- kaiser October 29, 2021 at 8:53 am
We’ve been using the SM200B in our products successfully for a while now, and have an upcoming project where it would be somewhat beneficial to move to the SM200C.
It seems like it should be a pretty easy integration, and that I’ll be happy with the C version. But I do have one question — our application currently relies on pretty quick delivery of the data from the unit. We’re sweeping at 3kHz, and using delivered sweep data to change parameters on our components for only 15-20 sweeps in the future. We typically synchronize two SM200B’s via 10MHz and then get data/queue sweeps in the same loop together to keep them as fully in sync as possible.
My general feeling is that the IP version will necessary have some additional latency just due to the IP overhead, framing, and then data transfer to RAM compared to the USB version, but I don’t know if that will be the case, and if so how much is that added latency?
Oops, thought I was posting in the SM series forum, not general discussion. Moving it.
Justin CrooksModeratorJustin Crooks October 29, 2021 at 11:15 am
Good news and bad news. In Windows, even though the median latency for the SM200C is a bit lower than the SM200B, the peak latency can be high (I have seen 80 ms delays when opening a PDF for example).
I do not believe the same problem exists on the Linux side.kaiser November 1, 2021 at 10:40 am
Thanks Justin, I appreciate the info. Glad to hear that it shouldn’t generally be an issue (we usually have a bit of extra processing oomph, and we’ll just make sure we’re good moving forward too).
Can you share typical expected latencies (and what the difference is from the B to C)? I’m very latency sensitive, so am wanting to test some new stuff on the C and see if I can tighten some loops if it’s usually a bit better.
Justin CrooksModeratorJustin Crooks November 1, 2021 at 1:39 pm
We ran a fresh batch of tests, with the API as it currently stands. I was a little surprised. A 2 GHz span (roughly 2 ms of actual sweep time), took from start to finish (not queued), on average:
SM200B = 4.4 ms
SM200C = 4.8 ms
There were no significant peaks in the 1000-sweep test. From this, it looks like the SM200C has a bit higher latency across the board (something like 2.8 ms vs 2.4 ms).
Obviously these numbers will vary from computer to computer and sweep to sweep, and peak numbers will vary based on CPU load, but it’s a starting point.kaiser November 2, 2021 at 6:32 am
Thanks Justin. Great and prompt answers, like always.
That latency looks like it’ll probably be in the noise to me, so we’ll go ahead and plan on trying out the C model.
You must be logged in to reply to this topic.