Vida CEM swapping

Post by **RickHaleParker** » 08 May 2021, 14:56

solitaire wrote: 08 May 2021, 12:22 but 3 bytes always in candidate 1-5

Hum ...
If that holds true for all six bytes.
6 bytes 5 candidates per byte. 5^6 = 15,625 possible pins.

15,625 is a short list compared to 100^6.

VTL: How long would it take to test 15,625 possible pins generated from six sets of five candidates. ?

Post by **vtl** » 08 May 2021, 19:55

RickHaleParker wrote: 08 May 2021, 14:56 VTL: How long would it take to test 15,625 possible pins generated from six sets of five candidates. ?

> 1000 pins in 640 ms, 1562 pins/s,

10 seconds.

Post by **RickHaleParker** » 08 May 2021, 20:07

vtl wrote: 08 May 2021, 19:55 > 1000 pins in 640 ms, 1562 pins/s,
10 seconds.

If the correct byte value always ends up a short list. Reduction by shortlisting might be the best compromise between speed and dependability.

Post by **vtl** » 08 May 2021, 20:14

RickHaleParker wrote: 08 May 2021, 20:07 If the correct byte value always ends up a short list.

But it does not. For the cmp+jmp instruction clusters on the same flash page/cache line/pipeline fetch only the first one yields a larger latency, the following has no latency spike, so it would never make it to the short list.

Post by **RickHaleParker** » 08 May 2021, 21:17

vtl wrote: 08 May 2021, 20:14
But it does not. For the cmp+jmp instruction clusters on the same flash page/cache line/pipeline fetch only the first one yields a larger latency, the following has no latency spike, so it would never make it to the short list.

Then how do we ever get bytes B1 & B2?

You are already doing the first three byte shortlist ... you call the ones on the byte shortlist candidates. Three more byte shortlist and you have everything you need to build a Pincode shortlist. Is there some reason a byte shortlist for B3 - B5 cannot be done?

jurresmikkelbeer · Post by **jurresmikkelbeer** » 08 May 2021, 21:30

T5Luke wrote: 24 Apr 2021, 08:08 Anybody has Dice, CODE and likes to try?
Only tested 1 time on bench and it worked. CEM allows only 100 write cycles so i dont like to do endless tests.

There are some small things i need to fix, if tests work 3 times i will upload here for all...

Hey,

I would like to try, can you send me a PM?

WhizzMan · Post by **WhizzMan** » 10 May 2021, 00:10

If voltage is so significant, would it be a good idea to read out the battery voltage before starting so you can at least print a warning that results may not be accurate below X volt? I imagine this will save a ton of questions that people will be posting later on.

Post by **RickHaleParker** » 10 May 2021, 08:09

Could use some more coders to fork the code, try out some of the other ideas and develop new features.
This is CEM swap not swamp VTL .

Post by **vtl** » 10 May 2021, 08:16

Yeah, I'm sorry, my free time has evaporated recently. Everyone is welcome to tweak the code and submit improvements, it is opensource.

Post by **RickHaleParker** » 10 May 2021, 08:24

vtl wrote: 10 May 2021, 08:16 Yeah, I'm sorry, my free time has evaporated recently. Everyone is welcome to tweak the code and submit improvements, it is opensource.

I figured that. I have too many pokers in the fire myself.

If I knew Arduino C better I would chip away at some of the other ideas but when it come to Arduino C or any version of C I am just a grasshopper. I contribute what I can. There got to be some C coders that can do way way more then I can.