STM32H5 - why only 250 MHz?

[Pavel A.]

Why there's no 480..500MHz variants of H5, like in the H7 line?

[Tesla DeLorean]

Probably because of the pipeline length and process. It's an M33 not an M7 core.

Although I'm using one chip that clocks a CM3 at 384 MHz, with all the code executed out of RAM without the drag of uncached FLASH. ​

[Pavel A.]

STM32H7 can execute code from RAM too.

Some performance figures are in data sheets to compare...

STM32H750, 480 MHz: 1027 DMIPS, 2.14 DMIPS/MHz (Dhrystone 2.1)

ADC:  up to 3.6 MSPS

STM32H503, 250 MHz: 375 DMIPS (Dhrystone 2.1) 1.5 DMIPS/MHz (Drystone 2.1)

1023 CoreMark (4.092 CoreMark/MHz)

ADC: up to 2.5 MSPS

So... the CM33 core does not automagically provide more bang for MHz.

[etheory]

I agree, the NXP i.MX RT600 is 300MHz and dual core. I wish the STM32 H5 series had such an option.​

[Foued_KH]

Hello @pavel A.​ ,

The new STM32H5 series comes to complete the STM32 High Performance MCUs portfolio.

It comes to complete the F4 series.

 

For more details, you can check the Product Presentation : STM32H5 MCU series Most powerful Arm® Cortex®-M33 MCU yet

Foued

[Pavel A.]

Thank you Foued. But my question is, why H5 clock rate is lower than H7. Is it a limitation of CM33 core vs. CM7, or has other reasons?

Users definitely like high performance figures. Even if they don't actually need or utilize it, potential to scale up looks good.

[Foued_KH · Accepted solution]

The STM32H5 comes to complete the STM32F4 not STM32H7 , STM32G4...

So , If you have a look on the portfolio you can find the STM32H5 after the STM32F4.

Foued

[etheory]

That's fine, but why is it clocked at 250MHz and not 300MHz like we know the M33 is capable of? And also, why does a 250MHz MCU support only FS USB and not HS USB?