OC causes texture artifacting [Archive]

MensaBoy

01-18-05, 09:14 AM

my AGP/PCI is locked @ 66/33, and spread spectrum is disabled

if i OC above 218 memory, i get strange horizontal "hits" in textures in games, like in the menus of doom3 - the game environment seems fine, no artifacts at all

in RTCW i get texture glitches in-game

this occurs even if i turn the video card OC off

if i set the ratio to 5:4, and OC the CPU to the same amount it doesnt happen, so it seems as if the memory OC is what is introducing the video artifacts

while i somewhat understand that this is possible, because of AGP texturing, i havent heard of anyone else talking about this

meanwhile, staying out of the 3D environment, i can OC up to 230 with prime perfectly stable, and can prolly go even more

anyone else seeing this relationship between memory clock and video artifacts?

jenko

01-18-05, 11:50 AM

I suffer from this aswell but i was told it was because of my 865 chipset.
I really do want a fix for this problem, as i dont want to use the 5:4 divider.

MensaBoy

01-18-05, 12:30 PM

you really think it may be the chipset?

something having to do with the way its passes texturing via the AGP?

nicknomo

01-18-05, 12:44 PM

Yes.. I had this argument a few days ago. The memory might be a factor, but the passing of data to the graphics card is being boggled somewhere.

MensaBoy

01-18-05, 12:54 PM

but only with overclocking....

i would think there would be a setting in the bios, or some sort of mod to fix this tho

jenko

01-19-05, 05:50 AM

Mensaboy does the glitching go away if you dont run your ram in dual channel?

MensaBoy

01-19-05, 10:33 AM

Mensaboy does the glitching go away if you dont run your ram in dual channel?

yes it does....

wtf?

jenko

01-19-05, 11:35 AM

yes it does....

wtf?

Same as me then, if anybody knows what causing the problem and a fix it would be great.
Wish i could lend another 2 sticks of ram to test ifs my ram at fault.

MensaBoy

01-19-05, 01:03 PM

i have 2 crucial ballistix on the way, should arrive fri or mon...

jenko

01-19-05, 02:57 PM

Cool can you let me know if it sorts the problem?
As i will stick my 1gb of twin mos in a budget rig and sell it and get some new ram.

MensaBoy

01-19-05, 03:03 PM

nah, i figger i'll keep it a secret

:)

Cyberwiz01

01-19-05, 03:26 PM

have you tried raising the chipset voltage. Just a suggestion. I don't have an 865 chipset, but that might help the problem.

MikeyLikesItSI

01-19-05, 03:43 PM

What is your AGP Aperture size set to? maybe its too low/high?

ropey

01-19-05, 03:48 PM

I have an ASUS P4P800 Dlx. I have had P4 2.8 @ 3.2, P4 3.0 @ 3.5, Celery D 2.4 (C0) @ 3.2 and Celery D 2.4 (D0) @ 3.8.

AIW Radeon 9800 pro with no problems whatsoever. However, if I overclock my video card over a 400MHz Core I do get artifacts.

R

MensaBoy

01-19-05, 04:16 PM

have you tried raising the chipset voltage. Just a suggestion. I don't have an 865 chipset, but that might help the problem.

AGP Voltage?
DDR Voltage?

tried 64, 128, and 256 aperature, no difference...

jenko

01-19-05, 06:16 PM

Well using aperature of 4mb seems to fix it, but 8,16,32,64,128,256mb i get the glitching?

Albuquerque

01-19-05, 06:57 PM

If your AGP apeture is at 4mb, it's probably effectively turned off at the driver level (since AGP texturing for only a measely four megs is worthless hehe)

I had this same problem until I turned up my AGP voltage. Most motherboards link chipset voltage to AGP voltage, so upping that voltage can help the northbridge cope with the higher overclock. I need the full 1.8vAGP in order to run my board reliably at 280+fsb's.

