Performance: run benchmarks on r6i.4xlarge instance (#622)

docs/pages/performance.md

@@ -8,7 +8,7 @@ classes: wide
 
 ## Method
 
-The results on this page are produced by running elastiknn on the ann-benchmarks benchmark on an AWS EC2 r6i.8xlarge instance.
+The results on this page are produced by running elastiknn on the ann-benchmarks benchmark on an AWS EC2 r6i.4xlarge instance.
 We use `task dockerStartBenchmarkingCluster` to start a single-node benchmarking cluster `task annbRunLocalFashionMnist` to run the benchmark.
 
 See the ann-benchmarks directory in the Elastiknn repository for more details.

docs/pages/performance/fashion-mnist/results.md

@@ -1,10 +1,10 @@
 |Model|Parameters|Recall|Queries per Second|
 |---|---|---|---|
-|eknn-l2lsh|L=100 k=4 w=1024 candidates=500 probes=0|0.378|377.213|
-|eknn-l2lsh|L=100 k=4 w=1024 candidates=1000 probes=0|0.447|320.000|
-|eknn-l2lsh|L=100 k=4 w=1024 candidates=500 probes=3|0.635|302.384|
-|eknn-l2lsh|L=100 k=4 w=1024 candidates=1000 probes=3|0.717|259.201|
-|eknn-l2lsh|L=100 k=4 w=2048 candidates=500 probes=0|0.767|335.978|
-|eknn-l2lsh|L=100 k=4 w=2048 candidates=1000 probes=0|0.846|287.923|
-|eknn-l2lsh|L=100 k=4 w=2048 candidates=500 probes=3|0.921|229.917|
-|eknn-l2lsh|L=100 k=4 w=2048 candidates=1000 probes=3|0.960|204.375|
+|eknn-l2lsh|L=100 k=4 w=1024 candidates=500 probes=0|0.378|374.760|
+|eknn-l2lsh|L=100 k=4 w=1024 candidates=1000 probes=0|0.447|317.451|
+|eknn-l2lsh|L=100 k=4 w=1024 candidates=500 probes=3|0.634|302.484|
+|eknn-l2lsh|L=100 k=4 w=1024 candidates=1000 probes=3|0.716|262.492|
+|eknn-l2lsh|L=100 k=4 w=2048 candidates=500 probes=0|0.767|333.057|
+|eknn-l2lsh|L=100 k=4 w=2048 candidates=1000 probes=0|0.846|284.019|
+|eknn-l2lsh|L=100 k=4 w=2048 candidates=500 probes=3|0.921|228.224|
+|eknn-l2lsh|L=100 k=4 w=2048 candidates=1000 probes=3|0.960|205.738|