Vertex 3 is way to slow!

[Keychain82]

A few days ago I received my new MBP 15", 2.2ghz quad with the standard Apple SSD equiped. Today my Vertex 3 240 GB arrived. I was really looking forward to the final speedbump, installed it and - felt no difference at all? Okay, let's run some benchmarks with xbench. Here are the results:

1st run

Disk Test 470.87
Sequential 274.96
Uncached Write 563.91 346.23 MB/sec [4K blocks]
Uncached Write 497.42 281.44 MB/sec [256K blocks]
Uncached Read 110.14 32.23 MB/sec [4K blocks]
Uncached Read 593.65 298.37 MB/sec [256K blocks]
Random 1637.85
Uncached Write 2489.78 263.57 MB/sec [4K blocks]
Uncached Write 935.92 299.62 MB/sec [256K blocks]
Uncached Read 2896.44 20.53 MB/sec [4K blocks]
Uncached Read 1595.27 296.01 MB/sec [256K blocks]

2nd run

Disk Test 480.94
Sequential 279.49
Uncached Write 613.54 376.71 MB/sec [4K blocks]
Uncached Write 499.56 282.65 MB/sec [256K blocks]
Uncached Read 110.36 32.30 MB/sec [4K blocks]
Uncached Read 617.66 310.43 MB/sec [256K blocks]
Random 1722.36
Uncached Write 2759.62 292.14 MB/sec [4K blocks]
Uncached Write 990.82 317.20 MB/sec [256K blocks]
Uncached Read 2972.24 21.06 MB/sec [4K blocks]
Uncached Read 1627.81 302.05 MB/sec [256K blocks]


This doesn't look like the specs at all. A firmware upgrade didn't change a lot..

Disk Test 480.45
Sequential 278.38
Uncached Write 548.52 336.79 MB/sec [4K blocks]
Uncached Write 494.69 279.90 MB/sec [256K blocks]
Uncached Read 112.28 32.86 MB/sec [4K blocks]
Uncached Read 618.17 310.69 MB/sec [256K blocks]
Random 1752.50
Uncached Write 2734.72 289.50 MB/sec [4K blocks]
Uncached Write 1036.12 331.70 MB/sec [256K blocks]
Uncached Read 2867.61 20.32 MB/sec [4K blocks]
Uncached Read 1658.58 307.76 MB/sec [256K blocks]


The numbers are just getting worse. Can anyone see a problem? The drive is listed at a 6Gbit/s connection.

Any help is appreciated.

 

[Orlandoech]

Did you do a clean install of OS X or did you clone? When I did a clone my Vertex 3 was slow too, I then tried a fresh install and it was blazing.

 

[Keychain82]

It was a clean install. I did not format the disk before, but I don't think, that this is the Problem. I startet the dvd, wrote the partitioning table to the disk and installed.

 

[Orlandoech]

Possibly a defective SSD then? Maybe try doing it one more time with a format and see if it makes and diff. If not then I would RMA the drive.

 

[Keychain82]

Okay, I'll give it a go. Format with the diskutility will be fine? I honestly don't like taken the book apart the third time only for formatting the disk. It's a shame it's not possible to boot from a livecd and update the FW, erase the drive etc.

 

[dagamer34]

Not defective. I get similar numbers. Do keep in mind that all SSDs advertise their speeds as "up to XXXMB/sec", not "always running at XXXMB/sec". It may not show up in this benchmark, but the drive is plenty capable.

 

[cluthz]

Xbench at ~500 is very good.
The stock HDD lists about 70.
I think the Intel x25 clock in about 200-250.

 

[Keychain82]

Interesting, thanks for your replies. I just can't feel any difference between the stock-ssd and the vertex 3. At least during the benchmarks it should it somwhere in the region the manufacturer posts i thought. During normal use there is absolutely no speedbump.. starting eclipse takes the exact same 5 seconds as it does with the toshiba drive. I try to get the benchmarks from xbench i did with that drive.. I think they showed something around ~350.

 

[dagamer34]

Interesting, thanks for your replies. I just can't feel any difference between the stock-ssd and the vertex 3. At least during the benchmarks it should it somwhere in the region the manufacturer posts i thought. During normal use there is absolutely no speedbump.. starting eclipse takes the exact same 5 seconds as it does with the toshiba drive. I try to get the benchmarks from xbench i did with that drive.. I think they showed something around ~350.

I doubt anyone really can tell, that's what benchmarks are for. You really have to push the drive with heavy server I/O-type actions in order to separate SSDs in the real world based on performance.

The great leap from HDD to SSD comes from taking a 4.37ms latency on the fastest HDD (WD Raptor drive @ 10k RPM) and just destroying it with your average SSDs latency of 0.1ms (a 43x improvement) on every single file request. But since latency hasn't improved that much, going from one SSD to another isn't that impressive. Higher bandwidth IS needed if you wanted to dump RAW un-encoded 1080p data from a RED camera pushing hundreds of megabutes of writes per second, but not many people are in that position.

 

[cinek]

not sure if you've heard but there are some issues with the latest ocz ssds - especially the sandforce 2 controller. That's what might be causing your problems.

http://translate.google.se/translat...-problem-med-sandforce-2-aterkallelse-i-sikte

 

[JronMasteR]

My new Vertex 3 which was delivered with firmware 2.06 is working perfectly. Cloned it via Carbon Cloner and put it in. My speeds are about 500MB/s read and write. I hope it stays that way.
My advice is, try another speed test tool like AJA System Test and see if you get other results. On xBench I'm getting similar results as you do...

