A1 vs A2 — Application Performance Class Explained
A1 and A2 are application performance classes. They measure random read and write IOPS — how the card behaves when a device reads and writes many small files at once, which is what happens when applications are installed on and run from the card. A2 is the higher tier and requires command queuing and cache features on both the card and the host to deliver its advantage. If your device stores photos, video or data sequentially, neither class is the specification that matters — sustained write and endurance are.
What each tier requires
| Class | Random read (min) | Random write (min) | Also requires |
|---|---|---|---|
| A1 | 1500 IOPS | 500 IOPS | Sustained sequential write of 10 MB/s |
| A2 | 4000 IOPS | 2000 IOPS | Command queuing and host-side cache support to realise it |
When it matters, and when it does not
- Matters: a device that installs and runs applications from the card — some Android-based devices, single-board computers, kiosks
- Does not matter: dashcams, IP cameras, data logging, boot storage, file transfer
- A2 only delivers its advantage if the host supports the features — on a host that does not, an A2 card performs like an A1 card
What to specify instead for most applications
For recording and embedded workloads, specify the sustained write rate, the endurance requirement and the temperature range. Those are measurable on your own device and do not depend on host features.
Which class, by what you run
The second column is the catch: A2 only delivers its IOPS advantage when the host supports command queuing and a write cache. On a host without them, an A2 card behaves like A1, so the tier you pay for is the tier you get.
| Your device | Specify |
|---|---|
| Runs apps from the card, host supports queuing | A2 |
| Runs apps from the card, host does not | A1 — A2 brings nothing without host support |
| Records video / logs data | V-class and endurance — A-class irrelevant |
| Stores files / giveaway | Capacity and grade lead; class not the constraint |
The mistake to avoid
Do not pay for A2 on a host that cannot use it. Confirm host support first, then specify the class — or skip the A-class entirely and specify sustained write and endurance, which apply on any host. We build cards to A1 and A2 and state the class on the quotation, but we will also tell you plainly when A-class is irrelevant to your product and the V-class or endurance figure is the one to write instead. Paying for a marking your device cannot exercise is the common, quiet waste in this category.
How we handle A-class on your order
When you place an order with us, we state the application-performance class on the quotation against the host you name, not as a generic label. If the host cannot use A2, we propose A1 or drop the A-class entirely and specify sustained write and endurance instead — the figures that hold on any reader. You get a number you can verify on the device, not a tier printed for show.
Not sure which applies to your product?
Send your workload — what writes, how often, at what temperature — and we will propose a configuration with the reasoning. We reply to business enquiries within 1 business day.
Request a quoteFrequently asked questions
Do I need A2 for a dashcam?
No. Dashcams write sequentially and continuously. Sustained write and endurance are the specifications that matter; the V-class is the relevant marking.
Will an A2 card be faster in my device?
Only if the host supports command queuing and the cache features A2 depends on. Otherwise you pay for capability your device cannot use.
Can you supply to A1 or A2?
Tell us the class you need and host you are using, and we will confirm the configuration to meet it.