 Ian Goldberg
|
82a40cd709
Remove code that's now a no-op
|
1 year ago |
|
Ian Goldberg
|
a70720716d
Convert global padding to per-storage-server padding in round 2
|
1 year ago |
|
Ian Goldberg
|
6f71b7a820
Correct sizes of Round 2 messages and buffers to match the paper
|
1 year ago |
|
Ian Goldberg
|
ad6a417880
Compact the padding messages to the end to preserve the sorted order
|
1 year ago |
|
Ian Goldberg
|
80a681a261
In round1c, keep only the highest-priority pub_in messages destined to any given receiver id and turn the rest to padding
|
1 year ago |
|
Ian Goldberg
|
7b90bbebdb
Report results meaningfully when SHRINK_TO_MEM and/or OVERLOAD_CORES are set to 1
|
1 year ago |
|
Ian Goldberg
|
13571fe1c5
Take "EXPERIMENT" out of full caps
|
1 year ago |
|
Ian Goldberg
|
b68cf9b944
Main repro script
|
1 year ago |
|
Ian Goldberg
|
ad7b01c95c
Script and docker wrapper for a short "kick-the-tires" test
|
1 year ago |
|
Ian Goldberg
|
f48645b344
Add a wrapper script to run epc_probe.py within the docker
|
1 year ago |
|
Ian Goldberg
|
ecd5711f3e
Address the TODO from the "extra 5 bytes" commit above
|
1 year ago |
|
Ian Goldberg
|
2ccb6a6b2a
Touch up some comments
|
1 year ago |
|
Ian Goldberg
|
d4e3e7f793
Obliviously sanitize padding messages before putting them in clients' mailboxes
|
1 year ago |
|
Ian Goldberg
|
b7e073b443
Formatting
|
1 year ago |
|
Ian Goldberg
|
b9627c972e
run_experiments.py doesn't need to be executable
|
1 year ago |
|
Ian Goldberg
|
c781321dd7
Update the EPC cutoffs to meach the new (lower) EPC memory requirements
|
1 year ago |
|
Ian Goldberg
|
66ed3d04d4
In round1b, send an extra 5 bytes at the end of the message
|
1 year ago |
|
Ian Goldberg
|
dc2c4f3c38
Update gen_enclave_config to track the updated buffer sizes
|
1 year ago |
|
Ian Goldberg
|
3fef8f3fe4
Don't optimistically use smaller WaksmanNetworks than was asked for
|
1 year ago |
|
Ian Goldberg
|
d210fca4b1
Ensuring the size of the blocks sent in round1b are a multiple of the number of routing nodes removes the ceilings and + 1 issues
|
1 year ago |
|
Ian Goldberg
|
e92d046286
Only send messages in one direction in round1b
|
1 year ago |
|
Ian Goldberg
|
8cd0015726
A little more minor formatting
|
1 year ago |
|
Ian Goldberg
|
d208b486b2
Just a bit of formatting and correcting comments about the sense of higher priority
|
1 year ago |
|
Ian Goldberg
|
8a5a63878e
"make clean" should clean the Client dir as well
|
1 year ago |
|
Ian Goldberg
|
7afcd457ec
Check the EPC and free RAM sizes before running the experiments
|
1 year ago |
|
Ian Goldberg
|
60e3633b3d
Add cpuid to the Dockerfile
|
1 year ago |
|
Ian Goldberg
|
a0037dae9f
The round1a buffer size needs to be a multiple of the number of routers
|
1 year ago |
|
Ian Goldberg
|
bf3f1304ba
More accurately compute the required enclave heap memory
|
1 year ago |
|
Ian Goldberg
|
17400d15bc
Some code to track heap usage
|
1 year ago |
|
Ian Goldberg
|
a022ff705c
Correct the heap size computation when N/M is small and M is large
|
1 year ago |
