Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

MASTER THREAD: USB drives that work with Sentry and TeslaCam

This site may earn commission on affiliate links.
I used the 128GB version of the Samsung Fit Plus for many months with no problems and then started getting errors after a Tesla update a while back. I tried reformatting it in different ways but the problem continued daily.

I switched to this combo a couple of weeks ago and not a single error so far so fingers crossed:

https://www.amazon.com/gp/product/B07NY23WBG/ref=ppx_od_dt_b_asin_title_s00?ie=UTF8&psc=1
https://www.amazon.com/gp/product/B07G5JV2B5/ref=ppx_od_dt_b_asin_title_s00?ie=UTF8&psc=1

I have it partitioned with a 32gb ext4 partition for TeslaCam and the rest as FAT32 for music and don't use a splitter or hub.

It seems the only reliable drives for this now are SSD or a high endurance microSD card with USB adapter. I chose the SanDisk model above because the Samsung version was only available through third party sellers at the time and there were several reports of counterfeit cards that didn't perform properly or failed quickly.
 
  • Like
Reactions: MrABRanch and jdw
I recommend using a micro SD with USB adaptor instead of a flash drive. Micro SD is more designed for video and photo recording while flash drives are more for file transfers. So a flash drive may have a lot higher peak writing speed while a Micro SD can offer a more stable continuous writing speed. To our cars, the continuous writing speed matters.

Keep in mind because our cars do none stop writing, no matter how high end the storage media we use, it will break a lot sooner than the ones for normal day-to-day use. For that, in general, Micro SD will last longer than a flash drive.

If you live in Canada like I do, micro SD offers another advantage. Micro SD is more weather resilient than flash drive, especially in extreme winter. Costco carries SanDisk Ultra Micro SD 128G with USB adaptor for dirt cheap. I have been using it and no problem with V10.

If you are in the States, I highly recommend Samsung high endurance micro sd card (this is not available in Canada other than online). It is specifically designed to be used in continuous writing devices and it is not much expensive than the regular ones.
 
  • Like
Reactions: WarpedOne and jdw
I recommend using a micro SD with USB adaptor instead of a flash drive. Micro SD is more designed for video and photo recording while flash drives are more for file transfers. So a flash drive may have a lot higher peak writing speed while a Micro SD can offer a more stable continuous writing speed. To our cars, the continuous writing speed matters.

The car is writing at 2 MB/s

The crappiest USB keys around do sustained write 2-3 times faster than that on the low end.... decent ones are 5-10 times faster on the low end.

Speed is not an issue on any properly working USB storage despite Teslas error message.



I
Keep in mind because our cars do none stop writing, no matter how high end the storage media we use, it will break a lot sooner than the ones for normal day-to-day use. For that, in general, Micro SD will last longer than a flash drive.

Also no.

They both use flash memory.

They both are available using several different types (QLC, TLC, etc).

A USB key and an SDCard using the same type of flash will be rated for the same # of write cycles.

The crappiest type is rated for 1000 write cycles.

The car writes 7.2GB of data per hour to the storage (with the 4th camera added)

At 9 hours a day of constant use (1 hour driving 8 on sentry at work), all 365 days a year, that'd use just under 185 write cycles on a 128GB storage device.

So 5.4 years to hit the minimum write rating of the crappiest USB storage on the market.

If you buy decent (TLC) stuff it's 3000 write cycles. So over 16 years.

Regardless of USB key vs SDcard.

The "high endurance" stuff IS indeed 'better'

It's rated for 5000 write cycles.

So instead of 5.4 years, or 16 years, it would be rated for just over 27 years.

Make it a 256GB key and even the very cheapest crap USB keys are good for almost 11 years... 32 for decent ones, and over 50 for high endurance.

I guess if you plan to keep the car for decades that'd be an issue- but most folks don't keep cars that long.


I
If you live in Canada like I do, micro SD offers another advantage. Micro SD is more weather resilient than flash drive, especially in extreme winter.


... what?

They're literally the same type of storage in a different form factor.

Samsungs SDcards and keys are both advertised as temp and shock proof for example.
 
  • Love
Reactions: nickhauenstein
The car is writing at 2 MB/s

The crappiest USB keys around do sustained write 2-3 times faster than that on the low end.... decent ones are 5-10 times faster on the low end.

Speed is not an issue on any properly working USB storage despite Teslas error message.





Also no.

They both use flash memory.

