A quick and non-micro benchmark to figure out how the ARM and AMD CPUs I have and what AWS has relate to each other in terms of single-thread performance.
Comparison is between
- AWS t4g-micro (1 CPU)
- IBM M75q (Ryzen 5 3400GE @ 3.3 GHz)
- NanoPi R2S (RK3328, 4xCortex A53 @ 1.3GHz, 64 bit)
- CubieTruck (AllWinner A20, 2xCortex A7 @ 1 GHz, 32 bit)
Below is the node-sqlite3 results from https://github.com/JoshuaWise/better-sqlite3 with Node.js v14.15 and sqlite 3.31 (average of 2 runs):
|read rows individually||19932||15249||2160||884|
|read 100 rows||1966||1405||320||144|
|iterating over 100 rows||207||155||24||10|
|inserting rows individually||17214||13085||1556||530|
|inserting 100 rows||250||191||30||13|
|tar tfv node-v14.15.1-linux-arm64.tar.xz||1.5s||1.4s||5.2s||10.5s|
The last one has its stdout redirected to /dev/null.