firedancer

firedancer copied to clipboard

fd_bmtree sucks because it's relatively slow. It does pairwise node hashing which does not leverage SIMD parallelism. fd_wbmtree sucks because it's wasteful in terms of space. It does pairwise hashing...

ripatel-fd

**Problem** The fd_quic_crypto APIs use separate buffers for the encrypted input and the decryption output. This is unnecessary because both header protection and payload encryption can trivially be done in...

ripatel-fd

Consider using memfd_secret for signer secret

1

Use memfd_secret to store the node's private key Not supported on all kernels, so needs a fallback https://lwn.net/Articles/865256/

ripatel-fd

mmcgee-jump

Most fd_quic users run a main loop like so: ``` for(;;) { fd_quic_service( ... ); fd_quic_process_packet( ... ); } ``` fd_quic_service's execution time is currently unbounded which can starve RX...

ripatel-fd

ripatel-fd

- Guard zstd dependency in fd_block_to_json - Only compile fd_stem when SSE is enabled - Only compile fddev when SSE is enabled

ripatel-fd

Currently live values are computed from the end of the prior leader slot. Change to a 1-second moving average. Also change queries for our slots to return max within slot,...

yufeng-jump

asiegel-jt

mjain-jump