how important is the video card?

[thekuai 0]

With software based dvr cards? I'm currently running a cheap Nvidia 7300 series video card. Is it worth upgrading to a nvidia 9500gt or is the cpu more important?

[Soundy 1]

CPU and RAM upgrades would make the bigger difference. The video card doesn't need to be anything special, as it doesn't do any processing outside its own display.

 

Most machines we build, we just use the onboard video chipset.

[bpzle 0]

We don't use anything special. Usually just a 512mb Nvidia. As soundy mentioned, it's only needed for good local display. Doesn't affect the recorded images or what you view remotley, even on the same LAN. If you're building a PC based DVR (we call them servers) and the viewing client will not be used on that machine, onboard video is fine. However with products like Avermedia where the viewing client/ recording application run in the same application I would definetley get one. We usually use ones around $50 and they work just fine.

[thekuai 0]

How come no one is developing software utilizing gpu acceleration? Even a budget card can encode faster than software that uses only the CPU. Nvidia's cuda application is one such example. Even the latest version of adobe Photoshop supports gpu acceleration.I find that my aging nvidia 8800gt encodes in half the time a CPU based app can on my phenom 2 955 does.

[bpzle 0]

Honestly, I don't know the difference. I'm just a part swapper aka technician and small business owner.

 

But, I can tell you that the CCTV industry is always a few years behind popular consumer products. It's all about supply and demand. Simple business. There simply isn't much demand in the CCTV world compared to everyday household electronics and computer products. It should only make sense that CCTV products won't have all the fancy (even common sense) R&D that your favorite gaming machine has.

[rory 0]

How come no one is developing software utilizing gpu acceleration? Even a budget card can encode faster than software that uses only the CPU. Nvidia's cuda application is one such example. Even the latest version of adobe Photoshop supports gpu acceleration.I find that my aging nvidia 8800gt encodes in half the time a CPU based app can on my phenom 2 955 does.

 

No need to do that when they have hardware compression DVR cards.

 

As long as the video adapter handles direct draw overlay then good to go.

Surprisingly some dellls from only a couple years ago have onboard video which cannot handle dd properly, and then need to use a separate video card, which in those cases were actually already installed by Dell. So basically most current onboard video will work fine.

[Sawbones 0]

Never had a problem with on-board video.

[thekuai 0]

I realize that there are hardware encoded dvr cards and that they are the mainstream. However, the aforementioned cards are ridiculously expensive. Even the most expensive single core graphics card is only $500. A sub $100 card should be more than suffice. Like I mentioned, gpu's are much more efficient at encoding video given the proper software. Just a thought...

[Soundy 1]

There's no point in dumping your money into a good GPU if the DVR software can't take advantage of it. Photoshop is designed to take full advantage of things like multiple CPU, multiple cores, and advanced GPUs to aid its processing; your average DVR software is not.

[Mnorman 0]

I thought the real work of the DVR cards was not in the encoding/decoding of the image streams, rather it is in the compression for storage. Obviously one can argue that the data streams of multiple cameras are being encoded while being compressed, but that does not involve the video card at all since the CPU cores have to process it first either way and then send the info to the PCI bus.

 

I tend to agree that a faster processor and 4+ gigabytes of ram would go farther to manage the resources than a video card. The bottleneck is not at the PCI-E 2.0 slot, it is before it.

[thekuai 0]

I thought the real work of the DVR cards was not in the encoding/decoding of the image streams, rather it is in the compression for storage. Obviously one can argue that the data streams of multiple cameras are being encoded while being compressed, but that does not involve the video card at all since the CPU cores have to process it first either way and then send the info to the PCI bus.

 

I tend to agree that a faster processor and 4+ gigabytes of ram would go farther to manage the resources than a video card. The bottleneck is not at the PCI-E 2.0 slot, it is before it.

 

There are two types of dvr cards. Hardware dvr cards do all the work and really lightly on the cpu. Software dvr cards rely heavily on the cpu. It depends what kind of card you get.