[WCCFTECH]Possible AMD Radeon RX 490 Performance Numbers Show Up in DX12 AOTS Benchmark – On Par Wit

[thesmokingman]

right, so I'm BSing because I wont buy a 480 to test myself, when there are tons of reviews on the internet showing Polaris 10's power draw to be north of 160w, to the point that AMD had to publicly address it. But apparently that's not proof, and some Youtube videos of Afterburner readings are?

Which do you think is more likely to be accurate, actual power draw numbers, or Afterburner?

You're bs'ing because you're not interested in any numbers just garbling random facts into a pile so you can throw crap on the wall. The issue with the power draw wasn't that the card drew over 160w, but because it drew over 75w from the motherboard, breaking pcie specs. But don't let the facts stop you from demanding crap you really have no interest in, cuz it will only get in the way of your tirade.

 

[hawtdawg]

You're bs'ing because you're not interested in any numbers just garbling random facts into a pile so you can throw crap on the wall. The issue with the power draw wasn't that the card drew over 160w, but because it drew over 75w from the motherboard, breaking siig specs. But don't let the facts stop you from demanding crap you really have no interest in, cuz it will only get in the way of your tirade.

It drew over 75watts from the motherboard, because it was drawing over 160 watts in the first place.

You aren't even making any sense anyway. I'm very interested in numbers, numbers that are real and measured. Since when has what afterburner reports been taken as real measured numbers?

Here are some real numbers, as in numbers someone measured with the tools to do so.
http://www.guru3d.com/articles-pages/amd-radeon-r9-rx-480-8gb-review,5.html
https://www.pcper.com/reviews/Graphics-Cards/Power-Consumption-Concerns-Radeon-RX-480

"my tirade" is based on something called "facts". "facts" are not something you seem to have.

 

[RussianSensation]

It drew over 75watts from the motherboard, because it was drawing over 160 watts in the first place.

You aren't even making any sense anyway. I'm very interested in numbers, numbers that are real and measured. Since when has what afterburner reports been taken as real measured numbers?

Here are some real numbers, as in numbers someone measured with the tools to do so.
http://www.guru3d.com/articles-pages/amd-radeon-r9-rx-480-8gb-review,5.html
https://www.pcper.com/reviews/Graphics-Cards/Power-Consumption-Concerns-Radeon-RX-480

"my tirade" is based on something called "facts". "facts" are not something you seem to have.

Not all 480s draw 160W. Computerbase's latest review shows 4 different 480 SKUs drawing 215-226 total system power.
https://www.computerbase.de/2016-12/radeon-rx-480-partnerkarten-vergleich-test/4/

Kit Guru shows 225-232W total system power for both 480s:
http://www.kitguru.net/components/g...us-rx-480-strix-gaming-oc-aura-rgb-8192mb/29/

Notice now Palit reference 970 and reference 1070 systems use more power? We know those GPUs use roughly 140-145W in reference form.

Remember that most AIB 1070/1080 cards use 20-40W more power than reference 1070/1080.

Founders 1080 = 184W
https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080/24.html

Vs.

Asus Strix 1080 = 203W
https://www.techpowerup.com/reviews/ASUS/GTX_1080_STRIX/22.html

Zotac AMP! 1080 = 213W
https://www.techpowerup.com/reviews/Zotac/GeForce_GTX_1080_Amp/21.html

Gigabyte G1 1080 = 216W
https://www.techpowerup.com/reviews/Gigabyte/GTX_1080_G1_Gaming/22.html

MSI Gaming 1080 = 217W
https://www.techpowerup.com/reviews/MSI/GTX_1080_Gaming_X/22.html

Palit GameRock = 224W
https://www.techpowerup.com/reviews/Palit/GeForce_GTX_1080_GameRock/23.html

Let's say RX490 offers 90% of 1080's performance and uses 230W of power. Is that the end of the world? What if it costs $399-499?

