SSD Random Read/Write Speed Survey

[Thiol]

Hi all. There's been a large emphasis on the sequential read/write speeds of SSDs, but in many day-to-day usage patterns, the random read/write speeds are a better benchmark of performance. I'd like to make a chart summarizing all the different SSDs out there and available. But I need your help. Can those of you with SSDs run a benchmark and report the numbers here so that I can compile everything? Here are the steps:

1. Download Xbench 1.3 at http://www.xbench.com/.
2. Open the program and run the "Disk Test" for your SSD drive.
3. Note the numbers under the Random test, for all four parameters of small and large writes and reads.
4. Repeat the test a second time to make sure that the numbers do not change. Accept the second test if the values are <10% different from the first test.
5. Screen capture the window by holding down command and shift and 4 and the spacebar. Click on the result window and that will save as a png file on the desktop; the default is Picture 1.png.
6. Report the values here with by attaching the screenshot of the result window. This will give us other needed information like computer and SSD model.

I realize that there are weaknesses to this survey. Xbench is only one benchmark. SSD speeds will vary based on usage. Etc. In the end, however, the variability should be sufficiently small that I think we'll be able to compare different SSDs well.

I'm going to get us started by posting my results with an Intel X-25M on a previous generation MacBook Pro. Ironically, my result will probably get thrown out because the SATA 1.5 GB interface actually limits the SSD read speed...

Thanks in advance!

 

[JtheLemur]

MacBook Pro, 2.93GHz C2D, 4GB RAM, Apple-provided Toshiba 256GB SSD. Shoot me in the head, 54.11.

For comparison, an older MacBook, 2GHz C2D, 2GB RAM, gets 35.83.

While I didn't buy the SSD expecting it to be a speed demon…*the bigger, biggest problem is that this drive can't handle multiple operations. The drives sloooows to a crawl, bringing the system with it. To really appreciate how dreadful it is you'd need to see it in person, no words can describe the experience.

Based on my experience with my old Air, which had the 64GB SSD, *this* drive is exponentially worse. Not necessarily on a raw performance basis, but the ability for the drive and/or the interface to handle the load. The Air would slow down uniformly but remain responsive - THIS machine takes a nose dive!

I don't know what to do with it. May pick up a cheap 320GB HDD to use while I try to exchange the SSD for a new one. Figure maybe something wrong. If it's the same experience with a new one, I'll promptly flip the hell out at the *epic* waste of money this drive is.

 

[alphaod]

233.61

Enjoy:

 

[bugout]

have fun - new one first..

and the older one..

 

[bcaslis]

What's the difference between your older and newer one? Are they both the X25-M?

 

[bugout]

What's the difference between your older and newer one? Are they both the X25-M?

I'd hate to sound like cap't obvious here, but did you look under the system info?

The older one is an Early 08 MBP 2.5 with an OCZ Vertex.

 

[bcaslis]

I'd hate to sound like cap't obvious here, but did you look under the system info?

The older one is an Early 08 MBP 2.5 with an OCZ Vertex.

Doh! You're right, sorry temporary blindness.

 

[rbmclaughlin]

I'm running a 15inch UMBP 2.66, with 8GB ram, and a corsair (rebadged samsung) 256gb ssd.

 

[samiznaetekto]

300.40 - Crucial 128GB on MPB i7 17" 8GB

 

[MotherFinMike]

MBP 13" 2.53 8Gb Ram

 

[MBHockey]

MacBook Pro, 2.93GHz C2D, 4GB RAM, Apple-provided Toshiba 256GB SSD. Shoot me in the head, 54.11.

For comparison, an older MacBook, 2GHz C2D, 2GB RAM, gets 35.83.

While I didn't buy the SSD expecting it to be a speed demon…*the bigger, biggest problem is that this drive can't handle multiple operations. The drives sloooows to a crawl, bringing the system with it. To really appreciate how dreadful it is you'd need to see it in person, no words can describe the experience.

Based on my experience with my old Air, which had the 64GB SSD, *this* drive is exponentially worse. Not necessarily on a raw performance basis, but the ability for the drive and/or the interface to handle the load. The Air would slow down uniformly but remain responsive - THIS machine takes a nose dive!

I don't know what to do with it. May pick up a cheap 320GB HDD to use while I try to exchange the SSD for a new one. Figure maybe something wrong. If it's the same experience with a new one, I'll promptly flip the hell out at the *epic* waste of money this drive is.

wow, are your sure that's an SSD?

