PDA

View Full Version : Nforce 4 problems, What are they?


eMaRe
04-03-2007, 09:57 PM
I've heard that the Nforce 4 chipset gives audio users problems. Does anyone know the specifics on this? I've got an Asus A8N premium with that chipset and I'm getting some issues and wondering if that's the cause.

pipelineaudio
04-04-2007, 12:11 AM
allegedly nforce 4 can be problematic with some audio interface companies' firewire drivers

Im running nforce 4 and FLYING

eMaRe
04-04-2007, 05:33 AM
I did some research and I saw some accounts of pci problems with creative cards, because the pci buss was being made to wait to be adressed. I have an EMU 1820M (which is Creative) and am having lots of dropouts at low latencies. The weird thing is that the CPU level is still at 30% or so. Shouldn't the CPU max out before I need to raise my ASIO buffer level?

pipelineaudio
04-04-2007, 10:09 PM
Shouldn't the CPU max out before I need to raise my ASIO buffer level?

Apparently not. When subtle little changes are made to reaper, sometimes really ridiculously low latencies can go from working to making little bits of trouble. Poor justin gets an earful of my buggery as I see trouble even at like 5 % cpu, but he always gets it worked out.

JBM is also using your card I think, we should see what his setup is

scottdru
04-04-2007, 11:46 PM
It's actually PCI bus hogging by video cards that seems to be the greater problem with the Nforce 4, and the seemingly unexplained crackling and popping you are getting is a typical symptom of this problem.

Some more info on this here:

http://en.wikipedia.org/wiki/NForce4

http://www.rme-audio.com/english/techinfo/nforce4_tests.htm

What video card are you using? Sometimes high performance graphics cards can have a higher default PCI latency, or indeed may force that higher latency even when rendering lower demand 2-D stuff, which can cause conflicts with other cards on the PCI bus that are operating at a lower latency. Some of the Radeon video cards, as well as some NVidia cards, are a bit notorious for this kind of issue.

This might be worth looking into if you have such a video card, in which case you may want to think about getting a less high performance graphics card (for example, Matrox Millenium G550 or something). Or you might be able to adjust the PCI latency of all cards on the PCI bus to match, which MAY be able to at least help improve the situation caused by the Nforce 4 chipset issues.

I posted some links to resources on this a while back here:

http://www.cockos.com/forum/showthread.php?p=48319&highlight=latency#post48319

There is a freeware application called DoubleDawg that enables you to adjust the PCI latency settings of all the cards on your PCI bus. That may be worth trying, to see if it will help.

http://mark-knutson.com/t3/_index.html

I haven't worked with that particular chipset, so I don't know whether that will fix the problem entirely if it is related to the less reliable PCI latency timing that has been noted with the Nforce 4. But it certainly couldn't hurt to try this in the event your video card is indeed hogging the PCI bus.

eMaRe
04-05-2007, 08:35 PM
Thanks a ton scottdru! I will check your links.

I have an old PCI (not PCI-e) 64 mb GeForce MX440 twin head card. Not exactly high performance. But my soundcard is PCI as well. Plus I do get more glitches when I maximize a window, particularly when maximizing and minimizing is animated. So it sounds like this might be it. Perhaps I could be getting much better latency if I worked this out. It bugs me because I use Guitar Rig 2 and Amplitube, which are hard to use with high latencies!

scottdru
04-05-2007, 10:54 PM
I had a GeForce MX440 card as well at one point, and it did give me some of those kinds of problems occasionally (especially when I still had an old Fraudigy card, which I used mainly for MIDI stuff along side my RME Digi 96/8 PAD, which didn't have MIDI I/O, and didn't offer as low latency as the Fraudigy did). That said, I didn't know to check and/or fix PCI latency back then, so it might have been fine if I had done this.

I still have the RME, but I use my 1820M for 99.999% of things these days.

Where are you in NYC? (I'm in Brooklyn, near the Williamsberg/Bushwick border.) I actually have an extra Matrox Millenium G550 laying around somewhere that I'm not going to be using for a while. If you aren't able to fix this by adjusting the latency with DoubleDawg, I might be willing to loan you my extra G550 in case you want to try it to see if it helps. (Assuming I can find it!)

That said . . . does your motherboard have an AGP slot? The G550 I have is an AGP card.

Either way, try DoubleDawg first, to see what your PCI latency is for both the video card and for the Emu 1820M. Another thing you might want to check is to make sure that the two PCI cards are not installed in slots that are on a shared bus or interrupt. You'll have to check the manual for your motherboard for this. But I suspect this is probably NOT the case, because it's usually the dedicated video (AGP or PCI-e) that would be sharing with one of the other PCI slots.

Whatever the case, I can tell you that the G550 plays nice with the 1820M straight out the box, without having to adjust latency, because they have the same default latency (64 clock cycles, if memory serves, or maybe it was 32). It's great for my music partition (I have a multiboot setup with a dedicated bootable artition for my DAW), though sometimes I wish I had a bit more hardware grunt for video when I'm asking it to do other more demanding video stuff in my general use partition.

