Radeon HD 3870X2 RMA

[Wills]

Hi Scan,

Posting on behalf of a friend who recently (March) purchased said graphics card from you for use in a new pc build he was building (Gigabyte GA-X38-DS4). The machine Worked fine until about 2 weeks ago when, the pc suddenly won’t post. We noticed that excessive heat was being pumped out of the gpu cooler even pre/during POST, however this could be normal and due to the power management stuff of the GPU not kicking in until after post (correct?). Anyway, we RMA'ed the card last week, only to have it tested (By Lee I believe) and confirmed working. The card was received back from SCAN yesterday (after paying £££ for your testing), and upon trying to use it again, the same issues occurred. Okay, so, here's what we have done to test ourselves:

1) Took card out of pc, and tested another graphics card in said PC, Posts fine, and windows boots, Left it running for an Hour + and working fine
2) Took suspect 3870x2 and put it in a completely different machine (Asus A8N, Linux, but not relevant). And everything worked fine, Did 10 + reboots (with about 30 secs of running after each reboot). (This points the card being fine however).
3) We then proceeded to take the suspect graphics card and took a spare Gigabyte motherboard and took the CPU (Q6600), Ram, Hard disk, and connected it all up to the spare gigabyte, once all was connected we turned it on, and it posted, but Vista would only work in safe mode (or VGA mode), even then the machine only worked for 10 mins then the card started blowing out more vast amounts of heat, then the PC rebooted, and this time, failed to post (GPU Cooler still screaming).
4) Fiddled with 1 stick of ram instead of 4 (we have 8gb in total), tried all ram sticks individually, tried bios resets. So we tested the spare graphics card (that works) in the pc we had assembled (with the spare Gigabyte motherboard) and it posted, and vista worked fine in normal (non-safe) mode.
5) tried the HD 3780X2 with a different PSU in the original PC (GA-X38-DS4) and machine didn’t post.

So we rang scan this morning and reported the issue, explained what we have explained again above and was issued another RMA number and plan to rma again, after we are happy here that it is definitely faulty.

I’ve been a scan customer for a few year and you’ve been great, just seem to be experiencing a bit of a hic-up here. We are almost certain that what we are experiencing here is a graphics card issue, so any thoughts / ideas?

Thanks,

Regards
Wills
P.S If you want the original RMA number I can PM you with it if it would be useful).

[Wills]

Come on guys ...

[wesleyaldred]

Please PM me your *NEW* RMA number I will make sure when it arrives my most senior technician works on this.

Wesley

[Wills]

Thanks sending a PM now

[Wills]

Sent

[Wills]

Okay, so we did some further testing over the weekend in a completely different Gigabyte based machine. Heres what happened Link . . .

We then tried it in the same Asus Linux box as before, and it worked ok.

Conclusion: The card is warped(bend) and asus just happen to have a tighter fitting PCI-E Slot than Gigabyte.

From the video you can quite obviously see something its seriously wrong, and this behaviour can be replicated in 3 different motherboards (All gigabytes I might add).

Where do we go next with this scan ?

[Wills]

Bump...

[Lee H]

Wills,

I've alerted the returns team that you need an update/assistance but in all honesty the easiest and quickest way to do this in future is to use the OLQ system and use that for communications with the testing team.

Best Regards,

[wesleyaldred]

Wills,

We cannot do anything further now until you have returned the card, please use the RMA number already issued to return the card to Scan.

Wesley

[Wills]

Cheers, Ill get it posted.

[sleepyhead]

Okay, so we did some further testing over the weekend in a completely different Gigabyte based machine. Heres what happened Link . . .

We then tried it in the same Asus Linux box as before, and it worked ok.

Conclusion: The card is warped(bend) and asus just happen to have a tighter fitting PCI-E Slot than Gigabyte.

From the video you can quite obviously see something its seriously wrong, and this behaviour can be replicated in 3 different motherboards (All gigabytes I might add).

Where do we go next with this scan ?

Did you try an install outside of a case? It could be a situation whereby the motherboard/case alignment is causing the card not to seat correctly when screwed in.