They both are available using several different types (QLC, TLC, etc).

A USB key and an SDCard using the same type of flash will be rated for the same # of write cycles.

The crappiest type is rated for 1000 write cycles.

The car writes 7.2GB of data per hour to the storage (with the 4th camera added)

At 9 hours a day of constant use (1 hour driving 8 on sentry at work), all 365 days a year, that'd use just under 185 write cycles on a 128GB storage device.

So 5.4 years to hit the minimum write rating of the crappiest USB storage on the market.

If you buy decent (TLC) stuff it's 3000 write cycles. So over 16 years.

Regardless of USB key vs SDcard.

The "high endurance" stuff IS indeed 'better'

It's rated for 5000 write cycles.

So instead of 5.4 years, or 16 years, it would be rated for just over 27 years.

Make it a 256GB key and even the very cheapest crap USB keys are good for almost 11 years... 32 for decent ones, and over 50 for high endurance.

I guess if you plan to keep the car for decades that'd be an issue- but most folks don't keep cars that long.





... what?

They're literally the same type of storage in a different form factor.

Samsungs SDcards and keys are both advertised as temp and shock proof for example.
The car is writing at 2 MB/s

The crappiest USB keys around do sustained write 2-3 times faster than that on the low end.... decent ones are 5-10 times faster on the low end.

Speed is not an issue on any properly working USB storage despite Teslas error message.





Also no.

They both use flash memory.

They both are available using several different types (QLC, TLC, etc).

A USB key and an SDCard using the same type of flash will be rated for the same # of write cycles.

The crappiest type is rated for 1000 write cycles.

The car writes 7.2GB of data per hour to the storage (with the 4th camera added)

At 9 hours a day of constant use (1 hour driving 8 on sentry at work), all 365 days a year, that'd use just under 185 write cycles on a 128GB storage device.

So 5.4 years to hit the minimum write rating of the crappiest USB storage on the market.

If you buy decent (TLC) stuff it's 3000 write cycles. So over 16 years.

Regardless of USB key vs SDcard.

The "high endurance" stuff IS indeed 'better'

It's rated for 5000 write cycles.

So instead of 5.4 years, or 16 years, it would be rated for just over 27 years.

Make it a 256GB key and even the very cheapest crap USB keys are good for almost 11 years... 32 for decent ones, and over 50 for high endurance.

I guess if you plan to keep the car for decades that'd be an issue- but most folks don't keep cars that long.





... what?

They're literally the same type of storage in a different form factor.

Samsungs SDcards and keys are both advertised as temp and shock proof for example.

Dude I have a drawer of flash drives (2.0) are too slow for our cars (freebies from job affairs). Some flash drives work before V10 update while doesn't after because of the 4th camera from V10. The speed of those flash drives can not accommodate the addition write.

In terms of speed, I have already pointed out that there is a difference between continuous writing speed and peak writing speed. Almost all manufactories (no matter USB drive or Micro SD) advertises peak writing speed. Some of them only advertise "read" speed. However, if you really use a professional speed test software to test the drives, you will see a significantly difference. Say you you a device peaks at 20M for a second, drops to 2M for a second and keeps alternating v.s. a device stables at 5M per second, which one your car prefers? I will love the first one for file transfer coz on average it is faster but not for a recording device.

In terms of the life span of the device, my friend your calculation is absolutely right but you have overlooked a very important factor: level wearing. Just for the sake of argument, in your example, the 128G memory may consist 64 small 2G chips inside (this is just an example, not necessary to be true). The device breaks when the FIRST 2G chip hits the full cycles, not ALL of them. Cheap device writes at random. An extreme example for simplicity would be it keeps writing on the same 2G (therefore, small files size but high file numbers hurts the device a lot more than a big single large file). Well high end device writes at the unit with the least amount of cycle one it. This is call level wearing. You calculation basically assume a dynamic level wearing, which is the most expensive type of controller.

You are right for the cheap or expensive or flash drive or micro SD, they are essentially the same if the storage type is the same. However, the controller is significant different.

For temp, I am talking about surviving -25C for a week.

I just want to point out a simple fact. Given how small and how cheap flash drives are, for my knowledge, still no large brand name uses it as the storage for their recording device, can't help thinking why.

Another fact I can point out is that if you read the warranty terms very carefully, some of them will specifically say “warranty voided if used in continuous recording device”. I have seen such exclusion for both sticks and micro sd from most big brands. Again what is that suggesting?