Obviously the usuals at HardOCP, TechReport and AT forums will make a mountain out of 30-40W differences should RX490 use 230-240W.

Your other point about AMD requiring a 300W card to compete with 1080Ti/Titan XP is fair. But again, AMD already made 290X and Fury X. 300W isn't a big deal, and as you know a 1450-1500mhz 980Ti uses 340-350W! I don't remember hundreds of NV users complaining regarding 980Ti's or 780Ti's or even Fermi 580's horrible power usage once overclocked. Sapphire Nitro Fury is quieter at load than GTX580/780/780Ti/980Ti are at idle. There is no limitation for AMD to not make a 275-300W TDP flagship (and I hope they do).

NV will continue to lead in perf/watt but the gap this generation will be far closer than 970/980 vs. 290/290X/390/390X/Fury.

NV will likely respond with price cuts and/or refreshed Pascal cards but at least we could finally see some competition.

Don't forget AMD can get easy upgrades from RX 470/480/390/390X/Fury/Fury X users who are currently GPU mining. These users will offload their existing AMD GPUs on the used market and replace them with Vega 10/11. The reason this didn't work during 390/390X/Fury/Fury X days is because for 290/290X/295X2 owners those cards were too small of a leap. Now if AMD releases cards 50-80% faster than 470/480/290/290X/390/390X, there will finally be incentive to upgrade for those users.

 

[littleg]

Unless the card is using utterly ridiculous amounts of power I really don't care. A difference of 30,40,50,60W is nothing. It's one fairly dim light bulb yet its somehow made out to be a massive selling point and hugely important metric. Price/perf is the key, nail that and it'll sell.

 

[advt.naveen]

Unless Vega has a drastic efficiency improvement, they will not be able to build a GP102 competitor with it. They'd be looking at a 350+ watt TDP.

GTX 1060 power is around 120w , were as RX 480 power is around 160w . As u said when a HBM2 is swapped in place of gddr there might be 15w less, then power usage will be around 145w . then its around 20% more power to compete with its counter part.

If we apply this 20% to NV titan's AMD competitor , 20% more of 250w is 300w . By the time vega hits the market 14nm process also gets matured bit more than how it was on polaris release. i don't think big vega will need 350w or even 300w.

As far as performance if the leaked benchmarks are for small polaris then it will be very good for AMD.

 

[ThatBuzzkiller]

Not much hope left for AMD considering DX12 and Vulkan are floundering in adoption ...

Console deals aren't good enough if AMD can't find a way to translate that fast enough seeing as how shader model 6 is still isn't out and that they haven't pushed for spec changes with Microsoft more aggressively to their favour ...

 

[SlickR12345]

Titan XP is a cut down gp102 471mm die with gddr5x. Full die has 30 SM's. Nvidia also sells a gp100 610mm die with 16GB HBM2 and a 250W TDP. Full die has 60 SM's. Previous generation Titan XP utilized a 601mm die. You really believe AMD is going to announce a card that is twice as fast as the current Titan XP in the next 6 months? Because that's what Nvidia is already producing. Certainly they don't want to be forced to sell them to gamers at much lower margins than Tesla's. But if they are forced to, they have much bigger guns in reserve than the Titan XP.

Even just using a full gp102 die plus HBM2 along with any tweaks Nvidia has made over the last year will result in a performance gain of well over 5-7%. Nvidia has played their hand? Please at least attempt to argue with facts.

You are not right at all. You need to get checked out! There is ZERO evidence, ZERO even rumors saying there is a 30sm Titan X, that is utter bullcrap! But not only that, your next sentence is where you go into cook land, saying Nvidia has a 60sm refresh, that is INSANE! That is certifiably insane!

Nvidia already released their newest architecture, the 1000 series IS IT! There is no "secret" uber, ultra 60sm magical unicorn card that has double the performance of Titan X2. That is just absurdly INSANE! Even a new generation from Nvidia, even their next gen Volta high end GPU is unlikely to be 50% fasten than the Titan XP2.

