You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Now the degree bit for shrinking the Stark verifier circuit in prove_root is assigned manually, the range of the degree bits is too large and lead to large memory consuming.
The text was updated successfully, but these errors were encountered:
First idea: Determine the relationship between the segment size and the original_degree_bit of each table.
Currently, the degree_bits_range for each table is:
Below are the tables of segment size and degree bit of 6 module tables
sha2-rust-program
Segment size
Arithmetic
Cpu
Poseidon
Poseidon Sponge
Logic
Memory
2^10 = 1024
16
11
11
11
6
17
2^11 = 2048
16
11
11
11
6
17
2^12 = 4096
16
11-12
11
11
7-10
17
2^13 = 8192
16
12-13
11-12
11-12
9-11
17-18
2^14 = 16384
16
14
12
12
11
18
2^15 = 32768
16
14
12
12
11
18
2^16 = 65536
16
14
12
12
11
18
2^17 = 131072
16
14
12
12
11
18
2^21 = 2097152
16
14
12
12
11
18
revme-program
Segment size
Arithmetic
Cpu
Poseidon
Poseidon Sponge
Logic
Memory
2^15 = 32768
16
15
12-14
12-14
6-14
18-19
2^16 = 65536
16
16
13-14
13-14
10-15
19-20
2^17 = 131072
16
16-17
13-14
13-14
11-15
19-21
2^18 = 262144
16
18
15
15
15
21
2^19 = 524288
18
19
15
15
16
22
2^20 = 1048576
18
19
15
15
16
22
2^21 = 2097152
18
19
15
15
16
22
It seems that with different segments of the same size, the number of each type of instruction varies. This makes it challenging to fix the degree bit for each table, particularly for the memory and logic tables. So this idea is difficult to implement.
Now the degree bit for shrinking the Stark verifier circuit in
prove_root
is assigned manually, the range of the degree bits is too large and lead to large memory consuming.The text was updated successfully, but these errors were encountered: