AMD once again violating power specifications? (AMD RX-480)

[spartaman64]

6 minutes ago, patrickjp93 said:

It's AMD covering its ass. No one sends things directly to the consortium for testing. They have the equipment to do it in-house.

so pci-sig didnt actually test it?

[KeltonDSMer]

3 minutes ago, Majestic said:

On the PEG connector, which falls on the PSU, which won't have any issues dealing with it unless it exceeds a certain average value.

It does not spike to 155W on the mainboard causing interferrence, also can't be accomodated for with better PSU. Big difference.

 

Hey would you mind responding to my post on the last page? You keep mentioning the mobo as having to "deal" with the extra power instead of the PSU. I am asking if the mobo conditions the 12V from the PSU in any way. I was under the impression it does not...that the traces are direct from the 24pin to the PCIe socket. If they are direct, why does any of this matter? The Mobo would not have to "deal" with anything except a few watts of heat off the traces, no components to burn out.

 

Again, I am not saying that equal draw from slot/external is a good thing, I just don't really see this as a problem if the 12V traces at the slot are direct runs from the 24pin.

 

Can anyone answer this: are 12V traces at slot direct from 24pin or is the power conditioned in some way?

[laminutederire]

11 minutes ago, Trixanity said:

The difference is that the 1080 does it primarily over the auxiliary connector(s) and the 480 does a mostly equal split between auxiliary connector and the pcie bus. The latter being the key word here. The bus is more sensitive. So the RX 480 should never have exceeded 75W (although less would be even better) and instead have drawn more from the auxiliary connector. So it would have been better if it took 50W over the bus and whatever else it needed from the cable.

 

Although I'm uncertain why neither (but particularly Nvidia) isn't getting shit for these power spikes. Maybe I'm just not an engineer but it seems that these spikes are too big despite also being so short that it probably doesn't do much harm.

I don't deny there is an issue, I'm just saying that the spikes aren't what we need to focus on.

The spikes shouldn't matter I think. The idea is that a power is measured in Watts, 1W is equal to 1J/s, therefore if you have a huge power in a very very limited time, you'll still will have produced the same energy, and that's the energy you need to look at when yiu want to know if everything is going to melt. As long as the spikes are short enough and not too powerful, there isn't too much energy to be dissipated before making the wires to melt.

I think their solution is to change the settings for it to draw the power from the PSU, since an 8 pin isn't so much powerful compared to a 6 pin, even if they have a massive rating difference.

[Dabombinable]

23 minutes ago, Sintezza said:

Ah AMD has launched a decent performing budget oriented card.

On which Nvidia has no aswer asof yet.

And we allready see tons of BullShit like this showing up.

 

 

Look at the price compared to the GTX 970...then look at the performance and power consumption. Nvidia really doesn't need to answer it-and as it is there will be more than the current 3 models

[Trixanity]

9 minutes ago, patrickjp93 said:

Tom's hardware went back and tried overclocking. It took out the test bench Asrock OC board.

Source please. Their review has, from what I can tell, not been updated. It still states they did not push the overclocking due to concerns.

[wolfsbane3083]

5 minutes ago, KeltonDSMer said:

 

Hey would you mind responding to my post on the last page? You keep mentioning the mobo as having to "deal" with the extra power instead of the PSU. I am asking if the mobo conditions the 12V from the PSU in any way. I was under the impression it does not...that the traces are direct from the 24pin to the PCIe socket. If they are direct, why does any of this matter? The Mobo would not have to "deal" with anything except a few watts of heat off the traces, no components to burn out.

 

Again, I am not saying that equal draw from slot/external is a good thing, I just don't really see this as a problem if the 12V traces at the slot are direct runs from the 24pin.

 

Can anyone answer this: are 12V traces at slot direct from 24pin or is the power conditioned in some way?

I believe the traces are direct, but they are a lot lower gauge of copper compared to the PSU connector, hence why you can draw up-to 150w from a 6-pin before it melts.

The traces, being a lot thinner gauge will burn out at a much lower wattage.

[zMeul]

8 minutes ago, KeltonDSMer said:

The Mobo would not have to "deal" with anything except a few watts of heat off the traces, no components to burn out.

not that simple

 

have you ever wondered what happens when you put way too many fans on a FAN header with a max 1A rating? thgey melt like ice cream in the sun

[laminutederire]

11 minutes ago, zMeul said:

 

Come on read what I said, I was making the point that the 1080 also strained something with those spikes, and then argumented on why spikes aren't the problem, the problem being the average strain on the motherboard.

[zMeul]

1 minute ago, laminutederire said:

Come on read what I said, I was making the point that the 1080 also strained something with those spikes, and then argumented on why spikes aren't the problem, the problem being the average strain on the motherboard.

what are you talking about? what spikes does the GTX1080 shows on the PEG power draw?