Judging from the past DECADE of performance from high end parts, even a completely new generation doesn't gain such massive performance! And we all know Volta is late 2017, early 2018. The best Nvidia can come up is a refresh somewhere in late Q2 with slightly improved core and faster clocks, maybe GDDR5x on 2080TI, 2070.

AMD has the primary supply of HBM2, they co developed the memory, of course they are going to be buying up the supply, Nvidia would be lucky if they release an HBM2 card in mid Q3 2017!

 

[SlickR12345]

GTX 1060 power is around 120w , were as RX 480 power is around 160w . As u said when a HBM2 is swapped in place of gddr there might be 15w less, then power usage will be around 145w . then its around 20% more power to compete with its counter part.

If we apply this 20% to NV titan's AMD competitor , 20% more of 250w is 300w . By the time vega hits the market 14nm process also gets matured bit more than how it was on polaris release. i don't think big vega will need 350w or even 300w.

As far as performance if the leaked benchmarks are for small polaris then it will be very good for AMD.

[redacted]

http://www.guru3d.com/articles_pages/msi_radeon_rx_480_gaming_x_review,5.html
http://www.guru3d.com/articles_pages/msi_geforce_gtx_1060_gaming_x_review,8.html

Here is MSI gaming X 1060 6GB consuming 140W, compared to MSI gaming X 480 8GB(notice how 480 has 2GB more memory, so that is at least 15W more purely from memory) consuming 170W. 170-15=155W

So the 1060 6gb uses 140W, while the 480 8gb uses 155W. That is a 15W difference, this is not eve a LED bulb level of difference. Literally one LED bulb. If you are lifeless turd who games 8 hours a day, for 6 days a week, for a whole year your parents will have to pay few dollars more in electricity for the year if you gamed on RX 480 8GB as compared to 1060 6GB.

Unless the difference is 50W or more in the mid range segment that its not as issue, its pointless to argue power draw! At the high end unless the difference is 100W for the same performance its meaningless. Even 100W is literally just 1 light bulb change, go from a normal 100W light bulb to a 15W led one and whoala!

Infraction issued for inappropriate language.
-- stahlhart

 

[parvadomus]

Hope its not HBM based, it would be stupid. AMD can easily lauch a 48Rop/~3000 cores/384-bit part that be in the 1080 performance territory.
AND about power consumption, it would be at most 50/60 watts more.

 

[tential]

Hope its not HBM based, it would be stupid. AMD can easily lauch a 48Rop/~3000 cores/384-bit part that be in the 1080 performance territory.
AND about power consumption, it would be at most 50/60 watts more.

Does AMD have a track record of doing things like this?
Fury X for example?

 

[zinfamous]

Uhh..no, considering the 2nd sentence of his post attacked nvidia users as "NV loyalists who will never buy AMD," the clear implication is that people who buy Nvidia are not objective and just buy based on brand.

you are having reading comprehension problems.

anyway--this thread is based on wccftech article. Is linking that site not yet a bannable offense?

Infraction issued for derailing thread.
- stahlhart

 

[Glo.]

110w? lol. The 480 pulls north of 160w at stock speeds. HBM2 would save, at most, 15w off of that. A 1080 is around 1.8x faster than a 480 and a GP102 is 2.3x or so faster. You do the math.


If AMD pulls magic out of a hat, they'll have a GP102 competitor with about a 300w TDP. Vega was never going to be anything more than a big Polaris GPU with HBM2 (that it needs for no other reason than for power consumption). It was never going to compete with anything faster than the 1080. HBM2 is the only reason we didn't see this card months ago. You people are kidding yourselves if you think otherwise.

