discuss-gnuradio
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] Dropping samples "D" when CPU at 60%


From: CEL
Subject: Re: [Discuss-gnuradio] Dropping samples "D" when CPU at 60%
Date: Wed, 28 Mar 2018 17:19:25 +0000

Dear Bakshi,
On Tue, 2018-03-27 at 18:00 +0000, Bakshi, Arjun wrote:
> 
>  After writing a cpp block for detecting the trigger and piping some samples, 
> the cpu utilization jumped up to +90% on all cores, and I can now support 
> 6.25MSps x 3 streams. 

This nicely demonstrates that CPU utilization alone really tells you
nothing about whether your processing works out in real time.

> But not much more. For some reason I expected that my cpu could handle the 
> full 1Gbps connection.

As you can see: read my above statement

> The CPU is an older (Sandy Bridge/gen-2) address@hidden 4 cores, running on 
> Ubuntu 16.

As you can see: read my above statement

>  Memory wasn't an issue, as the memory usage for the flow hovers around 
> 100MB, with ~10GB to spare. 

That also doesn't mean much: In many scenarios, processing is limited
by RAM *bandwidth* rather than RAM *consumption*.

> 
> The ethernet connection is able support higher sampling rates(8Msps x 3 
> streams) with direct USRP source to null sink connections. So the 
> ethernet/network should be fine.

No, you also can't infer that. Your PC is a complex system! If, for
example, your CPU is busy a lot, your OS might not be able to handle
network packets in real time, and "D" occurs. Or, if something else
causes a lot of interrupts (timers, GPUs, storage devices,…), then the
OS might not be able to serve the NIC in time, and "D" occurs. Or if
something needs a lot of PCIe bandwidth …

What you really need to understand is that your PC is a complex system
with a lot of shared resources. Just because things work in isolation,
doesn't mean much about them still being real-time capable in
coordination.

> Only concern now is if its harmful to have 90%-100% cpu utilization for an 
> hour or so.

No? Unless your PC's cooling isn't sufficient. If it works, it works.

Stochastic says that if you run an experiment with some probability of
failure for prolonged time, it will more likely fail. But that's kind
of an obvious statement.

Best regards,
Marcus

Attachment: smime.p7s
Description: S/MIME cryptographic signature


reply via email to

[Prev in Thread] Current Thread [Next in Thread]