Project

General

Profile

Meeting

Date: June 13, 2014 at 9:15 AM US CST (Fermilab local time)
Place: 1-866-740-1260 (ReadyTalk line) (Host : Soon)

Participants:

Krzysztof Genser
Soon Yung Jun
Boyana Norris

Discussion

(Summary by Krzysztof)

We discussed more of Boyana's results related to the introduction of a
helper class G4xyd in G4PhysicsVector comparing it to v10.0.

The new results are located at:
https://docs.google.com/spreadsheets/d/1jSiwafZZd0gIHX5kR2QCyjymD6ZMK3qGjrKzGxps0Dk/pubhtml

One should mainly concentrate on the tab titled Comparison (Inc.)
Which is inclusive, sampling based, (and includes the initialization
phase). One should treat the tab Comparison (Excl.) with some reserve
as the numbers there are less accurate.

The overall result is about 0.6% slowdown in the G4xyd case (although
the uncertainty of this number is large and the result is probably
closer to 0%).

G4PhysicsVector::Value() did become faster and had less L2 cache
misses, but other functions became slower.