RX 480 Die consumes 110W of power. It has been proven in number of reviews. All of the rest of the board with memory consumes the rest 50W of power. 2 stacks of HBM2 will consume 10W. So whole GPU package, without the board, if we would make Polaris 10 HBM2 would consume around 120W of power. And Small Vega is most likely around 175W TDP GPU. Memory in RX 480 consumes around 37-40W of power. So here is where you get biggest difference. Cores are actually pretty efficient, especially on lower core clocks.

You have misunderstood my point. GPU die WITHOUT BOARD AND MEMORY consumes 110W. HBM2 consumes around 10W of power. So whole GPU package, because memory is integrated in the GPU would consume 120W of power. Whole board therefore would have not more than 130W TDP, at nominal clocks.

Yes, you are correct that it will save around 30W only. But its not that AMD GPUs are so inefficient that they cannot compete with Nvidia. 3072 GCN core chip, with 1.5 GHz, and 8GB, 512 GB/s HBM2 will be on par with GTX 1080. And it most likely can have 175W TDP.

 

[OatisCampbell]

This card is a GTX 1070 competitor.
They are simply showing the card in its best light, which yes, allows it to compete with a card a tier above it. It's not like the 1070 and 1080 are far apart.
If the RX 490 was as fast as a 1080, that would leave a gap in the space that the 1070 occupies.... There is no way AMD does that.

This is really just common sense.

That would be a mixed bag for us.

It would be great to have some choices at last gen top tier level, with Freesync. That's a card that should sell for $350..

Unfortunately the Freesync is all that it brings new to the market, as there are currently $350 cards with that performance. With Rx480s over $200, there isn't much room to price compete as margins are already on the thin end.

If this is the "big Vega" (possible) it would mean AMD is a full generation behind NVIDIA in technology this round. (bad for everyone but NVIDIA)

VRAM "silver bullets" don't hold much sway with me as while higher bandwidth is great it only matters if the bandwidth needs are exceeded on the other solutions.

Maybe a lower clocked, binned and water cooled 490CF on a card $800 card for high end competition.

 

[Glo.]

I was watching this video from AdoredTV, when it got me. AMD is doing exactly the same thing, as back then, but with BOTH: CPUs and GPUs.

If 4C/8T Zen truly can compete with 4790K at fraction of the cost, and much better efficiency, if 8 core/16 Thread CPU can compete with 5960X at half the cost, and if they can bring competitor for GTX 1080 at lower price tag and around similar efficiency they have created a go-to gaming platform not only for high-end gamers.

TAM. That is what they might want to go for. For 1700$ USD you would be able to buy only 5960X+ GTX 1080. AMD may be aspiring to provide CPU, motherboard, and DUAL GPU setup in which single GPU can be as fast as GTX 1080, for the same money.

Project Quantum.

Again, disclaimer. This is only what I think might happen, and nobody has to agree on this, with me.

 

[Keysplayr]

I was watching this video from AdoredTV, when it got me. AMD is doing exactly the same thing, as back then, but with BOTH: CPUs and GPUs.

If 4C/8T Zen truly can compete with 4790K at fraction of the cost, and much better efficiency, if 8 core/16 Thread CPU can compete with 5960X at half the cost, and if they can bring competitor for GTX 1080 at lower price tag and around similar efficiency they have created a go-to gaming platform not only for high-end gamers.

TAM. That is what they might want to go for. For 1700$ USD you would be able to buy only 5960X+ GTX 1080. AMD may be aspiring to provide CPU, motherboard, and DUAL GPU setup in which single GPU can be as fast as GTX 1080, for the same money.

Project Quantum.

Again, disclaimer. This is only what I think might happen, and nobody has to agree on this, with me.

All due respect, Glo, I think if we here in the forums hear another "If AMD can......" comment, I think we will all collectively jump off the nearest high building. We are looking to get past "if" and see some actual progress from them, which isn't likely due to their declining history and super low (comparatively) R&D budget. AMD is the next Cyrix, if not already. I truly do hope Zen is good all around, not in just one or two ridiculously rare instances or apps. That would be refreshing.