That been said, I am perfectly okay if I know exactly when that stick or sd card is going to stop working, no matter how short the life span is. But do u wanna one day when u need the footage for something big only to find it happens to be the same day the flash disk stop working, because the stick that is 10 bucks cheaper?
 
Last edited:
I'm using SanDisk SDCZ880-256G-G46 Extreme PRO 256GB USB 3.1 Solid State Flash Drive .

Since your error message says "slow" so I use faster 3.1 protocol and not just a regular flash drive but it's actually a solid-state drive in the form of a flash drive.
Unfortunately the USB port on our car is usb 2. USB 3 stick will only work at usb 2 speed. So it doesn’t worth the extra money from usb 2 to usb 3 for the car but most likely ur computer has usb 3 port and downloading files to ur computer will be super fast.
 
Dude I have a drawer of flash drives (2.0) are too slow for our cars (freebies from job affairs). Some flash drives work before V10 update while doesn't after because of the 4th camera from V10. The speed of those flash drives can not accommodate the addition write.

The additional write is 0.5 MB/sec.

See previous threads on this where there's links to a database of benchmarks on over 600 USB keys from cheap to expensive.

none tested a sustained write of less than ~5 MB/sec.


In terms of speed, I have already pointed out that there is a difference between continuous writing speed and peak writing speed. Almost all manufactories (no matter USB drive or Micro SD) advertises peak writing speed.

That's great. But I've shown you benchmarks of both. And plenty of previous threads have even more.


However, if you really use a professional speed test software to test the drives, you will see a significantly difference

What you'll see is they're all plenty fast enough to sustain 2 MB/s of sequential writes.

In terms of the life span of the device, my friend your calculation is absolutely right but you have overlooked a very important factor: level wearing. Just for the sake of argument, in your example, the 128G memory may consist 64 small 2G chips inside (this is just an example, not necessary to be true). The device breaks when the FIRST 2G chip hits the full cycles, not ALL of them. Cheap device writes at random. An extreme example for simplicity would be it keeps writing on the same 2G (therefore, small files size but high file numbers hurts the device a lot more than a big single large file). Well high end device writes at the unit with the least amount of cycle one it. This is call level wearing. You calculation basically assume a dynamic level wearing, which is the most expensive type of controller.

Not sure why you don't think most USB keys have wear leveling as well? Because they do.

So again- what I wrote remains true for either type of storage at the size/quality we're talking about... (maybe an old 2GB key you found in a drawer, not so much- but then old SDcards often lacked it as well).


You are right for the cheap or expensive or flash drive or micro SD, they are essentially the same if the storage type is the same. However, the controller is significant different.

<citation needed>


For temp, I am talking about surviving -25C for a week.

Ok...and I'm telling you Samsung lists both their pro SD cards and their comparable keys as temperature proof.

I haven't lived in Canada for a while though so I don't have much of a horse in that race- it doesn't spend a ton of time below 0 C here.



I just want to point out a simple fact. Given how small and how cheap flash drives are, for my knowledge, still no large brand name uses it as the storage for their recording device, can't help thinking why.

Because it's physically larger- and dashcam makers typically don't want something sticking OUT of the dashcam.

Pretty easy answer there.

This isn't relevant on a Tesla where it's a hidden compartment and there's no SDslot already built into the device.


Another fact I can point out is that if you read the warranty terms very carefully, some of them will specifically say “warranty voided if used in continuous recording device”. I have seen such exclusion for both sticks and micro sd from most big brands. Again what is that suggesting?


That I'm correct in the difference being less than you claim between sticks and micro sd :)

That been said, I am perfectly okay if I know exactly when that stick or sd card is going to stop working, no matter how short the life span is. But do u wanna one day when u need the footage for something big only to find it happens to be the same day the flash disk stop working, because the stick that is 10 bucks cheaper?

Nope... which is why I got a quality 128GB stick- since it'll be good for ~15 years, write-cycle wise.

I've never owned the same car for 15 years so I think I'll be ok...but just to be safe I'll put a reminder in my calendar for 12 years from now to replace it :)
 
The additional write is 0.5 MB/sec.

See previous threads on this where there's links to a database of benchmarks on over 600 USB keys from cheap to expensive.

none tested a sustained write of less than ~5 MB/sec.




That's great. But I've shown you benchmarks of both. And plenty of previous threads have even more.




