Yeah since Intel came out with Core M and pushed that IP over to desktops, I have only had two major upgrades to my systems, and one of them was because of the change in BIos types. I really didn't want to upgrade my dual xeon system but the new graphics cards just wouldn't boot up.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
RX 480 is apparently killing pcie slots
- Thread starter Ieldra
- Start date
Yeah since Intel came out with Core M and pushed that IP over to desktops, I have only had two major upgrades to my systems, and one of them was because of the change in BIos types. I really didn't want to upgrade my dual xeon system but the new graphics cards just wouldn't boot up.
Lord Risky
AMDFanboy EchoChamber Member
- Joined
- Jun 30, 2016
- Messages
- 124
My sarcasm sucks, but anyone still rocking an LGA775 or AM2+/3 chipset realllllly needs to upgrade.
BadmAu5
Weaksauce
- Joined
- Oct 10, 2011
- Messages
- 101
Lord Risky
AMDFanboy EchoChamber Member
- Joined
- Jun 30, 2016
- Messages
- 124
Probably AM3+ as well
Yep yep
pendragon1
Extremely [H]
- Joined
- Oct 7, 2000
- Messages
- 52,288
My am3+ is still running great. Might look at am4 whenever it shows up but I'd still try a 480 on my current board.
Gonna be one hell of an upgrade for you from 280X. Let us know how you like it!
Lord Risky
AMDFanboy EchoChamber Member
- Joined
- Jun 30, 2016
- Messages
- 124
^
To an AM4, huge upgrade.......the 480 will just be a big upgrade.
To an AM4, huge upgrade.......the 480 will just be a big upgrade.
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
Correction: The board has 4/6 phases on the 6pin, not 3/6.
sutyi
n00b
- Joined
- Oct 23, 2010
- Messages
- 37
If they can modify / patch the how this little f*cker in the picture bellow distributes power with a firmware / BIOS update it's all good. If not, well... sh!te.
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
If they can modify / patch the how this little f*cker in the picture bellow distributes power with a firmware / BIOS update it's all good. If not, well... sh!te.
Yeah, I don't know how you could work with that thing, way too small...
Nenu
[H]ardened
- Joined
- Apr 28, 2007
- Messages
- 20,315
Its not a solution but it is helpful. However it is dangerous.
In its default state, half of the VRMs are connected to the PCI-E bus. These feed 1/2 the GPU and memory which will potentially allow more current to flow from PCI-E than the 6 pin connection. However this can be compensated using the voltage controller.
So while bridging the power inputs can "help" reduce saturation of the PCI-E bus power, it still doesnt guarantee that it wont be exceeded because current will now be divided equally between both power inputs.
If the card draws more than 132W + 9% (144W), the PCI-E 12V power delivery will be outside spec.
(PCI-E 12V is specced to 66W +9%)
Bridging the power connections is dangerous because AMD have not implemented the sense pin on the 6 pin power connection, they appear to be sensing actual power availability to detect a non connection.
(This assumes they do have a method of detection, we have to assume they havent completely lost the plot)
Once the power rails are bridged, both the 6 pin and the PCI-E power are connected to everything on the card simultaneously.
The card is no longer able to realise if the 6 pin power connection is missing because there will always be voltage to all circuits.
Should the 6 pin power connector come loose or is left unplugged, the full power draw will come through the PCIE-E bus power connections only.
It will fry the power connections.
I was really wondering about the way that guy just bridged it all together, if I understood what he did. Now the card has no control of the source of the power. He should unplug the pcie cable and see if it dies, lol. Since it'll run that way now.
I orderd the sapphire 8gb won't get it until Friday. Will not even open box. Newegg will refuse to refund offer exchange, will inform Newegg all cards flawed. They will deny, I will call CC and dispute.
Thanks AMD for finally seeing the error of trusting you. Oh by the way here's polaris launch..
Thanks AMD for finally seeing the error of trusting you. Oh by the way here's polaris launch..
Dreamerbydesign
Supreme [H]ardness
- Joined
- Feb 3, 2008
- Messages
- 6,238
No one releases RMA numbers anyways, so that does not matter.
You know I was really excited about the 480. Now thanks to Amd for failing to deliver. I get deal with Newegg BS. Last F@#ing time for AMD with me. In fact I am going out of my way recommending no one buy this JUNK......Fare thee well AMD you were great once.......
