Random Read Performance

Random 4 KB read performance is mixed among Intel's new drives. There are lot of newer SSDs out there able to trounce even the 240 GB SSD 525. Not even compressible data helps put the SandForce-based mSATA drives in a special place. Fortunately for Intel, the reality of most client-oriented storage solutions is that they won't see a lot of constant high-queue depth random I/O.
Between the capacities, results at a queue depth of one are constant at around 24 MB/s. Why are they all going that fast? Performance at those settings is governed almost exclusively by the flash, and even drives with different controllers perform the same when they utilize the same NAND. As more requests stack up, the controller, memory, firmware, and flash translation layer all come into play.

Not much changes when we switch to random data. The advantage attributable to moving compressible information isn't as pronounced when the access pattern is random, so the consequences aren't as severe when that advantage is lost. As the queue depth depth increases, larger SSDs do benefit, though. The 60 and 30 GB models are once again left in an unenviable position: slow and slower.
Random Write Performance

The smallest SSD 525 can even reach 300 MB/s with 4 KB random writes using repeating data. The other four models bunch up under 350 MB/s, hitting an overhead-induced bottleneck.

Testing with random 4 KB data looks a lot like the sequential workload on the previous page. Aside from the two smallest capacities, the 120, 180, and 240 SSD 525s continue punching above their weight.
The 180 and 240 GB models are neck and neck, just over and below 250 MB/s. The 30 GB drive still can't get over the 50 MB/s hurdle, while the 60 GB versions stalls at the 100 MB/s mark.
- Intel SSD 525: Intel Goes 6 Gb/s With mSATA
- Test Setup And Benchmarks
- Results: 128 KB Sequential Performance
- Results: 4 KB Random Performance
- Results: Comparative 4 KB Random Performance
- Results: Comparative 128 KB Seqential Performance
- Results: Storage Suite v1.0, PCMark 7, And Write Testing
- Power Consumption
- SSD 525 Is Pretty Pricey, But Also Powerful
Interesting, if some benches weren't Intel only, but all included the relavent competitors.
This is not something manufacturers do to just to p*ss off users who buy the smaller capacities.
A small drive has fewer memory chips than a large drive. The controller has then fewer chips to efficiently spread the data to... and this leads to decreased performances. There's nothing immoral to that.
It's not the same story like for example, a couple of years ago, Yamaha selling a 2x CD writer and a 4x CD writer at double the price ... and by removing one resistance, your 2x writer became a 4x model ;-)
evaluating price per performance as it is frequently offered at around $.60 or less per GB.
It's a surprisingly good drive, and performs very well on boards that only have SATA2.
I recently upgraded my brother's P55 system with an 840 250GB; the main game he
plays atm now loads in just a few seconds, instead of the more than 3 minutes it took
with the old mechanical disk (and that wasn't exactly a low-end drive either - a WD VR
150GB 10K SATA). He is, as one might expect, very happy indeed.
In addition, I bought him an internal Startech storage unit that holds 4 x 2.5" devices
(it takes up one 5.25" bay) and a couple of 2.5" drives (1TB for general data, 2nd-hand
250GB for backup of the 840). He bought another 1TB for backup, so the Startech now
holds the 840, two 1TB and the 250GB. The end results looks rather good, and the
performance with the 840 is excellent (I bought one for my 3930K setup).
I have a lot of OCZ drives (more than 40, various models); what impresses me the most
about the 840 is the way it maintains top performance even after being hammered with
an 80GB full clone from an old disk, lots of Windows and driver updates, game installs, etc.
Testing with HDTach, AS-SSD, etc. show performance almost identical to an original clean
state. None of my OCZ drives behave this way - the HDTach graph shows significant
variance, while the 840 graph is smooth across the range. Beats me how Samsung has
achieved this, but I like it.
Modern SSDs may be saturating the SATA3 interface, but they bring an amazing new lease
of life to older SATA2 systems.
Ian.
I have an ASRock Z77E-ITX back from RMA that I haven't yet put back into service that has a mSATA slot on its underside. It can be used to build a very small system. That these slots are only 3Gb/s hardly matters when comparing them to the speed of a mechanical HDD.
You are confusing msata with mini pcie. A drive is a drive is a drive, sata is sata is sata. Connect any msata drive to an actual msata port (not mini pcie which has the same connector) and it can become your C drive. No one is forcing you to use Intel SRT\RST to use an msata drive as cache.
If you purchased a 2.5" ssd and now feel your msata port is useless thats on you. If you had purchased an msata drive you could have used a 1tb in that 2.5" bay instead.
to add some clarification: the confusion stems from some laptops using a mPCI-Express as a multipurpose slot allowing either mPCI-Express or mSATA cards. while i have not seen this on desktop motherboards, maybe ddpruitt's experience comes from spotty documentation from laptop makers on whether their combo port supports mSATA? otherwise, you are very correct that the mSATA should appear to the system as any other SATA drive and be usable as such.