scheherazade

be careful with this board. MANY users (including me) have agp problems with it.

check out the msi furums : http://forum.msi.com.tw/

(search about neo2)
http://forum.msi.com.tw/search.php?s... 0cbe766e4f2ca

agp gart drivers from nvidia break a lot of games. Some users (like me) suffer from a couple-second pauses every few seconds, others suffer from network problems with punk buster (yes it's agp gart related, uninstalling the agp gart driver fixes it)
not using the agp gart driver also makes your video card act like a pci card, having crap pci bandwidth...

check out the issues with this board.
also, there has been no word about anyone fixing the problems. nvidia with their agp driver, or MSi with sme bios or something...

i would suggest the reviewer for this board test farcry and some punk buster using games, and making sure to have the nvidia nforce agp gart drivers installed. you might get an unpleasant surprise.

although the problems (as far as i know, from tracking the issue all over the place) seem to be random. some people only experience problems with certain video cards (problems on both ATI and Nvidia vid cards, randomly... 2 people with the problem can have it work and not work with the same video card), some only when using punk buster, some not, etc. it's very random. but it's _ALWAYS_ fixxed by uninstalling agp gart drivers. and it is a pretty common issue.

if you buy this one it's a crap shot. you might be ok, you might be one of the many unlucky souls...

-scheherazade

p.s. this problem is only when using XP, or certain Linux versions. 2000 users have so far been immune to this problem. *suggesting* it's only driver issues. but from checking out the beta 4.4.0 unreleased agp gart drivers from nvidia, the problem is still there.

jmke

Hey Scheherazade, we are aware of this problem, it was brought to our attention and posted for our readers here : Gainward releases 6600 with 1.6 nanosecond DDR3 memory

There were several other things that were brought to my attention -

* Problem does not occur on non-PB servers
* This problem seems specific to WinXP - this does not happen on Win2k
* Removal of the Nvidia GART Driver clears this problem up
* Reverting to ATI's 3.9 seems to clear this up for some people
* Reducing the AGP aperture size reduces the amount of delay
* Decreasing the PB scan frequency with pb_sleep decreases the frequency of the freezes
* Changing from an ATI to Nvidia card seems to have eliminated problems for many users

__________________

jmke

I've updated the article with the extra info provided by scheherazade

http://www.madshrimps.be/?action=get...&articID=22 8

thank you

__________________

scheherazade

i read your article.

a few other things i've read in other forums.


- reducing FSB *may* remove or reduce the pausing

- for *some people* it happens in single player too (not just PB, my farcry and painkiller pause and go repeatedly)

- for *some people* it happens regardless of PB or non-PB servers

- disabling onboard components of all sorts *may* reduce or remove the problem (depends on the instance)

Note : i read one post about the onboard ac'97->OFF fixing the issue. But to my knowledge only one person reported this.
i personally have tested all-unnecessary-things-off. all onboard off, all ports off, all unused ide controllers off, etc. it shortened the pause durations. i assume it's only bacause of a general lower-stress on the entire system. the benefit isn't really there, so i wouldn't recommend it as a 'fix' of sorta, from my experience.

- removing all other PCI cards *may* reduce the problem

- going into MSI's secret bios feature in corecell section (hit SHIFT+F2; then ALT+F3) and disabling 'Nvidia/ATI speedup' fixes the problem on *some* systems.

Note : this feature is described as a slight overclock to the video card. it [supposedly] happens when you use the automatic overclocking features of the motherboard. i personally don't use it, but other people have documented it.

- people moving from ATI->Nvidia video cards have reported the problem going away

- people moving from Nvidia->ATI video cards have reported the problem going away

Note : one person fixes it by making his system set to how it is set to not work for another person.
Example : 9800xt->6800gt fixed the problem for one person, while 6800gt->9800xt fixed the problem for another person.
*this dispells the notion of incompatability of chipset and certain makers, as the problem differs depending on the instance.

