AMD Radeon RX480 Announcement Live From Computex

If one card uses 150 watts and another card with the same performance uses 100 watts, which card do you think OEM's will push? For us its not a big deal (most of us build custom systems and take that into account for the performance we are looking for), for OEM's and system builders, its a huge deal the lower power consuming parts they can cut down on over all system costs.
 
MacLeod said:
How many people actually care about power? Is it really that many? I mean yeah if the thing is drawing gianormous amounts of juice then it's a problem but are the amounts these cards are at now a big deal? And is it the end of the world if one is making 150 watts? There was a post in one of these threads saying literally AMD is doomed because the 480 is a 150 watt card. Seems like some are getting carried away worrying about efficiency.
Click to expand...

Matters to me in relationship to how hot the card is and thus how quiet it is. Looking at my system power if I keep the video card low the PSU will also not kick in.
 
I just care about $ and performance ! if the medium range 490 ( if that`s gonna be the name ) will beat the 1070 in price and performance ?
 
Is it wrong that I'd be seriously interested in an 8GB version of this?

I'll hold off a little while and see if a 490 or 490X is forthcoming, but even at $250, the 480 actually is pretty tempting.
 
anthrex said:
Matters to me in relationship to how hot the card is and thus how quiet it is. Looking at my system power if I keep the video card low the PSU will also not kick in.
Click to expand...

I don't know brother. Look at my sig, I'm probably drawing 600 watts from the wall. I don't have a problem with heat or noise.

Don't get me wrong, I'm not saying efficiency isn't important, it obviously is. I just think all these cards are pretty efficient and it maybe it's just my imagination but it seems some are getting all worked up because the 480 is a 150 watt card and so is the 1070. That's a total non issue to me.
 
  • Like
Reactions: Zuul
like this
anthrex said:
Matters to me in relationship to how hot the card is and thus how quiet it is. Looking at my system power if I keep the video card low the PSU will also not kick in.
Click to expand...

you still don't understand. Look at the slide again. AMD didn't even give you specs for the card and what speed it is running at. They gave you the max wattage. The slide did say up to 2.8 per/watt due to enhancements. So let me tell you the base model if it is clocked at 5Tflops will get no where near to 150w, if it is clocked to hit 6tflops it might get close to 150w. We have leaks of the card running at 1266 that puts it at 5.83 I believe according to videocardz.com, but AMD is still calling it 150w. We don't even know if this is the only card or there will be two different ones clocked different. But they are fitting them all under 150 with a 6 pin connector. If they needed more than 150w to run any of those clock speeds they would have put 8 pin connector on there. So I am pretty confident the card is pretty damn efficient. AMD just is too damn quiet about going in to too many details.

And they probably just mentioned what they could at computex just to shut rumors of delays. its a hard launch on the 29th, same day as the NDA lifts. That's how it used to be before. Announce a date and all specs and launch at the same day.
 
1266 is the readout from techpowerup that is probably boost clocks too. AMD tends not to tell its power usage based on boost clocks but rather its base clocks, it still shouldn't be over 150 watts, but it can go above that. The reason why is because they want to stay with in the pci-e spec so if base clocks don't go above 150 watts, they only have to put a 6 pin connector. I expect boost clocks to stay within that range too for polaris.

Oh wait that seems to be from the driver readout? So no those aren't boost clocks.

PS I think this is the highest version of P10 as well (all ALU's active)
 
Last edited:
razor1 said:
1266 is the readout from techpowerup that is probably boost clocks too. AMD tends not to tell its power usage based on boost clocks but rather its base clocks, it still shouldn't be over 150 watts, but it can go above that. The reason why is because they want to stay with in the pci-e spec so if base clocks don't go above 150 watts, they only have to put a 6 pin connector. I expect boost clocks to stay within that range too for polaris.

Oh wait that seems to be from the driver readout? So no those aren't boost clocks.

