Popmann - the reason this thread is focussed on the graphics card (rather than HD issues) is that there is a definite link between an open VI GUI and pops & glitches - so it must be a graphics issue (of some sort).

I agree with you that it's unlikely to a VGA shadowing issue - because RAM is not the problem (only 550 MB after optimizing).

It's also not a soundcard issue either, because it still happens with a buffer size of 1024!

Frankly, I'm really pinning my hopes on simply replacing the graphics card - perhaps naively.

Would a 128MB 8x AGP suffice?

Many thanks,
Mike.

AGP--whatever max speed your mobo supports. It will autodetect. Usually the cards are backwards compatible...so, if you get one faster than you mobo, it will just kick it own in speed, which is fine.

the speed of AGP and video RAM and what not is not really that useful. You just want it off the PCI bus so it doesn't get in the way. I don't think there have been significant 2d performance gains in video cards in years. Go cheap, IMO. stick with ATI or nVidia chips, though.

not knowing which motherboard it is i thought i'll mention AGP versions, since AGP 2.0 (4x or higher) voltage is 1,5 V, below it is 3,3 V - both versions are not compatible and not all have the correct notches to prevent mismatch.

AGP 4x - 8x are compatible (in both directions) but i've not found a reliabl compatibility matrix for 16x - 8x - one should assume it's similar there.
christian

OK - finally, here are my specs - sorry for the delay.

Intel Pentium 4 CPU 3.00 GHz (Socket 478 mPGA)
Bus speed 199.5 MHz
Rated FSB: 798.0 MHz
Cores: 1
Threads: 2
L1 data cache: 8 KBytes
Trace cache: 12 kuops
L2 cache: 512 kbytes
Motherboard model: Springdale PE
Chipset: Intel i865P/PE/G/i848P rev.A2
Southbridge: Intel 82801EB (ICH5)
Sensor: Winbond W83627THF
Graphic interface: AGP version 3.0
Transfer rate: 8x
Max supported: 8x
Side band: enabled
NVidia GeForce4 MX 4000 (Location: PCI bus 1, device 0, function 0)
Memory: dual DDR 2gig (Performance mode: disabled
Intel 82801 Ultra ATA Storage Controllers (location: PCI bus 0, device 31, function 1)
Boot drive: 7200rpm Seagate IDE (Location: 0 (0))
VSL drive: 7200 Seagate IDE (Location: 1 (1))
LAN card: Realtek Gigabit Ethernet NIC (PCI bus 2, device 0, function 0)
Soundcard: MAudio Delta 2496 (PCI bus 2, device 1, function 2)
Midi-Over-Lan3

I have changed the PCI latency settings to the following:
Bus1 (GeForce4 MX4000) from 248 to 32
Bus2 (Gigabit Ethernet adaptor) to 32
Bus3 (ICE1712Envy24 PCI Multi-channel I/O Controller) from 32 to 248 (I assume this Bus is the one carrying the audio streaming, hence the highest setting).

With 6 instances of VI playing moderate tempo perf-leg instruments, I get glitches at buffer setting 512. Fewer glitches at 1024, but still there.
If I add one more instrument, it goes really crazy.
RAM loaded is only 600 gigs (after optimizing).

Am I asking too much from this system?
If so, what do I need to upgrade?

Thanks,
Mike.

i think i've found your motherboard, it is possibly a GA-8ISXT in a fujitsu computer.
http://vsl.co.at/upload/users/449/SpringdalePE.png">
if we talk about this board (or a similar one) we have to play the *find the right PCI-slot* game, because the respective manual does not provide an IRQ-table (meaning: which slot is using which IRQ and which of them are shared, possibly actually with onboard devices)

common IRQ routings on such boards are: AGP is shared with PCI1 and PCI2 with PCI5 - this means no bandwidth sensible devices should sit in the respective slots if the correlating one also holds a sensible device.
tell us in which slot which card is sitting for further suggestions.

in any case an AGP card will improve the overall performance on your system (this board supports 4x 1,3 V as well as 8x 0,8 V ... try to get a simple ATI card, they have the most reliable drivers.

in a second step you need to place your delta into an unshared PCI slot (we have to find out which would be the best) - make sure in your BIOS IRQ assignement is set to *auto*, everything else makes things worse usually. possibly the problem disappears already with replacing the PCI VGA with an AGP version.
christian

ps: please keep also the PCI latency settings at their default value - this kind of tweaking is a science by its own ....

Thanks very much Christian [:)]

I will look for an ATI AGP card today.

I will post the PCI slot allocations later today

I will not change the PCI latency settings (they reset to default at boot up anyway).

Appreciated,
Mike.

I can find an ASUS N6200 128MB (8x).

is this OK?

Mike.

Here is how my PCI slots are allocated:

AGP slot (brown) - graphics card (soon to be ASUS N6200 if you approve)

PCI Slot 1 - Gigabit network card
PCI slot 2 - Delta audio card
PCI slot 3 - empty
PCI slot 4 - empty
PCI slot 5 - empty (this slot is furthest away from the AGP slot)

BIOS IRQ assignments set to "AUTO".

Thanks,
Mike.