 

[bobm]

2010 MBP 13 8G 512G ssd

I get numbers that vary depending on the time of day (it seems).

Here is just the disk test, I have firefox and parallels running but parallels is paused.

Results 259.31
System Info
Xbench Version 1.3
System Version 10.6.3 (10D2125)
Physical RAM 8192 MB
Model MacBookPro7,1
Drive Type APPLE SSD TS512B
Disk Test 259.31
Sequential 177.74
Uncached Write 267.63 164.32 MB/sec [4K blocks]
Uncached Write 297.02 168.05 MB/sec [256K blocks]
Uncached Read 79.21 23.18 MB/sec [4K blocks]
Uncached Read 360.10 180.98 MB/sec [256K blocks]
Random 479.28
Uncached Write 273.70 28.97 MB/sec [4K blocks]
Uncached Write 400.99 128.37 MB/sec [256K blocks]
Uncached Read 1539.40 10.91 MB/sec [4K blocks]
Uncached Read 645.66 119.81 MB/sec [256K blocks]

 

[MacBird]

I'm running a 15inch UMBP 2.66, with 8GB ram, and a corsair (rebadged samsung) 256gb ssd.

I have the same SSD on a 2.53 uMBP with 4GB and Snow Leopard and get 205.

 

[JoshGlzBrk]

This is with my new 15" i7, 256GB SSD (Apple's Toshiba)

 

[JoshGlzBrk]

I have the same SSD on a 2.53 uMBP with 4GB and Snow Leopard and get 205.

You didn't even run the disk test...

 

[MacBird]

You didn't even run the disk test...

Actually, I did. I'm wondering if the difference is due to SL.

 

[jhajduk]

2x Vertex LE

 

[PaulDCarlucci]

Folks who are putting up the numbers for the 2010 MBPs with Apple SSDs... try filling up your drive with crap and useless junk and then rerun. It looks like those things might be crawling after they've got a bit of grime under their nails.

I'm 90% sure I'm going to just get a mechanical drive and then wait for the 3rd Gen Intels. Seeing that score at the top pretty much pushed me over the edge.

 

[expost]

Now we are talking, thanks to everybody for hard data (whatever hard is with benchmarks)!
I'd like to second PaulDCarlucci, THIS is the most interesting figure everybody in the market for Apple SSDs should ask. Would be great!

 

[spooky69]

The performance degradation over time is what I am interested in. With Snow Leopard not having any TRIM it seems that all drives will degrade more than they should. It will be interesting to see which drives degrade slowest but surely OSX should have TRIM by now given that they are supplying SSDs and have been doing so for some time? I wonder if we are any closer to being able to get an SSD at a reasonable price that OSX can actually manage as it should?

EDIT:
I am also interested in whether or not the new MBP still has the 1.5Gb SATA interface limit, as this will surely mean that the performance of the drive is limited?

 

[MBHockey]

Last gen MBPs were also SATA II...

 

[spooky69]

So by 'also' you are saying that they are still limited to 1.5Gb? I thought they were a 3Gb interface but limited to 1.5Gb?

 

[5883662]

Results 285.43
System Info
Xbench Version 1.3
System Version 10.6.3 (10D578)
Physical RAM 4096 MB
Model MacBookPro5,5
Drive Type OCZ-VERTEX
Disk Test 285.43
Sequential 240.81
Uncached Write 252.75 155.19 MB/sec [4K blocks]
Uncached Write 298.42 168.85 MB/sec [256K blocks]
Uncached Read 142.47 41.69 MB/sec [4K blocks]
Uncached Read 437.84 220.05 MB/sec [256K blocks]
Random 350.33
Uncached Write 123.48 13.07 MB/sec [4K blocks]
Uncached Write 550.99 176.39 MB/sec [256K blocks]
Uncached Read 2497.97 17.70 MB/sec [4K blocks]
Uncached Read 905.84 168.09 MB/sec [256K blocks]

 

[Mark Booth]

FWIW, my friend Robert Morgan at barefeats.com suggests that XBench is a bit long in tooth and recommends using QuickBench 4 instead. He feels that QuickBench 4 is better suited for such comparison testing.

Mark

 

[stjohnswell]

FWIW, my friend Robert Morgan at barefeats.com suggests that XBench is a bit long in tooth and recommends using QuickBench 4 instead. He feels that QuickBench 4 is better suited for such comparison testing.

Mark

Did you run the test?