NVIDIA nForce 4 = disaster

[swaaye]

You know, there sure seems to be a lot of love of NVIDIA's mobo chipsets. Well, my joy with this nForce4 really is coming to an end here.

I've been doing some research on Active Armor, and whether it can work at all or not. I also have been having pops and clicks out of my Audigy 2, reminiscent of a PCI Latency issue. What info I've run into looking into these issues is rather shocking and will probably make me not buy another NVIDIA chipset in the future.

http://www.xtremesystems.org/forums/showthread.php?t=67302

All one needs to do is browse the nVidia forums, http://forums.nvidia.com,
to find many alleged problems with the nForce4/AMD chipsets. The following
points are brought into focus:

1. nVidia is completely silent in responding to end user complaints about
defects.
2. The MAC Ethernet controller has issues (NAM, network access
manager/firewall; ACTIVE ARMOR, hardware acceleration).
3. The IDE/SATA controller has issues (shares irq with NAM; data loss).
4. After infrequent new bios releases the problems still exist.
5. It is now more likely that these problems are unfixable hardware bugs.

This one is sorta a tell-all. It brings up bus hogging issues NF4 has between PCIe and PCI, which is probably the cause of Audigy and X-Fi issues on the board. The bus is hogged by the video card (a latency issue) which causes audio problems in games. It doesn't happen in situations where the video card isn't being hit hard.

And, of course, Active Armor is said to not work right at all.

http://forums.nvidia.com/index.php?showtopic=11914&hl=nvtcp
Here we have folks trying to solve corruption issues with the buggy NV Ethernet and NV's Active Armor. One can solve the corruption and BSOD issues by disabling TCP offloading, as reported at Tech Report, but that's hardly the only issue with NAM/AA.

http://www.nforcershq.com/forum/nvidias-intrest-in-fixing-the-active-armor-problem-vt66765.html
This is a many page thread with people trying again to get Active Armor working. This is way more recent than Tech Report's post about NV solving corruption thru disabling TCP offload.

Check out this quote of a reply from NVIDIA's Norman Chen of Networking. This is from last month. Gotta cringe at what he says about marketing being in control. So much for quality drivers.

Hi Jeremy,

This is Norm from NVIDIA and I wanted to get back in touch with you about the corruption issue. I know I definitely dropped the ball on communication here -- I really apologize for that.

Since I last emailed you in May, we went through two releases of nForce5 drivers, and the corruption issue got bumped down low in our priorities. Because our product cycle is short, we get too caught up implementing new features sometimes that we overlook existing issues and support for the older drivers. As an engineer, I am also concerned whenever our products, old or new, have issues and don't work correctly for end users. Yet it's not always easy to work on it because at the same time, our marketing/sales is always pushing newer features. Regardless, I should have given you some heads-up about our progress. Thank you for your worries and for supporting NVIDIA during this time.

The good news now are that we are looking at this issue again and that your previous memory dump was helpful. If you remember, I added some code to trace the HW status/commands in the debug driver I sent you. Well, we were able to spot some traces in the memory dump that correspond to the corruptions in the file. They point to some potential code in the driver where the bug might be, but we need some more info to pinpoint the issue.

So now I would like to send you another debug driver that has more traces to capture the bug. Are you still willing to help us with collecting another trace dump? It will be pretty much the same steps as below.

Thanks,
Norm

Norm and Jeremy go back and forth for a while. Jeremy helps them actually solve a corruption bug, allowing TCP offload to be enabled. I tried out these debug drivers and while corruption seems to be gone, my comp BSODed during the night running uTorrent. Never saw that without Active Armor.

http://forums.nvidia.com/index.php?showtopic=8171&hl=nvtcp
Here's a thread about data corruption reminiscent of the same problem with VIA's 686B southbridge used way back when with KT133 and KT133A.

-----

I've been running this DFI NF4 Ultra-D board for about 8 months now. It works pretty well as long as you don't install NVIDIA's network software (the ethernet driver alone is OK). I gave the Active Armor software a shot last night, figuring it would at least work now that NV has admitted problems and that there have been driver revisions since then.

Not so.

• I tried a download of a ~200MB game demo. Corrupted.
• BSOD after a couple hours of uTorrent.
• The web-based UI locks up the entire system if I try to disable TCP Offload in the Ethernet settings.