What you'll see is they're all plenty fast enough to sustain 2 MB/s of sequential writes.



Not sure why you don't think most USB keys have wear leveling as well? Because they do.

So again- what I wrote remains true for either type of storage at the size/quality we're talking about... (maybe an old 2GB key you found in a drawer, not so much- but then old SDcards often lacked it as well).




<citation needed>




Ok...and I'm telling you Samsung lists both their pro SD cards and their comparable keys as temperature proof.

I haven't lived in Canada for a while though so I don't have much of a horse in that race- it doesn't spend a ton of time below 0 C here.





Because it's physically larger- and dashcam makers typically don't want something sticking OUT of the dashcam.

Pretty easy answer there.

This isn't relevant on a Tesla where it's a hidden compartment and there's no SDslot already built into the device.





That I'm correct in the difference being less than you claim between sticks and micro sd :)



Nope... which is why I got a quality 128GB stick- since it'll be good for ~15 years, write-cycle wise.

I've never owned the same car for 15 years so I think I'll be ok...but just to be safe I'll put a reminder in my calendar for 12 years from now to replace it :)

I totally agree with you that there are sticks do great job. If I have infinite amount of money, I will choose an SSD stick which has desktop longevity and performance (but not temp resilience).

All I am trying to say is that:
1. sticks are not the only option or the best option. I wanna share a likely more cost efficient approach I took that works well for me so far. (I have not mentioned a small wattage external hard drive because it won't work in low temp. In your city, that seems to be a great alternative to explore.)
2. cheap stick won't work and won't last so please do not use the numbers in theory to say any stick can work. When Sandisk can't warranty some cards with continuous writing for 2 years, I really doubt a generic brand can last for more than 6 months. The key is "quality", no matter if it is a stick or a card.
3. when our car tells us the device is too slow, sometimes it is a glitch indeed, sometimes it is telling you the truth. I highly doubt the car only need 2M/s but I can't prove or disprove it.
4. again sticks are designed for large file transfer while micro sds are designed for frequent video/photo savings. I believe at some point the difference won't matter anymore. For now, at similar price point, within the same brand name, a micro SD is likely to give you better performance than a usb stick, for the purpose of the cams only.
 
I totally agree with you that there are sticks do great job. If I have infinite amount of money, I will choose an SSD stick which has desktop longevity and performance (but not temp resilience).

All I am trying to say is that:
1. sticks are not the only option or the best option.

I never said that so not sure who you think you're arguing with.

2. cheap stick won't work and won't last so please do not use the numbers in theory to say any stick can work. When Sandisk can't warranty some cards with continuous writing for 2 years, I really doubt a generic brand can last for more than 6 months.

Sandisk won't do that because they assume you might use it in a 4k multi-camera setup.

Which will be writing many times more data than Tesla does to the device on a constant basis.

A key that'd last 5+ years at Teslas write rate would last far far less time in such a dashcam.

Hence they don't warranty it for dashcam use.

But Teslas dashcam is vastly less heavy duty than some commercially available ones.


3. when our car tells us the device is too slow, sometimes it is a glitch indeed, sometimes it is telling you the truth. I highly doubt the car only need 2M/s but I can't prove or disprove it.

You can though. With simple math.

Each clip is 30MB and 1 minute of video.

That's 0.5 MB per second.

4 cameras is 2 MB per second.

4. again sticks are designed for large file transfer while micro sds are designed for frequent video/photo savings.

<citation needed>

Flash is flash.

As already cited both are much faster than needed in professional benchmark tests for this application. The Samsung key for example the slowest it was doing sustained writes of any kind (even random ones) was over 12 MB/s... more than 6 times the speed the car actually writes at. Sustained sequential writes benchmarked at ~60 MB/s.



I believe at some point the difference won't matter anymore. For now, at similar price point, within the same brand name, a micro SD is likely to give you better performance than a usb stick, for the purpose of the cams only.


<citation needed>
 
  • Like
Reactions: Kevy Baby
I never said that so not sure who you think you're arguing with.



Sandisk won't do that because they assume you might use it in a 4k multi-camera setup.

Which will be writing many times more data than Tesla does to the device on a constant basis.

A key that'd last 5+ years at Teslas write rate would last far far less time in such a dashcam.

Hence they don't warranty it for dashcam use.

But Teslas dashcam is vastly less heavy duty than some commercially available ones.




You can though. With simple math.

Each clip is 30MB and 1 minute of video.

