PCI bus overflow on upgrade

Sam Glass · Post by **Sam Glass** » Tue Nov 06, 2001 3:18 pm

Hi all:

I've got this project which uses a DSPDev reverb and since I've upgraded from 2.0 to 3.0 the bloody thing keeps PCI overflowing on me. I've sorted a AGP VGA card at last but even with that and a PCI latency of 127 clock cycles I'm still getting overflows.

Does anyone know if theres been some fundemental change in the software which might cause this?

FYI: I've got a Pulsar v1.01 on an Asus P2BS

Cheers gents. I will contact Creamware too but if anyone else has experienced this ...

Post by **garyb** » Wed Nov 07, 2001 12:01 am

somethings screwy.i would go over the optimization tips on the infinite vortex site just to check-see if something has changed on your system......3.0 and 3.01 both work fine on my system.(even with lousy via chipset on asus k7v.)
well, you're probably doing the best thing checking with cw.

<font size=-1>[ This Message was edited by: garyb on 2001-11-07 00:02 ]</font>

Sam Glass · Post by **Sam Glass** » Tue Jan 29, 2002 12:08 pm

Got this last week from Paul Tanti. Not too fucking swift these creamware.com bods! I think this does suggest PCI bus handling was altered in v2 -> v3.

-------

Hi Sam,

sorry for the delay...

we are currently working on 3.1 I do not know if there will be a release
before this one.. ( sorry)

Best regards
Paul Tanti
Technical Support Manager

CreamWare Support
USA and Canada

Gordon Gekko · Post by **Gordon Gekko** » Wed Jan 30, 2002 3:00 am

Same here, since i upgraded to luna version 3.01, pci overflowing madly. will try to tweak it with wpcredit+wpcrset tools, never know

dxl · Post by **dxl** » Wed Jan 30, 2002 4:01 am

the overflow message just happens on me few min ago,
huh....... recently i can't even shut down PC properly faster using pulsar......... maybe it's the overflow thing

why is that creamware??
why that never happen before? but recently??

One2Waziru · Post by **One2Waziru** » Wed Jan 30, 2002 9:21 pm

I have the same problem by adding an xtc card to my pulsar1 on an asusp3b ,PIII 850,512ram,matox marvel agp and matrox milenium pci graphic card for dual sreen.
Pci overflow:no more sound ,swapping between slave and master sound comes back for a second and again pci overflow even with little dsp consuming projects.
I threw out all the cards also the matrox mil.pci,just working with puls1,xtc and matox marvel agp:same problem.
By the way:in cubase the latency has changed from 13 to 25 ms while ulli is on 13?????

dxl · Post by **dxl** » Wed Jan 30, 2002 10:39 pm

give one chance for 3.1
they better do a good job on it
other wise.................

Sam Glass · Post by **Sam Glass** » Wed Feb 06, 2002 11:59 am

hmph.

Post by **garyb** » Wed Feb 06, 2002 11:25 pm

hmmm i'm still working away happily with 3.01......20+ tracks lots o plugs.

Gordon Gekko · Post by **Gordon Gekko** » Thu Feb 07, 2002 1:29 am

Hi gary,

what's your ulli setting, motherboard, cpu? Just curious...

Gordon Gekko · Post by **Gordon Gekko** » Thu Feb 07, 2002 1:30 am

sorry i just saw your motherboard up there but ulli?

digitalboy · Post by **digitalboy** » Wed Feb 13, 2002 3:50 am

The very first thing you should try is to turn off DMA access on your hard drives. (Reply to this message if you need help doing this.) Creamware recommends a bunch of things that, in my experience, don't do anything to address the PCI overflow problem.

I couldn't even use my 3.0 Pulsar II system under Windows XP because of PCI overflow problems. However, as soon as I disabled DMA access, my PCI overflow problem went away. Of course, disabling DMA will drastically reduce the performance of you hard drives (important if you are running a bunch of tracks from disc). However, if you are mainly using your Pulsar setup with the STS series samplers and reverbs, the lack of PCI problems will be worth the hit to your disc performance.

Hope this helps!
Todd

subhuman · Post by **subhuman** » Wed Feb 13, 2002 9:43 am

digitalboy - disabling DMA only works on poorly written IDE drivers such as VIA. On an Intel, the IDE ports do not travel through the PCI bus (the proper way to do a chipset, IMHO), and therefore disabling DMA won't change anything on systems with a chipset like that (i8xx). It's interesting to see that this actually causes less PCI, something I must test with the RD2 next time I get a VIA based system into the lab (might be awhile, I've banned all VIA from the lab for new machines for now

)

Post by **garyb** » Wed Feb 13, 2002 5:29 pm

ulli @13ms (i only have first gen hardware. it won't go no faster)as you may have noticed, i have a crappy via chipset.(asus k7v)

digitalboy · Post by **digitalboy** » Thu Feb 14, 2002 11:10 pm

subhuman - if the DMA issue is a problem with VIA chipsets only, then why are so many people still having PCI overflow problems with their (non-VIA chipset) Creamware systems? It seems to me that if people can't get their Intel chipsets to work even after trying all of the tips recommended on the Creamware site (such as running the PC in "Standard" mode and disabling USB, etc.) then these users are basically screwed. If the DMA tip doesn't apply to non-VIA chipsets, then what else can a person do besides buying a new MOBO? And what guarantee do you have that a new motherboard will correct the problem?

I'd rather have a VIA chipset that will work under certain (less than ideal) conditions than an Intel chipset for which there is NO solution!

DB

subhuman · Post by **subhuman** » Fri Feb 15, 2002 9:26 am

<i>subhuman - if the DMA issue is a problem with VIA chipsets only, then why are so many people still having PCI overflow problems with their (non-VIA chipset) Creamware systems? </i>

Good question(!!!), but on an Intel chipset it is totally possible to fix this, it is 100% configuration, starting from i815 onward, all Intel chipsets share the ICH2 hub design leaving the IDE port NOT sharing PCI with everything else. Even the latest VIA KT266A has the IDE port sharing PCI resources. And now with the newer VIA latency patches, you can get a lot more PCI performance from a VIA, too (still not as good as Intel but usable for making music!) But the 'old' 440BX still has great PCI performance if you follow any of the advice I've given over & over in the past, you can avoid PCI overflows on those, too!

braincell · Post by **braincell** » Fri Feb 15, 2002 10:17 am

I agree with Subhuman. Itel is the way to go for CW cards. I am very happy with my system. I'm using 98Lite and about to try the "sleek" version which combines win95 and win98.

remixme · Post by **remixme** » Sat Feb 16, 2002 4:35 pm

Can someone clear this up for me;what is this w98lite and sleek version?
Whats the diff to w98SE?

dbmac · Post by **dbmac** » Sun Feb 17, 2002 6:15 am

http://www.98lite.net/98lite.html

remixme · Post by **remixme** » Sun Feb 17, 2002 12:32 pm

Thanks