Skip to content
Branch: master
Find file History
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
..
Failed to load latest commit information.
audit MLPerf inference v0.5 results Nov 3, 2019
code MLPerf inference v0.5 results Nov 3, 2019
measurements MLPerf inference v0.5 results Nov 3, 2019
results
systems MLPerf inference v0.5 results Nov 3, 2019
README.md MLPerf inference v0.5 results Nov 3, 2019

README.md

Habana's open submission

Habana's open submission aim to demonstrate Goya low latency capabilities. Our open submission follows the closed submission rules with one exception - strict the latency constrains for Multi-Stream scenario.

  • ResNet50: we managed to reach 20 Samples-Per-Query (SPQ) under latency constrain of 2ms and 40 SPQ under 3.3ms latency constrain. Thus, up to 25 fasters than the required latency for the closed division (50ms)
  • SSD-large: we managed to reach 4 SPQ under latency constrain of 16.8ms and 8 SPQ under 30.8ms latency constrain. Up to than 4 time faster than the require latency for closed division (66ms)

We present our results using two submissions:

  • Goya_fast_latency, for 2ms (resent) and 16.8ms (ssd-large) constrains.
  • Goya_medium_latency, for 3.3ms (resent) and 30.8ms (ssd-large) constrains.
You can’t perform that action at this time.