[Wills]

In 4 or 5 cases...... Bit unlikely isnt it?


Anyway the issuse is now resolved, and Lee has confirmed a fault, thanks guys. The customer in question will ring you later regarding a replacement

[wesleyaldred]

Wills

As per the telephone call with Amanda in my department just now, the card has not been tested as Faulty, we were trying to help in offering as a gesture of goodwill a brand new card which did out perform the original card.

As per the instructions on the telephone however, we understand that the customer in question did not wish to accept the better performing card and required his original one returned.

This will be actioned for him today.

Wesley

[Wills]

Hello,

During the telephone call, the customer was offered the 4850 (the memory size was not specified) so he assumed that this would be the 512mb version which he declines as after some quick research he concluded was not of equal performance to his current card so he declined the offer,the only other option that was give was to return his old 3870x2 which he accepted. The customer rang back 20 minutes later to confirm what had been offered (as aparently the during the telephone call the rep wsnt very specific) and was told that actually he had been offered the1gb version (this was not what was expected, he thought you had offered the 512mb version). So he then proceeded to accept the offer, at which point he was told that it was too late and that his old card had already been returned to him...
So, having paid £272 just over 6 months ago now to be offered a ~£130 card the customer didn't beleive that this was a reasonable offer.

This asside, the root of the problem still remains. You deemed the card "not faulty" and after the customer questioned how you tested the card, he was told it was tested in an ASUS motherboard and that you refuse to test in it a gigabye board because you said if it works in one it should work in any?
As far this SHOULD be true, as we have described in the numerous posts above, including the video, it doesn't function as correctly in AL LEAST 3 gigabyte boards. Okay, so maybe one of even at a stretch 2 peoples boards which we used to test could have been faulty, but this is very unlikely especially as said boards work perfectly with other hi end ATI and Nvidia graphics cards. The customers pc its self has been running perfectly since he brought the card (and the rest of his PC) in March (ish).
So the question i suppose is if you are sure it will work in any motherboard why do you refuse to test it in a Gigabyte board you know is working fine?

Another point to add, on your returns tracking website, it states that a fault HAS infact been found during testing by Lee. This contradicts what you have said during the phone calls?

The customer would be quite happy to accept the 1GB 4850 just to end the trouble, however he was told this is no longer an option, so we are stuck, again, with a large red paperweight!!!!

Please respond to the above issues,

Thanks,

Regards
Wills (Only the Messenger...)

[wesleyaldred]

Hi Will's

Actually I was listening in on this call it was Amanda that rang him, during the telephone call Amanda specifically said to the customer "would you like me to give you the LN number of the item so you can have a look at it", to which he replied NO. In all honesty there was no reason for him to make an assumption and certainly no reason for him to make the wrong one.

During the original call he was specifically advised we were making this offer as a gesture of goodwill and that we were offering him a better card which out performed his old one.

We did not advise him that his only option was to have his original card returned, he specifically rejected our goodwill offer and advised us to return his "faulty" card and he would sell it and buy himself another.


The customer was advised that if the card runs perfectly fine in any 1 specific motherboard then the card is NOT physically faulty, if you are experiencing problems with GIGABYTE boards then the only remaining obvious explanation is the card is not compatible with the Gigabyte board, this does not itself deem the card faulty.

Ultimately anyway despite all of this we were simply trying to help the customer by STILL offering to replace the card that was not physically faulty with a brand new, better specification card, that out performs the original card in all benchmarks.

This offer was rejected by the customer and following his request to return his original we actioned this. I'm sorry that he rang back later realising that we were being sincere in our offer and attempting to now accept the offer but ultimately it was to late, we had already actioned HIS specific request.


Wesley

[Wills]

Hi Wesley,

Ok, The only point we can make is if as you say it is a compatability issue, the customer was not informed at the time of purchase that it was not compatible with Gigabyte boards, and it has been working for over half a year fine, So there is a deeper issue here.

Also, why does your returns tracking website say a fault was infact found?

Is there nothing we can do to make you try a gigabyte board, even if it means sending the card back again?