Performance difference between native C implementation has increased

[shepmaster]

XxHash32/1024 bytes/offset 0
    time:   [194.75 ns 201.79 ns 209.05 ns]
    thrpt:  [4.8985 Gelem/s 5.0746 Gelem/s 5.2580 Gelem/s]
XxHash32/1024 bytes/offset 1
    time:   [200.74 ns 205.56 ns 210.85 ns]
    thrpt:  [4.8613 Gelem/s 4.9865 Gelem/s 5.1062 Gelem/s]

XxHash32 (C)/1024 bytes/offset 0                        
    time:   [129.31 ns 132.99 ns 137.30 ns]
    thrpt:  [7.4583 Gelem/s 7.6997 Gelem/s 7.9187 Gelem/s]
XxHash32 (C)/1024 bytes/offset 1
    time:   [130.69 ns 132.33 ns 134.12 ns]
    thrpt:  [7.6423 Gelem/s 7.7460 Gelem/s 7.8429 Gelem/s]