topic Enable FPU on KW45 in Wireless MCU

Enable FPU on KW45

Larry3 — Thu, 09 Jan 2025 06:02:26 GMT

Hello,

I am currently working on a project with KW45 , and I would like to enable the Floating Point Unit (FPU) for improved performance in floating-point calculations. Could someone guide me on the following:

What are the steps to enable it in the IDE (e.g., IAR)?
Are there any specific settings or configurations I need to consider?

Any advice, documentation, or examples would be greatly appreciated.

Best regards

Re: Enable FPU on KW45

Christine_Li — Thu, 09 Jan 2025 09:54:24 GMT

Hi, @Larry3

Thanks for creating case to us.

Please check our KW45's Reference Manual, it mentioned that below register:

11.4.2.9 Processor X Configuration Register 3 (CPxCFG3)

and

11.4.2.17 Processor 0 Configuration Register 3 (CP0CFG3)

There is one bit: bit0 for above 2 registers are using for FPU.

Floating Point Unit
This field indicates if hardware support for floating point capabilities are supported in the processor.
0b - FPU support is not included.
1b - FPU support is included.

And also it mentioned, the reset value:

Reset values for the Processor X Configuration Register 3:
• For CPU0, the CPxCFG3 value read is 0x00000231
• If the read access is not from a CPU, then the value read is 0x00000000

Reset values for the Processor 0 Configuration Register 3:
• CP0CFG3 = 0x00000231

So, it means, in default the FPU feature should be enabled.

Best regards,

Christine.

Re: Enable FPU on KW45

Larry3 — Tue, 14 Jan 2025 09:42:38 GMT

Then how can I make sure I have enabled hardware acceleration when using CMSIS-DSP with KW45? When I use CMSIS-DSP for some calculations, the operation speed does not meet my expectations. Could there be something I have overlooked?

Re: Enable FPU on KW45

Christine_Li — Wed, 15 Jan 2025 09:21:13 GMT

Hi， @Larry3

Can you please let me know your read results for CPxCFG3 and CP0CFG3？

In our KW45's RM, it mentioned:

Floating Point Unit
This field indicates if hardware support for floating point capabilities are supported in the processor.
0b - FPU support is not included.
1b - FPU support is included.

So, from your register reading results, we can see whether the hardware support for floating point capabilities.

Best regards,

Christine.

Re: Enable FPU on KW45

Christine_Li — Mon, 20 Jan 2025 14:39:51 GMT

Hi, @Larry3

I have confirmed with our internal expert team:

FPU is enable in default if the registers(mentioned in my previous comments) you read is supported.

If it could not meet your requirements, please help to create a new case and list your detailed requirements or how do you use it in details. So that we can help your further.

This case is only used to discuss how to enable FPU.

Best regards,

Christine.