That's 0.5 MB per second.

4 cameras is 2 MB per second.



<citation needed>

Flash is flash.

As already cited both are much faster than needed in professional benchmark tests for this application. The Samsung key for example the slowest it was doing sustained writes of any kind (even random ones) was over 12 MB/s... more than 6 times the speed the car actually writes at. Sustained sequential writes benchmarked at ~60 MB/s.






<citation needed>
If you ever get the too slow message, your car will tell you it needs a minimum of 4M/s on the screen. So again I do not know where the 2M/s is from.

You said the "slowest Samsung"... So it is a Samsung. Again "quality" is the key. Also, like I said before, does the slowest Samsung give you level wearing? I doubt it.

If you can recommend a couple of nice stick options and list why you love them and where to get them, you are adding values to this post This is a place where we share our experience and knowledge.

You jump on my post saying any cheap stick will do a good job than the sd cards I recommended, because any sticks are faster and have long time span, I find it misleading.

Feel free to comment on my post again and this is my last reply on this debate. I think we (including me) have wasted too much space on the page.

Good luck with your stick and thank you for your knowledge.
 
If you ever get the too slow message, your car will tell you it needs a minimum of 4M/s on the screen. So again I do not know where the 2M/s is from.

The fact that's the amount they're actually writing to the drive

You can tell by just looking at the files on the drive.

It records one 30MB clip per minute per camera.

30MB times 4 is 120MB. Divided by 60 seconds is 2 MB per second.


Also, like I said before, does the slowest Samsung give you level wearing? I doubt it.

Then you continue your streak of being wrong about stuff :)


If you can recommend a couple of nice stick options and list why you love them and where to get them, you are adding values to this post This is a place where we share our experience and knowledge.

I have used several different ones. Primarily a 128GB Samsung Fit Plus- but also a few much smaller/cheaper ones... currently my backup stick (what I swap in if I'm taking the Samsung out to review footage for a bit) is a 32GB Kingston DataTraveler G4 that's years old.

Both work flawlessly, and have for over a year now. I wouldn't trust that 32GB key long-term since it'd burn through write cycles 4x faster than a 128GB key, but given I only use it 1-2 days every few months it's not an issue and should last for many years to come as the temp backup key.


Because again, the actual speed requirements for this application are quite low and nearly any stick can handle them. See again the database of over 600 different USB sticks that all manage a minimum of ~5 MB/s on the very lowest end for sustained write speeds.



You jump on my post saying any cheap stick will do a good job than the sd cards I recommended, because any sticks are faster and have long time span, I find it misleading.

I'm...not really sure that sentence parses very well or matches what I actually wrote.

You were the one who repeatedly claimed SDcards were "inherently" better than sticks, and then refusing to support the claim in any way when asked repeatedly to do so.
 
I used the 128GB version of the Samsung Fit Plus for many months with no problems and then started getting errors after a Tesla update a while back. I tried reformatting it in different ways but the problem continued daily.

I switched to this combo a couple of weeks ago and not a single error so far so fingers crossed:

https://www.amazon.com/gp/product/B07NY23WBG/ref=ppx_od_dt_b_asin_title_s00?ie=UTF8&psc=1
https://www.amazon.com/gp/product/B07G5JV2B5/ref=ppx_od_dt_b_asin_title_s00?ie=UTF8&psc=1

I have it partitioned with a 32gb ext4 partition for TeslaCam and the rest as FAT32 for music and don't use a splitter or hub.

It seems the only reliable drives for this now are SSD or a high endurance microSD card with USB adapter. I chose the SanDisk model above because the Samsung version was only available through third party sellers at the time and there were several reports of counterfeit cards that didn't perform properly or failed quickly.

Same here, although I opted for the Samsung 128GB which is a more expensive but has a longer warranty.
 
  • Like
Reactions: superblast
A good read on MicroSD cards for use in dashcam applications. While Tesla's dash cam is not high definition, it is four cameras instead of two and it is still simultaneously writing, deleting and copying files plus the card is in a tougher heat/cold environment. While TeslaCam may have a lower data rate than a real dashcam the file operations are the same - constant read/write/copy/delete. High endurance MicroSD cards built for dashcam use don't cost much more but they do work better for the purpose - constantly overwriting files.

https://www.amazon.com/gp/product/B07NY23WBG/ref=ppx_od_dt_b_asin_title_s00?ie=UTF8&psc=1
https://www.amazon.com/gp/product/B07G5JV2B5/ref=ppx_od_dt_b_asin_title_s00?ie=UTF8&psc=1
 
Last edited:
A good read on MicroSD cards for use in dashcam applications. While Tesla's dash cam is not high definition, it is four cameras instead of two

Which is still significantly less data than 2 4k cameras. Or even ONE 4k camera.

