AMD Polaris Thread: Radeon RX 480, RX 470 & RX 460 launching June 29th

Page 41 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.

flynnsk

Member
Sep 24, 2005
98
0
0
Sure, I agree completely it's speculation. My point is that everything on this is - 110W is no more or less credible than close to 150W. They're just guesses based on very limited information, that's actually what I said.

As already pointed out, the 970 and 1070 are capable of using up to 225W (75W via PCIe, + 150W via 2x6pin/1x8pimn). When you look at actual power consumption numbers, typically cards only draw about 30-40W through the PCIe slot, and the rest coming from the 6/8pin connectors, and almost never at max. When looking at the 1070 power numbers across various sites, the average is about 145W (max goes over 160). That would be about 40W from the PCIe slot and about 105W from the 8pin, so about 70% of max from connector and about 50-55% from PCIe.

When looking at the 480, it is a much smaller part, (nearly 30% smaller) and has only one 6 pin (75W) and the PCie (75W). Again given most cards use only 30-40W max from the PCIe slot, even if the 6pin delivered 100% of its rate (75W), that would put the 480 about about 115W. Again, almost no previous cards (other than dual GPU) ever use their max. So if we use the same % from the GTX 1070 (75%) then Polaris comes in at about 96 W, (40W from PCIe + 56.25 W from 6 pin).
 

Sweepr

Diamond Member
May 12, 2006
5,148
1,142
131
Radeon RX 480 8GB listed in UK

£194.15 - £232.98 inc. VAT
www.lambda-tek.com/Sapphire-21260-00-20G~sh/B3292129

£235.18
www.kikatek.com/P788949

Interesting piece from TPU's Geforce GTX 1070 review:

AMD's upcoming Polaris cards will be nowhere near the GTX 1070 in terms of performance. Rather, expect RX 480 to perform about 20-30% slower. But AMD's $199 pricing for the 480 could stir things up, so if you don't need a new card immediately, maybe wait a few weeks and see how things pan out, which would also allow you to see how the custom GTX 1070 designs by board partners turn out.

Just a guess? Or he already has the card?



That's a custom card. Absolute maximum for FE in games was 151W according to TPU (154W stress testing), considerably less than what Geforce GTX 970 got (191-201W).
 
Last edited:

Mopetar

Diamond Member
Jan 31, 2011
8,011
6,455
136
Sure, I agree completely it's speculation. My point is that everything on this is - 110W is no more or less credible than close to 150W. They're just guesses based on very limited information, that's actually what I said.

Adored did a video with some comments on the efficiency based on information AMD released. His argument was that their claim that 2 480s in crossfire being called more powerful and "more efficient" than a 1080 meant that they were probably drawing around 95W.

I wouldn't be surprised if AMD was doing some cherry picking to find a best possible case of performance / watt, but it's certainly possible for the 480 to draw a lot less than the 150W it can draw. I don't know if that translates into good overclocking, but I think it does suggest that we'll eventually see a very capable 480X at ~$300.
 

xthetenth

Golden Member
Oct 14, 2014
1,800
529
106
You have a good history of tech predictions...

Since you predict AMD will lose marketshare, it's almost assured they will have a brilliant Q3 and Q4 this year.

He's a broken clock in the truest sense. It doesn't matter when you ask, the answer's going to be the same.

It's a shame too because if he were actually always wrong he'd be so much more informative than he is now.
 

sm625

Diamond Member
May 6, 2011
8,172
137
106
You know, it would make me extremely happy if cards just stopped pulling any power from the slot unless they're actually slot powered. It's silly to pull 12V power from the MB connector through the MB and then through a card edge connector when you could just draw current from a connector actually designed for such a task.

The PCIe connector has five +12V connections. It would be silly not to use them.
 

littleg

Senior member
Jul 9, 2015
355
38
91
You know, it would make me extremely happy if cards just stopped pulling any power from the slot unless they're actually slot powered. It's silly to pull 12V power from the MB connector through the MB and then through a card edge connector when you could just draw current from a connector actually designed for such a task.

Is there any tangible benefit from drawing from the slot as well as the connectors?

I mean in terms of smoothing, ripple control etc?
 

Qwertilot

Golden Member
Nov 28, 2013
1,604
257
126
That slide is the baffling thing - if it doesn't draw something like 150w then why put it there in the big reveal presentation?! They must know what the TDP is like and if it was say 110, it'd look much better than 150.

Guess there were quite a few odd things about that presentation, so maybe they just missed it.
 

crisium

Platinum Member
Aug 19, 2001
2,643
615
136
They also gave an inconclusive TFLOP rating. Maybe they really don't have a finalized clockspeed yet, and don't want to set the TDP yet. But in that case, why bother with saying 150W at all. AMD being AMD.
 

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
Interest, whats your source on this?

Both RX480 and GTX 1070 is specced for 150W. Yet they are by no means in the same performance class. In other words, Pascal looks to have substantial perf/watt advantage. And GP106 that competes more directly is 90-100W or so according to leaked specs.