 

[MikhailT]

Interesting, thanks for your replies. I just can't feel any difference between the stock-ssd and the vertex 3. At least during the benchmarks it should it somwhere in the region the manufacturer posts i thought. During normal use there is absolutely no speedbump.. starting eclipse takes the exact same 5 seconds as it does with the toshiba drive. I try to get the benchmarks from xbench i did with that drive.. I think they showed something around ~350.

The latency is pretty much the same between the stock SSD and the Vertex 3, you won’t see any major performance gains in your daily usage and you won’t feel it unless you pay extra super attention to what you’re doing. The only speed benefit you’ll see is when you’re transferring huge files over a long period of time on the same drive or to another SSD on the same machine.

Xbench is not the best app to test your speed, it is fairly old and inaccurate with the latest SSDs. Even so, your benchmarks looks to be right for that SSD.

The 500Mbps read/write speed isn’t entirely accurate, it is only for transferring files sequentially and on the same drive. The random performance is much lower than that, ~200-300MBps and even lower for write.

 

[imorton]

The latency is pretty much the same between the stock SSD and the Vertex 3, you won’t see any major performance gains in your daily usage and you won’t feel it unless you pay extra super attention to what you’re doing. The only speed benefit you’ll see is when you’re transferring huge files over a long period of time on the same drive or to another SSD on the same machine.

Xbench is not the best app to test your speed, it is fairly old and inaccurate with the latest SSDs. Even so, your benchmarks looks to be right for that SSD.

The 500Mbps read/write speed isn’t entirely accurate, it is only for transferring files sequentially and on the same drive. The random performance is much lower than that, ~200-300MBps and even lower for write.

I wanted to upgrade my Vertex2 on my 2011 Macbook Pro 13" to a Vertex3 because the specs (read/write) of the SataIII Vertex3 SSD's seem so impressive, yet I get the feeling that I wouldn't see much of a difference in general real world performance?

 

[couto27]

ocz had a few problems with some vertex 3 models

check their ocz forum, yours is most likely defective.

free replace.

 

[MikhailT]

I wanted to upgrade my Vertex2 on my 2011 Macbook Pro 13" to a Vertex3 because the specs (read/write) of the SataIII Vertex3 SSD's seem so impressive, yet I get the feeling that I wouldn't see much of a difference in general real world performance?

You won’t. Is there anything you’re doing that requires that much speed? Majority of people won’t notice a difference between Vertex 1 and Vertex 3 because they don’t anything to take advantage of the speed. The app’s launch speed is determined by the latency first, not entirely by the actual sequential read/write speed.

 

[imorton]

You won’t. Is there anything you’re doing that requires that much speed? Majority of people won’t notice a difference between Vertex 1 and Vertex 3 because they don’t anything to take advantage of the speed. The app’s launch speed is determined by the latency first, not entirely by the actual sequential read/write speed.

You can see some speed differences from a Sata-I bus 2008 Macbook 4.1 and regular Hard Drive (first column) , to a Vertex2 SSD in the second column.

Than in the last column you can see the speed difference going to a 2011 Macbook Pro 13" with a Sata-III bus with a Vertex2 SSD and 8 gig of ram.

 

[MacsOnAnabolics]

These are my numbers...


I/O Transactions
Frame MB/sec
Read Write
00.0 447.3 488.0
01.0 528.2 494.6
02.0 529.7 321.0
03.0 529.2 479.9
04.0 529.8 490.9
05.0 529.0 482.0
06.0 529.1 486.4
07.0 530.9 499.7
08.0 530.8 485.7
09.0 528.3 496.3
10.0 527.3 482.5
11.0 527.9 491.5
12.0 527.3 482.9
13.0 527.8 484.0
14.0 529.9 490.9
15.0 531.4 488.1
16.0 527.6 329.5
17.0 529.6 485.8
18.0 531.0 487.6
19.0 530.4 483.7
20.0 526.9 494.0
21.0 529.8 484.9
22.0 528.3 486.8
23.0 529.7 481.7
24.0 530.9 492.2
25.0 528.9 488.5
26.0 528.1 498.3
27.0 527.9 485.5
28.0 525.2 490.6
29.0 526.8 321.1

 

[lankox]

I also upgraded from the stock Apple SSD to the Vertex 3 and can tell no speed difference between them either. In hindsight, I wish I would have just stayed with the Apple drive.

 

[imorton]

I also upgraded from the stock Apple SSD to the Vertex 3 and can tell no speed difference between them either. In hindsight, I wish I would have just stayed with the Apple drive.

That's interesting what you say, that you don't notice any real difference between a SataII Apple SSD and the SataIII Vertex3 SSD.

I guess the typical boot, application launch, and general Internet, mail, calendar, and other general day to day work doesn't see any substantial increase in speed?

I guess the big difference in noticeable speed is when someone upgrades from the stock HD to a SataII or SataIII SSD

 

[JR1993]

That's interesting what you say, that you don't notice any real difference between a SataII Apple SSD and the SataIII Vertex3 SSD.

I guess the typical boot, application launch, and general Internet, mail, calendar, and other general day to day work doesn't see any substantial increase in speed?

I guess the big difference in noticeable speed is when someone upgrades from the stock HD to a SataII or SataIII SSD

Yes but I imagine that someone who runs servers or does a lot of read/write on their drive will notice the difference between SATA II and III.