- *some people* with identical systems have reported the problem being there for one person, and not being there for the other

- people have reported that certain versions of nvidia's 4.x.x gart drivers have no problem (i've read of a mention of Nvidia GART 4.05 being 'ok')

- disabling AGP-Fastwrites very often removes the problem (this makes sense, as no AGP GART driver means no fastwrites)

Note : AGP fastwrites for video cards is similar to DMA for hard drives. It allows the CPU to directly address the video card, without having to use main memory as a stepping-stone. not using fastwrites under typical conditions will not affect your framerate. BUT, it is very important in conditions where a lot of new large data is being written to the card. Such as up close multiplayer fighting where a lot of people _start_ firing off weapons. you will see a large drop in framerate (like 60->20) in times like this without fastwrites in new intense games, like doom3. With fastwrites, the drop would be more like 60->45.

- upgrading PSU's has fixxed this problem on certain systems, but is most likely a secondary cause because people with large PSU's report the same problem.

note : i use a (~$200 when i bought it) 550 watt antec, 5v/40a 12v/24a 3v/32a. well above the recommended needs of the bfg 6800 ultra oc i use, or my neo2 plat motherboard. i also tested my psu with a multimeter, and the power output is good. i also use an RS1500 APC UPS, and i have never let a power surge hit my system.

- upgrading bios bersions among neo2 users has not fixxed it. i tried 1.1, 1.2, 1.3. all have the issue. ALTHOUGH, some neo(1) users have mentioned a beta bios making things better.



most the issues i've read about revolve around the MSI boards. THIS IS LIKELY because i have an msi board. when i search for solutions and problems, i do it including my mobo make (MSI) and model (neo2). All the comments i posted here are related to what MSI users have posted in various forums. How many non-MSI users have this issue i can't make any comments on, because i really don't know. I have met a few posts about an asus a8n (or something like that, i don't remember the exact model name for sure) reporting the same problem (it is also Nforce3 250 chipset). i definately know that there are *some* instances of non-MSI boards with this issue, so most likely there are others. i just don't have the experience to say if this is an MSI problem, or an Nforce250 problem. I just know there are a lot of MSI boards with this issue, it isn't exclusive to MSI, and that it happens across Nforce3 250 chipset mobos. i am under the personal impression that MSI has more of the problems, but i could just be under that impression after readond about the MSI side of things.

If possible, i think it would be very useful to the end users if someone could organize a sampling of Nforce3 250 (and 250gb) chipset motherboards. maybe 3-5 samples of each make+model. store bought would be best, so makers don't slide in pre-tested motherboards for review. run these boards under the 'poblematic conditions' that people report (PB, w/ gart, quake3, hl, farcry, with PB on and OFF, etc...) and see if any of the problems can be replicated. IF it's a chipset problem, it is very likely that a good few of these samples would turn up problematic. A larger sample would be good, but nobody can expect someone to just buy up a ton of hardware. I expect most the boards 50%+ would be ok, but i suspect that 1/10 -> 1/20 would be plain aweful, and maybe ~1/5 would have some issues (reason being that when you search forums for the problem, you will find a lot of topics on it, and even though not all topics have a whole lot of posts, the views [in many cases] will be over a thousand per thread. This suggests to me that people all over are searching for solutions to this problem, and it is in fact common. i even suspect that people with this issue may simply think they were one of an unlicky few, and write it off as a loss, and return the mobo or just deal with it. one guy on the MSI forums said he RMA'd the same board a couple times already, and each time he got it back the same problem was there. doesn't that suggest a systemic fault somewhere?).

