GTX460 hangs - power supply problem or mobo needs bios update?

[zoran.zorro]

Hello,

here are the specs:
MSi Cyclone gtx460/768MB, OC edition
mobo MSI NF520T-C35
http://eu.msi.com/index.php?func=proddesc&maincat_no=1&prod_no=1880#
AMD Athlon II 930, quadcore (2.8GHz, not overclocked)
2HDD's, Creative X-Fi Music, wifi card, bluetooth dongle, DVD-RW, PCI modem (dont ask...), 2 case fans
4x1GB DDR2, mainstream (NPC)
Power supply: Cooler Master 450w Real Power (2x 12V rails, 12A/10A at continuity, 18A/16A at peak 60 sec). So, because of this "real power" mark, it is stated on supply that it have 264W total at 12V rails, in continuity. Approx. 4 years old supply, not sure.
OS: XP Home, SP3

What happened: until now, I had Asus 9800GT Matrix which worked flawlesly. Before GTX460 instalation, I did the usual old driver deinstalation, then cleaning with Driver Sweeper. GTX460 into case (both 6-pin extra power adapters connected), system entered the Windows, driver instalation (ones from MSI CD), then asked for a restart. New boot, Windows came to the logo screen, keyboard initialized, but mouse NOT (which usually happens immediately after keyboard). Then, after a few seconds, grey screen and the system hanged. Fans are working. Nothing I can do, except reset.

I have booted into safe mode, deleted drivers, restart, boot into "normal" Windows, installed newer drivers from NVidia site, rebooted - exactly the same hanging. Then, the whole night of different combinations, but always exactly the same ending. I can always reach safe mode, sometimes mouse will not initialize even there, sometimes it will. But I can log into safe mode via keyboard. I even tried instalation of old drivers (last ones used with 9800GT), but they dont recognize GTX460 and asks for a newer driver...

At the end, I returned the 9800GT in case, and it is working without problem with the mentioned latest NVidia drivers from their site.

OK, the usual suspect in such cases is a power supply, but I'm doubting it because, if I'm right, during the login process vga is not drawing much power, it is not in excessive 3D - even if the power supply is weak for a full load, it should at least logon in Windows?

The only other thing that is coming to my mind is that maybe the mobo is somehow not compatible with new card. The latest available mobo bios is installed, it is from may 2010. I have seen a post somewhere by somebody who said that he needed to update mobo bios for GTX460 to work on his rig, but without mentioning which mobo he is having.

Resume: if a new supply is solution for sure, I will obtain a new one. But I will be very sad if this don't solve the problem, because then the only solution left is to change the mobo (+ reinstall Windows & everything), and I simply dont have a time & nerves to do that. In that situation, I will rather return GTX460 and stay with 9800GT.

Any advice, words of wisdom, more or less similar situations seen?

Thanks in advance!

 

[Nirreln]

It's the power supply that thing is a POS, only 264 watts from the 12v line is bad it's really only good for about 300 watts. And before you purchase a new power supply seek out some advice on what to get.

 

[Rifter]

agree, crap PSU is not up to the task, buy a quality 400W like the Corsair 400CX should work fine, 360W on the 12v. Or if you feel like some breathing room get a quality 550W unit.

 

[zoran.zorro]

...so I went to a shop and bought this power supply:

http://www.coolermaster.com/product.php?product_id=6641

44A on a single 12V rail. Nothing changed, the problem remains exact the same.

 

[zoran.zorro]

Went today to the shop where I bought vga, they tested it in their PC. It is working like charm. They were gentle enough to even replace 1st with another Cyclone card. It is behaving exactly as the 1st one.
Jesus, this vga aquisition was supposed to be a present for me from myself, now it becames a worst enemy's wicked torture...

 

[faxon]

are you sure it isnt still the PSU? those GX PSUs are also pretty crap, the 750w goes way out of ATX spec around 450w and it just isnt a stable platform for long term reliability. i was gonna post about it before but didnt think it was that big of an issue until you said that the card is working fine in another system. if it isnt that, i would consider looking at your board for the problem, though i dont see why that would be the cause. bad ram maybe?

 

[zoran.zorro]

are you sure it isnt still the PSU? those GX PSUs are also pretty crap, the 750w goes way out of ATX spec around 450w and it just isnt a stable platform for long term reliability. i was gonna post about it before but didnt think it was that big of an issue until you said that the card is working fine in another system. if it isnt that, i would consider looking at your board for the problem, though i dont see why that would be the cause. bad ram maybe?

