Startup

SpyderTracks

We love you Ukraine
This is a splitter:

51iX67l0p2L._AC_SX466_.jpg
 

ChaAli

Bronze Level Poster
I got off technical support. They think it might be the GPU. So they will send a replacement GPU out. Appreciate the help.


A splitter is just that, it's one cable that leads to 2 rather than 2 individual cables.

It's fine if it's 2 cables with just 1 part of the splitter in each port, but if it's only a single cable with both ends of the splitter plugged in then that's your problem.
It two cables. One is a splitter with one part hanging out and one inserted into the mother board.
 

SpyderTracks

We love you Ukraine
I got off technical support. They think it might be the GPU. So they will send a replacement GPU out. Appreciate the help.



It two cables. One is a splitter with one part hanging out and one inserted into the mother board.
No, the one into the motherboard is the CPU power, that has nothing to do with the GPU.

We need to know if the cable feeding both ports of the GPU is a splitter cable ie coming off one cable and then split, or if it’s 2 individual feeds from the PSU

If it is, as soon as you plug in your new gpu, it’s going to fry again.
 

ChaAli

Bronze Level Poster
I have images above of the GPU. It looks like a single wire with two 8 pins into the GPU. The main core of the wire runs into one 8 pin that is linked across to a second 8 pin with an addiitonal wire.

There are images above. The one to the far right shows the GPU wiring.
 

ChaAli

Bronze Level Poster
A splitter is just that, it's one cable that leads to 2 rather than 2 individual cables.

It's fine if it's 2 cables with just 1 part of the splitter in each port, but if it's only a single cable with both ends of the splitter plugged in then that's your problem.
Honestly - thats what it looks like. A single wire with two 8 pins connected to the GPU.
 

SpyderTracks

We love you Ukraine
I have images above of the GPU. It looks like a single wire with two 8 pins into the GPU. The main core of the wire runs into one 8 pin that is linked across to a second 8 pin with an addiitonal wire.

There are images above. The one to the far right shows the GPU wiring.
Yes, I know, that's why I was asking for you to confirm what I thought I was seeing.

In that case it's a splitter.

That's a bad thing.

1 PCIe PSU lane can carry a maximum of 150W, that's weather its just a single cable or a splitter serving two ports.

The motherboard PCIe lane can carry a max of 75W BUT there are other devices like SSD's that require PCIe power, so that number immediately reduces.

You've got a 350W GPU in that build.....

So there's a shortfall of 125W that the GPU is going to demand from the build, that's going to overload the PSU, the motherboard and likely the GPU will get a surge leading to a fry.

It's not your fault at all, PCS have already been told about this because it's happened a few times.

Luckily you've got a very high end PSU in your build which will limit the overload, but it's still not a good setup as it currently is.

It needs a second PCIe power cable run from the PSU to the second 8pin on the GPU. This should be in your welcome pack, sounds like you've already installed it. Make sure you use both cables to feed the new GPU.

The PCIe outputs on the PSU are clearly labelled, they'll look something like this:

1612473094219.png
 
Last edited:

PapaSmoke

Member
Yes, I know, that's why I was asking for you to confirm what I thought I was seeing.

In that case it's a splitter.

That's a bad thing.

1 PCIe PSU lane can carry a maximum of 150W, that's weather its just a single cable or a splitter serving two ports.

The motherboard PCIe lane can carry a max of 75W BUT there are other devices like SSD's that require PCIe power, so that number immediately reduces.

You've got a 350W GPU in that build.....

So there's a shortfall of 125W that the GPU is going to demand from the build, that's going to overload the PSU, the motherboard and likely the GPU will get a surge leading to a fry.

It's not your fault at all, PCS have already been told about this because it's happened a few times.

Luckily you've got a very high end PSU in your build which will limit the overload, but it's still not a good setup as it currently is.

It needs a second PCIe power cable run from the PSU to the second 8pin on the GPU. This should be in your welcome pack, sounds like you've already installed it. Make sure you use both cables to feed the new GPU.

The PCIe outputs on the PSU are clearly labelled, they'll look something like this:

View attachment 22012
Thanks for your help. I jumped in this conversation because I have same issue ( bought in PCS ). I just don't understand how this setup could pass test and few days after no signal on the screen.
 

SpyderTracks

We love you Ukraine
Thanks for your help. I jumped in this conversation because I have same issue ( bought in PCS ). I just don't understand how this setup could pass test and few days after no signal on the screen.
Sorry, I'm getting really mixed up as this is the 3rd dead GPU I've been toing and froing from o_O

In this instance, and by the sounds of it in your instance as well, the GPU hasn't been wired correctly which COULD have led to an overload on the GPU line which could have fried the GPU. Weather that is the root cause of the GPU was just a dud and gave up no one can say.