You just delivered a spin, didn't you?

AMD and NV do not define TDP the same.

Not to discredit NV's excellent improvement in perf/watt with 1070 but just like last generation with 970/980, most after-market 1070 cards will use above 150W. Only the reference card will abide by the TDP spec.

MSI Gaming 1070 X uses 186W of power for example.
http://www.guru3d.com/articles_pages/msi_geforce_gtx_1070_gaming_x_review,8.html

Even if they did manage that though, I think they'd still drop overall market share.

I disagree. AMD has literally been MIA for new laptop and desktop OEM design wins from 2012 until late 2015. All NV had to do was release great products to keep renewing OEM design wins already won since Fermi/Kepler generations. Also, based on the horrendous paper launch of GTX1070/1080, it seems NV has no volumes in the market at all. I wouldn't be surprised if NV pulled off a GTX1060/Ti paper launch before Polaris 10/11 drop but what difference does it make if you cannot walk into a store and buy it same day?

We already saw people making claims in 2015 that AMD will never go above 20% dGPU market share and now the NV PR/online NV focus group members are trying to position GTX1070/1080 against P10 and desperately trying to align the price/performance of 1070 against P10 in favour of the NV card (despite ignoring the same metric entirely during GTX750/750Ti vs. R9 270/270X, R9 380 vs. GTX950, R9 280X/380/380X vs. 960, R9 290 vs. GTX960/960 SL comparisons).

As long as AMD delivers even a $249 RX 480 8GB in actual large volumes, they will take away NV's market share. If otoh, AMD also paper launches and it takes until August/September to be able to easily buy $199-249 RX 480 cards, then their strategy was poorly executed.

:thumbsdown:

You don't need a source but others do ?

Please stop with this nonsense.

I don't understand how he hasn't gotten banned yet. Every single topic in the NVIntel category vs. AMD is trying to talk garbage about AMD. Still hasn't admitted how awful 500 Euro GTX680 and 550 Euro 980 were though. :sneaky:

Since you predict AMD will lose marketshare, it's almost assured they will have a brilliant Q3 and Q4 this year.

That's true. :thumbsup:

145W typical gaming, 154W maximum according to TechPowerUp.

For a $449 reference card, which most of us will never touch. Here is a dose of reality for the types of cards we'll actually want to buy.

http://www.guru3d.com/articles_pages/msi_geforce_gtx_1070_gaming_x_review,8.html

This is the reverse of reference HD7970Ghz/R9 290X where NV loyalists will quote on days end all the poor metrics of power usage, noise levels, temperatures associated with those cards but many of us would never buy reference.

gtx 1070 uses 161w apparently...
http://www.guru3d.com/articles_pages/nvidia_geforce_gtx_1070_review,8.html
...hence the the pcie +8pin for a 150w tdp :\

Yup and it's more for AIB cards we'll want to purchase. If 1070 and RX 480 used a similar amount of power, then why is the 1070 not a 6-pin card?

This argument is useless anyway because who the hell cross-shops a $199-249 GPU with a $379-449 one? I find the hypocricy stunning after the same people who are bashing RX 480 have ignored the tremendous value R9 280X/380X/290 offered over the 950/960 cards, despite the AMD have having a much smaller premium than 1070 has over the RX 480.

AMD is going to be much more healthy as a company once they're completely rid of Global Foundries.

They cannot because Mubadala Development Company is one of the largest investors in both firms. When nearly 20% of your stock is owned by a single investor dropped an additional $10B in 2014, you do what you are told so to speak.

"ATIC owns unlisted GlobalFoundries, having completed a buyout of joint venture partner Advanced Micro Devices Inc in March 2012. ATIC is controlled by Abu Dhabi state investment fund Mubadala. Abu Dhabi's Advanced Technology Investment Co (ATIC) plans to invest up to $10 billion over the next two years in GlobalFoundries' upstate New York semiconductor factory, its chief executive said on Friday."

"GLOBALFOUNDRIES, which is 100% owned by Mubadala...

"Mubadala will purchase 58 million shares of AMD’s common stock..."

As of 2015:

"Besides all the shares outstanding and shareholders, Mubadala Development Co. the investment arm of the government of Abu Dhabi, owns a 19% stake in AMD after helping it spin off its foundry business in 2009 into Globalfoundries."
http://www.itworld.com/article/2982...y-firm-rumored-to-buy-a-big-chunk-of-amd.html

Sorry, the above is off-topic but just wanted you to be aware of the context. There is a lot more to it how AMD has to run its business and its supplier agreements than the HardOCP rant how AMD failed with next gen cards since P10 cannot compete with GP104, thus AMD was forced to cut prices. If AMD doesn't buy a certain amount of wafers from GloFo a year, they will be stuck with tens or hundreds of millions of penalties for breaching their supplier agreement. It means even if Vega and Navi are the best chips since 9700/9800Pro, unless AMD can start selling large enough volumes, their graphics division will continue to lose $. That's why AMD needs certain products to just sell in large volumes to meet the wafer quota.

