Implement SHA256 in Miden Assembly #123
Conversation
itzmeanjan
commented
Feb 22, 2022
•
itzmeanjan
commented
- SHA256 functions; see here
- SHA256 message schedule preparation; see here
- SHA256 digest compute; see here
|
Current cost of executing SHA256 related functions on Miden VM
|
|
VM Cycles required to compute SHA256 2-to-1 hash: 29861 |
|
See itzmeanjan#1, where I improve SHA256 implementation |
| popw.mem # write to mem msg[24, 25, 26, 27] # | ||
| pushw.local.4 |
There was a problem hiding this comment.
This could be replaced by something like:
storew.mem
loadw.local.4
This change should save 8 cycles.
There was a problem hiding this comment.
Noted, will revisit sha256 soon.
| popw.mem # write to mem msg[32, 33, 34, 35] # | ||
| pushw.local.4 |
There was a problem hiding this comment.
This could be replaced by something like:
storew.mem
loadw.local.4
This change should save 8 cycles.
| popw.mem # write to mem msg[40, 41, 42, 43] # | ||
| pushw.local.4 |
There was a problem hiding this comment.
This could be replaced by something like:
storew.mem
loadw.local.4
This change should save 8 cycles.
| let inputs = ProgramInputs::new(&i_words, &[], Vec::new()).unwrap(); | ||
| let trace = execute(&script, &inputs).unwrap(); | ||
| let last_state = trace.last_stack_state(); |
| popw.mem # write to mem msg[36, 37, 38, 39] # | ||
| pushw.local.4 |
There was a problem hiding this comment.
This could be replaced by something like:
storew.mem
loadw.local.4
This change should save 8 cycles.
| popw.mem # write to mem msg[44, 45, 46, 47] # | ||
| pushw.local.4 |
There was a problem hiding this comment.
This could be replaced by something like:
storew.mem
loadw.local.4
This change should save 8 cycles.
…nning using newly defined macro
|
Thanks for reviewing it @bobbinth ! I'm keeping a note of optimization suggestions you've shared, in next few weeks I'll revisit both blake3 & sha256. Also I've moved both blake3 & sha256 tests to use new |
itzmeanjan
force-pushed
the
itzmeanjan/sha256
branch
from
399d1fa
to
5c93bef
Compare
itzmeanjan
force-pushed
the
itzmeanjan/sha256
branch
from
c7f42e3
to
fa33733
Compare
