AMD Gives Statement on the PCI-Express Overcurrent Problems

Pls watch this https://www.twitch.tv/buildzoid/v/75850933

00:53:15 till end...

 

Science Studio said the issue affected an old motherboard, but was fine on a newer Asus gaming board. I would also think motherboards that have supplemental power connectors for the PCI-E slots like the Gigabyte GA-Z77X-UD5H and GA-Z97X-UD5H would be fine too.

 

You are here one day and you already jumped only on AMD forum (as the NVidia user) for bashing AMD cards and attacking old members?

 

Is it 14800 GPU score ? I can't see it.

 

yeah i watched it.

So what exactly is this , an overclockers page?
I see a random around 15-16 year old boy with an interesting understanding of electrostatic safety during being wired with random microphone/audio tech and simultaneously screwing around on a Radeon 480 and yeah the moment I saw the tower... I mean taste is taste. But this doesn't look like a very organzied test bench.
Yeah and the results, i am missing the results...
Or are you saying I should ignore the 20 main reviews and listen to his 5 second thumb down "review" in the end?

PS: so i actually just browsed shortly before posting this. and it seems this guy has a history of destroying his own cards by mishandling.
but after seeing that workplace i am not surprised.

https://www.youtube.com/watch?v=ghoiSZtqw-c

PPS: i am not the kinda guy that says: ESD wrist band in ANY case...
but this? with all the chaos he had going on and moving around without touching ground probably loading up.
considering this as professional... if people even dont accept jay or kyle or whoever here... is gonna be hard.
but yeah i am the kind of guy that says if you handle your hardware like they are melons on a fruit market... then you shouldnt be surprises components may fail.

 

That is interesting. So AMD went a weird route where they split the vcore control between the 6pin and PCI-E slot.

The 6pin has 3 grounds too, just why would they do that?

 

Do you think your post is going to change his "influence" to what you consider "positive"?I don't use to share vase opinions in general and specially his 480 defense but your comment is only adding gasoline to a flame. :flame2:

You are certainly not an expert making friends or resolving arguments.

If you are really concerned about the positive or negative "influences": don't worry.

Mods are not very far ready to advice or zap anybody account (new, old or even a "reborn" Hydra head) if the situation goes beyond limits.

After only (?) 24 hours in these forums it seems that you have made some friends...and "friends". Don't count me in neither side.

 

Whats making all this worse now the truth is being explored, AMD gave the 6pin 8pin spec Never used it for the Vcore because the card draws the extra from the PCI-E slot.

It's totally messed up when you realized the actual 6pin is an 8pin spec yet they never put an 8 on there and instead broke Compliance on the mobo slot.

 

Just to enter your assessment: I don't understand it.

If they give the 6pin an 8pin spec (overspec it), why does the 6pin not draw the majority of the power and the slot would stay at 75W?


on another note. i just uploaded this and since the main parts of official pci-sig specs are not open source, maybe this document has possibly any hints for us?
it is a whitepaper of the company that supplies PCI-SIG with oscilloscopes apart from the german company rohde and shwartz. and this one is called teledyne lecroy.
and no unfortunately there are no directly stated power draw tolerance levels described in there. else i would have screamed out loud already with the newly found insight. there is only some pointing into which tests were added for 3.0 spec logo test in april 2013 for example.
uh yeah did i mention that they mention a new v0.9 seg electrical spec test starting in 2013?
but i wont come to any more chewed for thoughts and just leave this here:

http://www.pdf-archive.com/2016/07/03/pcie-3-0-gen-compl-test/

there are many references to this document so sending it through robots/crawlers may yield some "underground" or even archived spec document?

if anybody has revision 2.0 or higher (i know 2.0) exists for sure
of this one
http://read.pudn.com/downloads166/ebook/758109/PCI_Express_CEM_1.1.pdf
can he link it?

 

Go to the Errata for the PCI Express Base Specification Revision 3.0 and look at page 17.

https://pcisig.com/specifications/pciexpress/specifications/ECN-WirelessDisable_a.pdf

 

Erm...

http://composter.com.ua/documents/PCI_Express_Base_Specification_Revision_3.0.pdf

I mean it's neat you're morphing into an electrical engineer... but lets not forget just recently you thought the motherboard controlled the power delivery to the GPU. I mean that in a nice way...as a professional in the field.

 

:thumbup:


do you / do we have the CEM (Card Electromechanical) specification not the base spec in a higher revision also? (3.0)
because ... yeah... because its in there.
we base our tolerance level of 8% of the 12V still on an outdated revision jpeg that was flooded into the forums since this topic has broken out.

at least we could clarify for boards after july 2013 it seems. if we can get it.


from their download section, it is inaccessible ofc, as all their newest spec rev.

"PCI Express Card Electromechanical Specification Revision 3.0
This specification is a companion for the PCI Expres...view more
3.x Specification July 21, 2013"

 

For what purpose? So you can make another dozen posts that are mostly irrelevant noise trying to obscure the base issue? AMD already conceded. You should do the same.

 

Miners beware, esp with multi-GPU systems. Some guy mining without issues for 3 years on 3 280x burnt out the 12v pins on 24-pin mobo connector after mining for 6 hours on 3 480s. Must use powered pcie risers to avoid potential issues.

https://bitcointalk.org/index.php?topic=1433925.msg15438988#msg15438988

 

i dont get why people are concerned, who the heck is going to buy a reference card on guru3d? even if for some reason(compact case) you like a blower fan, AIB partners will do it better

 

That's kinda weird. I'd say it's more likely that over the 3 years his mobo connector probably developed some resistance on the pins. There's no reason 3 of these should fry a 24 pin. 280x mining runs around 230 watts or so right?

 

OMG! What AMD did is insane. Heck according to this person that AMD Violated PCI-E spec with the grounding for a 6 pin connector. I feel that AMD can't fix this via software aka drivers but maybe they could find a workaround via the BIOS. I cant see the AIB partners using a reference design and just slapping their cooler on it. Man AMD really messed up big time with this. Heck even this person said that this card could have ran off a 6 pin without issues if they have not split the part of the Vcore to the PCi-E connector.

Very good question. Makes me wonder if AMD would have to recall all of the RX 480s because of this issue.

 

The 280x has more than 1 6-pin connector (I believe 8-pin + 6-pin). With 3 480s drawing close to 300w sustained thru motherboard, something has to give, esp on budget boards.