FPGA compatibility issue

19 views
Skip to first unread message

zzhan...@gmail.com

unread,
Apr 8, 2026, 1:44:07 PMApr 8
to Powder Users
Hi, 

I am trying to run a USRP command "tx_samples_from_file" but got this error:

Creating the usrp device with: ...
[INFO] [UHD] linux; GNU C++ version 11.3.0; Boost_107400; UHD_4.7.0.0+ds1-2emulab1
[INFO] [X300] X300 initialization sequence...
Error: RuntimeError: Expected FPGA compatibility number 39.0, but got 38.0:
The FPGA image on your device is not compatible with this host code build.
Download the appropriate FPGA images for this version of UHD.

I tried the suggested fixes but it didn't work. I am using gnuradio_ota profile and I was able to run the command in the past with no issues. Any help is greatly appreciated.

Thanks,
Zhenghao

Kirk Webb

unread,
Apr 8, 2026, 1:57:37 PMApr 8
to powder...@googlegroups.com
Hello Zhenghao,

Can you provide a link to your experiment so we can take a look?
Normally this issue can be resolved by loading the appropriate
firmware onto the SDR using the `uhd_image_loader` command, followed
by a power cycle of the device using the POWDER web UI.

BR,
-Kirk
> --
> You received this message because you are subscribed to the Google Groups "Powder Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to powder-users...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/powder-users/d84122b5-a3d4-4114-bb73-8e4a0680c951n%40googlegroups.com.
Message has been deleted
Message has been deleted
Message has been deleted

zzhan...@gmail.com

unread,
Apr 8, 2026, 2:49:05 PMApr 8
to Powder Users
Hi Kirk,

I tried to post the link of my experiment but my message got deleted 3 times.

I tried the commands and power cycle, which did not work. I ran into similar situations before and was able to get it to work with the commands and power cycle. I am using 5 nodes and only bes and hospital have this problem.

Thanks,
Zhenghao

Kirk Webb

unread,
Apr 8, 2026, 3:23:04 PMApr 8
to powder...@googlegroups.com
Hello Zhenghao,

I am not sure why your message would be getting deleted; were you
attempting to post via the Google Groups website, or by responding in
email? People have certainly been able to post links/URLs to the Group
throughout its existence, including in a recent message from me.

In any case, I have located your experiment and am taking a look.

-Kirk
> --
> You received this message because you are subscribed to the Google Groups "Powder Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to powder-users...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/powder-users/90bbad97-a1fe-424f-884e-61b5aff8cb6an%40googlegroups.com.

Kirk Webb

unread,
Apr 8, 2026, 3:36:38 PMApr 8
to powder...@googlegroups.com
I think I've identified the issue. The output from libuhd when there
is a firmware mismatch on these nodes is suggesting that you run:

"/lib/bin/uhd_image_loader" --args="type=x300,addr=192.168.40.2"

However, this path does not exist. I do not know why the installed UHD
tools/library are providing an incorrect absolute path to the
`uhd_image_loader` program. In any case, you can simply change the
command to omit the full path, like so:

uhd_image_loader --args="type=x300,addr=192.168.40.2"

And it should work. I have done this on your BES and Hospital compute
nodes, and have power cycled the corresponding SDRs. It looks like
these devices are now running firmware that is properly matched to the
tools on the compute nodes.

-Kirk
Message has been deleted

zzhan...@gmail.com

unread,
Apr 10, 2026, 10:43:36 AMApr 10
to Powder Users
Hi Kirk,

Thanks! Both nodes worked. I really appreciate it.

Zhenghao

zzhan...@gmail.com

unread,
Apr 10, 2026, 10:43:51 AMApr 10
to Powder Users
Hi Kirk,

This should be the link to my experiment:


I tried the commands and power cycle, which did not work. I ran into similar situations before and was able to get it to work with the commands and power cycle. I am using 5 nodes and only bes and hospital have this problem.

Thanks,
Zhenghao
Reply all
Reply to author
Forward
0 new messages