AMD Radeon R7 240GB Gaming SSD Review

[Guest]

Please feel free to comment about our story entitled "AMD Radeon R7 240GB Gaming SSD Review"

[parsec]

Re: AMD Radeon R7 240GB Gaming SSD Review

A very good and fair review of this SSD, which has all the best performance of OCZ's latest SSDs, plus a bit more. OCZ's somewhat write-centric performance is found in the AMD Radeon R7 too, of course. Its low latency is very impressive.

Forgive me my nitpick of one comment: "Still, the AMD Radeon R7 is faster at this time than the other OCZ products, but still lags behind newer SSDs that deliver over 10K random read IOPS at QD1.

Alas, the Samsung EVO is the only SSD that does give us over 10K random read IOPS at QD1, given the Samsung 850 Pro is not included in this review. Including the 850 still makes the 10K+ 4K QD1 IOPs an exclusive club, not even the SanDisk Extreme Pro makes it in. Over 9K random read IOPs at QD1 would be a more appropriate comment, IMO.

I can't help but smile at the sequential mixed workload tests results, showing us why SandForce based SSDs can have exceptional performance in some if not all real world scenarios, while appearing poor in some purely synthetic tests. Thanks for posting those results!

Interesting that the Indilinx Barefoot controller's firmware has a consistent break in performance at about the 70% read point. All the other SSDs shown except the Intel (SandForce) have a generally similar type of performance curve, an interesting phenomenon.

Does a random mixed workload test make any sense?

[chrisram]

Re: AMD Radeon R7 240GB Gaming SSD Review

Hello and thanks for the comments!

Samsung just sent the 850 Pro drives I need to finish my charts today (256GB and 512GB). 10K IOPS at QD1 Read is a milestone for sure but you are correct, I should stay grounded and use 9K.

I'm working on a random mix right now but I'm trying to find the right precon time and queue depth for the test. Since I write for the consumer market, it's not really fair to take a drive all the way down to 4K steady state and then run the test. I'm testing a handful of drives with different preconditioning before doing a the random mixed IO.