Show us your SSD performance

[Cr00zng]

I don't think we disagree. I was only pointing out the significance of access time versus R/W times for the OS. And then went on to say that R/W speeds will have a bearing once we use SSDs for large data transfers.

We can certainly agree that for the time being the access time is more important....

For images you run at the speed of the receiving disk which is most likely a HDD or an external drive. The best times I have achieved with a Vertex2 to HDD was 3.6 min for an image of a 21GB OS (about 11GB compressed).

Yes, it is a Seagate Barracuda 1TBs 7200.12 external HDD connected via the eSATA 3 port. The short image creation time could be due to a number of things, different processor, Sandy Bridge architecture, etc. It may not be the Sandy Bridge since the eSATA port is connected via the Marvell chip and that explains why it has only one internal SATA 3 port on the Marvell chip. As we know, the Marvell chip cannot support the SATA 3's 6Gb/s throughput, due to its limit of 5 Gb/s maximum throughput. The version of Macrium is the same on both machine.

 

[whs]

I did not have much luck with both eSata and USB3 for Macrium images. They were faster than USB2, but not earthshaking - maybe 30% better. I was quite disappointed. The CPU is an i7. Could be that my enclosures are no good, who knows.

 

[Cr00zng]

eSATA should be a lot faster than USB 2.0 that maxes out at 40 MB/s, although most actual devices max out between 15-30 MB/s. Your external drive doing around 40-50 MB/s is slow and you might be correct; the external case might be the culprit.
Here's the benchmark for my eSATA drive:



My system does have USB 3.0 port, but I have no faith in it and also heard, that its throughput is rather on the low end under the best circumstances.

The other drive connected to the internal SATA 2 port, same as the external one, has this benchmark:



That's a 20% performance increase. sequential read/write, when compared to the eSATA drive. I may just run a long SATA cable from the inside to the external drive

And might as well, here's Vertex 3 with CDM 3.01:



Once again it proves that the SSD blows away the HDD when it comes to 4K random read and write...

 

[whs]

Yeah, my Atto data looks promising too, but in real life it is not so impressive.

 

[Cr00zng]

Well, those numbers are pretty much what you can get with a platter based drive...

It's been bugging me as to why my eSATA drive is slower than the internal drive, same drives just plugged into different interfaces. I had a hunch that the position of the hard drive could be the reason. The previous benchmark posted was recorded with the drive on its edge, in the stand standing tall. I've removed the stand, laid the drive flat, and re-run the benchmark; here's the results:



That's pretty much the same as the internal drive's benchmark results and it's good enough for me.

It's interesting and in retrospect logical that the drive on its edge loses about 20% of its performance. After all on its edge, it has to fight gravity to keep heads in correct position, while laying it flat the gravity actually helps the heads staying in the correct position.

Maybe you should check the position of your external drive, even if it is "not so impressive"...

 

[essenbe]

In my system I don't see a huge difference. I have a samsung spinpoint internal and the same drive in an external enclosure. Although, the are the only drives I remember seeing that will write faster than they read.

This is the internal

 

[whs]

It's interesting and in retrospect logical that the drive on its edge loses about 20% of its performance. After all on its edge, it has to fight gravity to keep heads in correct position, while laying it flat the gravity actually helps the heads staying in the correct position.

This is an interesting point. I will check that next time I have the unit up.

 

[si8mon]

SSD Performance

I decided to rerun the same tests that I did on my ssd 3 months ago. I was expecting a performance drop, but nothing this big. I used HDTune and AS SSD benchmarking tools.

I have the infamous OCZ Vertex 2 25nm ssd. Is the performance drop because of this or all is it the norm for all drves?

Early Frebuary


Late April
​

 

[essenbe]

si8mon, I think some of the performance drop is because you were in ACHI mode on the first test and IDE mode on the second test. This is a tutorial that may help.

 

[si8mon]

si8mon, I think some of the performance drop is because you were in ACHI mode on the first test and IDE mode on the second test. This is a tutorial that may help.

Thanks for the help, I followed the tutorial and changed to ACHI mode. My results have improved but there's is still a noticeable drop in performance from february.

Early February (ACHI Mode)


Late April (ACHI Mode)
​

 

[Dave76]

That looks better.

If you write a lot of un-compressible files, including AS SSD benchmarking, it will throttle down for a while.

Leave your system sit at log on for at least 2 hours or more, at least once a week. It will let the garbage collection work and should improve the performance.

Only do one AS SSD benchmark per one log on session and not more than a couple a week.

 

[si8mon]

Leave your system sit at log on for at least 2 hours or more, at least once a week. It will let the garbage collection work and should improve the performance.

Thanks, when you mean leave your system at log on, does that mean leaving the desktop (as in logged on) idle?

 

[essenbe]

No, it means to not logon, just leave it at the logon screen
I also disable defrag, turn off Hiberfile. Your alignment is good.
I also set my HD to never sleep, but you can let your monitor sleep.

 

[Dave76]

Yep, that's what I meant

 

[si8mon]

Ok, thanks for clearing that up.

 

[Britton30]

New updates from the old one here.

This is with 2 x 50 GB OCZ Vertex 2 SSD in RAID 0

That just makes me dizzy Shawn.(d)

 

[essenbe]

What are the best ahci drivers to use for an X25-M msahci or iastor?

 

[Wordsworth]

Just for the fun of it I went from the msahci to iaStor driver. You can see a difference in AS SSD scores, but real world use not.

 

[essenbe]

Thanks Wordsworth, I benchmarked mine both ways also and came out with slightly better numbers with iastor also. But it seems not quite as fast in realworld use. I know my restart times went up by about 5 sec.

 

[Wordsworth]

Interesting. I didn't see a difference in restart times but I suppose it's just synergy, something behaving differently on your system when used in conjunction with the iaStor driver.