Real world experiences with the Sonnet Tempo Pro and 2 SSDs

[hfg]

Thanks for the quick reply. But is that really the reason? I am thinking of re-raiding as the user above did but am a bit unsure if it really was that which helped him.

Is there perhaps another testing software that uses a different method than incompressible data?

Try the other disk testing programs mentioned above for more representative results with your SSD.

SSDs with Sandforce controllers do not test well using the BlackMagicDesign "DiskSpeedTest" due to the test data used for the measurements (large incompressible video type files).

 

[AidenShaw]

Try the other disk testing programs mentioned above for more representative results with your SSD.

SSDs with Sandforce controllers do not test well using the BlackMagicDesign "DiskSpeedTest" due to the test data used for the measurements (large incompressible video type files).

If your workload mainly deals with compressed files, then Black Magic *is* representative.

 

[hfg]

If your workload mainly deals with compressed files, then Black Magic *is* representative.

Of course ... that is what it is designed for

But I gathered that pullman was concerned that his DiskSpeedTest results didn't measure up to others that he had seen which used other brands of SSDs and "appeared" to perform better. In normal OS X usage he will probably see no discernible difference.

DiskSpeedTest is very useful for "comparison tests" on these forums since it is readily available and free for all users. However, one must remember that the results can be skewed somewhat depending on the technology being tested.

 

[pullman]

Thanks very much for the further explanations. Thanks also 666sheep for suggesting Aja System Test and Xbench. I have now run them both and get the following.

I couldn't run AJA booted from the Sonnet's striped array but when booted off another SSD and run on the array I got reads around 850-1100 and writes approx 900 for most settings. For some settings, for instance the NTSC DV25 setting, the numbers dropped to around 500 on r/w.

Questions:

Why is Xbench's Unchached Read on 4K blocks so low?

Why do the numbers drop on some of AJA's settings?

About the compressed data thing - I mainly use the system for working on (sometimes rather large) Photoshop files (film scans actually). Do I then use "compressed files"? If so, I guess I shouldn't have bought Sandforce-based SSDs.

Overall I think i should be happy with the tests but honestly I don't really know what to look for so would appreciate your honest views.

Many TIA and again thanks for the further replies
Philip

 

[h9826790]

This is interesting, I have only one 840 Evo and mounted it on a Sonnet Tempo SSD. The Xbench result like this.



It seems the 4k and 256k performance has no relationship at all. compare to your result, my SSD perform better on 4k read/write, but much slower for 256k read/write.