LimeSDR DC spikes and CPU usage

All bug reports here please
timmy
Posts: 2
Joined: Sun Jul 11, 2021 10:33 am

LimeSDR DC spikes and CPU usage

#1

Unread post by timmy »

The console is great and my favorite SDR software. I have a few questions,

Question 1
LimeSDR, set RX Ant=NONE: There are 3 spikes on the spectrum: 415.314MHz, 435.000MHz, and 454.686MHz. It looks like 435.000MHz is like a DC spike. The other two are mirror images on both sides of the DC spike, 435.000+-19.686MHz?
Can software remove DC spikes? Can the mirror image on the edge be eliminated?

Question 2
My computer is a 2019 MacBook pro16". When it is running, the CPU usage is high, the fan runs fast, and the computer system becomes slower. Can the software reduce the resource usage?

Looking forward to your answer, thank you!

Timmy
Attachments
DC.PNG

timmy
Posts: 2
Joined: Sun Jul 11, 2021 10:33 am

Re: LimeSDR DC spikes and CPU usage

#2

Unread post by timmy »

Unfortunately,
it seems that software authors don't care about users' problems.

User avatar
G0OFE
Posts: 435
Joined: Thu Aug 06, 2020 7:33 am
Location: Sunny Sunny Bournemouth IO90BR
Contact:

Re: LimeSDR DC spikes and CPU usage

#3

Unread post by G0OFE »

Bit harsh.

Maybe he is out enjoying the warm sunshine, a commodity all too often lacking in Britain...
Jim, Bournemouth IO90BR
Image

jdow
Posts: 800
Joined: Mon Aug 10, 2020 8:17 pm

Re: LimeSDR DC spikes and CPU usage

#4

Unread post by jdow »

Um, Timmy, you do not have a problem in the part of the loop that runs from the keyboard through the computer up to the eyeball. If you exercise said eyeball and look towards the lower left if your picture it shows your bandwidth is 61.44 MHz. It's more or less astonishing that the CPU load is that small. Reduce the bandwidth if you want lower CPU usage. You also do not seem to have a GPU helping reduce your CPU load. That, too, would help.

{o.o}

jdow
Posts: 800
Joined: Mon Aug 10, 2020 8:17 pm

Re: LimeSDR DC spikes and CPU usage

#5

Unread post by jdow »

For further explanation I believe the FFT is a computation that requires roughly speaking a constant times the square of the number of FFT bins. So you'd be using about 1/40th the amount of CPU grinding out the full bandwidth of an AirSpy R2 spectrum.

I also believe the latest SDRC may have an option that turns the FFT bin count and computations per second down automatically on small computers. You can also use display options to reduce CPU usage.

{^_^}

jpsa
Posts: 15
Joined: Wed Aug 26, 2020 9:06 am

Re: LimeSDR DC spikes and CPU usage

#6

Unread post by jpsa »

jdow wrote: Tue Jul 20, 2021 10:04 am For further explanation I believe the FFT is a computation that requires roughly speaking a constant times the square of the number of FFT bins
FFT takes time proportional to N * log(N)

John

jdow
Posts: 800
Joined: Mon Aug 10, 2020 8:17 pm

Re: LimeSDR DC spikes and CPU usage

#7

Unread post by jdow »

Thanks - it was late and I was not sure. N log N corresponds with more reasonable numbers for his 61 MHz vs the usual usage for 10 Msps on a laptop.

{^_^}

Post Reply