DG NV tools

[MrVideo]

How fast is DGIndexNV supposed to be able to scan a 1080i MPEG-2 file that is 44 minutes long and is averaging around 36 Mbps for the bitrate?

Right now, it looks like it is going to take at least 40 minutes.

The old software method took about 5 minutes for a file like this.

[MuLTiTaSK]

@MrVideo

your GPU might be borked and it probably cant take the load so it's actually slower then DGIndex
are you using PureVideo Deinterlacer?

[Guest]

Originally Posted by MrVideo

How fast is DGIndexNV supposed to be able to scan a 1080i MPEG-2 file that is 44 minutes long and is averaging around 36 Mbps for the bitrate?

Right now, it looks like it is going to take at least 40 minutes.

The old software method took about 5 minutes for a file like this.

What do you mean by "the old software method"?

[MuLTiTaSK]

@neuron2

any good news about the crashes i posted?

[an3k]

@neuron2

is the behavior as described in http://forum.doom9.org/showthread.ph...07#post1361007 intended? Currently i have to index two times. Chance to get this changed? Thanks

[Guest]

Originally Posted by MuLTiTaSK

any good news about the crashes i posted?

It doesn't crash for me.

I will have access to a Win7 32bit system soon, at which point I can try to duplicate your crashes.

[Guest]

Originally Posted by an3k

is the behavior as described in http://forum.doom9.org/showthread.ph...07#post1361007 intended? Currently i have to index two times. Chance to get this changed?

It's not intended and would not be easy to change, but let me ask why you need to do this, versus not demuxing and just serving the video from the source stream?

[MuLTiTaSK]

@neuron2

that would be awesome plus stax76 is getting a compatible Nvidia GPU for use with DGIndexNV and he runs on Windows 7 so hopefully were be able to get this working as a team sooner then later

[laserfan]

Originally Posted by neuron2

* The Field operation is removed from DGIndexNV and added as the fieldop parameter to DGDecodeNV/DGMultiDecodeNV. This was done because the field operation has no effect on the operation of DGIndexNV and so it doesn't belong there.

* Backward stepping by GOPs (< button) is fixed for program streams.

* Lockup when serving MKV files and navigating near the end of file is fixed.

* Some performance improvements.

* Up-to-date documentation.

The changes are pretty extensive so beat on this one. I hope I didn't break anything.

Note that you need to remake your projects as the DGI file version was bumped. You also need to delete your INI file and let DGIndexNV make a new one.

http://neuron2.net/dgdecnv/dgdecnv.html

A belated "THANKS" and congratulations DG on the new release. Having just read the new docs front-to-back (good job as usual) I am struck by how far you've taken these tools wrt simplifying all manner of input types into one nice package.

A quibble might be that your QuickStart.html uses DGSource for the newcomer, and I might've guessed you'd tout instead DGMultiSource since it's simpler and you've said yourself that you use it instead of the DGSource/CUVIDServer combo. In any case once again...

[MrVideo]

Originally Posted by neuron2

What do you mean by "the old software method"?

Running DGAVCIndex or DGIndex, depending on whether the source was H.164 or MPEG-2.

[MrVideo]

Originally Posted by MuLTiTaSK

your GPU might be borked and it probably cant take the load so it's actually slower then DGIndex
are you using PureVideo Deinterlacer?

Borked? It is a brand new PNY GT240 VP4 card.

I do not use the GUI, I have a script and do everything via CLI. For DGIndexNV, all I do is provide it the input file and name of the output file.

[MuLTiTaSK]

@MrVideo

NVIDIA’s GeForce GT 240: The Card That Doesn't Matter

[MrVideo]

Originally Posted by MuLTiTaSK

NVIDIA’s GeForce GT 240: The Card That Doesn't Matter

Didn't see that before I ordered the card. Never expected a newer card to get worse reviews than earlier cards.

Even so. If another card was twice as fast, the result would still be slower than using DGAVCIndex/DGIndex on a 1080i file.

[MuLTiTaSK]

@MrVideo

download GPU-Z and post a screenshot of the GPU's sensors when using DGIndexNV

[InsulinJunkie]

Originally Posted by MuLTiTaSK

NVIDIA’s GeForce GT 240: The Card That Doesn't Matter

Game performance is what the review sites - and most of their readers - care about (and the GT240 *IS* overpriced from that perspective if you pay full MSRP, although prices have started to drop the last week or so), so they don't really dive into the video part of things.

A DDR5 card will have more than adequate memory bandwidth, and the separate video hardware chip (the 'VP4' end of things) is what primarily matters anyways.

[MrVideo]

Originally Posted by InsulinJunkie

Game performance is what the review sites - and most of their readers - care about (and the GT240 *IS* overpriced from that perspective if you pay full MSRP, although prices have started to drop the last week or so), so they don't really dive into the video part of things.

Ya, I read on NewEgg that as a gamer card it basically sucks. I don't do games, only video work.

[MrVideo]

Originally Posted by MuLTiTaSK

download GPU-Z and post a screenshot of the GPU's sensors when using DGIndexNV

OK, here ya go.

[MrVideo]

Originally Posted by MrVideo

I don't do games, only video work.

That said, I do tend to play solitaire while waiting for some jobs to finish.

I've noticed a definate slowdown in the drawing of the cards. When it does a nea deal, it is slower. When it turns over three cards is it slower. What really hit me is when I had a lot of cards on the playing surface and when I had a winning hand, I hit the right mouse button to aotumatically place the card on the four piles and I could actually see the cards building up on the piles. Even with the old 7600 graphics card it was instant. I'd hit the rodent and the cards were there. Not anymore. Strange.

[an3k]

Originally Posted by MuLTiTaSK

@MrVideo

NVIDIA’s GeForce GT 240: The Card That Doesn't Matter

They didn't used the "best" available GT240, there is at least one with 1GB GDDR5 for ~90 euro.

GTX295 has VP2, GT240 has VP5 (well, officially its called VP4 but the GT240 supports AVC-MVP which comes with VP5, thus GT240 is VP5). See http://en.wikipedia.org/wiki/Nvidia_PureVideo

I've compared both of my cards. I used The Dark Knight BluRay with DGIndexNV Build 2000, default settings (Enable Display, no Deinterlace, save project, no demux).
There was no noteworthy difference in speed using that BluRay since both cards can handle the source completely on the GPU.
If you use source with newer features, the GT240 is faster since it supports more features directly on the GPU!

Successive VPs have focused on adding support for additional video formats. VP2 had full H.264 decoding, and VP3 (which never made it into a GTX 200 series part) added VC-1 decoding. For VP4, NVIDIA has added support for full decoding of MPEG-4 (Advanced) Simple Profile, better known as DivX or XviD. With this addition, NVIDIA can now offload the decoding of most of the MPEG formats – the only thing not supported is MPEG-1, which as the oldest codec is trivial to decode on a CPU anyhow.

Gaming: GTX295
Video: GT240

[MrVideo]

When I started en editing job with VideoReDo and it said that it wanted to take 50 minutes to do an 8 minute job, I knew something had to be wrong.

THe indexer was slow, the card playing was slow and now editing a video was slow.

So, I did the normal Windblows thing and that was to reboot the system.

That fixed ALL of the timing issues. The indexer is now down to under two minutes (for the 45 minute H.264 file), the video editor only took the expected time and the cards moved as before.

The dumb thing is I have no clue as to what screwed up the system. There was no runaway process. The CPU was not peaked out and memory was not eaten up. The system was rebooted after Nvidia did its install of the video drivers.

But, it seems that the system needed another reboot after Nvidia finished doing what it did after the reboot.