 

[OatisCampbell]

All due respect, Glo, I think if we here in the forums hear another "If AMD can......" comment, I think we will all collectively jump off the nearest high building. We are looking to get past "if" and see some actual progress from them, which isn't likely due to their declining history and super low (comparatively) R&D budget. AMD is the next Cyrix, if not already. I truly do hope Zen is good all around, not in just one or two ridiculously rare instances or apps. That would be refreshing.

The FuryX is already pretty close to 980Ti/1070 performance, I can see AMD coming up with something 1080 level this round.

When I don't know, but AMD graphics is definitely a bigger player than AMD CPUs.

Those of us who have don't have as much disposable cash to blow on gaming PCs had better hope they do. 1080 sales show there is no shortage of people willing to pay $700+ for a mid-range card.

 

[Glo.]

All due respect, Glo, I think if we here in the forums hear another "If AMD can......" comment, I think we will all collectively jump off the nearest high building. We are looking to get past "if" and see some actual progress from them, which isn't likely due to their declining history and super low (comparatively) R&D budget. AMD is the next Cyrix, if not already. I truly do hope Zen is good all around, not in just one or two ridiculously rare instances or apps. That would be refreshing.

Well I write this because few weeks ago I have had the priviledge to talk with a game developer, who have had his hands on both Vega, and Zen CPU. However, I was not able to get out from him any information on performance. Nothing. Zero. He could not talk about it.

However what he have said is that 2017 year is for AMD brand reset. In terms of mindshare, and perception. Changed naming schemes, simplicity on the platform chipsets, etc, to attract people who were never interested in AMD platform(like me for example, I have absolutely no idea about the features of AMD chipsets, nor am I interested in performance of previous generations of AMD CPUs, because they are simply too weak, and too inefficient for my requirements). He also said that every clue that they have right now indicates that AMD is planning on launching gaming platform with Zen+Vega. And one of the key aspects for this will be APUs.

I was not writing this before, because I may have not been allowed to do this. Right now maybe its good time to post it.

 

[Mopetar]

The problem with videos like that (and a lot of predictions made in this thread) is that they assume everything goes perfectly for AMD and NV just sits around and does nothing in the interim.

This almost never happens in real life. HBM2 not being ready or yield/process issues at GF, or any number of issues happen all the time that completely ruin those type of outcomes. If AMD were that good or NV that inept, we wouldn't be having this discussion because NV would have gone out of business years ago.

 

[vissarix]

I think amd can give us a gtx 1080 level of performance, at 250w+ consumption, lower price, little to no oc headroom...not a great product since its late to the party by 6 months, but still better then nothing...nvidia can just lower prices and be competitive again..

 

[Glo.]

To all people believing that RX 490 will be based on Vega 10, a 250W GPU:

Have you ever seen the bigger GPU to be lower tier than the smaller one?

There are 2 GPUs in Vega stack. Vega 10, and Vega 11. Vega 10 is the bigger one. There is no reason to believe that bigger one, 12 TFLOPs GPU will compete with 9 TFLOPs GPU in Ashes, especially in DX12 mode. It would invalidate everything we know about current crop of GPUs.

 

[ultima_trev]

I'd be quite impressed if Vega performed at GTX 1080 level considering TSMC 16nm provides a huge advantage over Samsung 14nm. Considering the Fury series barely performed on par with an OC'ed GTX 980, I think a victory over the GTX 1080 would be unrealistic. So, in a sense this is good news if true. Unfortunately most high end buyers have already got their 600$ GPUs this gen, would've been nice if they had this performance available before ordering my GTX 1080.

 

[IllogicalGlory]

I'd be quite impressed if Vega performed at GTX 1080 level considering TSMC 16nm provides a huge advantage over Samsung 14nm.

It's not a huge advantage. 1050/1050 Ti are Samsung and they're very similar to other Pascal cards. The clocks might be moderately lower (10%-ish), but not huge.