A SINGLE 4k blackvue camera writes 3.125 MB/s. Two would be writing 6.25MB/s

The tesla system with all 4 720p cameras is writing 2 MB/s total.


and it is still simultaneously writing, deleting and copying files

It doesn't copy files at all. When sentry activates it moves them, which isn't the same thing and is virtually instant since all it's doing is updating the index instead of actually changing the files location or making another copy of the files.


plus the card is in a tougher heat/cold environment. While TeslaCam may have a lower data rate than a real dashcam the file operations are the same - constant read/write/copy/delete. High endurance MicroSD cards built for dashcam use don't cost much more but they do work better for the purpose - constantly overwriting files.

Again- the math on this is pretty easy.

They are "better" in the sense a 256GB one might last 25-30 years instead of 15-20 years. If that's worth a few extra bucks to you, go for it.
 
This has been repeatedly debunked, with actual math, over and over again by now.

Possibly you should read this and other threads on the topic before posting.

The short math though is even the cheapest/crappiest flash keys are rated for 1000 write cycles- and the decent cheap stuff is 3000 cycles.

On a 128GB key, for the amount Tesla writes, if you used sentry/dashcam 9 hours a day (8 hours parked at work, 1 hour driving) you'd get between 5-15 years before you hit the minimum write cycle limit.

Double that for a 256GB key.

That's without getting into debunking the "moving parts" and "shock resistance" nonsense someone else already covered.
I have no idea where you're getting your fuzzy math from. Doubling the size of the drive doesn't double the number of write operations! First of all, USB flash drives are good for between 10K and 100K write/delete operations, depending on the technology used. Are you under the impression that the dash cam is one continuous write for the duration of the drive, so it only counts as 1 write operation? Dashcam and sentry videos are 55-58 seconds a piece with 4 inputs (front, left, right, back). Each one is written to a separate file. So that is 4 distinct write operations per minute or 240 writes per hour. Every time sentry detects movement and records, it does so for ~1 minute, from 4 cameras, each into a separate file. Again that's 4 write operations per sentry event. So let's take the maximum of 100K writes/deletes, since you like math so much. 100,000 (max ops)/240 writes per hour of video, that's 416.67 hours of video before hitting the max write cycles. If you only drove 1 hour per day (we're not even including the writes for sentry mode, or when the Tesla OS deletes older videos when the device gets full), that would net you 1.14 years before hitting 100K writes (not decades like you're claiming). Very few USB sticks survive to 100K so it's much closer to the 10K number (which would net only 41 hours of video before they would begin to break down). Why do you think so many people are reporting that their USB drives are only lasting a few months??? But what would I know? I've only been a software engineer for the past 20 years, dealing with microcontrollers, pics, a plc's, and their internal flash memory!!!
 
  • Informative
  • Like
Reactions: brianman and jdw
A lot of talk about the Too Slow error. Are any of you encountering this?
...
My sandisk 128gb has been working flawlessly for months. But lately I encountered two of these errors. I reformatted the drive to fat32 and it works for a week or so and then the error returns. Seems to be happening post v10 update.
In case it helps diagnose...

I got the same message with a 2TB Samsung T5 formatted FAT32 to its full capacity.

Workaround for me (on Windows 10):
1. Remove the partition
2. Creating new partition using Computer Management of size 1048576 MB
3. Reformat partition as FAT32 using guiformat.exe (v 1.0.0.1 from Ridgecorp)

TLDR: I think this message is a general purpose message about the partition size or state.
 
I have no idea where you're getting your fuzzy math from

Nothing fuzzy involved, it's pretty simple math.

I
. Doubling the size of the drive doesn't double the number of write operations!

Of course it does.

Flash is rated for a total number of write cycles.

That effectively becomes the number of times the entire drive can be written to since wear leveling insures writes are done to all blocks to wear the cells evenly

So if a 32GB drive is rated for 3000 cycles (common for the most common decent cheap flash) that's 3000 times you can write 32GB to the drive.

So the total amount of rated writes is 96,000 GB.

