Have past cards from nVidia also violated PCIe power specs?

Overclockers UK Forums - View Single Post - Radeon RX 480 "Polaris" Launched at $199

Power Consumption: Gaming - GeForce GTX 750 Ti Review: Maxwell Adds Performance Using Less Power

This is just nVidia's last generation while Toms did this type of testing. Im not sure if Toms or anyone else did this type of testing before, but I would suspect that it wasnt an issue that was better in some of their less effecient generations.

Seems odd that this suddenly became one of the most important things ever...

EDIT: plz merge into existing thread. Cant delete.

jamesgalb said:

Overclockers UK Forums - View Single Post - Radeon RX 480 "Polaris" Launched at $199

Power Consumption: Gaming - GeForce GTX 750 Ti Review: Maxwell Adds Performance Using Less Power

This is just nVidia's last generation while Toms did this type of testing. Im not sure if Toms or anyone else did this type of testing before, but I would suspect that it wasnt an issue that was better in some of their less effecient generations.

Seems odd that this suddenly became one of the most important things ever...

Click to expand...

You do realize 960 average or sustain power consumption is less than 60 watts compare to 480 where their average or sustain power consumption is 80 watts. So no, it isn't an issue for 960 whereas it maybe an issue for 480 depending on motherboard.

Chimpee said:

You do realize 960 average or sustain power consumption is less than 60 watts compare to 480 where their average or sustain power consumption is 80 watts. So no, it isn't an issue for 960 whereas it maybe an issue for 480 depending on motherboard.

Click to expand...

The 960/750ti also have spikes/variations that are much larger, which is also against spec (by 20x?) and potentially more damaging to components.

So they exceed both the max power spec, AND the 'maximum current slew rate" spec...

Worse, yet no uproar?

is not the same a microsecond spike than a sustained load, the problem isn't the spike, even the most power hungry GTX 980 TI reached nowhere near to 75W sustained load on the PCI-E slot, the RX 480 have sustained loads over 14% of the maximum current tolerance by the PCI-E certification... so again as has been mentioned. spikes? no issues, extended periods with a sustained out spec current can cause serious damage to the motherboard power traces, which it's amazingly worse once you add a second RX 480 for crossfire... and that's at stock levels. with a 3% overclock the card jump in 33% power consumption up to the 200W mark draining the most of it from the PCI-E slot so we are talking over 100W in a 75W+5% maximum tolerance, it's a serious issue.

jamesgalb said:

The 960/750ti also have spikes/variations that are much larger, which is also against spec (by 20x?) and potentially more damaging.

So they exceed both the max power spec, and the 'maximum current slew rate" spec... Worse, yet no uproar?

Click to expand...

And why should there be an uproar if the average is below specs, you are complaining to the wrong company about AMD problem which you seem to don't acknowledge as a problem. I get it, you love AMD and will defend it to the end, but even you have to see they eff up on reviewers card.

Chimpee said:

And why should there be an uproar if the average is below specs, you are complaining to the wrong company about AMD problem which you seem to don't acknowledge as a problem. I get it, you love AMD and will defend it to the end, but even you have to see they eff up on reviewers card.

Click to expand...

If the 'uproar' is about 'spec violation', the 960 and 750ti violate 2 specs...

if the 'uproar' is about potential component damage, the spikes/variations would likely be worse for components...

Why would there be uproar about the 480 and not the 960 or especially the 750ti? Average power consumption doesnt matter unless you are concerned about components or just concerned about 'being within spec', and in which case nVidia violates both of those ideas worse.

Araxie said:

is not the same a microsecond spike than a sustained load, the problem isn't the spike, even the most power hungry GTX 980 TI reached nowhere near to 75W sustained load on the PCI-E slot, the RX 480 have sustained loads over 14% of the maximum current tolerance by the PCI-E certification... so again as has been mentioned. spikes? no issues, extended periods with a sustained out spec current can cause serious damage to the motherboard power traces, which it's amazingly worse once you add a second RX 480 for crossfire... and that's at stock levels. with a 3% overclock the card jump in 33% power consumption up to the 200W mark draining the most of it from the PCI-E slot so we are talking over 100W in a 75W+5% maximum tolerance, it's a serious issue.

Click to expand...

I would assume that more dramatics fluctuation in the power on the 960/750ti, which is a spec violation within itself, would be far more damaging then a more consistent power draw above spec.

Do you have something that suggests otherwise?

If there is a 'serious issue', it would be component damage correct? And if its just about 'specs', the nVidia card violates 2 specs technically... What would make the 480 a more serious issue?

the fact that you are assuming prove that you just want to guess what's good and what's not good in a power circuitry trace from a motherboard. sustained load out of specs produce overheat in the traces which are far pronounced on low-end motherboards with less PCB layers, the heat produce degradation of the traces and also will draw the power from other sources of the motherboard to keep up with the demand, that's where the most issues with bad audio, bad GPU signal, RAM troubles begins, think in a typical city highway when new everything it's flat and smooth but the higher traffic the faster degradation the asphalt will produce, in a similar way works the traces in a motherboard, the more and more sustained overcurrent loads will produce more damage, without take in consideration that the PCI-E slot itself will work hotter, and you must know, those issues with bent PCI-E slots, with loose ports are just aggravated, without take in consideration the extra strain putted in the motherboard logical controllers (yes, the chipset) even worse yet for the public aimed by this kind of GPU, it's the public that doesn't care of specs, doesn't care of power, just care about money and probably are using the cheapest possible motherboard. want to keep with the examples? keep thinking on the degradation suffered by a highway full transited of civil cars versus the same highway full of 10ton trucks. that's why specifications are made for to avoid issues, and the RX 480 go above 14% of the tolerance supported.

jamesgalb said:

If the 'uproar' is about 'spec violation', the 960 and 750ti violate 2 specs...

if the 'uproar' is about potential component damage, the spikes/variations would likely be worse for components...

