Droped frames for no reason?!?!

[White Thunder]

Have you enabled DMA?

[Lmano]

Yes, DMA is active for all UDMA drives on my system.

I will have to clock down and perform a series of tests to determin if the problem is O/C related.

I will have the results in about an hour.

I guess this is the crux of the issue.

It will be interesting to see if the O/C is responsible for the problem.



------------------
-=*Lmano*=-

[Dr Mordrid]

The "more humble" setups consisted of variously;

FIC PA-2013 R2
K6-2 300
64 megs PC-100
Win95/Win95-OSR2/Win98

Asus P5A/B
K6-2 400/500
64 megs PC-100
Win98/Win98SE

Abit BH6
C300a/C333/C466
128 megs PC-100
Win95-OSR2/Win98

Asus P3B-F 6/1/1
C466
128 megs PC-100
Win98/Win98SE

These are the storage units that have been passed around between these and the other systems in the lab:

Fasttrak33 w/four DiamondMax 8.4g (2 RAID0)
Fasttrak33 w/four DiamondMax 5.7g (1 RAID0)
Fasttrak33 w/four DiamondMax Plus 10g (1 RAID0)
Fasttrak33 w/two IBM 7,200rpm 18g (1 RAID0)

Capture cards:

RRS
Marvel G200 w/DVD card
Marvel G400

ISA audio:

SB-16
SB-AWE32
SB-AWE64 Gold

Since AVI_IO came out I've used it on just about all of these setups with all but the PA-2013 being able to throw 100,000 frames without a drop. This was likely due to the VIA/Apollo chipsets poor PCI support rearing its head.

More often than not if there were drops they were in the low single digits and mostly invisible because of the copying of the previous frame.

Dr. Mordrid

[Lmano]

After testing both the default 100Mhz FSB and the OverClocked 112Mhz settings on my MoBo I discoverd the following:

The default 100Mhz setting dropped more frames in less time.

The Overclocked setting maintains a 33Mhz PCI clock and doesnt have any adverse timing issues related to running the FSB at 112Mhz. In fact I was able to capture longer periods of time with less frame drop.

Conclusion:
Overclocking the system helps my capture rate and is an advantage over my default settings.


Doc, based on your system info I would say that In order for me to pull this off I will have to have a Raid0 setup! Well I would sure love to have a Raid0 setup and am going to purchase the Promise Ultra DMA66 controller and perform the FastTrack mod on it. I should then have the Ca$h for one more 15Gig Maxtor Diamond Max Plus 7200 2Mb H/D.

With this setup what will my Sustained Data Rate (SDR?) average at? Any idea?

Also I understand that IRQ 11 is used by the Promise card and that SoundBlasterLive (POS IMHO) wont work.
What sound card does work well with the Promise?
Does PCI work well or would ISA reduce PCI overhead?
I use Midi sound Equipment alot and dont want to jump back in performance in the Audio realm yet am not to happy with my OEM SoundBlaster PCI 128.

I would like to have 48Khz Record&Playback and support for AC3 audio etc. etc. suggestions?


------------------
-=*Lmano*=-

[Sciascia]

Lmano

I use the Diamond MX300. Has AC3 support and I also run a un-modded Fasttrak 66 (RAID 0) with no problems. But if you really want to optimize that PCI bus, I would suggest using a good old ISA SC (SB AWE 64). If you've got the extra slot and/or resources you could probably install both (just an idea I came up with) and use the ISA one just for captures. The quality should be good enough and I think that the Doc would agree. I personally don't have any problems though with dropped frames w/ the MJPEG. I didn't have a problem before I bought my RIAD though either.

My system:

Tyan 1830S MB
OC Celeron 450a
256M PC100
13G WD 5400 RPM on MB IDE
2 13 G WD 7200 on RAID0
Promise FT 66
Diamond MX 300 SC
3Com ISA 56k Modem
Marvel G400 (of course)
ISA Scanner card

[Captain Marvel]

Also your source can cause what are considered normal drops...e.g. weak source, to bright, gap between scenes...etc...not a hardware issue.

[Dr Mordrid]

Yes, an ISA audio card for video editing is a major advantage. They use very few resources and this keeps the PCI bus utilization way down.

As for the RAID0 it shouldn't be necessary if you're sticking to MJPeg. I only use one (well, several on assorted systems) because I do a lot of YUY2 and RGB captures.

Yes...the source can really have an adverse affect on dropped frames and the like. Some sources don't generate very good video synch and colorburst signals and this can cause dropped frames and low quality.

This is why I frequently use a video preprocessor inline when using analog sources (Elite Video BVP-4 Plus: ~$750). It restores the synch and colorburst signals to broadcast standards in addition to giving perfect blacks and higly optimized color resolution. It's also great for smooging the signal before sending it to VHS for distribution. VHS needs a LOT of signal to get good copies.

This can also be handled at lower cost by a gadget made by SIMA: the SCC, albeit with a ton fewer features. It runs about $129 at VideoGuys.

Dr. Mordrid


[This message has been edited by Dr Mordrid (edited 24 June 2000).]

