New video card!

[C38368]

Sadly, the title belies the truth. I've been happily running my super-awesome half-won rig for about 18 months now. Sexy black-and-red colour scheme and an HD4870. Not the best card on the planet any longer, but still competent enough for my needs.

Until Thursday.

I'd gone out, as I often do, for ThuNBaR, and got home at around 1 in the morning, again as I often do. My desktop, oddly, was off. Strange, because I don't turn it off, but it happens now and again that a power surge will trip it, and it won't power back on after a mains loss. "OK," think I, "no problem," as I go for the power button.

And then it happened.

My video card exploded.

Fire, smoke, boom. Well, more of a pop, really, but it computer terms, it might as well have been an explosion.

Not sure if there was any collateral damage to the rest of my machine or not. The standby LED on my motherboard remained on; a hopeful sign. I just hope that it was, in fact, my video card failing and not (as I recently suggested in another thread) my power supply. That would really ruin my day...

Anyway, here's a pic of the unfortunate card:


Should have an HD5870 here by Thursday. Not sure what I'll do for cooling just yet.

[Sheepeep]

As a 4870 owner, I am neither surprised nor dismayed to see a dead one. There seem to be all manner of problems with them, and mine is no exception. Now it seems that yours can join the club.

Still, both cards were a pretty serious investment, so I'll feel for your wallet instead.

[Mastershroom]

My venerable 8800GT has cranked out its last pixels in the past week, although its death was more slow and peaceful than yours. Unfortunately, pretty much any new card these days is double-slot, which my motherboard definitely won't accommodate.

[>Omen<]

I always suspect a bad PSU and/or one that doesn't have very good protection features built in when I hear stories like that. Where I live we don't seem to have very stable current in the wall sockets. I've seriously considered getting a line conditioner.

[*The.Doctor]

My 4870 crapped out on me a few months ago. Worked fine one day, shutdown, turned it on the next morning and nothing. Just black screen and my mobo showing a video error code on the LED display. I got it replaced and still have the new one in a old computer as i'm rocking it with a GTX 470 now.

Have fun with that 5870 though!

[kow_ciller]

5870 shouldn't need any additional cooling if its reference design.

Looks like one of the VRM's overheated or something. You'll have fun with a 5870 though

[C38368]

The 5870 won't need additional cooling, no. But I now have a somewhat pointless water loop in my desktop, and that just seems like such a bloody waste...

With respect to this card, it was one of the first 4870s to hit the market. You can see that giant Vitec inductor in the picture I posted. Those were replaced very quickly in the second revision, as they would regularly hit 100ºC and stay there. Keeping them cool was a massive issue. In this case, I suspect that it wasn't a PSU issue. The MOSFET that blew is one of the ones that, along with the inductor, controlled core voltage. As I'd mentioned in another thread, the card was consistently lighting up the D601 LED: critical core power fault. I'll be checking the PSU as best I can, to be sure, but I think this was just a card whose time had (violently) come.

Sad to see it pass, annoyed to have to replace it now that I'm back in school, but what the hell, right?

[>Omen<]

I would never buy a GPU that reaches anywhere near those temps. In fact I don't even like the 85c-95c range I'm hearing regarding Fermis.

My idea of a comfortable GPU range is more like 65c-75c, 80c tops, and only if playing a high resource game for extended length.

[C38368]

Originally Posted by >Omen<

I would never buy a GPU that reaches anywhere near those temps. In fact I don't even like the 85c-95c range I'm hearing regarding Fermis.

My idea of a comfortable GPU range is more like 65c-75c, 80c tops, and only if playing a high resource game for extended length.

Not the GPU. This was all in the voltage circuitry, which runs hot to begin with. Granted, this was on the extreme side, but is hardly in the realm of inconceivable. The core itself idled at around 35ºC and loaded to about 48ºC.

I did a little more investigating today, and it turns out that the MOSFET that blew wasn't exactly well-contacted by the waterblock I had on. I'm wondering now if it just got too hot, too often and finally just gave in. The proverbial ticking time bomb.

[>Omen<]

Well in my haste I used the term "GPU", but I really meant the card in general. It isn't very hard to get such info right after or even before their debut. Anyways, I'm sure you'll like the 5870 better, but I myself am waiting for the 475.