time an actual boot up sequence, which is fine as far as that goes. But the differences are usually 1 second or less. I don't know about you, but even if those measurements are accurate (one wonders if they might fluctuate +/- 1 sec if they repeat the test many times), I'm not very concerned about 1 second of boot time.

So I tend to look for real-world tests that approximate what I spend my time waiting for with my computer. In my case, that is video processing. The only times I'm really waiting for very long is when MKVmerge or whatever is processing some video or other on one of my scratch SSDs (before I copy the final video to HDD for long-term storage). And for me, the fastest SSDs for that sort of work are the Plextor M3P or the Samsung 830, since both of those have high sequential write speeds for incompressible data. The Intel 520 has a significantly slower write speed for incompressible data than those two.

But for you, obviously it depends on your workload. If you run a lot of games from your SSD, then you might want to look at real-world tests where they measure level-loading times. But in that case, there is little difference in speed between most of the top SSDs.

It all comes down to the workload that you consider "real-world".

[MaxT]

Thanks for your reply johnw, makes sense. Too bad most reviews I've seen don't really show true real world benchmarks. I've only seen synthetic and the one's i've linked.
Guess I'll have to do more research.

[Christopher]

JW -- Amen.

Peep out this review I chanced upon last week. I'm not familiar with this cat, but he would appear to be the new [H] SSD editor. HardOCP hasn't jumped off with any SSD reviews since last August IIRC, and the test bench is kinda cool actually. Check out the 520's read degradation at even pseudo-steady state.

Steady 4K Read kicks off our Steady State testing. For many SSDs the read speed is the same even after heavy usage, usually only the write speeds and mixed read/write speeds are affected. This measurement would usually not be included, but the Intel 520 does suffer some degradation of read speed after being subjected to a workload for an extended period of time. This certainly isn’t typical behavior. The Corsair does hold the same level of read speed even after heavy usage.

Just sayin'.

http://hardocp.com/article/2012/06/07/corsair_performance_series_pro_256gb_ssd_review/8


It's actually a Performance Pro review (how apropos....) but I'll be damned if I've ever seen any review bring up read degradation. To me, this little nugget is far and away the most interesting tidbit. The 520 and the M4 are tossed in for comparison's sake. It's not true steady state, but rather somewhat steady (I think 2x capacity seq. write precondition, then some WIP-C/WDP-C, but not Enterprise SNIA). Either way, I'm going to see what's up. I don't have a 520, but I have a couple other 2281s, and if I can replicate this, I have to ask... why hasn't anyone noticed?

Actually, now that I think about it, I remember SF1200s doing this. But that's a different ball of yarn.

[johnw]

That [H] reviewer does do some interesting tests. But he also makes some off-the-wall comments like this:

The M4 and the Corsair do not use over-provisioning, which is an extra amount of NAND that is used to keep performance steady under heavy workloads.

Of course the M4 and the Corsair PP do have reserved space, about 7%, since they have 128GiB or 256GiB of flash on board, but only available capacity of 128GB or 256GB.

All SSDs have some reserved space, since it is virtually impossible for an SSD to perform well without it.

Hopefully that guy learns some more of the basics of how SSDs operate before he makes too many more comments like that.

[MaxT]

So it looks like Plextro M3 Pro is the one to get. Guess I will be selling my 520 to cover my expenses and will wait for the M3P to get in stock at Newegg.

Thanks for your feedback!

[Christopher]

That [H] reviewer does do some interesting tests. But he also makes some off-the-wall comments like this:





Of course the M4 and the Corsair PP do have reserved space, about 7%, since they have 128GiB or 256GiB of flash on board, but only available capacity of 128GB or 256GB.

All SSDs have some reserved space, since it is virtually impossible for an SSD to perform well without it.

Hopefully that guy learns some more of the basics of how SSDs operate before he makes too many more comments like that.

I'm slightly less bearish on that . Every drive has ~7 percent spare area, but few drives have OP baked in. But I understand what you're saying. The Spare Area vs, OP stuff is pretty much beyond the patience of many readers, so trying to explain it is difficult under the best of circumstances. If that's the worst sin committed, then the dude's head and shoulders above the rest.

Seriously, I read a "review" recently in which the author claimed that SF drives got faster as capacity increases because "they have more cache". WTF? I'm not going to ding the H guy for a relatively minor gaffe in the grand scheme of things.

[johnw]

Spare area and overprovisioning and reserved space are the same thing. The only difference is that if the user leaves some space unpartitioned, then it is usually called overprovisioning. But it is all the same thing.

The reviewer got it completely wrong. And that is not the only example of strange statements he made.

[Christopher]

I understand. Drives don't make a distinction between the two, except for remaining spare block counts which don't increase with OP (in my experience this never happens, but I should look into this). I wonder how much editing had to do with some of the stranger sentences though. The editor may have trimmed or added or changed some of the authors words with reckless disregard. How many editors are both good at editing and proficient with accepted terminology/nomenclature?

[Zaxx]

Since I have picked up a 240GB 520 for 209.99 from newegg...

You must be referring to british pounds not dollars...either that or NewEgg screwed up big time.

[MaxT]

You must be referring to british pounds not dollars...either that or NewEgg screwed up big time.

Heh, nope. 329.99 - one day 20 instant coupon off - 2 day 100 rebate + free shipping = 209.99

[canthearu]

Not sure why you would sell the Intel 520 though. It is a perfectly good drive, will probably never fail in a hundred million years (my Intel 520 is boring as bat to endurance test)

[Christopher]

I agree, you should not sell the 520 unless you can do so at a profit. It is nothing if not reliable, and while it's not the absolute fastest, it has a lot going for it.

canthearu -- you should live dangerously and get an MTRON to test!!

[canthearu]

canthearu -- you should live dangerously and get an MTRON to test!!

I would, but I'm afraid the MTRON's never reached this little outpost of civilization.

[Christopher]

I would, but I'm afraid the MTRON's never reached this little outpost of civilization.

I bet the Battleship MTRON could make it to Australia, provided I'm not the Captain.