hawtrawkr

01-19-05, 07:10 PM

this is a result of your chipset. i had this and it almost drove me mad. i had even gone so far as to underclock my 9800p because of this the cpu and hte memory and basicly everything id been running in my max-3 got transfered over to a newely purchased ai7. when i oced my cpu/ram/gpu all back to where i had them on the max3 everything worked fine i ran dual instance of prime to check heat levels and make sure i mounted my water block correctly and everything worked and ran fine......... and then i loaded planetside on the menus i had massive amounts of taering and artifacts everywhere. typically i would associate this with a gpu/vram clock thats too high so i went and checked my cards clocks and they where where i had used them on my max3 without any probelms so i went through all the steps made sure the fan was running on the card turned the clock down but even underclocked i still had the artifacts and taering and they happend instantly but on the desktop nothing. finally i started trouble shooting and ran it back to the northbridge. i took everything out and put it in my p4c800-e and instantly put the clocks to where i had been using them on my max3 and loaded planetside.... it worked just fine.

ive talked to other people that have this unexplained amount of artifacts and nothing they can do will seem to help but i can garuntee if you leave your videocard clocked where you had it and turn your cpu/mem oc down and watch the artifacts go away.

some 865 boards will do this and some wont. ive heard people with ai7s who have thiers clocked over 250mhz and dont get the artifacts and ive talked to people who have its not a one manufacturer thing over another though at least from what ive observed.

MensaBoy

01-19-05, 07:57 PM

this is a result of your chipset...

ive talked to other people that have this unexplained amount of artifacts and nothing they can do will seem to help but i can garuntee if you leave your videocard clocked where you had it and turn your cpu/mem oc down and watch the artifacts go away.

some 865 boards will do this and some wont. ive heard people with ai7s who have thiers clocked over 250mhz and dont get the artifacts and ive talked to people who have its not a one manufacturer thing over another though at least from what ive observed.

so....
what the heck do we do?
im of course gonna try the memory, but if thats not working, then im stuck at 218-219 fsb because of the video artifacting? i dont know what other motherboard to go to, i need AGP, SATA, and socket 478...this board seemed ideal

disclaimer:
please understand that im more than happy with the speeds at which i can run, for the most part, i never gave OCing a second thought till i found this forum when troubleshooting another issue i had, but this mad desire to get more and more is like a freakin virus...thanx a lot :rolleyes:

hawtrawkr

01-19-05, 08:05 PM

try upping your vagp there should be a option that allows you to set it in your bios that could help out sometimes a little extra voltage to the northbridge will get you some extra fsb.

MensaBoy

01-19-05, 08:36 PM

try upping your vagp there should be a option that allows you to set it in your bios that could help out sometimes a little extra voltage to the northbridge will get you some extra fsb.

northbridge? is that the same as AGP voltage?, or is that in addition to?

running @ 225/3.6 right now, tried upping AGP to 1.85 (max) and it settled the artifacting way way down

this artifacting has only to do with textures in 2D screens, for the most part, the in-game textures are rock solid, should i be able to stay with this, or should i lower FSB 'till it goes away completely?

i mean do you forsee any harm to the 6800 from this, or is simply aesthetic?

MensaBoy

01-19-05, 08:39 PM

MensaBoy

01-19-05, 08:50 PM

should i just get a p4c800e ???????

hawtrawkr

01-19-05, 11:26 PM

jenko, are you trying any of this, similar results?

also, hawtrawkr, if this is a quirk of the chipset, do you mean many people with p4p800e's cannot go over 220 or so? or does it seem to only have an effect on nvidia or 6800's?

its a problem with the 865 chipset most people ive read about will get the problem around 250mhz but it is the chipset would happen no matter what videocard you had in. my problems like from my personal fiasco all occured while using a 9800p so its defianantly not a nvidia or 6800 issue.

and i wouldnt just jump right in and purchase a p4c800e but thats up to you the 875 chipset would help you out some but it might not be worth it to you to spend the money it would take to get the newer board. also the p4c800-e really only overclocks well when its been droop modded and had a vdimm mod to allow more volts to the ram.

Albuquerque

01-19-05, 11:44 PM

As I mentioned earlier, my Albatron needs some AGP voltage to stay stable as the FSB's increase. I'm not sure it's so much the 865 chipset, but rather just a limitation of the motherboard as a whole. You have to consider that you're overclocking a LOT of things when you up your front side bus speed.

