Might be just a lucky pass, but this is the best benchmark so far with the current...

diff --git a/benchmarks.md b/benchmarks.md
index bd678546deed225de28d6947a2955626daa46c50..9a72d0b219469dff596f496873b5da4790f93811 100644 (file)
--- a/benchmarks.md
+++ b/benchmarks.md
@@ -408,6 +408,19 @@ NanoPow (WebGPU) 3070 (dispatch 0x800 workgroup 256)
        "geometric": 179.3126486381509
 }
 
+NanoPow (WebGPU) 3070 (dispatch 0x800 workgroup 64)
+{
+       "count": 64,
+       "total": 13121.699999928474,
+       "rate": 4.877416798154878,
+       "min": 9.700000017881393,
+       "max": 1030.5,
+       "arithmetic": 205.0265624988824,
+       "truncated": 142.53124999860302,
+       "harmonic": 63.87965522037639,
+       "geometric": 117.9588307327579
+}
+
 
 
 NanoPowGpu: Time to calculate proof-of-work for a send block 32 times