major graphics problem
- Thread starter Perkomate
- Start date
Okedokey
Well-Known Member
LOL a 7+ year old thread on a PCI based system where the OP gives up! You're out of your mind. Secondly how could he be in the operating system with the condition you outline (ie how does he take the screenshot). Thirdly, modern graphics cards assign memory onboard. Fourth, it happens with multiple graphics card, fifth, this is absolutely a 12V droop condition. In 2005 when the guy above thinks it’s an issue, computers were nowhere near as dependent on the 12V rail.
That PSU is garbage, Chinese made and either has massive ripple or 12V rail droop.
Motherboard fires up first time, capacitors and other reactive electronics cannot be fully charged (read out of ATX spec +/-5%), but on second time you try, they're fine because they were still partially charged from the first attempt. That makes perfect sense - in fact its as expected. The motherboard is not returning a good status to the PSU because the PSU is not providing sufficient power (pulse and peak loads at startup), to do so.
Secondly, PCIe lanes are an extension of the CPU, which surprise surprise is on the 12V rail. Thirdly, the PCIe lane also has to provide power to the graphics card. Surprise of all surprises, the 12V rail.
Add in ripple issues and long term under voltage conditions and you may have even damaged the graphics card, but the symptoms (the pic at the beginning), is one of two things, over heating or failing VRAM, or 12V droop.
You said there is no heat, so the next most probable cause is 12V droop.
But it doesn't matter, the quality of that PSU alone requires its replacement, even for some miracle it’s not the PSU, then you have to replace it now before it is.
Either way, you need a new PSU, if you want to try looney-tunes idea go for it, or if you want to replace your motherboard, yes, it may even fix it, but in a short period problems will arise.
As the OP is in Australia where it has been very warm recently, this also supports the PSU issue as heat derates that PSU by around 4W/1oC increase above 25oC (rated temp). That means recently a further 40W (3/4A) is being lost due to heat, reducing efficiency and stability.
Either buy a multimeter and test the PSU, or replace it, because if you don't you're going to have ongoing issues.
Last edited:
Man seven years ago. On a socket 7 board with windows 98. 98 had big time IRQ and memory assignment issues. Plus that was AGP which really ran off the PCI bus. Has a VIA chipset which at the time had big issues with the AGP bus. Using a AGP 1.0 board with a AGP 2.0 card. Something that was never brought up was, the problem he was having could alot of the time be fixed by manually setting the AGP port to X2 instead of leaving it set to auto. Had some boards at the time that did the same thing.
Last edited:
The PCIe controller for the slots on your system is on the processor. Never saw a board that all the PCIe slots went out. Do you have or know someone that has a power supply you could try. Not saying 100% thats what it is, but one with a low or going out 12V will do exactly the same thing.
Take this dude advice above. If you have a socket 7. Running windows 98. Running on a AGP port. Using a AGP 2.0 card on a AGP 1.0 board. Plus if you have a VIA chipset with screwed up drivers and if you can change the AGP speed from auto to X2. If none of the above applies to you. COMPLETEY IGNORE HIS POST!
Right, spend money, instead of checking to see if there are memory assignment conflicts. I suppose if you are so rich that you have money to burn, that is the right thing to do.
The guy I was trying to help, at first had his machine booting randomly; but after he assigned non conflicting memory assignments to his video card, it would invariably fail on the first boot but succeed on the second boot. That is because there were other conflicting memory assignments in the default position, and it would boot to those settings first, then when that failed, it would boot to the alternate settings he has assigned. Problem was he was too stupid to "Remove" the conflicting assignments, so he never did get it to boot properly every time. As for Win98-SE, Windopes is WINDOPES, and stuff like that seldom gets fixed, - because they are too busy monkeying around with SECURITY UPDATES.
The guy I was trying to help, at first had his machine booting randomly; but after he assigned non conflicting memory assignments to his video card, it would invariably fail on the first boot but succeed on the second boot. That is because there were other conflicting memory assignments in the default position, and it would boot to those settings first, then when that failed, it would boot to the alternate settings he has assigned. Problem was he was too stupid to "Remove" the conflicting assignments, so he never did get it to boot properly every time. As for Win98-SE, Windopes is WINDOPES, and stuff like that seldom gets fixed, - because they are too busy monkeying around with SECURITY UPDATES.
Last edited:
Getting real tried of this. After 95/98/98SE/ ME the IRQ and memory assignments changed dramatically.
Thats like telling someone to check his holly carburetor float levels and check his point gap. When the guy has Fuel injection and a electronic ignition.
Thats like telling someone to check his holly carburetor float levels and check his point gap. When the guy has Fuel injection and a electronic ignition.
Okedokey
Well-Known Member
Exactly, or telling the Spitfire pilot to go full afterburner. Nonsense ever since 2005.
Stick to diagnostics mate, the same thing happens with different graphics cards. Logic fail.
Perk mate, Strangle is right, your motherboard is just underpowered or been damaged from brownout(s) or continuous low voltage stability. As Strangle said, there is very few times that all PCIe lanes are dead. If thats the case, its more likely to be CPU issue than motherboard.
But the PSU is garbage.
Since you are getting a PSU anyway, wouldn't you just try that first?
Last edited:
Perkomate
Active Member
I have an old Antec 300w, how do you think that would go with a Radeon 2400 and the same CPU, just stock clocks and undervolted? Enough?
-edit-
I mean just for testing that it works.
m3incorp
New Member
Since you said you plan to buy another PSU anyway, why not just go ahead and buy and try. If it fixes the problem, then no money thrown away. If it doesn't fix the problem, you are replacing motherboard and still have that new PSU that you were still going to buy, so still no money thrown away.
No. I don't see why he'd need a whole new PC.