Benchmarks

All figures below are measured on real hardware, with the same benchmark suite, across every listed toolchain/RTOS combination. Compiler options are the highest optimization level, optimized for time (GCC at -O3, Keil/IAR at their highest settings).

• Cortex-M0+ is evaluated on an STM32G071RBT6.
• Cortex-M4 is evaluated on an STM32F429ZIT6.
• Cortex-M7 is evaluated on an STM32H743ZIT6.
• RV32IMAC (RISC-V) is evaluated on a CH32V307VCT6.

Flash and SRAM are guest image footprints in KiB (code + data / data + bss); all other columns are CPU clock cycles. These are typical values exercising the full system benchmark — minimum sizes seldom mean anything in practice.

Commemorative ports

Single-combination memorial ports: Cortex-M7 + GCC + STM32H743ZIT6.

Machine	Toolchain	RTOS	Flash (KiB)	SRAM (KiB)	Yield	Mail	Sem	Mail/I	Sem/I	Mem
Cortex-M7	GCC	DjyOS	27.49	63.38	1017	1488	892	1873	1220	288
Cortex-M7	GCC	KLite	11.65	20.27	291	1402	379	1964	996	2751

Practical ports

Cortex-M0+ — STM32G071RBT6

Toolchain	RTOS	Flash (KiB)	SRAM (KiB)	Yield	Mail	Sem	Mail/I	Sem/I	Mem
Keil uVision 5	RMP	16.71	3.88	465	783	681	1677	1572	N/A
...	FreeRTOS	12.74	9.57	433	949	1023	1789	1808	843
...	uC/OS-II	11.86	6.73	N/A	1062	1039	3800	1484	2638
...	uC/OS-III	13.63	8.09	863	1117	1143	2125	2156	363
...	RT-Thread	18.69	8.27	950	1581	1317	2760	2464	13221
IAR EWARM	RMP	12.38	3.88	511	876	770	1764	1642	N/A
...	FreeRTOS	11.29	9.57	432	944	1078	1804	1859	944
...	uC/OS-II	10.62	6.73	N/A	1089	1067	4899	1072	2737
...	uC/OS-III	11.89	8.08	892	1154	1187	2161	2189	360
...	RT-Thread	13.15	8.26	1031	1680	1415	2859	2589	13861
GCC	RMP	22.72	2.88	507	883	773	1746	1680	N/A
...	FreeRTOS	15.38	8.57	455	986	1108	1859	1926	1012
...	uC/OS-II	13.08	5.75	N/A	1111	1090	1526	1097	2876
...	uC/OS-III	15.38	7.09	871	1144	1185	2188	2246	396
...	RT-Thread	18.57	7.27	1080	1755	1450	3042	2694	14899

Cortex-M4 — STM32F429ZIT6

Toolchain	RTOS	Flash (KiB)	SRAM (KiB)	Yield	Mail	Sem	Mail/I	Sem/I	Mem
Keil uVision 5	RMP	22.72	37.06	336	519	452	1278	1176	391
...	FreeRTOS	14.73	41.57	308	612	776	1394	1428	527
...	uC/OS-II	13.09	7.73	N/A	671	647	2303	773	1425
...	uC/OS-III	15.50	9.09	504	691	713	1570	1588	196
...	RT-Thread	20.34	8.27	532	1036	824	1948	1743	7307
IAR EWARM	RMP	14.29	36.07	376	602	551	1512	1438	403
...	FreeRTOS	11.48	40.57	316	648	833	1592	1667	549
...	uC/OS-II	10.75	6.73	N/A	708	696	2483	824	1521
...	uC/OS-III	11.65	8.08	601	809	829	1852	1868	207
...	RT-Thread	13.78	7.26	612	1053	852	2231	2020	7646
GCC	RMP	24.20	36.07	416	709	552	1434	1344	382
...	FreeRTOS	15.44	40.57	372	726	882	1544	1611	724
...	uC/OS-II	12.48	6.75	N/A	729	693	2629	849	1597
...	uC/OS-III	14.67	8.09	580	812	849	1759	1820	227
...	RT-Thread	17.63	7.27	648	1173	893	2236	1975	7764

Cortex-M7 — STM32H743ZIT6

Toolchain	RTOS	Flash (KiB)	SRAM (KiB)	Yield	Mail	Sem	Mail/I	Sem/I	Mem
Keil uVision 5	RMP	22.73	37.06	244	384	370	920	811	284
...	FreeRTOS	15.06	41.57	229	494	595	954	987	347
...	uC/OS-II	13.35	7.73	N/A	436	445	1503	584	849
...	uC/OS-III	15.75	9.09	410	590	594	1227	1252	141
...	RT-Thread	20.52	8.28	457	798	651	1537	1391	5197
IAR EWARM	RMP	14.37	36.07	272	459	412	1030	985	284
...	FreeRTOS	11.50	40.57	220	526	609	1122	1109	363
...	uC/OS-II	10.83	6.73	N/A	541	512	1674	654	938
...	uC/OS-III	11.70	8.08	411	592	620	1282	1313	151
...	RT-Thread	13.85	7.26	424	829	675	1585	1441	5710
GCC	RMP	24.29	36.07	248	476	414	886	939	283
...	FreeRTOS	15.29	40.57	225	511	596	1020	1036	354
...	uC/OS-II	12.49	6.75	N/A	489	468	1544	569	909
...	uC/OS-III	14.66	8.09	404	561	578	1216	1291	150
...	RT-Thread	17.75	7.27	465	819	674	1570	1435	6402

RV32IMAC (RISC-V) — CH32V307VCT6

Toolchain	RTOS	Flash (KiB)	SRAM (KiB)	Yield	Mail	Sem	Mail/I	Sem/I	Mem
GCC	RMP	29.61	25.29	322	499	437	1719	1652	360
...	FreeRTOS	17.55	24.77	319	580	630	1778	1827	440
...	uC/OS-II	16.28	6.75	N/A	571	559	2591	570	1337
...	uC/OS-III	16.72	8.09	478	606	611	2050	2079	182
...	RT-Thread	19.45	9.77	533	846	701	2429	2280	6003

Metric definitions

Metric	Meaning
Yield	Time to yield between different threads
Mail	Mailbox communication time between two threads
Sem	Semaphore communication time between two threads
Mail/I	Time to send to a mailbox from an interrupt
Sem/I	Time to post to a semaphore from an interrupt
Mem	Time for a memory operation, e.g. allocation/free

• Yield is N/A for uC/OS-II: with one task per priority, a yield between equal-priority threads is not a meaningful measurement.
• Mem is N/A for RMP guests: RMP exercises its own memory subsystem outside this benchmark.
• Context switch and interrupt response for a virtualized FreeRTOS are about 4x the bare-metal original, and still roughly 25x faster than RT-Linux — see Why Virtualization.