TeslaCam
writes about 2GB per hour, (<--great article) creating at minimum of 180 files from the three cameras in an hour. If it writes 3 files every minute, that is .5965 MB per seond or 36MB of data per minute. So somewhere between 36MB and your 90MB is probably reality, depending upon motion/compression and lighting.
Might want to double check your math there....
Your .5965 and 36 numbers would be per file, and there's 3 files (one from each camera)... (and they are based on a 2GB start, which isn't correct either).
We know exactly how much it writes. You can tell by just looking at the files.
It writes 3 30MB files per minute of video. That's 90MB/minute. Or 1.5MB/second. Or 1.8GB an hour if you prefer.
I'm glad to see they finally updated that article though until a few weeks ago it was badly out of date with some serious misinformation in it.
They're still kinda misinformed on their write speed claims though.... they're suggesting a "minimum" spec on your device of roughly 13 times faster than the car itself is writing data and claiming if you use less you get the issues that...even folks with MUCH FASTER devices up to and included SSDs have seen.
Because the problem isn't write speed it's bad software by Tesla.
There are
user benchmarks of common USB cards and drives. The best "peak"result for 128GB drives is the SanDisk Extreme Pro USB 3.1 128GB at 263MB/s - that is a best case sequential write of a single large file. That same drive has a peak write speed of only 12.2MB/s on smaller files - TeslaCam file size is somewhere in between these numbers. This on USB 3.1 and a computer - USB 2 in a Tesla will be slower.
So a few problems here....
One- you are misreading the results.
The 12.2 isn't "on small files"... the 4k write is
random writes... that is, non-sequential... where it has to jump all over the drive to write.... such as what an OS might do where it's writing something to a temp folder, than writing something else to a data folder for some app that's running, then writing something else to a cache for a browser, then writing something else because you saved a document...all in different random places from different random sources. It's not the size that's the problem it's the random part.
That's the exact opposite of what the cameras do.
If you highlight the column and click the question mark it'll explain what the measure is about
The other write number (the much higher one) is peak sequential writes which is the far more relevant benchmark for recording video files one after the other.
Second-
Why do you think it would be slower than the 12.2 MB/s on USB2?
12.2MB/sec is significantly less than a USB2 bus can handle.... (which is anywhere from ~45 to 60 MB/sec depending how much overhead is going on)
And again remember, the
actual cameras are only writing at 1.5 MB/sec
To me, it seems within the realm of possibility that a good many USB drives are not up to being used with TeslaCam
That's because you misunderstood what the 4k benchmark was testing.
The temperature rating on all the stick style flash drives is between 32 and 95-140 degrees.
Assuming you run cabin overheat to protect all the OTHER electronics in the car the cabin isn't getting above 105 anyway, leaving quite a bit of leeway there.
The Samsung Pro Endurance MicroSD card is built for video recording and dashcams and sells for $35 and has a temperature rating from -13 to 185 degrees.
To be fair- you'll also need to buy a MicroSD to USB adapter to make that work. They're not expensive but they're not free either. That said, obviously that card will work great. It's certainly way overkill on the write-cycle front though.... even "regular" non-endurance flash will be good ~5-10 years at 128GB given how little data the cameras write. (the math has been shown in a number of threads, maybe even this one, by now)