Performance Consistency

Starting things off, our performance consistency test saturates the drive with 4kB random writes for a full hour, with a queue depth of 32, the maximum supported by the AHCI protocol used by SATA and most PCIe drives. This puts the drive's controller under maximum stress and writes enough data to exhaust all free space and spare area on the drive. This is an unrealistic workload for any client use, but it provides a worst-case scenario for long-term heavy use, and it sheds light on how different SSD controllers behave and if their performance will hold up as they fill up.

The average of the last 400 seconds of the test gives us a steady-state IOPS rating that is usually very different from what the manufacturer specifies for a new, empty drive. We also quantify the consistency of the drive's random write performance, and provide plots of the performance over the course of the test.

Steady-State 4KB Random Write Performance

The BX200 is off to a poor start, with very low steady-state IOPS where the BX100 managed to place closer to the middle of the pack.

Steady-State 4KB Random Write Consistency

The BX200's consistency is also at the bottom of the chart, indicating that there's wide variability between its best and worst performance even after entering steady-state.

For a brief moment, the BX200 performs almost as advertised, and then for a few minutes it performs well for a budget drive, but when it runs out of cache and spare area, performance hits the floor.

Save for the periodic but infrequent excursions to 9k IOPS and 20-25k IOPS, the BX200's steady-state hovers between 200 and 700 IOPS: better than a hard drive, but not what we want to see from a SSD.

Introduction, The Drive & The Test AnandTech Storage Bench - The Destroyer
POST A COMMENT

86 Comments

View All Comments

  • FalcomPSX - Tuesday, November 3, 2015 - link

    SLC NAND stores one bit per flash cell. MLC stores two bits per cell. and TLC stores three bits per cell. Reply
  • zeeBomb - Sunday, November 8, 2015 - link

    Less bits the better? Reply
  • Dahak - Tuesday, November 3, 2015 - link

    So I had primarily just skimmed the article, but for a successor to the BX100 its not.
    I wonder if there is some firmware bug that might be causing an issue.

    I have used the BX100s fine in general office settings, but i many need to pass on these and start to look for an alternative
    Reply
  • mczak - Tuesday, November 3, 2015 - link

    Honestly I think the BX200 as a cheap SSD makes sense. The problem with the BX100 was that it was "too good". There was just about no difference in performance to the MX200 (faster in some benchmarks even), there was some feature differences (for instance no AES encryption, albeit this seems to have been an artificial firmware limitation as the controller supports it) but who cares about that in client ssds. But on the downside, there was just about no price difference to the MX200 neither, some shops selling some capacities of the BX100 actually for more than the same capacities for the MX200 sometimes. So the Crucial BX100 and MX200 were really pretty much the same in nearly all aspects.
    But with BX200 this changes - the MX200 now clearly is the better product, but the BX200 definitely is cheaper (or at least retail prices really need to be cheaper). Albeit the benchmarks are definitely concerning - even with tlc nand it shouldn't be THAT bad, maybe indeed the firmware is to blame.
    Reply
  • hojnikb - Tuesday, November 3, 2015 - link

    Actually, TLC is pretty bad and is mostly the sole reason why this drive performs so badly. Remove the SLC cache and it would be much worse (especially on paper). Reply
  • mczak - Tuesday, November 3, 2015 - link

    I don't quite agree. There's other tlc nand ssds in the benchmarks. Even if you exclude the 850 Evo (with its 3d nand) there's still the (toshiba tlc nand based) ocz trion 100. And this performs quite ok - sure it's not exactly leading but it doesn't trail by huge amounts in some benchmarks like the bx200 does. I'd have expected the bx200 to perform close to that rather than a league of its own as far as slow ssds go... I don't know though if that's due to differences between toshiba tlc and micron tlc or if that's due to controllers. Reply
  • Drumsticks - Tuesday, November 3, 2015 - link

    It can't be *just* the NAND. The other TLC based drives in the charts perform much closer to "mainstream" than "absolutely terrible at everything" Reply
  • hojnikb - Tuesday, November 3, 2015 - link

    its pretty hard to make good tlc. so yeah, tlc from micron simply sux. no other way around it.

    its no controllers fault, as it should perform just fine on its own, since its based off *46
    Reply
  • DanNeely - Tuesday, November 3, 2015 - link

    Assuming the street price does settle in at a reasonable discount to MSRP (and thus is cheaper than most other budget drives), these look like a reasonable option for office work. It look ~4 minutes of sustained full writes for the 480GB model to choke. Outside of photo/video editing the only time a normal user is going to hit that long of a sustained write is for OS or very large software installs. The rest of the time, most writes will be a few seconds at most before an idle period gives the drive time to catch up and flush the ram/SLC caches; meaning they should mostly sit in the (short) burst where the drives IO is similar to what much higher end drives achieve; and most longer operations will still fit in the middle plateau which is similar in performance to most budget drives on the market. Reply
  • beginner99 - Tuesday, November 3, 2015 - link

    Too expensive or too low performance. I think everyone can afford that $20 to step up to a MX200 or similar. And this $20 is more than worth it given the extra performance. Off topic but I would also like how older SSDs fare in these benches just as a comparison. Say I have an intel G2, will I actually gain anything noticeable from moving to a new SSD? Reply

Log in

Don't have an account? Sign up now