Doubt about xcore processor ？

[SUNROC1]

1、xmos offical datasheet declare that its I/O speed can reach up to 60Mhz ，is this fast enough ？
2、STMicroelectronics declare its new stm32f43xx series IO speed up to 90MHZ
3、xcore per core only 125mhz ，but stm32f439 could reatch 180mhz , so there's room for improvement ,perhaps limit to 125mhz could keep low power consume .if improve core frequency ,could reatch 2000mips maybe

haha

personal view

welcome your point !

[octal]

Hi SUNROC1,
Xmos is a hardware parallel architecture, i.e. when STM32F4xx is servicing an interrupt, the main program is completely stoped, while in XMOS you can dedicate a core to service a task when the main task is still executing.
The main advantage in favour of XMOS, from my point of view, is determinism. STM32 with its cache and other stages for prog execution makes it completely undeterministic.
XMOS ports does automatically data sampling, and data serialisation/deserialisation depending on internal or external clock. On STM32 this is not possible (I know you can fiddle with DMA to do some of that, but it's not as flexible as XMOS ports - I know it because I'm using STM32F4 since more than 2 years in professional environment in my daily work).
XMOS is not the chip only, it's the entire devlopment toolchain. While I'm using UlinkPro for tracing on STM32F4 (using Keil entreprise edition), I can say that there is no equivalent to XMOS XTA in all ARM envirnment (mainly because of non-determinism of the arm architecture)!!! the whole XMOS toolchain, the free IPs and open source nature of tools, makes XMOS far beyond other silicon tools.
Now I'm not saying that XMOS can replace STM32F4, or FPGA or any other thing. Every architecture and tools answer a special market, and special situations.

[SUNROC1]

Hi SUNROC1,
Xmos is a hardware parallel architecture, i.e. when STM32F4xx is servicing an interrupt, the main program is completely stoped, while in XMOS you can dedicate a core to service a task when the main task is still executing.
The main advantage in favour of XMOS, from my point of view, is determinism. STM32 with its cache and other stages for prog execution makes it completely undeterministic.
XMOS ports does automatically data sampling, and data serialisation/deserialisation depending on internal or external clock. On STM32 this is not possible (I know you can fiddle with DMA to do some of that, but it's not as flexible as XMOS ports - I know it because I'm using STM32F4 since more than 2 years in professional environment in my daily work).
XMOS is not the chip only, it's the entire devlopment toolchain. While I'm using UlinkPro for tracing on STM32F4 (using Keil entreprise edition), I can say that there is no equivalent to XMOS XTA in all ARM envirnment (mainly because of non-determinism of the arm architecture)!!! the whole XMOS toolchain, the free IPs and open source nature of tools, makes XMOS far beyond other silicon tools.
Now I'm not saying that XMOS can replace STM32F4, or FPGA or any other thing. Every architecture and tools answer a special market, and special situations.

yes,I agree with your point,its hard for stm32 to process multi tasks using interrupt ,its undeterministic . xmos is flexible to process multitask ,I hope xmos could be more stronger,
i fascinate about multi core processor , maybe someday we can use xmos chip to build a multi-core server

[pstnotpd]

maybe someday we can use xmos chip to build a multi-core server

You can, it is sort of done now.
Swallow many core research project