Optimize GCM

By allowing multiple blocks for clmul, slight speedup there though still
far behind optimum.

Precompute a table of multiples of H, 3-4x faster on systems without clmul
(and still no secret indexes).

Refactor GMAC to not derive from GHASH

1 files changed, 1 insertions, 0 deletions

diff --git a/src/cli/speed.cpp b/src/cli/speed.cpp
index 83bb29efa..01a84d7c9 100644
--- a/src/cli/speed.cpp
+++ b/src/cli/speed.cpp
@@ -1024,6 +1024,7 @@ class Speed final : public Command
 
             const Botan::SymmetricKey key(rng(), mac.maximum_keylength());
             mac.set_key(key);
+            mac.start(nullptr, 0);
 
             Timer timer(mac.name(), provider, "mac", buffer.size(), buf_size);
             timer.run_until_elapsed(runtime, [&]() { mac.update(buffer); });