Make it a 64GB drive and it's now exactly double the amount of data for the same 3000 cycles.




First of all, USB flash drives are good for between 10K and 100K write/delete operations, depending on the technology used.

This is true if you're posting from like 2005 and paying thousands of bucks for large amounts of enterprise grade flash.

Today? Not so much.


QLC, the most common cheapest type of flash, for example is usually only rated for 1000 cycles.

TLC, the most common used on decent/mainstream USB flash drives, is only 3000-5000 cycles.

Even 10k is crazy higher than nearly anything on the consumer market except the most expensive SSDs using MLC.

100k cycles is crazy expensive enterprise class stuff nobody is using in their Tesla with SLC flash.



Are you under the impression that the dash cam is one continuous write for the duration of the drive, so it only counts as 1 write operation?

Are you under the impression writing ONE block on the drive uses an entire drive write cycle?

Because it doesn't.

Again, each cell is rated for that number of writes- and wear leveling insures writes are spread evenly.

So the entire capacity of the drive is one write cycle if you write to each cell once, even though that involves many many write operations


Dashcam and sentry videos are 55-58 seconds a piece with 4 inputs (front, left, right, back). Each one is written to a separate file. So that is 4 distinct write operations per minute or 240 writes per hour.

Write operations- yes.

Drive write cycles? No. Because that's only using ~7.2 GB of the blocks on say a 128GB drive.

You'd need to write for a little under 18 hours straight to use one entire drive write cycle assuming even wear leveling of the cells.


Every time sentry detects movement and records, it does so for ~1 minute, from 4 cameras, each into a separate file.


Nope.

Sentry doesn't "record" anything.

It's always the dashcam. Which simply keeps running when Sentry mode is engaged.

The only thing Sentry does is, when it goes to alert mode, it moves (NOT copies- those are very different from a drive wear perspective) the last 10 minutes of dashcam from recent to saved.


If Sentry didn't start recording until it went to alert it'd be impossible for it to have the previous 10 minutes of footage after all.

Unless you think the car can time travel....


Again that's 4 write operations per sentry event. So let's take the maximum of 100K writes/deletes

That would be dumb since nobody is running expensive enterprise-class SSDs in their car.

1000 on the low end, 3-5k on the high end, for mainstream consumer flash in the current market.

I can deluge you with sources from flash memory makers if you'd actually be willing to admit your error- but you don't seem the type.


Thus the rest of your math is totally nonsensical (even more because you think writing a single 1 minute file uses up an entire write cycle for the whole drive)


've only been a software engineer for the past 20 years, dealing with microcontrollers, pics, a plc's, and their internal flash memory!!!


That's a frightening and very hard to believe claim unless you retired 15 years ago or something.

It'd explain why your info is so grossly wrong and out of date.
 
  • Like
Reactions: Kevy Baby
I have no idea where you're getting your fuzzy math from. Doubling the size of the drive doesn't double the number of write operations! First of all, USB flash drives are good for between 10K and 100K write/delete operations, depending on the technology used. Are you under the impression that the dash cam is one continuous write for the duration of the drive, so it only counts as 1 write operation? Dashcam and sentry videos are 55-58 seconds a piece with 4 inputs (front, left, right, back). Each one is written to a separate file. So that is 4 distinct write operations per minute or 240 writes per hour. Every time sentry detects movement and records, it does so for ~1 minute, from 4 cameras, each into a separate file. Again that's 4 write operations per sentry event. So let's take the maximum of 100K writes/deletes, since you like math so much. 100,000 (max ops)/240 writes per hour of video, that's 416.67 hours of video before hitting the max write cycles. If you only drove 1 hour per day (we're not even including the writes for sentry mode, or when the Tesla OS deletes older videos when the device gets full), that would net you 1.14 years before hitting 100K writes (not decades like you're claiming). Very few USB sticks survive to 100K so it's much closer to the 10K number (which would net only 41 hours of video before they would begin to break down). Why do you think so many people are reporting that their USB drives are only lasting a few months??? But what would I know? I've only been a software engineer for the past 20 years, dealing with microcontrollers, pics, a plc's, and their internal flash memory!!!
The vast majority of consumer grade flash drives don't include wear level tracking and algorithms because they aren't needed for the intended purpose of the device (infrequent writes and frequent reads) Commercial/industrial grade flash does. SSD drives do as well and that's why I was recommending them to people that were having repeated issues with their USB flash drives after a few months.