-scheherazade

scheherazade

one other thing,

the sevarity of the problem ranges from ~10 second freeze-> ~>1 second run-> ~10 second freeze (rinse and repeat)

to

play for 20+ minutes, and have a single 5 second pause, and then another 20+ minutes of play.



also,
the severe pausing seems to be absolute [to the cases in which they occur], while the short pauses with long playtimes in between seem to be related mostly to PB users in multiplayer.

as both single player and multiplayer pauses are fixxed by removing the GART driver, it is possible that these are actually 2 separate problems, that share a common root cause. possibly the gart driver is enabling something(s) that cause trouble somewhere down the line in the system. that trouble spot may be different among different motherboards, explaining the large variance in what causes and fixes the problems.
or perhaps the single player pausing problem is just a more aggravated instance of the multiplayer pausing problem...


another tidbit worth noting : my neo2 had trouble detecting ps/2 mouse on boot. it wouldn't complain about a missing mouse in windows, the mouse would simply not work (light was on, but no reaction to moving the critter). some boots it was there, some boots it wasn't. i simply switched to USB mouse early on and let the issue go. later on when reading forums i read a mention of '[a problem like this is usually] a bad motherboard'.


**
as a side note, PS/2 users who like the 200hz port refresh setting, look into the USB refresh rate driver hack. you can hex edit your usbport.sys for XP to allow refresh rates above 125hz (up to 1000hz, but most mice only work well up to 500hz). usb spec states 1000hz compliance. so this isn't anything deadly, and by spec it is required to work fine. 2000 and linux supports higher than 125hz refresh rates on usb HID's. only XP has an artificial 125hz lock on it. just search for "XP usb refresh rate" and you'll find instructions on how to hex edit your driver.

-scheherazade

Sidney

Holy cow! Did anybody contact MSI directly?

__________________
lazyman

Opteron 165 (2) @2.85 1.42 vcore AMD Stock HSF + Chill Vent II

scheherazade

aside from trolling MSI forums (among many others), i've not dealt with MSI yet.

i've just searched for a solution for a long time, so i can finally play farcry 'properly'. i can't turn off fastwrites or not use the gart driver because it simply irks me that it doesn't work when it should. and i won't RMA it [yet] because i've read about RMA's not doing anything to fix the board (they send it back 'fixxed', but the problem is still there). i simply don't want to disable my computer for 2 weeks for what very likely could be for no benefit.

i've also got a feeling that this is a design flaw that will get a driver or bios work around. so i'm waiting it out (which i can afford to do since the only things i can't do so far are play farcry and play painkiller. not exactly critical applications). ALSO if it's really a major flaw, then a rev 2.0 board will be out soon. If i RMA it then, i might get a 2.0 returned to me (if the 1.0's are irreparable).





another 'fix' i forgot to mention

**IF you are an ATI user**
(I say "IF" because i'm using nvidia, and i've not been able to successfully use this 'fix', while a good few ATI users have reported success)

1)goto your boot.ini file (c:\boot.ini)
2)look for the entry "/NoExecute=<something>"
3)make sure <something> is AlwaysOff
i.e. "/NoExecute=AlwaysOff"

This will turn off Data Execute Prevention.
it's that fancy 'stack overflow hack' prevention tidbit in the new cpu's.
It's possible that somehow the ATI driver has DEP exception (like WINE has in the latest linux distros with DEP and pre-linking)

ATI people might be able to trade a security feature for a working system. (and who knows, it might work for some nvidia user somewhere)

-scheherazade

scheherazade

I tried the msi k8n neo2 (7025) 1.37 beta bios.
problem seems to be slightly essened.
less time to first pause, shorter pauses, but still there

currently i'm all up to date. newest drivers and all those along the way didn't fix the pausing.

beta 6.11 nforce w/ beta 4.4.0 gart driver
beta bios 1.37
beta forceware driver 70.41

(70.41 is reported to have problems on some systems, but mine works as well with 70.41 as with WHQL 66.81. used all the beta drivers since last release)

-scheherazade

scheherazade

(sorry, board won't let me edit my last post)

i meant to say :

I tried the msi k8n neo2 (7025) 1.37 beta bios.
problem seems to be slightly lessened.
MORE time to first pause, shorter pauses, but still there

-scheherazade