While the 875 chips may take a little more abuse before showing artifacts, the truth is that the 865 and 875 chips are the same silicon that has simply been speed-binned by Intel. Overclocking your FSB from 200 to 250 is a 25% overclock of those northbridge chips, and they (just like your processor) need adequate cooling and power to stay stable at higher speeds.

jenko

01-20-05, 10:23 AM

hawtrawkr

01-20-05, 10:43 AM

Nope upped my agp voltage frome 1.5v to 1.8v and with a fsb of 220 i still get the glitchs.
The only thing i have to found to stop it putting my aperature to 4mb but im not sure how this will affect my games.
I have removed my northbridge HS twice to apply new paste and have also put a 40mm fan it and is very cool.

id think it would hurt your performance in games that needed to use more texure memory than what your video card has.

aperature is basicly just system memory that you set aside to be used by the videocard to act as a buffer between hdd and vram since its quicker to call upon the system memory than it is your hard drive.(if my understanding of this is correct someone pls correct me if im wrong) if your not using any aperature or in your case 4mb which is basicly not using it youll simply go from your drive to your videocard when you need memory for texures and crap.

if adding more vagp didnt help you the only thing you can do would be to replace your mobo or just not overclock it over 220mhz. (well i guess you could try freezing it :burn: )

another thing to try would be if say you had 2x512mb only run it with one memory stick in without a divider and see if you cant gain a few more fsb of oc since there would be less strain on the chipset. this of course isnt feasable to run all the time since dc is more valuable than a few fsb but it might be another way to show you the chipsets whats causing your headaches.

MensaBoy

01-20-05, 12:02 PM

to be honest, all this overclocking that ive been recently experimenting with has been a bit scary anyway, running on the ragged edge of errors and artifacting isnt something im entirely comfortable with

my volts arent very high and my temps are real good, so my current plan is to find the absolute limit, then back everything down just a bit for everyday running

my PCmark and 3Dmark scores are above average for the clocks im running, so im confident that i dont have any major bottlenecks anywhere killing my day-to-day performance

its just that in single channel mode, i can get almost 10% more clocks across the board, and i'd love to find a way around that drawback

the new crucial memory im getting has much tighter timings than the kingston, so i'll see what happens then, i've read that this mobo is picky about mem timings, so that may be a big step in the right direction

Albuquerque, hawtrawkr, your help has been invaluable in my understanding some things - jenko good luck, i will update when i get the memory

(im not going anywhere, just wanted to make sure you guys know you've been help)

mtb856

01-20-05, 12:11 PM

hawtrawkr

01-20-05, 12:19 PM

Albuquerque, hawtrawkr, your help has been invaluable in my understanding some things - jenko good luck, i will update when i get the memory

im glad to help out anyway I can. I think youll find these forums are filled with people that are wanting to help out others the way they themselves have been helped along.

theres alot of diffrent hardware out there some good, alot more of it bad but with the wealth of knowledge going around these forums you can really save yourself some money and headaches from reading the experiences of others.

MensaBoy

01-20-05, 02:13 PM

I'd consider the Abit IC7-G Max II if you're looking at a 875 chipset motherboard..... the new one's actually handle more power than the IC7-MaxIII (and cost quite a bit less), so you can overclock Prescott's on them without having to deal with droop mods and other mods as soon.

I've never had a problem with artifacting on this motherboard..... my AGP aperture is at 256MB and I've run up to a 317FSB on it without any problems.

why do you think the max3 has poorer power handling if it has a ducted fan cooling all the power electronics?

im not debating, just curious if theres a reason...

and how would you tell a "new" max2 compared to an "old" one?

mtb856

01-20-05, 02:27 PM

I'll see if I can find the info on it, but it was in the Search for 4 gigs thread in the Intel cpu area.... one of the big abit guys mentioned that the IC7-G can handle 20W more power (or so) within its ciruitry than the IC7-MaxIII. I'm pretty sure that it was woutertal that posted it, so maybe he'll post in here and clarify.

I may have set myself up to sound really stupid, but I'll do a search to make sure that's what he really said.

