«

[Halindir]

Originally Posted by Japo

First of all, if the computer is new and within warranty, you should return it immediately and demand a 100% trouble-free replacement, no questions asked. I encourage you to do this, and ignore the rest of my post... Hardware problems are a pain in the ass, even just identifying them. And don't you think that you can solve the problem by taking it to a computer repair shop, they will only try what any of us could, sometimes they don't know a lot actually, and if the crashes are random and can't be reproduced, you're liable to get the computer back and being billed and told that it's now good, only to find later that the problem remains.

Hey I have a GT520 too, although mine only has 1 GB.

So does this happen only with games? Are you experiencing any other problem--small as it may seem in comparison to total system crashes?

250 watts sounds like very little to me. I have 375 W, and only my GT520 recommends at least 300 W (and that's if the rest of the parts aren't specially demanding). But I'm no expert, you could list your hardware at the forum over at http://www.tomshardware.com/ for example, and get a recommendation of minimal power rating. Also depending on the manufacturer of the power supply, I heard, some of them list in the specs a power rating that they can give constantly, whereas others like to list peak values that they can't sustain. Anyway you should be able to see the power requirements of your parts on the website of each manufacturer. For example my nVidia was made by Gigabyte:

http://www.gigabyte.com/products/pro...px?pid=3826#sp

Even if your power supply is poor, it needn't be the problem here, specially if these crashes aren't being caused by high load (?). My experience (see below) biases me towards RAM, RAM, RAM. And the only way to be sure may be to run with different sticks for long enough periods. RAM is such a basic part of a computer, that a problem or conflict with it is capable of causing problems in other parts, that use RAM, and the symptoms will point to those parts which are good, instead of the real culprit. Just like when I was dead sure there was a problem with my old ATI...

As Horseman says, even if the computer starts with each RAM stick--as it starts with all of them together--there may be a fault. Memory test programs that run inside Windows are limited because they can't access all the memory.

http://download.cnet.com/Ultimate-Bo...0_4-51584.html

Even if scans find no defects, for example I recently had problems because even though all my RAM was good, two halves of them (not even in dual channel) had slightly different undocumented characteristics, so they caused problems only when used at the same time. But if your computer is new and you haven't added or replaced RAM, this shouldn't happen, unless you got the computer at a shop where some guy assembled it from separately bought components.

The computer isn't new new, my dad assembled it for me. The only thing I had to do, were to buy a graphics card - which has a 3 weeks warranty.

My dad will be sending me, in about 4-5 days, 2 1GB RAM - In case one or more of the ones in my pc are damaged - and a not-so-good graphics card - to test if the freezing occurs on that one too.

And yes, it only occurs in games. Everything runs smoothly aside from that.

[Halindir]

Originally Posted by The Fifth Horseman

250 W is positively too low for the kind of build you describe.
A computer's power supply unit MUST be picked out with taking into account the sum total of each component's peak power consumption - and you should add a 20% to 25% overhead, because PSUs can't sustain a 100% load for extended time periods (and as they wear down with age and use, the max output only goes down).

Faulty power supply can cause damage to your machine's components.
It sounds like he built the thing himself?

..So I shouldn't really be using this pc at all, until I get a new PSU, or just stay away from games until then?

My PSU has 250Watt, 12V and 19A. And GT520 recommends a minimum of 300Watt, 12v and 22A

[Japo]

Voltage isn't minimal or maximal, but right or not. As long as there's a standard (e.g. PCIe) you're good. The question is if the supply keeps it stable enough at that level. Then power and amperage are directly related.

I'm not sure, if there's any risk in using an undersized power supply. I think if anything could get damaged, it would be the supply. Don't take my word on this.

[The Fifth Horseman]

It's safer to not touch the machine until the PSU is replaced.

[Halindir]

Turns out my PSU is a SCP350-V2..so its 350Watt I guess..So it might not be it after all.
A friend of mine told me that should be enough to power my graphics card and the rest of the system.

[Japo]

Now that makes more sense. Anyway the wattage and other specs should be written on a plate on the supply itself. What about amperage?

If you're going to have spare RAM and gfx card to test, that's what you should do, there's really no other way. But remember that if the faults are unpredictable, you'll have to run with alternative hardware configurations long enough to make the problem re-appear, or to be reasonably sure that it has gone away.