[Majestic]

4 minutes ago, wolfsbane3083 said:

I believe the traces are direct, but they are a lot lower gauge of copper compared to the PSU connector, hence why you can draw up-to 150w from a 6-pin before it melts.

The traces, being a lot thinner gauge will burn out at a much lower wattage.

@KeltonDSMerThis pretty much, i'm not sure if it has some extra phase. But regardless, the traces are very small. It's not designed for 25-50% the power draw I rekon.

[laminutederire]

1 minute ago, zMeul said:

what are you talking about? what spikes does the GTX1080 shows on the PEG power draw?

Spikes are at 300W for the whole package, which is above what they should have used, which then strains something if spikes were dangerous. However they're probably not the issue, the issue being that the 480 draws on average too much. That's it.

I just wanted to give a better frame to the discussion because it seems like it goes in every direction when it shouldn't,  the issue is precise enough.

[Bouzoo]

10 minutes ago, Trixanity said:

Source please. Their review has, from what I can tell, not been updated. It still states they did not push the overclocking due to concerns.

Guru3D had luck pushing it even at 1375MHz and they got 10% more performance, and according to this, the AIB cards are hitting 1480 with a bit of luck going to 1600. 

Guru never said anything died. And custom do have 8 pins so there is that (at least some).

[Sintezza]

1 hour ago, patrickjp93 said:

Yes

yes, Tom's hardware went back and tried overclocking. It took out the test bench Asrock OC board.

Tom's Hardware, TechReport, a French review site, and 2 German ones.

 

Lmao yeah toms hardware.

 

Let them firstly officialy proof that the board went out because of pci-e peak power overshoot.

Let them proof it, they could verywell have messed up something else.

 

But yeah this is typical Toms

 

Its realy the last thing i have to say about it.

The cards are getting tested for standards, on which they have to pass.

Power overshoot normaly gets handled by the psu.

 

[zMeul]

5 minutes ago, laminutederire said:

 for the whole package

so .. you're talking bullcrap and "instructing" Tom's how to do their PEG power draw testing 

 

we're done here

[Majestic]

2 minutes ago, zMeul said:

so .. you're talking bullcrap and "instructing" Tom's how to do their PEG power draw testing 

 

we're done here

I've lost his argument aswell.

 

6 minutes ago, Sintezza said:

Its realy the last thing i have to say about it.

Ok, bye.

[Trixanity]

10 minutes ago, Bouzoo said:

Guru3D had luck pushing it even at 1375MHz and they got 10% more performance, and according to this, the AIB cards are hitting 1480 with a bit of luck going to 1600.

Yeah, I'm not saying overclocking isn't possible or anything like that. I was questioning Patrick's statement that Tom's review sample apparently fried their test bench due to the excessive power consumption over the pcie bus. I can't find it myself, so either 1) I'm blind 2) it's not in the article 3) Patrick has inside sources 4) Patrick has a vivid imagination. You be the judge.

 

I have seen others OC it a little bit but it's clear that the reference cooler can't handle much more than it was designed for: namely running at stock speeds.

 

Although it does beg the question: why AMD did not pull off a Wraith-equivalent cooler for their graphics cards? Would have been a sweet reference card.

[Misanthrope]

8 minutes ago, Sintezza said:

 

Lmao yeah toms hardware.

 

Let them firstly officialy proof that the board went out because of pci-e peak power overshoot.

Let them proof it, they could verywell have messed up something else.

 

But yeah this is typical Toms

 

Its realy the last thing i have to say about it.

The cards are getting tested for standards, on which they have to pass.

Do you remember the xfx R9-295 X2 with dual 8 pins, drawing more then 450W?

It did not burn out any motherboard pci-e slot either right?.

Power overshoot normaly gets handled by the psu.

 

You do realize that stressing the PCI cables isn't the same as stressing the mobo itself right?

 

How much extra wattage you put on a heavy CPU overclock? 15 to 20 watts tops? And look at all the extra power phases and shit those boards come with.

 

Now grab a B150 cheap as shit motherboard and add the same wattage, possible more if overclocking? Let alone Crossfire...

 

[Bouzoo]

5 minutes ago, Trixanity said:

Yeah, I'm not saying overclocking isn't possible or anything like that. I was questioning Patrick's statement that Tom's review sample apparently fried their test bench due to the excessive power consumption over the pcie bus. I can't find it myself, so either 1) I'm blind 2) it's not in the article 3) Patrick has inside sources 4) Patrick has a vivid imagination. You be the judge.

 

I have seen others OC it a little bit but it's clear that the reference cooler can't handle much more than it was designed for: namely running at stock speeds.

