 Ian Goldberg
|
5f23b6d5d5
Move method definitions for MPCIO and friends from mpcio.hpp to mpcio.cpp
|
2 years ago |
|
Ian Goldberg
|
e48ee72b6f
Have MPCPeerIO's dump_stats also show the stats of number of precomputed values used
|
2 years ago |
|
Ian Goldberg
|
099228a946
Gather and dump stats on number of messages, number of bytes, wall clock and CPU times, Lamport time
|
2 years ago |
|
Ian Goldberg
|
3478cbf398
Fix the code to pass the test
|
2 years ago |
|
Ian Goldberg
|
198ce0e216
Update the Lamport clock on received messages
|
2 years ago |
|
Ian Goldberg
|
08ce010f6d
Optionally add headers to all messages containing length and a Lamport clock
|
2 years ago |
|
Ian Goldberg
|
6aadfb5a57
Put recording of IO traces (the sizes of every message sent and received) behind an ifdef
|
2 years ago |
|
Ian Goldberg
|
77f977367e
Remove unused function for recv into a vector of buffers
|
2 years ago |
|
Ian Goldberg
|
ee02e44109
Expose a routine to reset the counts of precomputed values used
|
2 years ago |
|
Ian Goldberg
|
37e906047e
Computational peers can report the number of precomputed values they used in the online phase
|
2 years ago |
|
Ian Goldberg
|
7af573e9da
More merging of computational peer and server execution paths
|
2 years ago |
|
Ian Goldberg
|
bba5036964
Merge the computation peer and server online execution paths
|
2 years ago |
|
Ian Goldberg
|
36c4f65049
Make MPCIO variants for computation peers and the server into subclasses
|
2 years ago |
|
Ian Goldberg
|
49d2ba07c0
Remove some debug output
|
2 years ago |
|
Ian Goldberg
|
33a6a54fe4
The server now also opens one socket per thread to each computational peer
|
2 years ago |
|
Ian Goldberg
|
995b8c3fdf
Improve the API a little bit
|
2 years ago |
|
Ian Goldberg
|
8077f953e7
Start filling in some MPC operations
|
2 years ago |
|
Ian Goldberg
|
146cdf32dc
Move defined types into types.hpp and make the basic value type flexible as to whether it is 32 or 64 bits
|
2 years ago |
|
Ian Goldberg
|
06e69c1c77
Load the half-triples at runtime
|
2 years ago |
|
Ian Goldberg
|
3a16491f86
Preprocessing can create half-triples as well
|
2 years ago |
|
Ian Goldberg
|
342faa921c
Each thread will also need its own storage for precomputed values
|
2 years ago |
|
Ian Goldberg
|
426344a717
Add a flag to specify the number of threads the computational peers will use
|
2 years ago |
|
Ian Goldberg
|
247a5522c5
Have queue() automatically send() if we have a bunch of full packets of data already
|
2 years ago |
|
Ian Goldberg
|
f9824f70d7
Start on the oblivds framework code
|
2 years ago |