BUT best practice going forward is to have 2 individual PCIe cables running the ports.

It's not good, PCS have been pulled up about this by us a few times, and this shouldn't be passing QC. You can get away with it with lower end cards, but IMHO as soon as you're at a 3080/3090 or 6800XT / 6900XT then you really HAVE to have 2 individual PCIe cables running to the GPU, otherwise it's going to overload the whole thing.

They should have that as one of the first checks on any high end build at the QC stage.
 

ChaAli

Bronze Level Poster
Thanks for your help. I jumped in this conversation because I have same issue ( bought in PCS ). I just don't understand how this setup could pass test and few days after no signal on the screen

I appreciate your help - alot. I have paid a lot of money and both our computers are high end. It shouldn't be happening at all. I was offered a replacement GPU and accepted with the option to decline at any point. I accepted mainly, as the alternative would be sending the computer back and waiting 2-3 weeks.

But to be honest - it is unacceptable. I accept wait times in view of the current pandemic. I am very disappointed. In view of the above and the price I have paid. I'll be sending the computer back. Otherwise I might as well have built it myself. Not going to be tinkering with wires.
 

ChaAli

Bronze Level Poster
Sorry, I'm getting really mixed up as this is the 3rd dead GPU I've been toing and froing from o_O

In this instance, and by the sounds of it in your instance as well, the GPU hasn't been wired correctly which COULD have led to an overload on the GPU line which could have fried the GPU. Weather that is the root cause of the GPU was just a dud and gave up no one can say.

BUT best practice going forward is to have 2 individual PCIe cables running the ports.

It's not good, PCS have been pulled up about this by us a few times, and this shouldn't be passing QC. You can get away with it with lower end cards, but IMHO as soon as you're at a 3080/3090 or 6800XT / 6900XT then you really HAVE to have 2 individual PCIe cables running to the GPU, otherwise it's going to overload the whole thing.

They should have that as one of the first checks on any high end build at the QC stage.
I appreciate your honesty and help.

It shouldn't be happening. Period.
 

PapaSmoke

Member
Sorry, I'm getting really mixed up as this is the 3rd dead GPU I've been toing and froing from o_O

In this instance, and by the sounds of it in your instance as well, the GPU hasn't been wired correctly which COULD have led to an overload on the GPU line which could have fried the GPU. Weather that is the root cause of the GPU was just a dud and gave up no one can say.

BUT best practice going forward is to have 2 individual PCIe cables running the ports.

It's not good, PCS have been pulled up about this by us a few times, and this shouldn't be passing QC. You can get away with it with lower end cards, but IMHO as soon as you're at a 3080/3090 or 6800XT / 6900XT then you really HAVE to have 2 individual PCIe cables running to the GPU, otherwise it's going to overload the whole thing.

They should have that as one of the first checks on any high end build at the QC stage.
Only 2 dead GPU's 🤬 me and CHAALI. I will send mine tomorrow. I'll use 2 8pins for new GPU.
 

SpyderTracks

We love you Ukraine
Only 2 dead GPU's 🤬 me and CHAALI. I will send mine tomorrow. I'll use 2 8pins for new GPU.
There was another on a lower end build as well, I was getting really confused who had what system and kept having to delete comments because it was the wrong thread :)
 

PapaSmoke

Member
I appreciate your help - alot. I have paid a lot of money and both our computers are high end. It shouldn't be happening at all. I was offered a replacement GPU and accepted with the option to decline at any point. I accepted mainly, as the alternative would be sending the computer back and waiting 2-3 weeks.

But to be honest - it is unacceptable. I accept wait times in view of the current pandemic. I am very disappointed. In view of the above and the price I have paid. I'll be sending the computer back. Otherwise I might as well have built it myself. Not going to be tinkering with wires.
I am very disappointed too. I thought this should be good end for amount we paid, connect all cables and play some games. It's end up dead GPU because someone didn't connect GPU as suppose to be. unacceptable.
 

ChaAli

Bronze Level Poster
In the OPs case, I am not convinced it is being caused by the lack of the second PCI-e cable...sure, it doesn't help...but, for it to happen when turning the system on for the first time (least I assume it was the first time) meaning the GPU under no load makes me think it really is an inherent weakness somewhere in those cards, whether a capacitor or transistor or something that is prone to breaking on movement.

I recall saying the same thing on another thread yesterday about another GPU failure of some description

I have called technical support - I'll be sending it back to them to figure out what the root cause is.
Mentioned the issue with the wiring.

Many thanks for all the help.
 
Top