Hiksemi is a brand that I discovered while browsing through microSD cards on AliExpress. Upon doing further research, I discovered that they’re the storage technology division of Hikvision. (Ok, cool — we have a major manufacturer on our hands.) I discovered them by sorting my search results by price from lowest to highest; the 8GB version came up as one of my results (as it was being sold for $0.01 — before shipping — at that point in time) — and after being impressed by how well that one performed, I decided to go back and try the 32GB and 128GB versions as well.
As a whole, performance was slightly above average in sequential read and random read speeds, and slightly below average in sequential write and random write speeds. Only one sample had sequential write speeds that were above average; the other two were below average. Interestingly, out of the three Hiksemi NEO models I purchased (along with the 8GB and the 128GB versions), this model performed the worst out of the three. These cards carry the V10 mark, and performance was good enough to qualify for that mark.
On the endurance testing front:
Sample #1 was initially plugged into a SmartQ Single reader. I’ve had problems with these readers randomly failing every few days — and this one was no exception — so I eventually moved it over to one of the JJC CR-UTC4AC’s. Its first legitimate1 error was a four-sector wide address decoding error that occurred during round 7,941. It continued to chug along for another couple thousand rounds until round 9,450, when large swaths of sectors just started returning all
ff‘s — and then going back to normal during the next round. This continued until round 10,894, when the problem had affected over 50% of the total sectors on the device. Here’s the graph of this card’s progression through the endurance test:- Sample #2’s first error was a 64-sector wide data verification error during round 1,940. It continued — largely without issues — until round 5,166, when a large number of sectors (about 1.1% in total) read back as all
0xff‘s. It continued to do well — with the failed sectors verifying correctly on subsequent rounds — until round 7,001, when the card stopped responding to commands entirely. Up to that point, only about 1.1% of the card’s sectors had been flagged as “bad”. Sample #3’s first error was a four-sector wide address decoding error during round 1,324. It kept going, unremarkably, for quite a while afterwards — until round 8,686, when a large swath of sectors (about 7/8 of the total sectors on the card) read back as all
0xff‘s. Here’s the graph of this card’s progression:
Overall, I’d say this is about an average card in most respects — performance was average, and endurance was average. On performance, it did about as well as the SanDisk Ultra 32GB — that is, except for random read speeds, where the SanDisk Ultra performed about twice as well. On the other hand, the SanDisk Ultra did about half as well in terms of endurance. So if you’re trying to choose between the two, it might just come down to a question of “what’s more important to you — endurance or random read speeds?” However, there are other cards that outperformed both of them in both performance and endurance — for example, the Kingston Canvas Select Plus — so if you don’t mind spending a couple extra dollars, I’d go for one of those instead.
1This card did technically suffer some errors earlier in the test. However, I have code to handle device disconnects/reconnects — and it took me a while to get this code right, so I chalked up these errors to issues with my code.
June 12, 2025