 

Although it does beg the question why AMD did not pull off a Wraith-equivalent cooler for their graphics cards. Would have been a sweet reference card.

Some issues are being sorted out, and I am disappointed in the stock cooler. We'll see how AIBs will do. From the custom ones with Arctic cooler they hit 63 while at 1350 iirc, so there is a lot of room for improvement. I for one am still considering a AIB 480 with an 8 pin, unless 1060 bests it to the ground at the same price. We will see. Also as I've said in another thread, drivers will help a bit, like the latest driver dropped the draw power in idle from 16 do 11 W, saw that on Videocardz.

[laminutederire]

4 minutes ago, zMeul said:

so .. you're talking bullcrap and "instructing" Tom's how to do their PEG power draw testing 

 

we're done here

Dude you aren't listening to anything I've said, are you illiterate or what?

I said the 480 has an issue. I also said the 1080 hasn't one so much. And I used a logic reflexion explaining why since obviously the 1080 has peaks beyond what's supposed to be handled by the 8-pin, and since it hasn't been an issue with that kind of strain, the real issue with the 480 is that it's drawing too much ON AVERAGE on the motherboard.

Is that clear enough? Nowhere I've said Tom's hardware was wrong, I just built a reflexion to make sense of what they are talking about.

[patrickjp93]

28 minutes ago, spartaman64 said:

so pci-sig didnt actually test it?

I sincerely doubt it. If you have the money to develop and produce tens of thousands of ICs this complex, you can spend half a million on a proper PCI-SIG test platform, or you can properly wire up an enthusiast OC motherboard. I know for a fact Intel and Nvidia just test in-house with proper equipment.

[spartaman64]

4 minutes ago, patrickjp93 said:

I sincerely doubt it. If you have the money to develop and produce tens of thousands of ICs this complex, you can spend half a million on a proper PCI-SIG test platform, or you can properly wire up an enthusiast OC motherboard. I know for a fact Intel and Nvidia just test in-house with proper equipment.

pretty sure nvidia also does pci-sig testing http://www.nvidia.com/object/IO_15266.html or maybe they dont anymore idk

[Sintezza]

10 minutes ago, Misanthrope said:

You do realize that stressing the PCI cables isn't the same as stressing the mobo itself right?

 

How much extra wattage you put on a heavy CPU overclock? 15 to 20 watts tops? And look at all the extra power phases and shit those boards come with.

 

Now grab a B150 cheap as shit motherboard and add the same wattage, possible more if overclocking? Let alone Crossfire...

 

 

Sure i do.

 

And like i said, as far as motherboards power cirquitry goes nobody has to tell me anything about it.

In Toms case they speak about a decent Asrock OC formula board.

And i think its very unlikely that their board crapped out because of a video card drawing to much power from the pci-e slot.

Very unlikely, unless they fucked up something realy badly.

Like i said, let them firstly officialy proof that the gpu overclock was the exaly cause of the board going out.

If they can proof that, then they have a concerning poit they proofed.

Not impossible, but highly unlikely.

 

 

[zMeul]

2 minutes ago, laminutederire said:

Dude you aren't listening to anything I've said

no, I'm not listening anymore

the issue discussed is that RX480 draws way too much power from the PEG slot compared to the PCI-SIG 75W standard! PERIOD! GTX1080 doesn't do this! PERIOD!

 

Quote

Nowhere I've said Tom's hardware was wrong

that's not what I said, I said you're "instructing" Tom's how to conduct their testing; you said:

Quote

the whole package, which is above what they should have used

no! PERIOD!

 

- The End -

[Trixanity]

4 minutes ago, Bouzoo said:

Some issues are being sorted out, and I am disappointed in the stock cooler. We'll see how AIBs will do. From the custom ones with Arctic cooler they hit 63 while at 1350 iirc, so there is a lot of room for improvement. I for one am still considering a AIB 480 with an 8 pin, unless 1060 bests it to the ground at the same price. We will see. Also as I've said in another thread, drivers will help a bit, like the latest driver dropped the draw power in idle from 16 do 11 W, saw that on Videocardz.

I imagine a 1060 being 5-15% faster at launch for $250-300. I also imagine that an RX 480 will be faster within a year or two.

I don't think Nvidia wants to sacrifice margins to stomp on AMD and it probably won't last either, so they'll end up in a position where they have a card that's already lower in price than they want to and could be forced to lower it further to stay competitive. Nah, they'll price it higher from the get go and lower it when the RX 480 starts to surpass it.

[patrickjp93]

2 minutes ago, spartaman64 said:

pretty sure nvidia also does pci-sig testing http://www.nvidia.com/object/IO_15266.html or maybe they dont anymore idk

That was a decade ago before you could just buy test platforms and maintain certification to do it yourself with yearly site inspections.