cancel
Showing results for 
Search instead for 
Did you mean: 

SPWF04Sx FW_1.1.0 SPI UDP socket throughput

Karsten Wiese
Associate II
Posted on December 08, 2017 at 12:02

Hi,

using newest fw and x-cube 3.1.0 in an UDP ping-pong application I can get no more than about 3kBytes/s.

I guess the reason is the '

Socket Query for SPI interface may stall

...' 'known limitation' mentioned in the Release Notes.

We need at least 10times that throughput: > 30kBytes/s.

Is it possible to achieve 30kBytes/s using UART?

Are there SPI fixes I could test?

Some github like beta repository?

tnx & best regards

#spwf04sx #uart #spi
3 REPLIES 3
Posted on December 11, 2017 at 10:55

The 'known limitation' only refers to SPI interface, and only refers to WIFI1 package.

I means, using SPWF04 by itself, you can easily reach 1Mbps using UART, and 5Mbps using SPI (UDP streaming). Use big chunks (80% or more of free_heap) for xSOCKxW commands.

As far as I know, WIFI1 package allows more than 2Mbps (in SPI + UDP socket client)

Posted on December 11, 2017 at 12:33

Please supply demo code showing that performance.

Using small 1100 Bytes UDP packets over  UART - WIFI1 I got 30kBytes/s.

I had to reduce  TimHandle.Init.Period from 99 to 19.

It doesn't run stably: Sometimes it sticks waiting for WiFi_Control_Variables.stop_event_dequeue == WIFI_FALSE and IO_status_flag.AT_Response_Received == WIFI_TRUE.

Posted on December 12, 2017 at 14:06

I tried SPI one more.

SPWF04S:

  • connected to WiFi as station.
  • UDP Server
  • under debugger

PC sends 10Byte UDP Telegrams on KeyHit only.

WiFi over standard Router.

Congestion Test:

  1. Breakpoint on UDP Telegramm callback to application.
  2. Send 10Bytes UDP from PC.
  3. Breakpoint hit.
  4. Send 10Bytes UDP from PC.
  5. Breakpoint release

Symptom after 5:

The expected 55 WIND only comes sometimes.

Sometimes after I sent another 10Bytes UDP from PC.

Sometimes the EXTI line stays passive also after further 10Bytes UDP Packets sent.

Maybe my test arrangement is to blame, maybe I've hit a bug in the SPWF04S.

Can anybody reproduce the symptom I saw ?

Or does anybody see SPWF04S acting correctly in a congestion test like the above?