The new IC7-G's will be labeled as IC7-G Max II Advanced.... it's really the replacement for the Max III since it was pulled from the market. They should be the same price as the IC7-G ..... I paid about 110 for mine.

hawtrawkr

01-20-05, 04:35 PM

its to do with the power circuitry that powers the cpu the mosfets/caps arent as good and the max3 cant provide the same amount of continous wattage as a max2 or typical ic7 can.

the otes cooling was little more than a gimmick on the max3 you get better performance from a suspended 80mm fan and no otes as all. one of abits favorite tricks was to take a picture of a max3 without the fan running but with the shroud still over the mosfets/caps. this would act as a insulator and keep the heat around that area of the motherboard. then theyd turn the fan on and show a thermal picture of it with otes fully running, of course the picture of the otes fan operating showed much cooler temps on the fets/caps
but what they didnt show you is a thermal picture with no fan and no otes shroud which would actually give lower temps than the otes system with a well ventilated case.

this was addressed in the intel cpu section under the quest for 4ghz woutertal also made mention of it in his ic7 max3 caps mod thread if memory serves me correctly.

hawtrawkr

01-20-05, 04:37 PM

mtb856

01-20-05, 04:40 PM

It was nice to have someone back me up though...... It'd been a while since I'd been to that thread :D

hawtrawkr

01-20-05, 04:43 PM

It was nice to have someone back me up though...... It'd been a while since I'd been to that thread :D

yeah same here use to read it everyday..... course that was before vapochill came into my life then it wasnt such a quest to hit 4ghz.

i posted then read yours and just kinda kicked myself a lesson learned in doing things in proper order i suppose :beer:

jenko

01-21-05, 04:22 AM

The game i have found that gets these glitches the most is Wolfienstien ET, if i turn of GL exstensions in et's options its fixed.
Guess im just cursed into using the 5:4 divider.

MensaBoy

01-21-05, 09:44 AM

The game i have found that gets these glitches the most is Wolfienstien ET, if i turn of GL exstensions in et's options its fixed.
Guess im just cursed into using the 5:4 divider.

same here, RTCW is the only game i found that artifacts while in-game

MOH, Doom3, HL2, Far Cry...all artifact free, in fact if it werent for the PDA screen artifacting in Doom3 i dont know that i would have even noticed

jenko

01-21-05, 01:36 PM

Wolfenstien enemy territory is much wost then RTCW, doom3 and farcry gets them just its really hard to notice hard to explain but its a diff glitch altoghter for them games.

MensaBoy

01-21-05, 01:43 PM

Wolfenstien enemy territory is much wost then RTCW, doom3 and farcry gets them just its really hard to notice hard to explain but its a diff glitch altoghter for them games.

does the amount of your overclock translate in a linear fashion to the amount of artifacting?

like, in the PDA screen i can almost count the number of "hits" per second, and when i clock up higher they get worse at that rate

MensaBoy

01-21-05, 07:06 PM

got the Crucial memory, along with an NV Silencer and some AS5
have installed the memory, and currently running tests, however it appears as tho all artifacting is gone :clap:

some questions, which i would prolly be able to figure out with some searching and time, but i want the answers to reflect my specific situation

1. if memtest-86 runs 1 pass without errors, is it accepted that it will continue to pass without errors, or must it be run for a long time to be sure?

2. same as above with Prime95, with the additional question, how do you know when P95 has done a complete cycle?

3. if memtest is error free, is this an assurance that the memory timings and speed are stable, or could other factors once booted to windows affect the stability of memory? - in other words, if memtest is stable, can stability issues then be definitively attributed to something other than the memory?

4. is there anything i should be aware of when installing the NV Silencer, or is it pretty straight foreward?

Lancelot

01-22-05, 03:01 AM

I did not read all the topics above but I want to point out it can also be a hardware problem like I experienced. On a P4P800 board I would always have artifacting when using the 1:1 divider!? Believe me I tried every which way settings back and forward. 5:4 heavily OCed was fine but 1:1 stock speeds always caused artifacts with both nVidia and/or Radeon cards. I also tried different sticks of RAM etc. etc.
When I upgraded to the rig in my signature I sold the board to my bro-in-law. If we run his Celly 2.4 at stock 1:1 we get artifacting! His system is OCed to 3.24Ghz 135FSB with the RAM running 3:4 (180Mhz) and we are gaming all night without any problems. Weird huh?! Really seems to be a hardware failure of the i865PE chiset here...

mtb856

01-22-05, 07:58 AM

For Memtest..... I personally run about 10-20 runs of the #4 test before I say it's error free.

For Prime.... I won't consider my overclock stable unless it runs for 6-12 hrs without crashing. Alternatively, you can also use Folding at Home as a stress test..... it runs your cpu at 100% 24/7, so if there is instability it will crash eventually. FAH is the best way IMO b/c it doesn't interfere with gaming or normal use of the computer like Prime does.