I'm aware about notoriety of PSU's for vga problems, but damn - 44A on a single 12V rail should be enough for a single GTX460? Nvidia states that it needs 24A on 12V rail(s), and they always calculate like 20 percent overhead to protect themselves. I have read a few tests of this PSU model before buying, and all of them agree that they are more or less up to official specifications. CM has a 5-year warranty on these series of PSU's, and as a salesman I have learned a golden rule - nobody gives a long warranty if they are not sure in their product. Look, for example, at a Toshiba laptops - in my country cheapest have 1-year warranty, middle class 2-year, and a bussines ones 3-year...
Old 450W CM supply with total of 22A on 2x 12v rails pumped 9800GT without a hiccups, and according to tests GTX460 uses like 20 percent more power than 9800GT in full load. And a PSU at a shop where they tested GTX460 was a poor 450W Chieftec. I really, really doubt that the PSU can be a problem.

I did all other tests that came to my mind - getting out all PCI cards and leaving PC to barebones:1 HDD, 1 DVDRW, 1 memory stick; then changed memory stick... Everything stays the same. Clearing CMOS, of course, too.

So I also came to conclusion that my motherboard is a bad guy. I have already contacted MSI, as well as posting the same question at their forums - nothing useful to this moment.

It is not a problem to buy a new mobo, but the million dollar question is - which one? Googling gave me info that the same problem happens not just on AMD/NForce mobos, but also on AMD/AMD mobos, as well as Intel ones. WinXP and Win7. And if it happens that the GTX460 is not working even on a new mobo that I choose, I will jump from a bridge. Literally.

 

[betasub]

Don't do it literally.

 

[edplayer]

try putting your 9800GT back in to make sure the PCI-E didn't get damaged

 

[happy medium]

Have you tried to update your chipset drivers?

 

[zoran.zorro]

Forgot to mention that - 9800GT is returned into case, I'm using it now.

try putting your 9800GT back in to make sure the PCI-E didn't get damaged

 

[zoran.zorro]

Yeah, de-installed them in safe mode, as well as vga drivers. +Driver Sweeper for extra security.
Then a fresh install, 1st the chipset, then vga.

Have you tried to update your chipset drivers?

 

[zoran.zorro]

Don't do it literally.

Eh, it is a internal joke for a friend who aimed me here and will probably read it today/tomorrow. I will jump for a weekend, really, few times. The thing is, this "bridge" is approx 4m above the water, so it is not as scary as it sounds. Local sport here

 

[cusideabelincoln]

Power supply probably isn't the problem, but you should really stop buying crappy units. That 450W is one of the worst power supplies. It can't even deliver 450W - gets unstable after 350W.

The same deal with their GX series. However your system won't be loading it that high so you should be fine.

I'd peg the problem on the motherboard, and some incompatibility between a Gen 1 PCI-e slot and gen 2 card.

 

[happy medium]

and some incompatibility between a Gen 1 PCI-e slot and gen 2 card.

+1

 

[heat901]

I'm aware about notoriety of PSU's for vga problems, but damn - 44A on a single 12V rail should be enough for a single GTX460? Nvidia states that it needs 24A on 12V rail(s), and they always calculate like 20 percent overhead to protect themselves. I have read a few tests of this PSU model before buying, and all of them agree that they are more or less up to official specifications. CM has a 5-year warranty on these series of PSU's, and as a salesman I have learned a golden rule - nobody gives a long warranty if they are not sure in their product. Look, for example, at a Toshiba laptops - in my country cheapest have 1-year warranty, middle class 2-year, and a bussines ones 3-year...

I had a CM 650 go on me in about a year and a half. So I wouldn't put longer warranty = better product. From what I know of business more companies will extent warranties when there are a lot of failures in a certain line of product. So that people like yourself feel assured that even if it does fail you can get it repaired for free.

Good example is hyundai car company they have the longest warranty so that people feel assured that if something does go bad its covered. Now would you say hyundai has better cars then honda because their warranty is only 36k miles?

 

[zoran.zorro]

I had a CM 650 go on me in about a year and a half. So I wouldn't put longer warranty = better product. From what I know of business more companies will extent warranties when there are a lot of failures in a certain line of product. So that people like yourself feel assured that even if it does fail you can get it repaired for free.

Good example is hyundai car company they have the longest warranty so that people feel assured that if something does go bad its covered. Now would you say hyundai has better cars then honda because their warranty is only 36k miles?