Another nice thing about the G550 video cards is that they have a passive heat sink -- no fan. I don't know about yours, but the fan on my GeForce was a loud, whiney little bastard. It was the loudest thing on my computer, even before I updated to Zalman heat sinks for the North Bridge and CPU, etc.

Jason Brian Merrill
04-06-2007, 05:03 AM
JBM is also using your card I think, we should see what his setup is

My setup is an antiquated p4 1.7, nothing like his.

no relevance, sorry :)

Jim Roseberry
04-06-2007, 01:36 PM
I've heard that the Nforce 4 chipset gives audio users problems. Does anyone know the specifics on this? I've got an Asus A8N premium with that chipset and I'm getting some issues and wondering if that's the cause.

NF4/AMD/PCIe is the problem child.
Poor performance... and forget using UAD1 cards.
Uli 1689 chipset mobos are the way to go as a replacement.

NF4/Intel seems to work fine...


Jim Roseberry
www.studiocat.com
jim@studiocat.com

scottdru
04-06-2007, 02:51 PM
NF4/AMD/PCIe is the problem child.
Poor performance... and forget using UAD1 cards.
Uli 1689 chipset mobos are the way to go as a replacement.

NF4/Intel seems to work fine...

]

Then the problem must be PCI latency of the MX440 card.

Just for my own curiosity, I actually just did a quick google search on "GeForce MX440 PCI latency", and actually the first link to show up is the site that I linked to for DoubleDawg.

Apparently, the GeForce MX440 forces a PCI latency timer value of 248 cycles by default. The Emu 1820M adopts the motherboard's default timer value (which may be something closer to 64 clock cycles, for example). So this is most likely where the conflict is, especially given the symptoms described. If you can reset the MX440 to a lower latency setting, that would be the way to go, if you want to be able to retain the best low latency from the 1820M.

The Matrox G550 (or even the G400 or G450) will use the motherboard's default latency timer value. My 1820M and my G550 are both running happily at 64 clock cycles right out the box, with no adjustments.

If you need something a little more up to date, the newer Matrox Parhelia cards will also run at the motherboard's default latency setting, and should be a good choice for running in DAWs with pretty much any sound cards.

eMaRe
04-06-2007, 04:09 PM
There's got to be a higher performance, and less expensive card I can get that will work.

Is there any patch that Nvidia has provided for this? I will search for these things and try Double Dawg as well.

And wait! Justin I believe had a similar system to me, with the same X2 4400+ processor. Justin, if you get a chance, could you tell me what chipset was in that computer?

pipelineaudio
04-08-2007, 01:27 AM
NF4/AMD/PCIe is the problem child.
Poor performance... and forget using UAD1 cards.
Uli 1689 chipset mobos are the way to go as a replacement.

NF4/Intel seems to work fine...


Jim Roseberry
www.studiocat.com
jim@studiocat.com

What about all the super performance tyan s2985 dual opteron systems? Arent they nf4?

I got one and I swear its nf4

ChrisADK
04-08-2007, 08:40 AM
What about all the super performance tyan s2985 dual opteron systems? Arent they nf4?

I got one and I swear its nf4

The Tyan s2895 uses the Nforce 4 Pro chipset and is integrated in to the motherboard design quite abit differently than the single socket boards using the NForce 4 Ultra and SLI chipsets.

The S2895 uses these chipsets
AMD-8131 HyperTransport PCI-X Tunnel
http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_4699_7980%5E4741%5E4745,00.html
NVIDIA nForce Professional 2200 (CK8-04)
NVIDIA nForce Professional 2050 (I/O-4)
http://www.nvidia.com/page/pg_20050120743342.html

The PCI-e and 32 bit PCI slot are on the Nfooce 4 chipset with the PCI having a bus that is shared with some of the onboard devices like firewire.
The AMD 8131 is seperated into to PCI-x busses. 2 slots on the 133mhz and 1 one the 100mhz.

The s2895 does have some UAD performance issues but that is becuase of the AMD 8131 chipset not the Nforce4.

If some one is using a Magma chassis then the issues go away or if they are using PCI-e UAD cards.


AMD 8132 chipset that is in a couple of the Super Micro boards fixed any UAD issues. But sense most al these all the OPteron and X2 and M2 chips are basically EOL and the Pro audio market is such a tiny tiny part of the computer market no one is gonna bother fixing any of these issues.

AMDs biggest asset and also their biggest Achilles Heal is that they let 3rd party people do the chip set controllers. They don't enforce compatibilty enough at the mopment I think. This will hopefully be changing some sense they purchased ATI. All of AMD bad reputation is based on 3rd party chipset makers making short cuts in thier designs.


Chris

ChrisADK
04-08-2007, 08:50 AM
NF4/AMD/PCIe is the problem child.
Poor performance... and forget using UAD1 cards.
Uli 1689 chipset mobos are the way to go as a replacement.

