In an all out effort to push maximum IOPS we loaded a 512b random workload 3 workers @ QD 128. The system spiked above 420,000 IOPS a few times but it was very difficult to capture a screenshot. This was a result of the limitation of our hardware under this extreme loading of my processor which was causing severe “choppiness”. I can confirm that it was a limitation of the hardware in use as server environments in enterprise applications use dual CPU systems. I am very curious as to how this would handle with a 6-core processor as a 990X would surely be able to get us up to the 460K IOPS. My highly overclocked 920 @ 4.65 was really struggling with this load!
512B random read, 128QD, 3 workers.
PASSMARK
I am using this benchmark to highlight server usage loads and the performance of this card. It is pretty amazing that, in a few of these tests, it outstrips the Areca 1880 by a large margin even though that controller was running entirely in cache! The brute force of the 9265-8i is simply amazing.
These types of mixed read /write file tests are very demanding on the storage subsystem.
FILESERVER
WEBSERVER
LINKED INDEX
INTRODUCTION ~ EXTERNAL CHARACTERISTICS ~ BENCHMARK PREP
ATTO/CRYSTAL ~ AS SSD ~ HDTUNE PRO ~ WEI/AIDA ~ IOMETER
PASSMARK ~ VANTAGE ~ CONCLUSIONS AND THOUGHTS
If you want more detailed information on the testing, parameters and configurations used in testing, feel free to hit the forums! I will be happy to answer any questions, and as long as I have the hardware, if there is anything I have missed, I would be happy to run some tests for you and there will also be some more detailed presentations on 4k scaling for you enthusiasts out there 🙂 ~ Paul