MensaBoy

01-22-05, 11:42 AM

For Memtest..... I personally run about 10-20 runs of the #4 test before I say it's error free.

For Prime.... I won't consider my overclock stable unless it runs for 6-12 hrs without crashing.

let memtest run all night (21 cycles), stable at 235, but one of the prime threads fails after about a half hour...

not too sure where to go from here, currently backed down to 230 and prime looks good so far, i dont want to volt too high, and im on air, so i guess this is the end of the road for my current configuration

jenko

01-23-05, 07:55 AM

Have the artifacts completery gone know with your new ram?
I havent had this ram that long and dont wanna replace it.

MensaBoy

01-23-05, 12:18 PM

Have the artifacts completery gone know with your new ram?
I havent had this ram that long and dont wanna replace it.

yeah, all gone...
tried 2-2-2-6, but its not stable, so ive backed it to 2.5-2-2-6
running @ 230 FSB now, memtest thru 20 cycles, and prime for 10 hours - no errors

some benchamrks:

PCmark 04: 5703
i get low scores in virus scanning compared to comparable systems, i wonder why this is?

3DMark 05: 5362 3DMarks, 5050 CPUMarks
my CPUMarks compared to comparable systems is high, usually much higher

jenko

01-23-05, 12:33 PM

Glad its sorted for you, im still up **** greek lol
Asus does state the twin mos i have is compatible with my board.
I will find it hard to shift this ram so for £100 i might just get another 2x512mb so i have 2gb of ram and just use the 5:4 divider and hopefuly i wont have to upgrade my rig again apart from the 3dcard in a years time?
What voltage have you tried with your ram?

MensaBoy

01-23-05, 01:46 PM

What voltage have you tried with your ram?

when i was searching for the top speed i had it at 2.85, but the memtest and primeing were done at 2.75

ill prolly back it down another click and check for stability tonight, it definately helps to find the high end with the voltages up high, but once i have more info about my system as a whole, im going to try and back all the volts down as much as possible

im convinced from reading several thousand posts on this forum that voltage is the biggest factor in chip longevity

MensaBoy

01-23-05, 02:57 PM

changed the DRAM Burst Length to 8 (was 4)

PCmark 04: 5757
a substantial gain for a single bios change

3DMark 05: 5375 3DMarks, 5075 CPUMarks
similar gains here, im pretty stoked

ive run memtest thru 4 cycles with the burst length up, i think i'll stop tweaking for today (yeah, right) and run memtest for 4 hours or so at these settings...

heres all my bios settings so far:

P4 3.2E

FSB - 230
mem ratio - 1:1 (400)
AGP/PCI locked @ 66/33
CPU Volts - 1.5
DDR Volts - 2.75
AGP Volts - auto
DRAM Timings - 2.5-2-2-6
Burst Length - 8
idle timer/refresh - auto
AGP Aperature - 256
Spread Spectrum - Disabled
MPS - 1.4

MensaBoy

01-24-05, 08:14 PM

blew the P4P up :(

visit here:
http://www.ocforums.com/showthread.php?t=360487

mtb856

01-24-05, 09:05 PM

Wow, that's horrible..... sorry bro :(

So do you have any idea of what happened yet, or is it still a mystery?

MensaBoy

01-24-05, 09:52 PM

no idea, i cant imagine why connecting a USB device would cause anything like that

it freakin sucks, im looking at an IC7 max 2 or another P4P...

+Torm+

02-09-05, 07:31 PM

I experience much the similar probs as you guys does here. Would you mind trying the following.
Install Prime95 and RTHDRIBL.exe on your comp.

Start RTHDRIBL.exe and just let it run in its window.
Now start up Prime95 - select "Torture test". Select "Small FFTs" and start the test.

This should work OK. CPU load should be at 100%.

Now stop Prime 95.
Go to Torture test again - but select "Custom". In the field that says "Memory to use" enter e.g. 100 Mb.
Start the test.

Voila - Instant artifacts in the RTHDRIBL window.

Now stop Prime95 - unless your comps frozen by now - if you stop Prime95 the artifacts should stop.

Least this is what happens on my system - check this thread for details http://www.ocforums.com/showthread.php?p=3441254#post3441254