I haven't said that the longer warranty means better product, just that the company thinks that their product is pretty relliable. I really dont get this logic "we have a bad product, let's extend its warranty so 1d1ots [like me?] will buy it to have free repairs". Yes, it can boost sales at the begining, but after the warranty period it will cost company more than they have earned from boost. Nobody smart can do that, I can accept that maybe somebody new in business can gamble with this concept and extend warranty from 1 to 2 years, but 5 years...Sorry, no way.

And a term "better product" is more complicated than this. Yeah, 50.000 euro BMW is probably better product than 25.000 euro Honda in terms of power, comfort, gadgets, whatever... But it also cost double than that Honda. Is this "betterability" worth double the price? For somebody yes, for somebody not, I'm sure you agree. So we can translate this into a power supply debate: 1000W Corsair is for sure stronger and better made than my 550W CM, and also the more expensive. But the question is - do I need a better supply than this CM? In mu case not, I'm sure this CM is good enough for my needs, and I will rather spend this ~50 euros price difference to buy a new bicycle or whatever to my daughter.

Another problem is making general conclusions based on a single experience. I'm sad that your CM supply died so fast, but on the other side - my CM served me well for 4 years, and it is still alive. What now, who's right? When I made this conclusion between reliability and warranty, I really speaked from a long-time sales experience: electrical products with longer warranty usually have less failures than ones with shorter warranty. But this is not a law, electrical products have a general failure rate of 2-3% durring the warranty period. And EVERYTHING can die, even this 50.000 BMW or 1000W Corsair.

 

[3DVagabond]

Manufacturers of any product don't "give" you any warranty. Whether it's a year or lifetime, it's factored into the price of the product. You pay for the warranty, and any other after sale support you get (trade up policies, for example), when you pay for the item. Not to burst anyone's bubble but the "It has a better warranty so the manufacturer has more faith in it" concept is a fallacy. It's all rolled into the cost of marketing the product.

 

[zoran.zorro]

Manufacturers of any product don't "give" you any warranty. Whether it's a year or lifetime, it's factored into the price of the product. You pay for the warranty, and any other after sale support you get (trade up policies, for example), when you pay for the item. Not to burst anyone's bubble but the "It has a better warranty so the manufacturer has more faith in it" concept is a fallacy. It's all rolled into the cost of marketing the product.

Of course, aftersales costs are calculated into the price. But if a moderately priced product has a 5-year warranty, then it is a good sign, isn't it?

 

[zoran.zorro]

...and back to the topic. People that mentioned PCI-e compatibility are probably in the right direction, but unfortunately seems that only MSI can solve this with a new BIOS. If they are willing, and if it is possible - I remember that crazy situation when they were technically unable to make some NForce mobos compatible with AM3 processors. Why? Because there were no space left in BIOS chips to add the code. At least thats what they said...

I have had a long discussion with MSI online customer support, after the usual questions (bios, power supply, drivers, etc) already mentioned here, we stuck at "We will check your issue then get back to you as soon as possible". That was yesterday morning.

 

[Scali]

Pretty sure it's related to the motherboard/chipset/bios in some way.
I run my GTX460 in a crusty old Core2 Duo system with an Intel P965 chipset (Asus P5B Deluxe), which only supports PCI-e 1.0, and that works fine.

 

[3DVagabond]

Of course, aftersales costs are calculated into the price. But if a moderately priced product has a 5-year warranty, then it is a good sign, isn't it?

I'm not saying that the longer warranty isn't a good thing, just that it isn't determined by the quality of the product. In the end it's all calculated into the final price.

 

[llee]

The root of your troubles is the motherboard. Good Day.

 

[zoran.zorro]

So I gave up on waiting for a miracle to happen, and bought a new motherboard (Gigabyte GA-M720-US3). Straight instalation, not a single problem, vga is fast as Usein Bolt... All the other components in rig are exactly the same, as well as Windows XP instalation (no, I did not reinstalled it), so it is pretty clear that for some reason MSI NF520T-C35 motherboard is not compatible with GTX460 cards. Is it a problem with this specific mobo or with a design of GTX4XX series, it is hard to tell - I have found many reports that people are having identical problem (hanging at a Welcome Screen) with GTX4XX cards from other manufacturers, on many different motherboards, in all MS OS's. But for the end user, conclusion is clear...

Big thanks to all the people who tried to help me! Problem not solved, rather overcomed.

 

[Scali]

This motherboard has a different chipset than the other, so the issue may be with that particular chipset (this is a newer generation).
But still, I would find it strange that an nVidia GPU would not work with an nVidia chipset.

That's why my guess would be that it's something in the MSI motherboard... either a design flaw in the hardware somewhere, or a problem with the BIOS.