Biostar: "if you want to stress the cpu while recording audio - go intel"

[niv]

Hay ,
Here is an interesting hardware/software issue:
Using biostar Ideq 200V (http://http://www.biostar-usa.com/ideqdetails.asp?model=ideq+200v) with windows 2000, I installed a winTV card (http://www.hauppauge.com/Pages/products/data_gofm.html)to use it as a Tivo. I connected the output of the TV card to the line in of the onboard sound card.

listening to the audio I hear that the sound level goes down ever so often - then comes back.

I RMAed the board to Biostar, and they said that it may be due to the lack of hardware sound encoding in via chipset. So when encoding sound via soft - cpu high usage bursts may cause sound problems.

The rep sugestted upgrading the board to an Intel one wich has a hardware encoder for 50$.

So I will have to buy a new Intel cpu. -> 190$ or more :-(

any ideas, that may help? does it happen Linux?

10x,
Niv

[ppera2]

You don't need hardware encoder in chipset for audio. What Biostar says is complete BS. Audio encoding by capture is only fragment of CPU usage in compare to video encoding.
Additionally, it should not change sound level, it should make distorsion, interrupts, maybe A/V async.

Of course much depends from used software, codec.

I suggest you that try some other capturing software, newer audio and capture drivers. If problem still appears, it's likely that board has some failure.
Typical reaction - instead admit problems they offer you something expensiver. It's maybe just theory, but I would avoid that brand after such 'expertize' .

Btw. I make perfect captures with AMD XP 1700 on VIA board - Mpeg2 video and MP2 audio, full PAL res (720x576).

[niv]

is your cpu load high during the capture?

[niv]

may I pls have the motherboard you are using?

[lark]

no, but i'm capturing to huffyuv with wav audio.
but sure you'll get high CPU utilization, if you use high resolution and compress that while capturing.

btw you could have used a bit more descriptive subject ;-)

regards
t :)

[niv]

sorry , wrong thread.
mod pls kill

[Arachnotron]

I agree with ppera2 that Biostar is talking nonsense. But even if they were correct, the cheapest option to gain 'hardware' sound would be to plug in a PCI soundcard and switch of the on-board AC97 sounddchip. Since the quality of on-boards is often so-so, any cheap soundcard would be an improvement. And certainly be much cheaper than replacing both motherboard and CPU.....

By the way, most stories about needing a heavvy PC for capturing are from the times the P-II was king and 5400 rpm harddisks were considered fast. Any new off-the shelf PC is already way overkill as far as capturing is concerned. It's the editing/filtering part afterwards that still eats up CPU cycles...

[jggimi]

I've merged the two threads together rather than killing the inadvertantly separate "makes sense..." thread. They merge based upon timestamp, so its a little out of order, but is still readable.

[homermaster]

I can see two possibilities.
1. The soundcard has an audio compressor to limit incomming sound and avoid clipping (not likely).
2. The winTV card has an audio compressor to limit incoming sound and avoid clipping. This is quite a common feature on standalone tv's and is even employed by some cable companies to regulate audio. You know ... when you are watching a show and the commercials come on blaring ... this is a tool that tries to keep everything at the same level.

I don't think biostar has anything to do with this problem. I would look into seeing if there is an option to disable the audio compressor on the winTV card. I looked through their website FAQ http://www.hauppauge.com/Pages/faq/support_faq_pci.html and manual http://hauppauge.lightpath.net/manuals/wintv-man-eng-v40.pdf and the only possibility I saw was page 3-3 of the manual refers to a "Pref button" and an Audio subheading that says "your audio settings are configured here."

Another option .... return it and buy a different card. If your first post relatively coincides with your purchase you should have no trouble doing this. However, I would try reading reviews of products you are looking at buying to see if there is a compressor before buying it.

[niv]

Again I am happy to see ppl are knowladgeable and will to lend a hand.
I would submit that I used the same card on two other PCs and they didn't show that problem AT ALL.
One was a Gigabyte i845/1.8Ghz and the other a MSI KT133.
That is way I thought it was Biostar fault.
As this problem shows only when stressing the CPU while recording, I do think its the main board.

however I will read into it, as you sugested,
Niv

[homermaster]

This is a very interesting problem if it was not present on other systems. Maybe a little info from you could help me understand better.