LOL you mad bro? What if they patch this thing you still gonna be mad? If you are mad about overclocking why not wait for AIB cards
Getting mad is one thing, but just give them a chance to fix it, or tell us what is going on which ever, its only one day away anyways. You might end up with the product that you wanted to begin with.....
Daniel_Chang
[H]ard|Gawd
- Joined
- Jan 4, 2016
- Messages
- 1,313
I once ordered a product from Newegg with a manufacturer only return policy. Product was defective. I followed it and the manufacturer wouldn't fix it. I then called Newegg and told them of the issue. They emailed the manufacturer on my behalf and got the same response. Newegg told me there was nothing that they could/would do. I said, "That's fine, I'll just issue a chargeback on my CC."
Newegg issued me an RMA/Refund with prepaid shipping label. So, follow their policies, and if that fails, assert your rights. Don't go nuclear as the first step
The way a chargeback works depends on the transaction network.
American Express and Discover are nice. They just return the funds to the purchaser and tell the two parties to work it out. They CAN rebill you, they just can't re-charge the same card without your consent.
Visa and MC are harsher. They (via the bank that issues your card) will "investigate" the issue. If they side with you, the money is taken from the seller, awarded to the purchaser (same as Amex/Disc), but then on top of that, the seller is fined. And while the seller can re-bill you, they can NOT charge you for the fine assessed by Visa/MC. This fine ranges from $20 to $2,000 depending on many variables.
For most purchases under $50, the bank will forgo the investigation and just credit you out of their own pocket. They'll write off the loss.
You know, Nvidia has had their fair share of muckups in the past. At least this should be patchable to make the PCIe issue a little bit more bearable.
I was actually going to test this in an older board... The rant just came out of me. I will watch what the fix is but doubt it will work without killing performance. It almost feels like when AMD asked the last engineer if everything was a go he said yes, They said your fired get out here. As he walks out of the building he starts laughing............
Daniel_Chang
[H]ard|Gawd
- Joined
- Jan 4, 2016
- Messages
- 1,313
Unless they screw you. They converted my 650i Ultra from lifetime to 2 year, stating that mine was never eligible. The product listing I ordered from on Newegg had lifetime warranty. Their product page on the EVGA website showed it as lifetime, and, the box for the product (the physical box!) had their old "Lifetime warranty upon registration" sticker! From what I gather, at least at that time, the guys who did GPU and mobo warranty were different departments. So, I avoid EVGA motherboards, but I still go for their other products.
We have a (99%) confirmation that this thing can be fixed through bios update. I'm pretty sure The Stilt is writing the bios for this card as I write this.
Bios is going to change the distribution ratio between PCI-E and 6-pin connector which is pretty much 50/50 atm, at least for vcore because three upper mosfets are tied to the 6-pin connector and three lower mosfets to the PCI-E slot.
Luckily voltage regulators can be programmed to get more power from 6-pin connector, which means that 3 upper mosfets need to work a little bit harder but it will not be a problem because the power drain will not reach even the conservative safe limits of the fets, not even if you change the distribution ratio 30.6% which is the max value.
Still wondering why AMD is blaming memory when its possible to just give the bios some fairy dust and boom, problem fixed
Oh well, its not the first time that guy outmanages the whole AMD engineering team... x87 AMD patch anyone.
Bios is going to change the distribution ratio between PCI-E and 6-pin connector which is pretty much 50/50 atm, at least for vcore because three upper mosfets are tied to the 6-pin connector and three lower mosfets to the PCI-E slot.
Luckily voltage regulators can be programmed to get more power from 6-pin connector, which means that 3 upper mosfets need to work a little bit harder but it will not be a problem because the power drain will not reach even the conservative safe limits of the fets, not even if you change the distribution ratio 30.6% which is the max value.
Still wondering why AMD is blaming memory when its possible to just give the bios some fairy dust and boom, problem fixed
Oh well, its not the first time that guy outmanages the whole AMD engineering team... x87 AMD patch anyone.
Nenu
[H]ardened
- Joined
- Apr 28, 2007
- Messages
- 20,315
That puts the 6 pin connector even further out of spec.We have a (99%) confirmation that this thing can be fixed through bios update. I'm pretty sure The Stilt is writing the bios for this card as I write this.