More importantly I suppose are the problems I've been having with my Audigy in games like Far Cry, Prey, and Quake 4. There are loud distortions, pops and clicks etc, in the audio very frequently. Honestly it reminds me of VIA's PCI issues with SBLive! way back when. Their PCI implementation had latency and bandwidth issues that only showed up with RAID and sound cards, and of course everyone just blamed Creative etc until a fellow named George Breese released a PCI Latency patch for the chipset and fixed everything. I haven't messed with PCI Latency tools on here yet.

The Xtremesystem's forum post has a link to a company named RME Audio that did some tests with NF4 and came to some disturbing conclusions about its viability for high-end audio work:
http://www.rme-audio.de/english/techinfo/nforce4_tests.htm

At present the NF4 single cpu chipset cannot be recommended for demanding pro DAW work. Multimedia users, recording hobbyists or semi-pro DAW users who use a limited combination of streamed audio tracks, samples and/or VSTis may not face the aforementioned limitations and performance may be as good as on any comparative platform.
.....
It thus so far seems as if the current breed of PCI-E/SLI bus/driver implementation creates artifacts which hog resources off the PCI-E bus as both the FireWire and PCI soundcards communicate and stream audio data via the PCI bus on the PCI-E/SLI equipped mainboards) - seemingly the more demanding PCI-E video cards cause more performance penalty on audio streamed via the PCI bus even when only used in 2D video modes.

Amazing how in almost 2 yrs NVIDIA can't solve the problem of corruption, but with a few weeks with Jeremy they get something accomplished. I feel the NF4 problems are basically being ignored. Active Armor should just be dropped completely from the package because it DOES NOT WORK. On install of the latest driver package I got data corruption and instability. Unless you know not to install the NVIDIA NAM/AA, you are going to be in for a rough ride.

I'm not sure what to think of the Audigy/X-Fi vs. NF4 issue. I find it ridiculous that PCI latency could still be a problem today. But it sure seems that that is the issue here.

I find it amazing that an NV engineer working with a home user can solve a corruption issue with Linux server connections that has been plaguing NF4 Active Armor for almost 2 years. What kind of Q&A do they bother with over at NV? This is unbelievable. And the engineer went incommunicado for something like a month when he was supposed to be getting back to Jeremy. He comes back saying marketing is taking up all his time implementing new features. Nice.

IMO, stay away from NF4 and NF5.

 

[Bobbler]

It works fine for me...

X-Fi, PCI-E vid card, same motherboard.

 

[karlotta]

It works fine for me...

X-Fi, PCI-E vid card, same motherboard.

what does? all the extras? the MB?

 

[INKster]

I have an Audigy2 ZS on an Nforce 4 SLI for well over a year now and i haven't had a single problem with either of them...

There's a PCI TV card (TV-Wonder) in there, too.

 

[swaaye]

Have all of you tried to use the Active Armor firewall? I've heard that some people have no problems with it at all, while others like me have rather terrible issues. The audio issue could be the same way. A lot of these issues could be board quality and/or BIOS related too.

I'm using a DFI NF4 Ultra-D.

 

[Rodéric]

I've had that problem (network corruption) too, when I bought my A8-SLI more than a year ago, it was solved by upgrading both the BIOS and the drivers.
It was quite annoying to get the drivers with a corrupt internet connection, hopefully I had a 3C905C-TX-M around to plug in my comp

Not a problem since then.

 

[Skrying]

I had to trouble shoot a system that couldnt play BF2 without being kicked by Punk Buster, searched about the chipset (which was NForce 4) and sure enough Active Armor had issues with BF2 and Punk Buster on a regular basis. Just reinstalled drivers for chipset with the Active Armor and worked fine.

Active Armor is terrible, the chipset as a whole is great though.

 

[INKster]

I believe that part of the chipset was made in collaboration with 3COM (not sure if it's really 3COM or Broadcom, though ).

Maybe it wasn't so easy to implement such a concept on a mass-produced desktop PC chipset, at least at the time ?

 

[BRiT]

Give up on ever trying to use Active Armor. That will solve a lot of your problems, not to mention return your sanity and prevent the loss of time spent troubleshooting issues as well.

My system works flawlessly with my X-Fi, X1900 XT, TV Tuner.