1. Regarding volume
a. how long does the drop last and it consistent?
b. does it return to normal gradually or suddenly?
c. does the change seem to happen at random times or when the volume spikes?
2. What compression are you using for encoding the audio and video?
3. Are you still using the 1.8 Ghz processor?
4. How much and how fast is your ram?
5. I'm not sure if win 2000 has the ability to view CPU usage like xp but if you can, can you view CPU usage and correlate drops in volume to CPU usage spikes?

I looked into some of the hardware on the mainboard (chipset http://www.via.com.tw/en/products/chipsets/k7-series/km400/ and sound card http://www.cmedia.com.tw/product/CMI9739.htm) from Biostars ideq 200v site http://www.biostar.com.tw/products/barebone/ideq/200v/index.php3 and I have a hard time believing that a lack of a hardware audio encoder would cause CPU usage spikes great enough to cause this anomoly. The sound card isn't anything special but it is 6 channel capable. With the chipset I looked at some reviews and nothing was said about this issue (not surprising as it is a fairly uncommon set of circumstances). I also went to another forum http://forums.sudhian.com/categories.aspx?catid=96&entercat=y that was recommended by one of the reviewers but I couldn't find anything like this either (but thought you might like to have it for future reference). :)

I saw your post at viaarena and agree with his reply that the newest drivers are a must. As for his recommendation of the PCI latency patch I would keep it as a very last option as it adresses IDE and sound blaster issues and you would most likely have other problems such as HD/CD-ROM read/write failure or file corruption and not just your problem.

At this point I would look for things that could be causing a bottleneck.
The cheapest solution that I can think of is, as Arachnotron stated, a new sound card. However, your only option with this is an external USB card as your tv card is in the only pci slot.
If it is really an issue of CPU spikes, which I doubt as I agree with ppera2 that it would manifest in other ways than a change in volume, you could try to get a faster CPU. It says that you can get up to an Athlon XP 3000+ boosting you up to 2.17 Ghz. Not a huge jump but for $100 at pricewatch.com it wouldn't hurt to upgrade anyway.

[niv]

Answers:
Originally posted by homermaster

1. Regarding volume
a. how long does the drop last and it consistent?
b. does it return to normal gradually or suddenly?
c. does the change seem to happen at random times or when the volume spikes?

sound sample (http://hl2.bgu.ac.il/users/www/34332/200v_snd.mp3)

Originally posted by homermaster

2. What compression are you using for encoding the audio and video?

MJPEG/WAV mono

Originally posted by homermaster

3. Are you still using the 1.8 Ghz processor?

Yes. infact I just got this:

---- Biostar tech support email quote ----
Checking with Alfred, our head technician, on his testing results. No problems found.

Configuration

Your motherboard P4VBA
2 x 512 Meg Corsair DDR400
Sempron 2200+
Flamenco PCI Win TV card
Windows 2000 sp/4
Drivers are from the CD that comes with the IDEQ units.

Basically ran SuperPI (memory benchmark program so the cpu is always at 100%)
Used the line to record audio for 2 1/2 minute. No fluctuations of any kind.
---- end of email ---

Originally posted by homermaster

4. How much and how fast is your ram?


512Mb PC3200

[homermaster]

The audio clip helps immensely. It sounds like the work of a compressor as described in my first post but definitely does not behave like one. I loaded the clip into Nero Wave Editor as I find it more objective to have something to look at and not just listen to.
Seeing this confirmed my suspicions from listening to the clip. The attack of the compression is sudden but not immediate (about .5 sec.) while the decay is more gradual in its release (about 3 sec.), which is exactly how a compressor works, but it dosen't follow any pattern a normal compressor would. While listening, I noticed compression occurred at quieter parts which will not happen with a compressor as its entire design is to make loud things quieter. I would not completely rule this possibility out but I think this evidence is pointing somewhere else.

I also am still having trouble understanding how an overtaxed CPU could be at fault. If volume changes were the result of spiking the CPU I would expect immediate changes both down and back up with a time interval that corresponds to the length the CPU was spiked. You are simply capturing a wav and not capturing and compressing into an audio format that would require much from the CPU. Plus the results from Biostar tech support showing no problems while running SuperPI (a program that calculates pi http://files.extremeoverclocking.com/file.php?f=36) and recording.
The best way I can see to rule this possibility out (and I am really hoping we can) is to view CPU usage while capturing audio and see if spikes in CPU usage correspond to when the audio compression happens. As stated in prevoius post I'm not sure if win2000 has this ability. If not here is a freeware one http://www.coolbeans.ws/cpumeter.shtml If they do correspond that means this is still a possibility, if they don't correspond this possibility can definitely be ruled out.

