Quickest way to get highest set bit on port in to a variable?

[Archangel]

Henric, never the less, not as quick as 8 rotates and 8 jump on carry in-line. Worst case, with no bits set in the target thats only 16 cycles.

rscor01, I didn't realize LOOKUP would do that. How would that go?

George

lookdown2 index, >= [%00000000,%00000001,%00000010],temp

[rsocor01]

lookdown2 index, >= [%00000000,%00000001,%00000010],temp

Nice! Very clever indeed . I have never used this command before, but now I can see it's potential.

[Tabsoft]

I may be wrong but I don't think this will work.
LOOKDOWN2 Search,{Test}[Value{,Value...}],Var

Lookdown2 compares the value of "Search" with the values in the list from left to right (index values 0 up to 255) with the "Test" comparison.

So if "Search" = 5 and "Test" = "=>" then using your values for the Value in the list, the logic tests should look like this.
(Values: 0, 1, 2, 4, 8, 16, 32, 64, 128) bits 0-7
Test 1: is 5 => 0? : Yes, (0 will be stored in Var)
The testing would stop right here for every value of "Search" (0-255) for bytes (0-65535) for words because the "Search" value will always be greater than or equal 0 (unless you're using Long variables)

Maybe I'm wrong?

As I see it, to use Lookdown2 you would still need to enter all the values from 0 to 255 inclusive into the list.
And "Test" would need to be set to "="
Lookdown2 supports up to 85 values in the list or up to 256 when using a PIC18.

[rsocor01]

I may be wrong but I don't think this will work.
LOOKDOWN2 Search,{Test}[Value{,Value...}],Var

Lookdown2 compares the value of Search with the values in the list from left to right (index values 0 up to 255) with the "Test" comparison.

So if Search = 5 and Test = "=>" then using your values for the Value in the list, the logic tests should look like this.
(Values: 0, 1, 2, 4, 8, 16, 32, 64, 128)
Test 1: is 5 => 0? : Yes, (0 will be stored in Var)
The testing would stop right here for every value of Search (0-255) for bytes (0-65535) for words.

Maybe I'm wrong?

As I see it, to use Lookdown2 you would still need to enter all the values from 0 to 255 inclusive into the list.
And
Lookdown2 supports up to 85 values in the list or up to 256 when using a PIC18.

No, I don't think that you are wrong. I think there is a minor change that needs to be done to Dave's code,

lookdown2 index, <= [%00000000,%00000001,%00000010,%00000100,%00001000,%00010000,%00100000,%01000000,%10000000,255],temp

[Tabsoft]

Thanks for the feedback.

I still think there are issues with this approach.
I ran a test and printed the output.
I ran a For/Next loop from 0 to 255.
e.g.
For i = 0 to 255
Lookdown2 i, <= [%00000000,%00000001,%00000010,%00000100,%00001000, %00010000,%00100000,%01000000,%10000000,%11111111], j
Next i

The output resulted in 247 wrong values out of 256.

I am attaching the output here.

Again, perhaps I'm wrong and missing something?
Lookdown2.pdf

[rsocor01]

Thanks for the feedback.

I still think there are issues with this approach.
I ran a test and printed the output.
I ran a For/Next loop from 0 to 255.
e.g.
For i = 0 to 255
Lookdown2 i, <= [%00000000,%00000001,%00000010,%00000100,%00001000, %00010000,%00100000,%01000000,%10000000,%11111111], j
Next i

The output resulted in 247 wrong values out of 256.

I am attaching the output here.

Again, perhaps I'm wrong and missing something?
Attachment 7724

Yes, you are right. We should always use truth tables just to make sure it is correct. We were just giving Matias a general idea. Try this now,

For i = 0 to 255
Lookdown2 i, <= [%00000000,%00000001,%00000011,%00000111,%00001111, %00011111,%00111111,%01111111,%11111111], j
Next i

[Tabsoft]

Thanks. That works great.

To find the port pin state using this, you would just need to decrement the return value by 1 if the value is >0 ( same for NCD).

When I get some time I will run this against NCD and check the instruction cycles to see which is quicker.

Thanks again.