Bios is going to change the distribution ratio between PCI-E and 6-pin connector which is pretty much 50/50 atm, at least for vcore because three upper mosfets are tied to the 6-pin connector and three lower mosfets to the PCI-E slot.
Luckily voltage regulators can be programmed to get more power from 6-pin connector, which means that 3 upper mosfets need to work a little bit harder but it will not be a problem because the power drain will not reach even the conservative safe limits of the fets, not even if you change the distribution ratio 30.6% which is the max value.
Still wondering why AMD is blaming memory when its possible to just give the bios some fairy dust and boom, problem fixed
Oh well, its not the first time that guy outmanages the whole AMD engineering team... x87 AMD patch anyone.
They need to reduce power use.
The ram speed drop wont net much power saving.
If they have over volted all cards then they can save some there but its likely some cards will suffer. They could allow RMA replacements of those cards.
Otherwise it will have to be clock drops to allow a safe voltage drop.
We have a (99%) confirmation that this thing can be fixed through bios update. I'm pretty sure The Stilt is writing the bios for this card as I write this.
Bios is going to change the distribution ratio between PCI-E and 6-pin connector which is pretty much 50/50 atm, at least for vcore because three upper mosfets are tied to the 6-pin connector and three lower mosfets to the PCI-E slot.
Luckily voltage regulators can be programmed to get more power from 6-pin connector, which means that 3 upper mosfets need to work a little bit harder but it will not be a problem because the power drain will not reach even the conservative safe limits of the fets, not even if you change the distribution ratio 30.6% which is the max value.
Still wondering why AMD is blaming memory when its possible to just give the bios some fairy dust and boom, problem fixed
Oh well, its not the first time that guy outmanages the whole AMD engineering team... x87 AMD patch anyone.
That is a partial fix and good for temporary reasons and even end users who have already bought the, but not for OEM's and system builders.
thesmokingman
Supreme [H]ardness
- Joined
- Nov 22, 2008
- Messages
- 6,617
The 6pin can supply a lot of voltage out of spec. You guys never realized how overclocking supplies so much power to gpus? Remembers the old days of gtx580s sucking down 800w on extreme cooling lol. That said ppl are guestimating on how AMD will achieve it... guys it's only a day away.
They only have to play with 10-15 watts and that is not under all scenarios. Pretty much scenarios that tax it heavily. It may not be that hard to patch it to meet specs. Now if you are overclocking then thats a whole different story. I wouldn't have recommended the reference design for overclocking anyways. Just not designed to cool much more or handle more power. it is designed to just get the job done at stock.
Do tell me, which one you prefer, 85w from PCI-E slot with puny pins or 100w from 6-pin connector which can handle +250w, at least on RX480 because it has all six pins there
Add yet it still won't be under PCI-e spec, they still have to fix the absolute problem.
Nenu
[H]ardened
- Joined
- Apr 28, 2007
- Messages
- 20,315
Its nothing to do with what I like, its not my problem.Do tell me, which one you prefer, 85w from PCI-E slot with puny pins or 100w from 6-pin connector which can handle +250w, at least on RX480 because it has all six pins there
Daniel_Chang
[H]ard|Gawd
- Joined
- Jan 4, 2016
- Messages
- 1,313
The ideal is to get it within spec. If the user willingly overclocks and goes out of spec, that is their choice, but ideally, a card should be within spec out of the box.
Also, I keep seeing you post that about a GTX 580 consuming 800W. That's obscene, so I finally decided to Google it. Does not compute.
GeForce GTX 580 review
After 5 days of testing, their average was 280W on the GPU (GTX 580) and 447W for the entire system. Can you link to one instance of this card drawing 800W?
Nenu
[H]ardened
- Joined
- Apr 28, 2007
- Messages
- 20,315
Its out of spec.
For a well set up system its no problem.
But for a bad connection it will cause damage faster and do more damage.
Selling a new product that doesnt conform to basic specs is painful to see.
thesmokingman
Supreme [H]ardness
- Joined
- Nov 22, 2008
- Messages
- 6,617
I already referenced it to overclocking. A lot of gpus will suck down two to three times their rated TDP when unleashed.
Yea, ofc its still out of spec but it's not that big a deal. I suppose some just want to scream about it its a fucking huge deal. Shrugs. I could care less it it drew more power out of the 6pin. And on that note, if they had diverted more power draw over the plug and used an 8pin... lol it would have been too easy it seems lol.