streaming 50MS/s

[S_Hong]

Hi Alann,

Unfortunately, we do not list system requirements for every different hardware for streaming applications. Moreover, since the PXI-5661 is a non-express card, most of our benchmarks are for Express cards due to the higher bandwidth of the PCI/PXI Express architecture. That being said, on our page, we recommend using the PXI-8106 controller (Core 2 Duo 2.16 GHz) with a RAID array for streaming with the PXI-5661.

As for optimizing your streaming application, I recommend having a look at the following example program on our NI Developer Zone.

NI-Scope Stream to Disk Using Win32 File IO 

http://zone.ni.com/devzone/cda/epd/p/id/5273

Here's also some extra documentation you can refer to for more information.

Benchmarking Data for Streaming with RF and IF Products

http://zone.ni.com/devzone/cda/tut/p/id/6266

Perform RF Streaming Successfully

http://zone.ni.com/devzone/cda/tut/p/id/7209 

[Alann]

Hello S_Hong,

I have PXIe-8130 controller, HDD-8263 and NI-8262. I believe that suffice the system requirements, then why cant 100MB/s streamed to disk?

[S_Hong]

Hi Alann,

As I mentioned, it is hard to say exactly what may be causing this issue as streaming is a highly system dependent performance metric. Therefore, it is possible that these speeds are simply not achievable with your current system. This is one of the caveats of streaming at high speeds close to the theoretical limits. Also, I would like to add that depending on what chassis you have, some slots may be better suited for streaming than others as data going through a various bridges can cause extra overhead and reduce the overall streaming speeds. More on this can be found on your chassis user manual.

What is the maximum streaming speed you are able to achieve with your system?

[Alann]

Hello S_Hong,

Thanks a lot for your patience and help. I am working with1065 chassis and 8262 connected to slot 7.

I have observed that with RFSA complex IQ acquisition, at 25MS/s we do stream 100MB/s, as each sample has width of 32 bits. When I use SCOPE to fetch I8 (binary sample width of the sample is 8 bit and fetch format is also I8) , I can stream the samples max at 50MB/s. If i set sample rate to be 100MS/s to stream 100MB/s, the samples are over-written. Is there any limitation of SCOPE fetch VI?

[S_Hong]

Hi Alann,

There are two things I would like to clarify:

1. Although we are below the theoretical limits, 100 MB/s may simply be above what your system is capable of streaming and therefore, this may be a hardware limitation.

2. I realized that we were so focused on streaming data at such a fast rate that I forgot to check and see if the data we were recovering made any sense. Now I would like to remind you that in our NI-5661, the downconverter has an instantaneous bandwidth of 20MHz. Due to the FIR filter used in the DDC of the NI-5142, we rate our response flatness to be within a bandwidth calculated as BW = 0.8 x IQ Rate (Shown in page 24 of the NI-5142 Specifications Document). Therefore, because we are limited by the bandwidth of the downconverter, the maximum IQ rate that we can set on our RFSA is 25 MS/s. Because we had our NI-Scope session separate from our RFSA session, we were allowed to set our IQ rate to be higher than 25 MS/s without any error, however, this will not make much sense since we are limited to only 20 MHz of data. So in short, I believe that we may have been too focus on streaming speed rather than looking at the validity of the data we were acquiring.

I hope this makes sense. Let me know if you have any questions.

Regards,

NI PXI/PCI-5142 Specifications

http://www.ni.com/pdf/manuals/374300d.pdf