PS I think this is the highest version of P10 as well (all ALU's active)
Click to expand...
AMD doesn't really have a boost clock, well not like Nvidia. The 290x was listed as 1000mhz but the base clock was 700 or so. It ran at the max clock of 1000 as long as the tdp/powertarget/temp stayed within range. so generally AMD lists the TDP of the cooler or the max power of the PCI-e PSU connection plus the PCI-e slot and not the avg wattage of the card during play. They do the same with their CPUs, well did I think they changed that recently in the mobile/laptop space.
 
The 480 is shown right on par with the Fury, above the 980 and 390x, far above the 970 and 390... at $200

AMD Radeon R9 480 3DMark11 benchmarks | VideoCardz.com

I was expecting that to be the 480x, but here is the post confirming it is the normal 480... the card that was announced as $200...

AMD Radeon RX 480 confirmed as Polaris 67DF:C7 | VideoCardz.com

So yeah... $200 for something Fury-level, better than a 980 or 390x, at 150w, and seemingly 8gb of vram...

nVidia may have a bit of competition for the 1070, when you get nearly the same performance at almost half the price... Fury-level vs 980ti-level, equal vram - $200+ vs $380+
 
  • Like
Reactions: N4CR
like this
razor1 said:
The connector doesn't tell you the TDP, TDP is what the cooler is rated to dissipate the heat at at certain RPMs of the fan. So the power consumption has nothing to do with the TDP. The 1070 from what have seen the board doesn't use more than 150 watts at load.
Click to expand...

Power consumption has everything to do with TDP. All the power 'consumed' gets turned into waste heat. If you 'consume' 150W of power you generate 150W of waste heat.
 
JustReason said:
AMD doesn't really have a boost clock, well not like Nvidia. The 290x was listed as 1000mhz but the base clock was 700 or so. It ran at the max clock of 1000 as long as the tdp/powertarget/temp stayed within range. so generally AMD lists the TDP of the cooler or the max power of the PCI-e PSU connection plus the PCI-e slot and not the avg wattage of the card during play. They do the same with their CPUs, well did I think they changed that recently in the mobile/laptop space.
Click to expand...

Correct. I think AMD cards just stay mostly around their stated clocks max. I think nano was an exception because of power limits. But these cards should stay under power envelope regardless so they should stay around their stated 3d clocks. Unless ofcourse AMD changed something and went Nvidia route, like you never know what fucking clocks you are actually getting lol, but you know it wont go below the minimum lol.
 
MacLeod said:
How many people actually care about power? Is it really that many? I mean yeah if the thing is drawing gianormous amounts of juice then it's a problem but are the amounts these cards are at now a big deal? And is it the end of the world if one is making 150 watts? There was a post in one of these threads saying literally AMD is doomed because the 480 is a 150 watt card. Seems like some are getting carried away worrying about efficiency.
Click to expand...

It's an indication of efficiency. And lower temps lead to better overclocking headroom (typically) But as this is FinFET and AMD had to push the 480 to the max to save face from 1070, I'm guessing there will be little in terms of overclocking headroom (like the Fury X had no overhead)

But all this is a educated guesstimate. We have to wait for the reviews. But given what we see, it doesn't look promising.
 
Okay here's the deal

The motherboard is designed to deliver 75 Watts
The 6 pin PCIe is designed to deliver another 75 Watts

THAT'S 150 Watts total. And computers turn just about all power into waste heat. (I'm talking IC's and power circuitry. Obviously the exception are things like fans and LEDs, etc)

So the upper TDP can't exceed 150 Watts or you are overloading the PCIe. I imagine when everything is maxing out on the RX480, I bet it does come close to the 150Watt limit....But the TDP will be slightly below that. Compared to the 1070 which could nearly double (possibly more) the performance of the RX480 for a similar power envelope*, that just doesn't look good.

*again we won't know for sure until the benchmarks are out. So settle down you bunch of spec junkies.
 