[*-)] Just got back from the PC retailer. He says I must already have an AGP card - because a PCI card will not fit into an AGP slot.

To be fair, I did not have the PC with me for him to look at when he said this.

My current GeForce card is plugged into the brown AGP slot.

Can you please clarify whether the GeForce4 MX 4000 is an AGP card?

Thanks.
Mike.

That's just a graphics chip...not the model of a card. But, I would assume they're correct. Literally--it's HARD to find a PCI video card. If you bought it new in the last few years, it's AGP.

So, that's not the issue. However...your PCI config may be.

On my mobo, the first PCI slot shares resources with the AGP. In general, I wouldn't put anything in there unless you absolutely had to. See if you pull the network card all together if it doesn't resolve the issue.

Thanks Popmann

1.Can't pull the network card because this P4 serves as a slave, and all midi data is via Midi-Over-Lan.
With that in mind, which slot would you recommend for the NIC?

2.Is there some way to see which slots share resources?

Sorry for the misleading info re the graphics card - I'm on the learning-curve from Hell.

Mike.

i should have seen this, of course an MX4000 is AGP ... it actually says PCI bus 1, whereas the other devices are on bus 0 and 2 ... although the N6200 should be slightly better, there is no reason to change it ...

also the motherboard must be different from the one i've found, because i don't see an onboard ethernet now (model is often printed beneath the RAM slots or between the PCI slots, sometimes near the edges of the board) - as long as we don't know the model we have to somehow grope in the dark ... (hopefully it is not an MSI board, i found them to be very poorly designed)

in any case you should move your gigabit card to another slot - my feeling says the network card should sit in slot 2 or 3 and your soundcard in slot 4 or 5

christian

CAUTION: always remove main power cable before changing cards in slots - a screw might accidently fall on the board and destroy parts leading standby voltage.

OK here are some test results.

I installed Cubase on this P4 so that I could test without using the network card, and used the offending midi sequence from the Host PC (ie. using the P4 as a host, rather than a slave).

a) With a buffer setting of 512, the Delta card played flawlessly on PCI slots 2, 4 and 5. Slots 1 and 3 broke up.

b) With the Gigabit card connected to PCI slot 3 (but no network activity), playback was still flawless.

c) When I play the sequence from the Host PC to the slave P4 viaMidiOverLan, bad audio breakup occurs.

So the problem seems to be caused by the Network (Realtek RTL8169/8110 Family Gigabit Ethernet).

I'm only using 6 midi channels with modwheel data, slow note tempo, nothing frantic - should be a piece of cake for the network to handle.

Any suggestions?

Thanks,
Mike.

Is there any audio "sync" going on? Word clock? Are you sure the break up is NOT digital misclocking? Clicky buzzy "breakup"?

I've got no experience wiht MIDIoverLAN. Know what it is...but, since I've found USB MIDI interfaces to suck compared to old serial boxes...I just figured MOL would just be another step in the direction of problems. [;)]

If 1 and 3 broke up...don't put ANYTHING THERE--probably shares with the USB controller or something else on the motherboard.

There's a physicality to the resource assignments...(based on what you've posted), I'd try the audio in slot 2 and the network in 4 or 5. I actually miss having ISA slots for network cards. You could always manually assign those resources elsewhere if they got in the way. Oh well...give that a shot.

My Maudio card has issues as long as my wireless Linksys card is in there. Never had an issue with the 3com NIC. FWIW.

Hi Popmann

1.No digital clocking issues with the slave P4s. Audio is hardwired back to the Host with balanced audio cables, no sync at all. This breakup is more of an audio-dropout type of breakup than the classic clocking clicks & pops.

2.I will swap the audio card to PCI slot 2, and NIC to slot 5 as you suggest.
Yes, I will put NOTHING in slots 1 and 3.

3.MOL works very well indeed. I've used it for years without problems. I've only had problems since I started using Vienna Instruments on the slave P4. I think the streaming demands of the perf leg instruments are testing the limits of the P4, with the NIC pushing it just over the edge. It's not a RAM issue, more of a buss bottleneck problem (I think).

I'm hoping that a higher-spec NIC will somehow solve this.

The significant issue here is that the P4 can handle the sequence fine on it's own, but not via the network.

Keep the ideas coming! I've got a studio to run... [:(]

No luck.

Put audio card in slot 2, NIC in slot 5.

Same breakup.

[:'(]

Plays fine without the network though...

years ago i've changed all network cards to 3Com (for performance reasons) 3C996B and 3C940, the newer ones are displaying broadcom (as newer intel NICs do btw), and we are using also 3Com gigabit switches (unmanaged) for all networks.

you should in any case look for updated drivers fro your realtec card (possibly even a firmware update) and please have another look for a label on your motherboard - without the IRQ table we have 200 possible combinations to equip the PCI slots ...
christian

Thank you Christian.

1.I will definitely try a driver/firmware update for the Realtek NIC.

2.First thing tomorrow, I will renew my efforts to find the label for this motherboard.

Mike.

This is all I can find on the mobo (close to the memory chips):

865PE-A Rev 2.0
15-H70-012004

Hope this is helpful.

thanks,
Mike.