 

[Pariah]

You are not right at all. You need to get checked out! There is ZERO evidence, ZERO even rumors saying there is a 30sm Titan X, that is utter bullcrap! But not only that, your next sentence is where you go into cook land, saying Nvidia has a 60sm refresh, that is INSANE! That is certifiably insane!

Nvidia already released their newest architecture, the 1000 series IS IT! There is no "secret" uber, ultra 60sm magical unicorn card that has double the performance of Titan X2. That is just absurdly INSANE!

AMD has the primary supply of HBM2, they co developed the memory, of course they are going to be buying up the supply, Nvidia would be lucky if they release an HBM2 card in mid Q3 2017!

So this is what it has come to? We have now gotten to the point where we are denying the existence of products that are readily available for purchase? The P100 was the first pascal product announced back in April.

http://www.guru3d.com/news-story/nv...ecap-full-gpu-has-3840-shader-processors.html

Then there is big Pascal, the big daddy Nvidia GPU developed under GPU codename GP100. This is the GPU that will empower (for the consumer side) the enthusiast class products e.g. the Titan etc. Make no mistake, this product will not launch anytime soon for consumers. Expect at the very best a launch late this year closer to the Christmas season, likely even later in Q1/Q2 2017 (we think).

This is from back in April. The unexpectedly early August release of the GP102 based Titan XP makes it highly unlikely that we will see a GP100 based Titan the first half of 2017. But that's not the point, which is the product exists now. It would have been pretty awesome if Nvidia had called it the "Secret Uber Magical Unicorn Card", but they instead went with the more bland name of Tesla P100 (slightly cut down to 56SMs). It sells in two configurations of 12GB and 16GB HBM2. Again, that's HBM2, and already for sale, so no, AMD does not have a monopoly on HBM2 production.

I did not say Nvidia has announced any new products beyond what they have out now. You said Nvidia had played their hand and had nothing left to offer. You're free to believe that, I don't care if you want to live in some alternative universe where products we can buy don't exist and it's blasphemy to suggest Nvidia is capable of an uncut GPU. However, the information is here for other people who may be interested in what Nvidia has in the stable to potentially respond to AMD's upcoming Vega based cards including whatever card is benched in the OP.

 

[hawtdawg]

[redacted]

http://www.guru3d.com/articles_pages/msi_radeon_rx_480_gaming_x_review,5.html
http://www.guru3d.com/articles_pages/msi_geforce_gtx_1060_gaming_x_review,8.html

Here is MSI gaming X 1060 6GB consuming 140W, compared to MSI gaming X 480 8GB(notice how 480 has 2GB more memory, so that is at least 15W more purely from memory) consuming 170W. 170-15=155W

So the 1060 6gb uses 140W, while the 480 8gb uses 155W. That is a 15W difference, this is not eve a LED bulb level of difference. Literally one LED bulb. If you are lifeless turd who games 8 hours a day, for 6 days a week, for a whole year your parents will have to pay few dollars more in electricity for the year if you gamed on RX 480 8GB as compared to 1060 6GB.

Unless the difference is 50W or more in the mid range segment that its not as issue, its pointless to argue power draw! At the high end unless the difference is 100W for the same performance its meaningless. Even 100W is literally just 1 light bulb change, go from a normal 100W light bulb to a 15W led one and whoala!

Infraction issued for inappropriate language.
-- stahlhart

Wow, there's someone who still thinks power consumption has something to do with your electricity bill?

 

[Pariah]

I think amd can give us a gtx 1080 level of performance, at 250w+ consumption, lower price, little to no oc headroom...not a great product since its late to the party by 6 months, but still better then nothing...nvidia can just lower prices and be competitive again..

If I had to put money on it, this is the most likely scenario. Though I hope Nvidia has a refresh ready to go so that mainstream gamers have something faster than the 1080 GTX a year after release that doesn't cost more than about $650.