I don't know. There seem to a few too many compromises made about this one. It may be dirt cheap, but the performance looks pretty bad for NVMe SSD. And the cooling, that's bad.

Actually quite impressive results considering how the drive is constructed. The cooling problem makes sense since with the smaller size, all the heat is also concentrated. If I ever get any M.2 drive, I'll buy a cooling element at the same time, no doubt.

good piece of kit for the price

Its time to ditch the old CrystalDiskMark results or retest older drives on the new build. It would also be a good idea to maintain a RS3 (1709 without updates) system with older BIOS to compare pre and post patch 4KQ1T1 results as these patches seem to hurt these numbers by a tangible percentage.

The crystaldiskbench latest version is used, indeed some older drives below the fold still show a screenshot that ran with the old version. Quick note: some of you are under the impression that everything shipped here for review stays here, that is not correct. Most SSDs are returned to the manufacturer or press-agency for media rotation. Ergo, we can't retest all and any old SSDs. Last but not least, each SSD is tested on up-to-date windows build. That means all patches are in place. Currently, this creates the biggest offset in heavily queued 4K perf compared to the older drives, but it's not something I can test nor update for the old drives. Worst case scenario there specifically is (sometimes) 100 MB/s difference back and forth on say what normally (Hypothetically) is a ~600 MB/sec 32Q4K value, and now does ~500 MB/sec. The Spectre/Meltdown patches perf hit is, however, feel mostly isolated towards that 4K perf with extremely large queues. That's such a massive workload that I doubt any of you could ever notice a difference, but yes it can be measured and visualized though. I could leave all older drives out of the result sets, of course, but that doesn't have my preference either. Not ideal, I agree. You may thank Intel for that though.

Fox2232:

If anything. Hilbert should test all drives in system which has intel patches. Because most end users will have them in place too. Be it just from MS or via BIOS too. There where it is complicated to get BIOS microcode update (like Ivy), you are unlikely to have M.2 slot too.

I think you misunderstood me. I was talking about maintaining a 2nd system for comparison. It is very easy to slap a new M.2 drive into the spare slot and bench it. I brought up the 2nd system for the purpose of demonstrating the change in performance pre and post patch. Some SSDs are hit a little, some are crippled by upwards of 25% and the consumer has the right to know.

Fox2232:

I got that. But for what audience. Most of systems with M.2 are patched anyway. Only thing that this achieves is a lot of angry people. Because it will show difference between what they paid for and what they get now from intel's system. I think there is a lot of negativity already over all those patches in tech crowd. Every time I bring up yet another new hole in intel's CPUs and its effects for discussion in work, it is: "$#*^, &\/(@, ..." Thing is, that you should make purchase decision based on post patch performance. Because you should always expect worst case scenario. And that's being in situation requiring BIOS update. (New GPU incompatible. Another vulnerability so bad, that you no longer dare run unpatched system...) Take this M2. SSD. It is 2~3 times as fast as SATA SSD, but it overheats and slows down. Hilbert revealed it. But it is relatively cheap. And good to 2nd M.2 slot in Ryzen systems, as that is slower on almost all MBs. But users should consider that worst case scenario and accept risk or look elsewhere.

I would agree with you if I trusted the hardware industry. If there are 2 scores (pre and post patch) and the spec sheet matches the pre patch performance the consumer has the right to see this deliberate attempt to misrepresent a product.

Great review but.....prices are so low atm that the 76c full load is deal breaker, even with a heatsink added this drive would still achieve high temps after 10 mins or so of read/write cycles, I`d rather pay a little extra for a secondhand Samsung evo and not have the heat to worry about whilst getting higher read/write speeds, to me this is about as useful as a 16gig Intel optane drive.....too little too late.

@thegoose Please show me consumers that will have more than a few min of "full" load on a drive... And most nicer boards come with a heatsink (which the fast drive already has), or i buy one for 10$. Ppl spend 1-5000 on a tv and it doesn't even come with a "20$" hdmi cord... At least i haven't seen ppl worrying about spending a few dollars more on a rig that already isnt "cheap" (2nd m2 slot means it isnt a value board; and two nvmes, where the 1st drive most likely a fast/more expensive one). This is the perfect drive for the 2nd (slower) slot to install games/software or as shadow drive when it comes to video editing, without spending a lot more to have double (or better) the performance of a ssd.