[Pytkin]

Hmm .. Everybody finding problems in audio, DMA, hard drives , Chipsets .. etc .. I thing problem is in Zoran compressor .. If Im capturing in MJPEG ... Im everytime dropping frames .. In 800k/s rate or 2M/s or 3M/s .. If I capture sound or not .. I got out Sound card from computer .. no way .. Still Drops .. If I perform an capture test in VirtualDub , than still dropping frames .. This is disk less operation .. Just capturing ..

But if Im capture in UYV2 in 5M/s stream , no drops .. Capture with sound (SB live) .. everything ..

It looks like ZORAN is everytime drop frames and is unusable for real works ..

[Fred H]

Agree with Captain Marvel and Doc.
Weak signal, intermittent signal break etc. causes frame drop and a "perfect" and fast computer doesn't help.
Different capure programs answer with a wide different # of droped frames with weak signals.

[wizz]

Lmano,

I think it is the Maxtor drive. I used to have a Marvel G200 with a dedicated 10 Gig western digital video drive and it worked like a dream, no dropped frames. Then i bought a Marvel G400 and a new drive a 27 Gig Maxtor diamondmaxplus with 2 meg cache (ATA 66) since then no matter what i tweak the system drops frames, i've tried everything except getting an ATA 66 controller just for that drive. I think forcing the drive to run at ATA 33 is giving it some problems, i've just about given up video editing on my system at the moment until i can buy an ATA 66 controller and see if that cures it.

Tony

[Dr Mordrid]

IF it's purely the Zoran chip then why can so many of us capture without drops? Example: my system can capture over 100,000 frames with no drops. So can many other cards that use Zoran chipset.

Software can also be a big issue here. The ability of AVI_IO to capture footage that trips up lesser programs certainly proves this.

You also have to understand the non-standard nature of a large percentage of consumer and low end professional video devices. If I had a dollar for every camcorder, vcr or TV I have 'scoped and found to have a non-standard signal I'd have quite a savings account.

Apparently some companies feel 1 volt peak-to-peak is a guideline and not a standard. Not to mention weak synch pulses and colorburst. This problem is definately non-trivial. Sources created by such hardware were enough of a problem for me that I got the BVP-4 Plus.

And yes, a bad signal can trip up the Zoran chipset. It could also trip up most other chipsets. No drops with YUV signals just means the absense of a chipset to trip up.

Don't get me wrong. The Zoran chipset does have its weaknesses. It's just that in this case you need to step back a bit and take a look at the whole forest.

Dr. Mordrid


[This message has been edited by Dr Mordrid (edited 27 June 2000).]

[Lmano]

I have spent some more time on this and I do think that the Doc, and Wizz/Tony may be on the right track. I think the Maxtor does have a few problems with running at 33 and not 66, but even more of the problem is related to the source as Doc Mordrid and many others point out. I belive that the source is the primary cause of single frame drops at irregular intervalls. I have sat infront of my terminal for quite some time observing and I think that is Exactlly the situation. Since all of my input is analog I can see how this can happen.

After reading the previous posts that pointed to the source I did a little test. Here is what I used and What I came up with:

1 GE VG4275 4 Head Hi-Fi HQ-VCR
1 Magnavox VRU262AT21 4 Head Hi-Fi HQ-VCR
1 Matrox Marvel G400

VHS tapes are horrible to capture and cause a lot of problems if the tape/source arn't in 100% perfect condition. Adjusting the source tracking can reduce/increase dropped frames. Finding the best traking setting for a tape isnt easy, often times the best tracking for one section of the tape isnt the best for other parts.

Also I have to say this, Composite signal is the worst posible signal source to work with. While I dont attribute the frame drops to the composite cables (I tested internal TV for many of my tests) I do have to say that when I work with DV cams and SVHS source the Zoran produces much higher quality results.

I have a short question about the MatroxVidTools application called:
VideoTracker.exe this tool seems to work with the PC-VCR prog but not with the other tools I use to capture. How can I use this tool to its full potential? I would also like to increase the slider range of this tool, only one of the four settings will produce clear frames and since its the last setting I assume it could be better. Perhaps if I change the starting value higher or lower I could get better results.

I would really like to know more about this VideoTracker.exe prog. Doc perhaps you know more about its function?

------------------
-=*Lmano*=-

[Frankie G]

I have to agree on the Maxtor acting squirrely at UDMA 33. I have a 92040U6
20.4MB DiamondMax as my boot drive on a Giga-Byte GA6BXE mobo. The BIOS
detects this drive as UDMA 66, but according to Maxtor this is an issue with the BIOS,
it cannot use UDMA 33 with this drive. So then Windows 98 has a hissy fit
and puts it into DOS compatibility mode. Maxtor has a utility that is supposed
to make the drive ID itself to the BIOS as UDMA 33, but it doesn't seem to work.
Odd thing is, I can go into the BIOS's "Integrated Peripherals" settings
and force the drive to Mode 2 LBA, and Windows 98 is then happy with it,
even turns on the "DMA" check box. I'm not too concerned, as it is
my boot-swap-temp drive and my IBM 16.8 G works fine for capture.

[Frankie G]

I proofread that post but as soon as I hit "Submit" saw
the typo... the mobo _must_ use UDMA 33, it _cannot_ use UDMA66...