g10/plaintext.c:do_hash() and g10/sign.c:sign_file() use iobuf_get for process data byte-by-byte. This causes significant processing overhead.
Use of iobuf_read here increases signing speed by 2.0x and verify speed by 4.9x (Ryzen 5800X, SHA256, 4.3GiB file).
Patches used for testing:
| Status | Assigned | Task | ||
|---|---|---|---|---|
| Resolved | jukivili | T5828 Improvements for gnupg data operation performance (enc/dec/sign/verify/enarmor/dearmor/etc) | ||
| Resolved | • werner | T5826 Improve detached signing and verification speed |
Patches look good for me.
Please go ahead.
Minor nitpicking: I'd prefer use of 32768 for both (now, [PATCH 1/2] uses 32768, while [PATCH 2/2] uses 1<<30)
I used "1<<30" by example of existing code in g10/free-packet.c, which is another place where iobuf_read is reading to NULL.
@werner Could these two patches could be backported to 2.2? These changes give same level of performance increase in 2.2 as seen in 2.3.