 

[Martin Eddy]

Active Armour and TCP offload sounds great but causes so many problems.
With those enabled I couldn't share media to my XBOX360 as well as other on again off again problems with file sharing on my home network.

Of course BSOD's whenever downloading Torrents were a pain in the ASS as well.

Since I've disabled them I haven't had a problem at all.

 

[{Sniping}Waste]

I have had problems with NF2, NF3 150 and NF3 250. The ethernet would at times corrupt files when downloading with the NV firewall not installed. Problems with sound cards with pops and cracks. The the worst was the corrupted harddrives on both PATA and SATA using NV ATA drivers. If I used windows drivers for the ATA then the corruption stopped but the speed dropped but damit, the NV ATA should work right in the first place and its been 3 years like this.

I flat out will not use NV chipset mobo any more having to replace mobo in family, friends, and customers so the systems would stop come back. I have found that VIA, SIS, and ATI chipsets are more reliable and stable the NV chipsests. The sad thing is if you want a good overclocking mobo about 95% are NV chipsets but I will buy a non NV chipset mobo and hard mod it before getting a NV chipset mobo.

 

[bloodbob]

Really the activate arm doesn't do much to protect you I find it does a better job protecting everyone else i.e. you can't hax ppl with it enabled.

 

[swaaye]

I have had problems with NF2, NF3 150 and NF3 250. The ethernet would at times corrupt files when downloading with the NV firewall not installed. Problems with sound cards with pops and cracks. The the worst was the corrupted harddrives on both PATA and SATA using NV ATA drivers. If I used windows drivers for the ATA then the corruption stopped but the speed dropped but damit, the NV ATA should work right in the first place and its been 3 years like this.

I flat out will not use NV chipset mobo any more having to replace mobo in family, friends, and customers so the systems would stop come back. I have found that VIA, SIS, and ATI chipsets are more reliable and stable the NV chipsests. The sad thing is if you want a good overclocking mobo about 95% are NV chipsets but I will buy a non NV chipset mobo and hard mod it before getting a NV chipset mobo.

I too have had trouble with NF2 and the NV IDE drivers. I don't install their IDE or SATA drivers at all, because they can cause the strangest issues, and performance increases zilch.

A friend of mine is having probs with his Audigy and a NF2Ultra-based Abit NF7-S 2.0 (one of the most popular NF2U boards). I honestly can't remember if I had probs with my NF2U-based Shuttle AN35 or not. He gets lots of audio problems in Far Cry at some points, like I do on this NF4U. The issues are definitely reminiscent of bus hogging by probably the video card. I've fixed problems like these in the past with PCI Latency Tool (on a notebook). But, c'mon, PCI has been around for over a decade. It's easy to blame Creative, but it's also true that they are the only game in town for hardware accelerated audio that really pushes PCI and needs reliable latencies from the bus.

 

[SugarCoat]

I went through a whole slew of problems that destroyed about 5 windows installs due to active armor. It would even corrupt windows security updates in addition to just about everything else. Dont use it, period. Its funny you should bring up the audio popping as well, as i've had a few games doing it to me. I know for sure its not an issue with the speakers, and music has no problems at all either, but for some reason some games pop like crazy. AOE3 and Rise of Legends are the latest that come to mind that did it badly.


Thankfully the board was cheap so to be honest i dont really care, but i'll be happy when i trash it if it solves remaining issues.

If you're that worried about computer security just invest in some nice third party software and/or a decent router. The only thing i have installed as far as board drivers is quite simply the ethernet driver, thats it.

Intel chipset stability are just worlds better then really everything out there. Truly a shame.

 

[L233]

Has there even been a nForce chipset that didn't have major problems? Over the years there have been problems with PCI latency and PCI throughput, USB problems, X-FI compatibility problems, Audigy compatibility problems, network data corruptions problems, IDE data corruption problems, features not working correctly or not at all... and that shit rarely gets fixed. Not every user suffers from these problems but they are frequent and reproducible, so they're not unique to one particular configuration.

What's even more irritating is that the big hardware test sites never even mention all the crap that's going on with the nForce chipsets. You have forums filled with thousands of postings detailing these problems but hardware sites never investigate, they instead run pointless benchmarks in order to find out which board is 0.2 FPS faster in CS:S.