Another test that could be helpful is if you can capture the same material twice through a DVD player or VCR connected to the winTV card and see if the results are the same.

And another couple of questions to try to narrow down what could be causing this.
1. What capture software are you using? Did it come with your winTV card?
2. Does this only happen when you capture using this software or is it consistent with other programs? (Capture a clip from a DVD player or VCR like the above test but change to sound recorder or similar software and capture the clip again to see if problem persists.)

[Arachnotron]

Maybe way off, but could this be caused by the channel frequency being just a tad off? I noticed with another WinTV the channel search function was not that stellar. When I manually inserted the frequencies from a table provided by the cable company, reception improved dramatically for some channels.

If so, captures from a VCR or DVD player should not be affected.

[homermaster]

Thanks Arachnotron for another good question
Was the sound sample above captured from a broadcast signal or from something else? My assumption is that it came from some media you own as it sounds like something you had prepared.

[niv]

The sound sample was captured from the tv card while in FM radio mode. The first time I heard this problem my intuition was to plug my headphones and listen to the audio source as the tv card supplies it. As we know our ears are more sensetive to volume change. I heard no variation in the audio source.
I also pluged another tv card to that board, and got the same resault.

I can see the CPU usage in windows 2000, I also use Cpuidle (http://www.cpuidle.de) to reduce the cpu temp.

[GrofLuigi]

Originally posted by niv
I can see the CPU usage in windows 2000, I also use Cpuidle (http://www.cpuidle.de) to reduce the cpu temp.

That might be your problem. I don't know exactly the 'insides' of this program, but earlier VIA chipsets had problems (like yours, sound related) with CPU cooling flag and PCI bus - that was related to the PCI latency issues mentioned above. PCI Latency Patch (http://www.georgebreese.com/net/software/) tried to address these issues in different versions with different solutions, but everything had to just "fit in".

In newer VIA chipsets (like yours), I think those problems are gone, and VIA was also notified of them and probably incorporated some 'fix' in their drivers, but one never knows...

Basicaly, I'd try without the CpuIdle program. Who knows what it messes with in your system. If you don't have sufficient cooling in your case, nothing can help you.

GL

[homermaster]

I was hoping this was going to be a software issue... usually easier and cheaper to change programs than hardware, especially with how many freeware programs are out there. :)

Kind of late and I've been doing too many of those lately so I didn't read much into Cpuidle (looks like a cool program) but it could be a good contender for being the culprit with the problems they talked about with VIA chipsets in the last paragraph here http://www.cpuidle.de/works.php.

Probably an easy test would be to turn Cpuidle off and try capturing.

Hope it works and hats off to GrofLuigi if it does. If not we'll keep digging.

Just as a side note, I might try it and even use it ... as freeware. However, I have had a 400 Mhz PentiumII that I got second hand from my parents that I just gave away (you don't wan't to know what they had) free of charge because it is so old I would feel bad charging for it. The system has been rock solid since the day I got it back in mid 1999. I don't really think this kind of "preserve your processor" thinking is really necessary. UNLESS you want to overclock. Then cooling and processor life come into play. I now have a 3Ghz P4 that can do two pass XviD encodig of movies using AutoGK in less time than it takes to watch the movie. I feel good about what I paid for it and what it will do, without overclocking. Sorry for you guys that always do it but I just never saw the need. I wanted my systems to do more but worked with what I had and was patient enough for what I wanted to come down to the price I set for myself and bought it with a tax return. I'm not putting anyone down that isn't patient but I plan on this system being as solid as the PII I just gave away.
Not really a side note but oh well. My brother is way into overclocking and he lives by what he can do with it. ;) He won't listen to me and I gotta get my two cents in somewhere. Of course he's probably right as he puts me to shame with tech stuff.

Nephi RULES!!!

[niv]

So it seems that I may be able to resume the problem without sending in my hard disk.
I already sent Biostar word about this new development.
I also found a KM400 in my Uni. Union. It just got a fresh XP instal. I saw no problem recording from the line-in even when bursting CPU usage. I hope to install cpuidle tommorow , and if I can hear the problem I reported.

update: I tried to replicate the problem on a Gigabyte 7vm motherboard which uses the KM400. I tried recording from the line in while stuning the cpu with superPI. no volume shift in the audio recording.
I installed cpuidle, and tried all options to see the problem while bursting the cpu usage.
NOTHING :-(