Card died shortly after connecting two seperate PCIe cables.

[Katze]

So I've been told that it is generally better to use two separate PCIe power cables to plug into a GPU, specifically power hungry ones.

 

I recently ordered a 3070, and got a buyer for my 1070 sooner than I anticipated, so I've been using an old 980 until my new card gets here.
I was running it on a single PCIe cable for a few days, then decided to plug in a second one, and turned over my PSU so it was face up while I was there.

 

I instantly noted more frames in Cyberpunk with an overclock, but after about an hour my PC rebooted and now there are artifacts on the screen... rip.

 

Could the two cables have killed the card, or was it just it's time? or maybe the OC, it was +150 mhz.

 

Also, can the second cable be bad? I'm a little nervous to connect my 3070 to it if that's the case.

 

 

[Skiiwee29]

Dial back the OC. Usually with Artifacts, its the Memory more so than the Core. If the core is unstable, it will just lock up or crash outright, not produce artifacts. 

[Katze]

Just now, Skiiwee29 said:

Dial back the OC. Usually with Artifacts, its the Memory more so than the Core. If the core is unstable, it will just lock up or crash outright, not produce artifacts. 

The artifacts are there even in bios, the OC isn't percipient.

[Skiiwee29]

Just now, Katze said:

The artifacts are there even in bios, the OC isn't percipient.

boot it to windows and try and run Furmark. 

[Katze]

Just now, Skiiwee29 said:

boot it to windows and try and run Furmark. 

I haven't tried to boot into windows yet, but the artifacts are there from the start, I added a picture.

[Skiiwee29]

Just now, Katze said:

I haven't tried to boot into windows yet, but the artifacts are there from the start, I added a picture.

looks like you have a dead card most likely with faulty RAM chips. I would RMA it. 

[Katze]

1 minute ago, Skiiwee29 said:

looks like you have a dead card most likely with faulty RAM chips. I would RMA it. 

It's like 5 years old at this point, can't really do much, it's an old card as it is.

[Skiiwee29]

Just now, Katze said:

It's like 5 years old at this point, can't really do much, it's an old card as it is.

ahhh, so this is happening on the 980? I was thinking it was the 3070. Then yea, you will just have to nurse it along until your 3070 arrives or just use the iGPU on your 8700k assuming thats the system this is on and not do any gaming until then. 

[Katze]

1 minute ago, Skiiwee29 said:

ahhh, so this is happening on the 980? I was thinking it was the 3070. Then yea, you will just have to nurse it along until your 3070 arrives or just use the iGPU on your 8700k assuming thats the system this is on and not do any gaming until then. 

Yeah sorry I might not have been clear. Yeah I'm using a backup, backup 660 ti now lol.

 

But there is no real way that using two PCIe cables would have been the cause right?

[Skiiwee29]

Just now, Katze said:

Yeah sorry I might not have been clear. Yeah I'm using a backup, backup 660 ti now lol.

 

But there is no real way that using two PCIe cables would have been the cause right?

No. The only different from using a daisy chain to 2 dedicated is the resistance to the power. There is a measurable level when using a piggy back and it actually does have a negative effect on FPS as a result. 

[Katze]

2 minutes ago, Skiiwee29 said:

No. The only different from using a daisy chain to 2 dedicated is the resistance to the power. There is a measurable level when using a piggy back and it actually does have a negative effect on FPS as a result. 

Yeah that's what I thought. I do however want to make sure it wasn't the cable, so I'm thinking about using both cables on my 660 ti that is currently plugged in.

That wont be too much for the 660 ti right? I'm assuming it will be fine.

[Skiiwee29]

Just now, Katze said:

Yeah that's what I thought. I do however want to make sure it wasn't the cable, so I'm thinking about using both cables on my 660 ti that is currently plugged in.

That wont be too much for the 660 ti right? I'm assuming it will be fine.

If the 660ti takes 2 cables, then no it wont be to much and is required. The card will not output if all the PCIe power isnt supplied that it is expecting. 

[Katze]

Just now, Skiiwee29 said:

If the 660ti takes 2 cables, then no it wont be to much and is required. The card will not output if all the PCIe power isnt supplied that it is expecting. 

Yeah that's what I thought, I've just always ran it off a single PCIe cable that has the piggyback plugs back when I used to use it.

[GamerDude]

Just outta curiosity, what PSU are you using?

[Katze]

1 minute ago, GamerDude said:

Just outta curiosity, what PSU are you using?

Corsair RM750x Gold 750W

[Loki0111]

My speculation is the OC eventually did it in. If the thermals were too high for an extended period that absolutely would cause damage.

 

This is one of the reasons I have never understood OC'ing parts, it is adding additional potential for failure over time.

 

If I need a faster part I can just buy something that meets that spec.

[Katze]

Just now, Loki0111 said:

My speculation is the OC eventually did it in. If the thermals were too high for an extended period that absolutely would cause damage.

 

This is one of the reasons I have never understood OC'ing parts, it is the potential for causing a failure over time.

 

If I need a faster part I can just buy something that meets spec.

Was literally the first time it had been OC'ed lol.

[Loki0111]

9 minutes ago, Katze said:

Was literally the first time it had been OC'ed lol.

So usually, artifacting is an indicator of an unstable OC, that or bad VRAM. Given this is a proven card I'm going to assume the VRAM is fine. If pushed way outside of the envelope, yes OC'ing it can damage parts. This is why carefully monitoring temps is so important when initially OC'ing a part.

 

I suspect this GPU probably has a decent amount of usage already on it which would make it even more susceptible to failure if pushed. If it was just the fact the OC was unstable it should have returned to normal function after removing the OC and returning the card settings to spec.

 

So I don't want to get into a whole thing about this but CPU's for example are usually guaranteed by manufacturers for 10 years when operated within spec. They will often last well past that if power and thermals are carefully controlled over their lives.

 

When you overclock you are effectively trading years of service for additional performance. A mild OC might knock a year or two off the chips lifespan, a major OC might knock 5 years off. If you are someone who upgrades every 5 years you probably won't care since you got your use out of that processor but that chip will be in much worse shape then one that has been operating within spec for that period of time.

 

This is also why you never want to buy used parts that have been operating OC'd for an extended period of time.

 

 I work with a lot of server gear and the most common cause of failure I regularly see is due to a parts operating outside of spec for thermals.