The Crucial BX300 (480GB) SSD Review: Back To MLC
by Billy Tallis on August 29, 2017 9:00 AM ESTRandom Read Performance
Our first test of random read performance uses very short bursts of operations issued one at a time with no queuing. The drives are given enough idle time between bursts to yield an overall duty cycle of 20%, so thermal throttling is impossible. Each burst consists of a total of 32MB of 4kB random reads, from a 16GB span of the disk. The total data read is 1GB.
The burst random read performance of the Crucial BX300 is better than the previous Crucial SSDs, but still trails behind quite a few other MLC SSDs and the two fastest 3D TLC SSDs.
Our sustained random read performance is similar to the random read test from our 2015 test suite: queue depths from 1 to 32 are tested, and the average performance and power efficiency across QD1, QD2 and QD4 are reported as the primary scores. Each queue depth is tested for one minute or 32GB of data transferred, whichever is shorter. After each queue depth is tested, the drive is given up to one minute to cool off so that the higher queue depths are unlikely to be affected by accumulated heat build-up. The individual read operations are again 4kB, and cover a 64GB span of the drive.
On a longer test of random read performance and with some higher queue depths in play, the Crucial BX300 ends up falling behind the MX200 but is substantially faster than the MX300 and most other TLC SSDs. Most of the MLC SSDs and the 3D TLC-based Samsung 850 EVO and Intel 545s significantly outperform the Crucial BX300.
The power efficiency of the Crucial BX300 when performing random reads is a bit below average for SATA SSDs, with a few planar TLC SSDs beating it.
 |
|||||||||
With a high enough queue depth, the Crucial BX300 delivers random read performance that is as good as any SATA drive of this capacity, but it also requires quite a bit of power to perform that well. At more modest queue depths, the BX300 underperforms most of its competition-especially the Samsung drives, which saturate at QD16.
Random Write Performance
Our test of random write burst performance is structured similarly to the random read burst test, but each burst is only 4MB and the total test length is 128MB. The 4kB random write operations are distributed over a 16GB span of the drive, and the operations are issued one at a time with no queuing.
The Crucial BX300's QD1 burst random write performance is tied with the Samsung 850 EVO for second place, slightly behind the ADATA SP550 of all things. Crucial's MX300 is only about 6% slower, while the Samsung 850 PRO is about 16% slower.
As with the sustained random read test, our sustained 4kB random write test runs for up to one minute or 32GB per queue depth, covering a 64GB span of the drive and giving the drive up to 1 minute of idle time between queue depths to allow for write caches to be flushed and for the drive to cool down.
With a longer test duration and higher queue depths, the Crucial BX300 holds on to second place, this time scoring just behind the Crucial MX200. The Samsung 850s are just a hair slower than the BX300, and the Crucial MX300 is the next fastest.
In terms of power efficiency during the random write test, the BX300 is again in second place, with the MX300 holding on to a safe lead. The BX200's efficiency was abysmal, and the BX300 provides four times the performance per Watt on this test.
 |
|||||||||
The Crucial BX300 scales to near saturation by QD4, but power consumption keeps increasing up to QD8. The performance curve for the MX300 is just below the BX300's, but the power consumption of the MX300 stays significantly lower and even the slowest drives end up drawing more power than the MX300.
Facebook
Twitter
RSS
90 Comments
View All Comments
lilmoe - Tuesday, August 29, 2017 - link
Question. This is provably unlikely, but is binning layers possible?lilmoe - Tuesday, August 29, 2017 - link
Probably*Billy Tallis - Wednesday, September 6, 2017 - link
3D NAND is not really built one layer at a time. The first stage of building the memory array is to make a tall stack of alternating materials, and then vertical strings of memory cells are formed through that stack by etching deep but narrow holes and filling them with the remaining components. That high aspect ratio etching step is one of the main limiting factors in scaling layer count. If you push the layer count too far, you end up with memory cells in layers near the top of the stack having significantly different properties from the ones near the bottom of the stack.It's relatively unlikely to have an individual layer somewhere in the middle of the stack be dead/defective across that entire layer. It's more common to see an entire vertical column fail, which involves a much smaller number of memory cells.
Radio-Zone - Wednesday, August 30, 2017 - link
Thanks for the information!!!Ej24 - Wednesday, August 30, 2017 - link
Congratulations Micron you're almost back to where you were 2 years ago in performance with the m550, Mx100 and mx200. I've always been a huge fan of crucial SSD's. Great bang for the buck for MLC drives. But the last year or so it's been hard to keep praising crucial.m16 - Sunday, September 3, 2017 - link
It's an interesting move, but all in all, due to the shortage, any SATA drive will do for anyone that is looking for a switch to SSD on the desktop, while power might be the top issue for laptops.There's the RAM caching on some of their drives which is very good all in all, especially for computers that have AMD CPUs, that can't use Intel's caching technology to speed things up.
keta - Wednesday, September 6, 2017 - link
Over two-and-a-half years ago (January 2015), I bought a 256GB MX100 for $95. That worked out to $0.371/GB, or a little less than what the BX300 is going for today ($0.375).I would be willing to pay the same rate if it meant better performance, but using the ATSB Heavy stats in Bench, it seems that my old MX100 outperforms the BX300 in both data rate and latency. Are the 2015 ATSB Heavy stats comparable to the 2017 stats? Is it really the case that SATA SSD price/performance is worse than it was 2.5 years ago?
Billy Tallis - Wednesday, September 6, 2017 - link
The average data rate and latency stats for the ATSB tests should be comparable between the 2015 and 2017 test suites. The workload didn't change, but the OS version and motherboard did. Next month or maybe late this month, I'll pull the MX100 from my gaming machine and put it through the 2017 test suite.keta - Wednesday, September 6, 2017 - link
Thanks! I'd be super interested in a 'long-view' piece that puts some of the older flagship SSDs (X-25M, Vertex 2, MX100) through the present-day latency/consistency analysis that AT has developed. And maybe throw in that old WD Scorpio as well, not just to see how far we've come from spinning drives, but also to put the differences between SSDs in perspective!Lolimaster - Thursday, August 9, 2018 - link
And now in Peru you can find the BX300 120GB for $35 xD.