Singularity_Survivor said:
If you plan to stick with 1080p gaming for awhile, then the $80 saved with a 480 might be the most logical choice. But if you have plans for 1440p gaming at ultra settings at higher than 60hz, then a 1070 would be the obvious choice. Just my opinion anyway.......
Click to expand...

Actually... I think I'm going wait for the non-founder's edition of the 1080. I've been saving for a while and with the $200 I got from of my 7950s I think it makes sense since I'm officially fed up with SLI/CF and want to move up to at least 2k if not 4k monitor.
I won't pay $100 more for a reference card with a leaf blower attached to it. Hopefully we see some better cooling designs from NV's partners in the next month.
 
Last edited:
DigitalGriffin said:
It's an indication of efficiency. And lower temps lead to better overclocking headroom (typically) But as this is FinFET and AMD had to push the 480 to the max to save face from 1070, I'm guessing there will be little in terms of overclocking headroom (like the Fury X had no overhead)

But all this is a educated guesstimate. We have to wait for the reviews. But given what we see, it doesn't look promising.
Click to expand...


You know you speculate and say it doesn't look good and then we need to chill and wait for reviews. So many assumptions in your posts and then finishing with lets wait for reviews but since you already know everything why wait? Lol
 
NKD said:
You know you speculate and say it doesn't look good and then we need to chill and wait for reviews. So many assumptions in your posts and then finishing with lets wait for reviews but since you already know everything why wait? Lol
Click to expand...

It's easy to take a good guess at where something sits. HOWEVER a guess isn't a final answer.

For example you can look at a 450HP V8 engine attached to a 3200 lb car and make a GOOD guess it's 0-60 time. You'll get in the ballpark, but you won't have a definite answer till testing.
 
  • Like
Reactions: N4CR
like this
DigitalGriffin said:
It's easy to take a good guess at where something sits. HOWEVER a guess isn't a final answer.

For example you can look at a 450HP V8 engine attached to a 3200 lb car and make a GOOD guess it's 0-60 time. You'll get in the ballpark, but you won't have a definite answer till testing.
Click to expand...

But guess what amd hasn't even officially announced the clocks yet. Basically all cards will be under 150w depending on how they are clocked. But you made an assumption it's worse than 1070 when it comes to power. When you have no idea at what clock speeds it will consume that much. Yes you are right we need to wait but to call it a power hog vs 1070 is ill advised because you don't know the exact specs.
 
There's a number of unknowns, but AMD said their intention is to leave overclocking headroom for AIB partners. I think 130W is reasonable guess for 1266Mhz clockspeed. GDDR5 at 256 GB/s uses a little over 24W of power, so that leaves 106W for the GPU. If you would simply crank the clockspeeds up 25%, you would be sitting at 133W, or 157W with the GDDR5 at 1582Mhz. That's assuming a linear clockspeed to power usage scaling of course.
 
variant said:
There's a number of unknowns, but AMD said their intention is to leave overclocking headroom for AIB partners. I think 130W is reasonable guess for 1266Mhz clockspeed. GDDR5 at 256 GB/s uses a little over 24W of power, so that leaves 106W for the GPU. If you would simply crank the clockspeeds up 25%, you would be sitting at 133W, or 157W with the GDDR5 at 1582Mhz. That's assuming a linear clockspeed to power usage scaling of course.
Click to expand...

6 pin power cable limits you to 150 watts
 
DigitalGriffin said:
6 pin power cable limits you to 150 watts
Click to expand...
Thats according to the PCI-E spec, but you can easilly squeeze more out if you have reasonably good quality PSU. The only physical limitation is when the cable resistence is too high that the PSU's rated amperage is too low to push the required current over the cable.
 
KazeoHin said:
Thats according to the PCI-E spec, but you can easilly squeeze more out if you have reasonably good quality PSU. The only physical limitation is when the cable resistence is too high that the PSU's rated amperage is too low to push the required current over the cable.
Click to expand...

