Thumb-2 Poly1305: implementation in assembly #7939
Conversation
|
This needs to be merged first: #7935 |
SparkiDev
force-pushed
the
thumb2_poly1305
branch
2 times, most recently
from
3b77224
to
eb76034
Compare
Implementation of ChaCha algorithm for ARM Thumb-2.
SparkiDev
force-pushed
the
thumb2_poly1305
branch
from
eb76034
to
e5ead65
Compare
| "LDR r3, [%[key], #4]\n\t" | ||
| "LDR r4, [%[key], #8]\n\t" | ||
| "LDR r5, [%[key], #12]\n\t" | ||
| "LDM r10, {r6, r7, r8, r9}\n\t" |
There was a problem hiding this comment.
This is a Cortex M7 (STM32H753ZI). This was with debug build.
arm-none-eabi-gcc "../Middlewares/Third_Party/wolfSSL_wolfSSL_wolfSSL/wolfssl/wolfcrypt/src/port/arm/thumb2-poly1305-asm_c.c" -mcpu=cortex-m7 -std=gnu11 -g3 -DUSE_HAL_DRIVER -DHAVE_PKCS11_STATIC -DSTM32H753xx -DDEBUG -c -I../Middlewares/Third_Party/FreeRTOS/Source/include -I"/Users/davidgarske/Projects/WolfSSL/STM/STM32H7/STM32H753/Middlewares/Third_Party/wolfPKCS11" -I../Middlewares/Third_Party/FreeRTOS/Source/portable/GCC/ARM_CM4F -I../Drivers/CMSIS/Include -I../Core/Inc -I../Drivers/STM32H7xx_HAL_Driver/Inc/Legacy -I../Drivers/CMSIS/Device/ST/STM32H7xx/Include -I../Middlewares/Third_Party/FreeRTOS/Source/CMSIS_RTOS_V2 -I../Drivers/STM32H7xx_HAL_Driver/Inc -I../Middlewares/Third_Party/wolfSSL_wolfSSL_wolfSSL/wolfssl -I../wolfSSL/. -I../wolfSSL -I../Middlewares/Third_Party/wolfSSL_wolfSSL_wolfSSL/wolfssl/ -I../Middlewares/Third_Party/wolfSSL_wolfSSH_wolfSSH/wolfssh/ -I../Middlewares/Third_Party/wolfSSL_wolfMQTT_wolfMQTT/wolfmqtt/ -I../wolfTPM -I../Middlewares/Third_Party/wolfSSL_wolfTPM_wolfTPM/wolftpm/ -O0 -ffunction-sections -fdata-sections -Wall -fomit-frame-pointer -fstack-usage -fcyclomatic-complexity -MMD -MP -MF"Middlewares/Third_Party/wolfSSL_wolfSSL_wolfSSL/wolfssl/wolfcrypt/src/port/arm/thumb2-poly1305-asm_c.d" -MT"Middlewares/Third_Party/wolfSSL_wolfSSL_wolfSSL/wolfssl/wolfcrypt/src/port/arm/thumb2-poly1305-asm_c.o" --specs=nano.specs -mfpu=fpv5-d16 -mfloat-abi=hard -mthumb -o "Middlewares/Third_Party/wolfSSL_wolfSSL_wolfSSL/wolfssl/wolfcrypt/src/port/arm/thumb2-poly1305-asm_c.o"
Implementation of ChaCha algorithm for ARM Thumb-2. Implementation of Poly1305 algorithm for ARM Thumb-2.
SparkiDev
force-pushed
the
thumb2_poly1305
branch
from
e5ead65
to
27033c2
Compare
|
retest this please |
There was a problem hiding this comment.
Wonderful improvement!
On an STM32H7 Cortex-M7 at 480MHz:
CHACHA: 50% faster
CHA-POLY: 85% faster
POLY1305: 276% faster
Before:
CHACHA 6 MiB took 1.000 seconds, 6.177 MiB/s
CHA-POLY 3 MiB took 1.004 seconds, 3.404 MiB/s
POLY1305 12 MiB took 1.000 seconds, 12.207 MiB/s
After:
CHACHA 9 MiB took 1.000 seconds, 9.204 MiB/s
CHA-POLY 6 MiB took 1.000 seconds, 6.299 MiB/s
POLY1305 34 MiB took 1.000 seconds, 33.789 MiB/s
Description
Implementation of Poly1305 algorithm for ARM Thumb-2.
Testing
Tested with QEMU.
With and without: -DWOLFSSL_SP_NO_UMAAL
Checklist