(In reply to Jason Newton from comment #14) > to say "well we shouldn't make it go any faster > because it's the wrong tool" is ignoring just how common place they are used > for these tasks. Let me reiterate to make clear what my position is: "The patches are not going into the stock package without a review." Please do read comment 12. If you can put them the patches into a separate package (say in a HPC repository) in OBS on your own, it will be welcomed. > The main problems with high throughput tend to be either > encryption cpu overhead or buffer sizes... there doesn't conceptually seem > to be much of a problem with that tweaking those or adding specialized > encryption modules. Things always get complicated when you start fiddling with cryptography code. It has to be clear, that current functionality isn't crippled by introduced changes.