Mainboard reviews are generally atrocious, not matter if it's Anand or Tom's or whatever. Remember that ULi chipset that used to be Anandtech's darling for a while? It had (and still has) severe memory compatibility issues. Were they even mentioned? Now we have i965-based mainboard reviews popping up all over the net and do they mention that currently most i965-boards have issues with RAM modules using Elpida RAM chips? Until recently, several ATI XPress3200 + ULi SB boards had severe LAN and IDE performance issues. Some of them have been fixed others still exist. Did the reviews mention these problems?

Mainboard reviews should be about RAM compatibility, stability under several kinds of load (CPU, I/O, Bus), network performance, IDE performance and stuff like that and NOT about frigging gaming performance because there' hardly any difference there at all.

 

[Npl]

I went from a Via KT333 Board to a Nforce2 in an attempt to throw out most PCI-Cards. I use the same IDE Devices and Graphiccard as before, but the Nforce2 has serious trouble with IDE whereas the Via just worked.
First, its impossible to burn a CD/DVD without the Burner suspending a couple times, luckily it can fix buffer-underruns.
Second, having any task stressing a harddrive will result in the whole system getting very unresponsive, of course the Via System also slowed down, but it wasnt nearly enough to be annoying. I guess thats the price you have to pay if you want top performance in benchmarks??
Third, changing CD/DVD turns out to take around 30 seconds, while locking apps that wait for the Drive (and with locking I meant that not even the GUI redraws in that App)
Using any IDE-driver from Nvidia result in random bluescreens (hitting "Save" in MS VS2003 turned out to almost always crash, was very painfull to expierience this) while fixing next to nothing.

Standard MS Drivers are the only ones that atleast dont result in random bluescreens, but had the remaining problems
NV-Filter drivers dint fix anything, but introduced random bluescreens.
Nvidia drivers fixed the CD/DVD Lag but are buggier than anything else.

It doesnt looks like this is ever going to be fixed, as the last driver for NForce2 dates back to 2004

 

[swaaye]

Yeah you should NEVER install NVIDIA's IDE/SATA drivers. They offer quite literally zero advantages other than a pretty name and a benchmark in Device Manager, but usually bring with them very odd showstopper issues.

I have had some luck with them myself, but if you have any weird HDD or optical quirks at all it's probably caused by those drivers. Optical drives especially are prone to weirdness.

And, yes, I agree that there has seemingly never been a working nForce. The original chipset had IDE issues, and so does NF4. Rather unbelievable. I'm not sure what chipset to look at. I guess if I were buying a Core 2 Duo I would buy Intel all the way.

That engineer's email and all the nonsense problems with "features" of NF4 gives me a strong vibe that NV is way too fascinated with the business tactic of "value-added features". And honestly that term has disgusted me since I had it jammed down my throat in a college business course. NV doesn't carry thru with quality of the features and they keep things quiet and ignore issues reported even on their own forum. The features are purely for initial market penetration IMO, and after that they are forgotten in favor of the next big thing.

 

[Npl]

I had good expierience with Via, but they arent without fault either and they seem to stagnate since their first K8-Chipset. ATI will be interesting to watch, AMD will go through great lengths to ensure they have a 100% troublefree Chip+CPU Combination for their OEMs.

Im pretty confident my next Chipset wont be from Nvidia though

Glad to see such a thread, it sometimes fells as Im dreaming this all up while the rest of the world keeps running their benchmarks.

 

[Sobek]

Yeah you should NEVER install NVIDIA's IDE/SATA drivers. They offer quite literally zero advantages other than a pretty name and a benchmark in Device Manager, but usually bring with them very odd showstopper issues.

I have had some luck with them myself, but if you have any weird HDD or optical quirks at all it's probably caused by those drivers. Optical drives especially are prone to weirdness.

Aint that the truth...

I honestly can't remember what happened (it was so confusing and stupid), but my Nvidia IDE/SATA drivers went on a rampage, refusing to install correctly, complaining about insufficient resources...you name it. After about the 6th attempt at reinstalling them after doing a good thorough driverclean, they just magically worked...haven't touched 'em since.

 

[vazel]

Funny i should come across this thread. I've suspected my nf2 chipset has been responsible for a few quirks on my system and this thread certainly doesn't alleviate that suspicion. No more nforce chipsets for me.