NTT docs #38
NTT docs #38
Conversation
|
|
||
| The twiddle factors are precomputed to save runtime and improve performance. | ||
|
|
||
| 4. **Bit-Reversal Permutation:** |
There was a problem hiding this comment.
to be more accurate splitting to even and odd is called decimation-in-time (DIT) and required input-reorder.
Output reorder is relevant for decimation-in-frequency where the coefficients are split first-half (lower degree terms) and second-half (higher degree terms). Output reorder is relevant to DIF.
docs/icicle/primitives/ntt.md
Outdated
|
|
||
| ### Mixed Radix | ||
|
|
||
| The Mixed Radix NTT algorithm extends the concepts of the Radix-2 algorithm by allowing the decomposition of the input sequence based on various factors of its length, not limited to powers of two. This approach offers enhanced flexibility and efficiency, especially for input sizes that are composite numbers, by leveraging the "divide and conquer" strategy across multiple radixes. |
There was a problem hiding this comment.
actually we are still limited to powers of two. Mixed-radix basically is composed of NTT blocks of sizes 16,32,64 compared to radix2 which is always splitting the problem such that we end with NTT of size 2. The NTT blocks in mixed-radix are implemented more efficiently based on winograd NTT but also optimized memory and register usage is better compared to the older radix2 alg.
There was a problem hiding this comment.
"The NTT blocks in Mixed Radix are implemented more efficiently based on winograd NTT but also optimized memory and register usage is better compared to Radix-2." i added this
No description provided.