NF4/Intel seems to work fine...


Jim Roseberry
www.studiocat.com
jim@studiocat.com


HI Jim,
I think it may be the a combination of how the motherboard designers intergreted the chipset into thier board deisign and the design of the PCI to PCI-e bridge. Don't forget that PCI-e is still a bridged device they aren't truely independant in the current PCI-e spec. I think that how the mobo makers deal with all this is what maybe cuase some of these issues.

Of course this is all pointless sense the all these chipsets and motherboard and soon processors will be EOL. Which means they will never be fixed. :)

The ULI and ATI based chipsets work far better for pro audio for sure. The funny thing I noticed with all these chip sets is that the better the industry benchmarks the more potential for issues with Pro Audio. LOL


Chris

ChrisADK
04-08-2007, 09:05 AM
There's got to be a higher performance, and less expensive card I can get that will work.

Is there any patch that Nvidia has provided for this? I will search for these things and try Double Dawg as well.

And wait! Justin I believe had a similar system to me, with the same X2 4400+ processor. Justin, if you get a chance, could you tell me what chipset was in that computer?

The double dawg stuff might make a slight difference. You already have a low end video card anything that is a more powerful video card will only make thing worse.

Getting a over priced Matrox PCi-e vidoe card will not help either. The money wasted ona Matrox card you could just get a better motherboard.
I'd recommend getting a ASUS A8r-MVP. They can be had new for around $60.00.
A Matrox P-650 pci-e will run around $250.00

If you get the new mothebroatrd you will be abel to use a much higher end video card with out the issues your having.

Only sort of draw back is that you will have to do a fresh windows install.

Chris

kLyon
04-08-2007, 11:19 AM
It's interesting that I read all of the nForce4 stuff -- after I already had gotten my nForce4 board, of course -- and attributed any problems I had to the chipset. But then, over time, I ceased having any problems. I run big, complex mixes with lots of convolution reverbs, Reason rewired right into the mix (as opposed to rendering to audio), and other vsti's, all at 64 or 128 samples latency. It seems the problems were ameliorated by getting a dual core processor and using a Lynx card. At this point -- and Reaper's efficiency helps, no doubt -- I really have nothing to complain about at all. (Of course, I don't have a UAD card.)

eMaRe
04-08-2007, 08:38 PM
Hey Chris!
Wow! I've read so many great reviews of your ADK computers, it's cool that you're on this forum giving me advice! I will check out that mobo. Isn't it always after the fact that you find these things out! Thanks for all the help! If I had to do it over, I would just get a computer from you!

ashbyrules
12-18-2007, 10:14 AM
PCI-E problems with firewire card. I have a Tyan S2895 and a Nvidia Geforce 7600 GS video card. The driver for the video card is the latest one off of Nvidia's site, version 6.14.11.6375. Other than the Nvidia card which is in PCI-E slot 3, there is a firewire card in the PCI slot which is slot 2. What we are experiencing is that the video display will go fuzzy (lines across the middle and what not) then the system will lock up and you will have to do a hard reboot. Is there a conflict between the firewire card and the PCI-E card? If so, I know you can adjust the IRQ's in the Bios, but what would we adjust it to? If it is a driver issue on the card what would the issue be?

Thanks for any help.

Steve

ChrisADK
12-18-2007, 11:15 AM
Hi Steve,
The video card should be in the top most 16x PCI-e slot which is the one closest to the CPUs.
Do you get the lines and machine lock up if you uninstall the nvidia driver and run with the standard vga driver?

It sounds almost like the video card has bad ram to me.

Thanks Chris


PCI-E problems with firewire card. I have a Tyan S2895 and a Nvidia Geforce 7600 GS video card. The driver for the video card is the latest one off of Nvidia's site, version 6.14.11.6375. Other than the Nvidia card which is in PCI-E slot 3, there is a firewire card in the PCI slot which is slot 2. What we are experiencing is that the video display will go fuzzy (lines across the middle and what not) then the system will lock up and you will have to do a hard reboot. Is there a conflict between the firewire card and the PCI-E card? If so, I know you can adjust the IRQ's in the Bios, but what would we adjust it to? If it is a driver issue on the card what would the issue be?

Thanks for any help.

Steve

ashbyrules
12-18-2007, 03:23 PM
Chris,

I popped it into the first slot and am testing it now. I needed to use the PCI slot that is #2 but can't with the video card in the first slot. If this fixes the issue I will have to look into alternate slots. The reason I mentioned the firewire conflict is that anything we plug in via firewire generates paging errors when Windows goes to do its drive check. Thought the two might be connected.

Steve

Hi Steve,
The video card should be in the top most 16x PCI-e slot which is the one closest to the CPUs.
Do you get the lines and machine lock up if you uninstall the nvidia driver and run with the standard vga driver?

It sounds almost like the video card has bad ram to me.

Thanks Chris