ESP32-S3 vs Nucleo G474RE
Xtensa LX7 (ESP32-S3) at 240 MHz against Cortex-M4F (STM32G474) at 170 MHz, on real hardware, same C kernels, clocks verified.
ESP32-S3
Xtensa LX7 (ESP32-S3)
- Family
- Xtensa
- Board
- ESP32-S3 DevKit
- Clock
- 240 MHz
- float / double
- 32-bit / 64-bit
Nucleo G474RE
Cortex-M4F (STM32G474)
- Family
- Arm Cortex-M
- Board
- Nucleo G474RE
- Clock
- 170 MHz
- float / double
- 32-bit / 64-bit
Typical gap
1.12×
ESP32-S3 faster, geometric mean of 30 cells
Biggest win for ESP32-S3
1.93×
double add
Biggest win for Nucleo G474RE
3.24×
double divide
Clock for clock
1.26×
Nucleo G474RE needs fewer cycles per op (1.41× clock gap)
Operation by operation
Every cell
How many times faster ESP32-S3 is than Nucleo G474RE, operations per second. Blue: ESP32-S3 ahead. Orange: Nucleo G474RE ahead.
| Add | Subtract | Multiply | Divide | Modulo | |
|---|---|---|---|---|---|
| uint8_t | 1.08× 12.7 M/s · 11.7 M/s | 1.08× 12.7 M/s · 11.7 M/s | 1.08× 12.7 M/s · 11.7 M/s | 1.28× 11.5 M/s · 8.96 M/s | 1.35× 10.9 M/s · 8.1 M/s |
| uint16_t | 1.08× 12.7 M/s · 11.7 M/s | 1.08× 12.7 M/s · 11.7 M/s | 1.08× 12.7 M/s · 11.7 M/s | 1.24× 10.1 M/s · 8.2 M/s | 1.29× 9.72 M/s · 7.56 M/s |
| uint32_t | 1.06× 13.4 M/s · 12.6 M/s | 1.06× 13.4 M/s · 12.6 M/s | 1.06× 13.4 M/s · 12.6 M/s | 1.25× 11 M/s · 8.73 M/s | 1.26× 10.5 M/s · 8.3 M/s |
| uint64_t | 1.04× 7.87 M/s · 7.56 M/s | 1.03× 7.81 M/s · 7.56 M/s | 1/1.22× 5.45 M/s · 6.67 M/s | 1.55× 1.8 M/s · 1.16 M/s | 1.50× 1.74 M/s · 1.16 M/s |
| float | 1.08× 12.7 M/s · 11.7 M/s | 1.08× 12.7 M/s · 11.7 M/s | 1.08× 12.7 M/s · 11.7 M/s | 1/1.85× 3.48 M/s · 6.42 M/s | 1.46× 1.95 M/s · 1.34 M/s |
| double | 1.93× 3.04 M/s · 1.58 M/s | 1.77× 2.79 M/s · 1.58 M/s | 1/1.11× 2 M/s · 2.22 M/s | 1/3.24× 516 k/s · 1.67 M/s | 1.78× 1.22 M/s · 688 k/s |