Companies would rather just throw an 8 pin on there instead of expecting users to exceed 150 and not melt anything.
 
Last edited:
Ocellaris said:
Companies would rather just there an 8 pin on there instead of expecting users to exceed 150 and not melt anything.
Click to expand...

Thats exactly the point. Most likely the BIOS wont let input voltage go over a certain point, and it will limit the maximum amperage running through it.

But a little BIOS mod can make all of that go away!
 
KazeoHin said:
Thats exactly the point. Most likely the BIOS wont let input voltage go over a certain point, and it will limit the maximum amperage running through it.

But a little BIOS mod can make all of that go away!
Click to expand...

ummm yeah and when it melts the wires on the six pin and starts a fire....you have a fine lawsuit on your hands.
 
Last edited by a moderator:
variant said:
That's assuming a linear clockspeed to power usage scaling of course.
Click to expand...

That's a huge addendum there. It actuality FinFET is nothing linear. It's more like a square. The faster you clock a chip the more current it uses. But to keep the current up, you have to start upping voltages. And FinFET doesn't respond well to voltage bumps. It takes a large voltage bump for a little more current.
 
Singularity_Survivor said:
If you plan to stick with 1080p gaming for awhile, then the $80 saved with a 480 might be the most logical choice. But if you have plans for 1440p gaming at ultra settings at higher than 60hz, then a 1070 would be the obvious choice. Just my opinion anyway.......
Click to expand...

Preordered the EVGA 1080 ACX 3.0 from B&H photo this morning. Hopefully it ships on the 16th *fingers crossed*
 
DigitalGriffin said:
ummm yeah and when it melts the wires on the six pin and starts a fire....you have a fine lawsuit on your hands.
Click to expand...

I've pushed 600 watts over a 8 pin. It got a bit warm but held up.

I'd be more worried about the PCB and VRMs especially without a waterblock.
 
Dayaks said:
I've pushed 600 watts over a 8 pin. It got a bit warm but held up.

I'd be more worried about the PCB and VRMs especially without a waterblock.
Click to expand...

I'm surprised it was stable. As you get closer to the current rating of the wire, internal resistance comes into play and you get voltage drops as the current is pushed. You really start to affect stability. If it did start a fire it wouldn't be the PSU or vid car suppliers fault but the person who hooked it up
 
Dayaks said:
I've pushed 600 watts over a 8 pin. It got a bit warm but held up.

I'd be more worried about the PCB and VRMs especially without a waterblock.
Click to expand...

Yeah you can push past the limits that are suggested by the companies, but best not to if you don't need to lol. Its always a safety issue.
 
DigitalGriffin said:
I'm surprised it was stable. As you get closer to the current rating of the wire, internal resistance comes into play and you get voltage drops as the current is pushed. You really start to affect stability. If it did start a fire it wouldn't be the PSU or vid car suppliers fault but the person who hooked it up
Click to expand...

It was actually for peltiers, just saying my EVGA P2 1600W held up to the strain of 600W on one 8 pin, which is a testament to their power supplies. I'd think it's more likely something on the video card would blow first.

I wonder if the vdrop would bother a video card at all, it's switching from 12V down to 1.3V (ish) anyways. You could probably give it an input a bit below 12V and still be fine.

razor1 said:
Yeah you can push past the limits that are suggested by the companies, but best not to if you don't need to lol. Its always a safety issue.
Click to expand...

Definitely.

Most I've put into a card is my current Titan X at 400W, nominal is 250W, but it has 2x8 pin. I am more worried about it only having a few phases. At one time I looked up the current maximum of the FETs and it didn't have a whole lot of headroom. I have it underwater though, which keeps them cool and gives them more headroom than the stock cooler.

Even on the stock blower with a modd'd BIOs I was pulling around 325W fine, just 100% fan...

Overall my point being is as an enthusiast I don't worry about going over stock power limits a little. The risk is obviously on you.
 
Last edited:
You must log in or register to reply here.
Back
Top