Why would there be uproar about the 480 and not the 960 or especially the 750ti? Average power consumption doesnt matter unless you are concerned about components or just concerned about 'being within spec', and in which case nVidia violates both of those ideas worse.

Click to expand...

Like I said, if you want to go complain, go complain to AMD. Right now you are just grasping straws defending your brand that are going over specs for a sustain period of time.

Average doesn't matter. It's all about does it exceed the spec at any point in time. If it does then the design needs to be fixed, and all of those poor people with damaged motherboards need to be compensated. Someone should sue Nvidia for making false claims about adhering to specs that are plainly laid out. You could burn someone's house down!

cageymaru said:

Average doesn't matter. It's all about does it exceed the spec at any point in time. If it does then the design needs to be fixed, and all of those poor people with damaged motherboards need to be compensated. Someone should sue Nvidia for making false claims about adhering to specs that are plainly laid out. You could burn someone's house down!

Click to expand...

The joke here is that i own a power supply that could legitimately catch a fire if i had hooked it up to rx480 and launch a benchmark. And i would be left puzzled because store page clearly wrote: 110W and 1 6-pin, just like my current card!

Two wrongs do make a right?

why is this under the amd section? it should be move to the nv forum.

Many cards in the past have violated this. Hd6990 did. It became a huge issue suddenly because people saw an opportunity to nail AMD for something, especially after Nvidia's latest slew of driver issues, 970 3.5gb, and GTX 1080 2.1ghz on air debacles that AMD fanboys gave Nvidia and its fans so much heat over.
I just wish the Fanboying would die out. I don't understand why to be a fan of anything anymore these days you have to hate the other thing and talk as much garbage as you can. I'm an AMD fan but as a competetive overclocker I've owned more Nvidia cards and I like them. This back and forth stuff is disgusting and destructive.

lolfail9001 said:

The joke here is that i own a power supply that could legitimately catch a fire if i had hooked it up to rx480 and launch a benchmark. And i would be left puzzled because store page clearly wrote: 110W and 1 6-pin, just like my current card!

Click to expand...

This is an outright lie.

Sheppard said:

This is an outright lie.

Click to expand...

What part of it? The fact that 55nm 9800GT is rated for 110W? Well, i was wrong, reference is rated at 105W TDP, my card however is factory OCd by at least 17% on core, so here are your 5W. Not to mention that stupid-me would not even consider those 5W as worth a worry.

The part that stores mention 110W? It's entirely true.

The fact that my PSU can catch a fire? Well, let's put it like this: it's $20 PSU that did not even come in a box, is rated at 450W but only comes with a single 6 pin and 20-pin mobo cable. I can entirely see it catching a fire and may even gift to someone willing to demonstrate it at full load after i buy a new PSU.

lolfail9001 said:

What part of it? The fact that 55nm 9800GT is rated for 110W? Well, i was wrong, reference is rated at 105W TDP, my card however is factory OCd by at least 17% on core, so here are your 5W. Not to mention that stupid-me would not even consider those 5W as worth a worry.

The part that stores mention 110W? It's entirely true.

The fact that my PSU can catch a fire? Well, let's put it like this: it's $20 PSU that did not even come in a box, is rated at 450W but only comes with a single 6 pin and 20-pin mobo cable. I can entirely see it catching a fire and may even gift to someone willing to demonstrate it at full load after i buy a new PSU.

Click to expand...

This post and your ridiculous claim disqualify you from any further discussion on the matter.

Sheppard said:

This post and your ridiculous claim disqualify you from any further discussion on the matter.

Click to expand...

Well, if you are unwilling to hear facts, sure. But i laid simple facts and i have actually described what experiences of people AMD markets this card towards to may look like in worst case.

In best case in similar scenario, it will simply result in computer being gracefully shut down every time this card is used beyond basic load, and another Nvidia convert after the buyer asks someone about his issues. Success!

Chimpee said:

You do realize 960 average or sustain power consumption is less than 60 watts compare to 480 where their average or sustain power consumption is 80 watts. So no, it isn't an issue for 960 whereas it maybe an issue for 480 depending on motherboard.

Click to expand...

As someone, who is not electrical engineer, I would be greatly interested in explanation why very frequent spikes occuring all the time that break the spec almost twice are not dangerous at all, just because the average happens to be within spec.

Looking at fuses at my mother-in-law place, where prior to turning vacuum cleaner, we need to turn almost all other electrical devices in her appartment, because of spike power at turning the device one, I am quite sceptical the average is what really counts.

ppi said:

As someone, who is not electrical engineer, I would be greatly interested in explanation why very frequent spikes occuring all the time that break the spec almost twice are not dangerous at all, just because the average happens to be within spec.

Looking at fuses at my mother-in-law place, where prior to turning vacuum cleaner, we need to turn almost all other electrical devices in her appartment, because of spike power at turning the device one, I am quite sceptical the average is what really counts.

Click to expand...

Because the real danger is connector and wire melting. For melting you need heat. Now imagine a nanosecond long 1 megawatt spike (the most batshit crazy case i can think of to make a point). It sounds scary, right? Well, it's only a thousandth of joule of energy generated (and consequentially heat generated). Now imagine a sustained load over a minute load difference of 20 watts. That's 1200 joules. That's enough to heat up about anything, in particular wires and connectors.

Spike power has other consequences and it is not pretty either. But modern stuff is mostly capable of handling that in short run, so the overheat becomes the the worrying part.

It has to be conceded that going out of PCI-E spec even in a spike, is still a violation of said spec, however.