It seems like were putting 4k random performance up on a pedestal as the be all end all of everyday SSD performance, much like we did with CPU clock speed in the past. With all the millions of dollars Crucial, Intel, Marvell etc invest in R & D do people really think that they're going to take a step backwards at this stage of the game?
Take a look at the Anand review of the 510, http://www.anandtech.com/show/4202/t...d-510-review/4 which uses the same controller as the c400, and even though the firmware will be different, it still follows the same trend of lower random 4k, higher sequential. It shows the c300 is almost 3 times faster at random 4k write performance (49 mb/s vs 141 mb/s) and almost twice as fast in random read performance, (44 mb/s vs 79 mb/s) yet if you look at the real world "Light Workload" benchmark, the 510 bests the c300 in every benchmark (as well as the Heavy Workload benchmark.) Even the Vertex 3, which has a massive 212 mb/s random 4k write, and a sequential read/write performance advantage, is beaten by the 510 in several heavy and light benchmarks, so pure speed cant be the only factor to consider, obviously firmware plays an important role too.
Maybe 4k random alone is not as important as we thought it was, or we have reached the limits of what it can noticeably give us, as we have already seen massive improvements on 4k random since the first gen SSDs. Maybe it is a combination of different parameters, including firmware that give the best results and now we are starting to find the sweet spot.
Maybe random 4k performance after a certain speed just isnt noticable anymore (eg 50mb/s) For example, if a file is loaded in .02 seconds on one drive, and .01 seconds on a different drive, are you really going to notice which is faster, even though technically its twice as fast? No, your not. So why not put extra focus into areas that will improve performance where you WILL notice it, which is what i think Intel and Crucial are doing.
All im saying is that im sure the guys making these drives know their stuff alot better than we do, so we should'nt put all our eggs in the random 4k performance basket.
Take a look at the Anand review of the 510, http://www.anandtech.com/show/4202/t...d-510-review/4 which uses the same controller as the c400, and even though the firmware will be different, it still follows the same trend of lower random 4k, higher sequential. It shows the c300 is almost 3 times faster at random 4k write performance (49 mb/s vs 141 mb/s) and almost twice as fast in random read performance, (44 mb/s vs 79 mb/s) yet if you look at the real world "Light Workload" benchmark, the 510 bests the c300 in every benchmark (as well as the Heavy Workload benchmark.) Even the Vertex 3, which has a massive 212 mb/s random 4k write, and a sequential read/write performance advantage, is beaten by the 510 in several heavy and light benchmarks, so pure speed cant be the only factor to consider, obviously firmware plays an important role too.
Maybe 4k random alone is not as important as we thought it was, or we have reached the limits of what it can noticeably give us, as we have already seen massive improvements on 4k random since the first gen SSDs. Maybe it is a combination of different parameters, including firmware that give the best results and now we are starting to find the sweet spot.
Maybe random 4k performance after a certain speed just isnt noticable anymore (eg 50mb/s) For example, if a file is loaded in .02 seconds on one drive, and .01 seconds on a different drive, are you really going to notice which is faster, even though technically its twice as fast? No, your not. So why not put extra focus into areas that will improve performance where you WILL notice it, which is what i think Intel and Crucial are doing.
All im saying is that im sure the guys making these drives know their stuff alot better than we do, so we should'nt put all our eggs in the random 4k performance basket.