[Halindir]

Originally Posted by Japo

Now that makes more sense. Anyway the wattage and other specs should be written on a plate on the supply itself. What about amperage?

If you're going to have spare RAM and gfx card to test, that's what you should do, there's really no other way. But remember that if the faults are unpredictable, you'll have to run with alternative hardware configurations long enough to make the problem re-appear, or to be reasonably sure that it has gone away.

Im not sure..But heres what was written on the side of the PSU:

| AC-I/P | 115-230 | 8/4A | 60-50Hz |
| DC OP | +5V | +12V | -12V | 5VSB |
----------| 28A | 19A | 0,8A | 2A |


The freezes are somewhat predictable, in that they only happen during games.

[The Fifth Horseman]

Your GPU's specs not just state a 300 W PSU is an absolute minimum but also recommend a 450 Watt one.
And yes, the under-specced amperage on your PSU's +12V line means that it cannot supply your GPU with enough energy under heavy load.

My dad will be sending me, in about 4-5 days, 2 1GB RAM - In case one or more of the ones in my pc are damaged - and a not-so-good graphics card - to test if the freezing occurs on that one too.

Chances are that if you add the RAM modules to the existing setup, the freezes will increase in frequency, and if you replace the graphics card instead - they will stop.

[Japo]

My money is either on the card, or RAM. It will have to be tested separately. Obviously if replacing the card solves the problem, there's no need to test further. This would be good news (in a sense since the problem is easier, but on the other hand one defective gfx card is a more expensive loss), but it is not certain.

Make sure you explore all avenues. As for drivers you already tried the Omega package. My first choice is usually to un-install the currently installed driver from the Windows Device Manager, and then instead of downloading another version from any website, including the manufacturer of the card, I like to get it from Windows Update--that way it's further quality-assured by Microsoft--or if applicable from the OEM's website--e.g. Dell, etc.

And if you could get your hands on another power supply, at least temporarily for testing, of course it would also be good. Although if your current supply has 350 W, I think that would be enough, if the requirements are the same as mine with 1 GB (though probably different manufacturer?). What's the exact part number of the card? Horseman, where did you get the 450 W figure?

As for RAM, the idea is not to add the new sticks, but to use them to replace the old ones. If replacing the gfx card doesn't solve the problems, and it comes to testing RAM, Halindir you should run with different RAM configurations, making sure that all the sticks have the same exact specs and come from the same manufacturer and exact model. First run tests from the Ultimate Boot CD overnight for all your sticks, but even if it tells you all are good, start using your computer in a stripped down configuration for testing. You'll have to make do with little run while you assess if the problem goes away or not and depending on what. It's best to run with one single stick, that way you're sure there aren't conflicts, which may appear among different sticks even if all of them are good. If you're absolutely sure that two or four sticks were bought together, you may use them together and be quite confident, but with one stick you can't be wrong. If the problem persists, try with a different configuration (e.g. with the new sticks you'll be receiving, instead of your current ones).

[Halindir]

Originally Posted by The Fifth Horseman

Your GPU's specs not just state a 300 W PSU is an absolute minimum but also recommend a 450 Watt one.
And yes, the under-specced amperage on your PSU's +12V line means that it cannot supply your GPU with enough energy under heavy load.

Chances are that if you add the RAM modules to the existing setup, the freezes will increase in frequency, and if you replace the graphics card instead - they will stop.

Originally Posted by Japo

My money is either on the card, or RAM. It will have to be tested separately. Obviously if replacing the card solves the problem, there's no need to test further. This would be good news (in a sense since the problem is easier, but on the other hand one defective gfx card is a more expensive loss), but it is not certain.

The idea is not to add the new RAM, but to use it to replace the old one. If replacing the gfx card doesn't solve the problems, and it comes to testing RAM, Halindir you should run with different RAM configurations, making sure that all the sticks have the same exact specs and come from the same manufacturer and exact model. Even only 2 GB should be enough to make testing comfortable. If the problem persists, try with a different configuration (e.g. with the new sticks you'll be receiving, instead of a pair of your current ones).

So I should definitely get a new PSU.
And I should test with the other graphics card first.

Im not sure what you mean by different configurations, specs and such..The four sticks in my pc looks identical, and I guess they're the same model.