compact vision system 1457 benchmarks

[hurban]

Hello

our company has been using a number of EVS-1464 RT in the past with good results.

Does anybody know performance benchmarks (eg. pattern matching) of the new cvs 1457.

How does it compare to the older EVS-1464 RT ?

I would be particularly interested in operating 2 or more Hi-Res Cameras (2MPix) on the same

CVS at a repetition cycle around 0.3 seconds.

Any hint is greatly appreciated

Thanks HU

[OEMa]

Hi hurban,

Regarding the specifications of both systems, the CVS-1457RT has 1GB RAM memory and 2GB Nonvolatile Storage Memory whereas the EVS-1464RT has 2GB RAM memory and 1GB Nonvolatile Storage Memory.

The performance of the EVS-1464RT (Intel Core Duo 1.66 GHz) is also about twice better than with the CVS-1457RT (1.6 GHz Intel Atom).

The performance benchmarks should be twice better.

The fact that the EVS-1464RT has a dual-core processor is also be an advantage since it is possible to separate the real-time code on the two cores (eg. deterministic loops and inspections on Core 1 and data storage or Network communication on Core 2).

Marc O

Application Engineer

National Instruments Germany

[hurban · Accepted solution]

Hi Marc

thanks a lot for the quick response, for the moment that should be all I need to know. Based on what you say I can simply do some benchmarks for the new project on our existing EVS and devide the performance by 2 - fine 🙂

Thanks again

HU

[BlueCheese]

I don't think it is really that simple... Even though the clock speeds are similar, the EVS uses a Core Duo processor and the CVS 1457RT uses a low-power Atom processor. The architectures are different enough that there are going to be performance differences depending on many, many factors. It is hard to even give a general benchmark for a specific algorithm, because even things like the image size can make a big difference depending on the CPU based on how much cache it has. You can likely make some general assumptions that the EVS is probably going to be faster in most cases, but beyond that it is really hard to say for sure.

[hurban]

Hi BlueCheese

and thanks for the extra information - frankly I was expecting some statement along these lines 😉

That is why I ordered a CVS as a loaner to do a few tests on my own here in the lab.

My plan is to install some 11 of these systems (with 2 cameras each) on an automotive production line.

This involves considerable investment so we need to make pretty sure we know what we are doing.

Thanks again

HU

[BlueCheese]

I'd be interested to hear about the performance differences you see between the two devices in your application. Getting real-world data about actual applications is always helpful to everyone!

Eric