That's a custom card. Absolute maximum for FE in games was 151W according to TPU (154W stress testing), considerably less than what Geforce GTX 970 got (191-201W).

Are you trying to be purposely misleading?

At Guru3D, reference 970 used 164W, reference 1070 used 161W, almost the same power usage.
http://www.guru3d.com/articles_pages/nvidia_geforce_gtx_970_and_980_reference_review,7.html

After-market 970 used 190-200W.
https://www.techpowerup.com/reviews/MSI/GTX_970_Gaming/25.html

Adored did a video with some comments on the efficiency based on information AMD released. His argument was that their claim that 2 480s in crossfire being called more powerful and "more efficient" than a 1080 meant that they were probably drawing around 95W.

This is because Cross-fire doesn't have 100% GPU utilization per card in Ashes of the Singularity. This doesn't prove that a stand-alone RX 480 uses only 95W-100W. Same reason that unless one is running distributed computing, mining, etc. workloads on a GPU, you never take a single GPU's maximum power usage and double it for 2 cards or triple it for 3 cards. Go to 4 min 33 seconds:
https://www.youtube.com/watch?v=UnS0xWtoRzk

I wouldn't be surprised if Pascal beats GCN 4.0 parts in perf/watt because of how great the deficit was for AMD during Maxwell days.

Regardless, I will continue to insist that the Engineer's perf/watt is a useless metric for consumers because you cannot use a videocard in a vacuum/stand-alone. Total system power usage is what consumers should care about when determining the Gamer's/Consumer's Performance/watt per each FPS produced. This is because all of the components, including the motherboard, the CPU, the memory, the SSD, the Graphics card work together to produce the output we see on the screen.

Test System Specs
Intel Core i7-6700K @ 4.50 GHz (Skylake)
Asrock Z170 Z170 Extreme7+
G.Skill TridentZ 8GB (2x8GB) DDR4-3000
Samsung SSD 850 Pro 2TB
Silverstone Strider Series ST1000-G Evolution
AMD Crimson Edition 16.5.3 Hotfix
Nvidia GeForce Game Ready Driver 368.22
Windows 10 Pro 64-bit





Any single-chip 2010->2016 modern graphics card with a Core i7 6700K @ 4.5Ghz will work perfectly fine on a 500W or even 450W high quality PSU.

This means the perf/watt superiority of 1070 over RX 480 is largely irrelevant for desktops where both cards will work in systems powered by the same PSU, but one will cost substantially less. Once GTX1060/Ti launches and we know its price, we'll have a better idea of how good RX 480 looks for next gen products. For now, trying to jam $199-249 RX 480 into a $379-449 performance category makes no sense.
 
Last edited:

Sweepr

Diamond Member
May 12, 2006
5,148
1,142
131
For a $449 reference card, which most of us will never touch. Here is a dose of reality for the types of cards we'll actually want to buy.

Here is a dose of reality for you as well. Only 14W difference in actual games according to Hardware.info, easily justified by higher average clocks, which also bring slightly better performance than Founder's Edition.

Also from your Guru3D review:

Guru3D said:
We noticed that the Gaming X 1070 did consumer a bit more power due to its tweaks and design, it was just over 180 Watts under full stress. That's OK though. Here again keep in mind we measure peak power consumption, the average power consumption is a good notch lower depending on GPU utilization
 

Mopetar

Diamond Member
Jan 31, 2011
8,011
6,455
136
But in that case, why bother with saying 150W at all.

If they treat it like the maximum it can draw, I suppose it's just letting you know where the ceiling is at.

My guess is that the 480 will have plenty of headroom to play with, but the scaling won't be all that great and that the efficiency goes to hell when you start using all of that 150W.
 

swilli89

Golden Member
Mar 23, 2010
1,558
1,181
136
If they treat it like the maximum it can draw, I suppose it's just letting you know where the ceiling is at.

My guess is that the 480 will have plenty of headroom to play with, but the scaling won't be all that great and that the efficiency goes to hell when you start using all of that 150W.

Yeah based on Hawaii, AMD might be leaving a bit more overclocking room on the table to ensure that its stock clocks achieve great power efficiency. Hawaii was probably 100-150mhz over its sweet spot so maybe that's why AMD is launching Polaris 10 at 1266mhz instead of 1500mhz.
 

flopper

Senior member
Dec 16, 2005
739
19
76
If they treat it like the maximum it can draw, I suppose it's just letting you know where the ceiling is at.

My guess is that the 480 will have plenty of headroom to play with, but the scaling won't be all that great and that the efficiency goes to hell when you start using all of that 150W.

waterblock baby and then weeeeeeeee
 

crisium

Platinum Member
Aug 19, 2001
2,643
615
136
Here is a dose of reality for you as well. Only 14W difference in actual games according to Hardware.info, easily justified by higher average clocks, which also bring slightly better performance than Founder's Edition.

Also from your Guru3D review:

So are you still saying 480 and 1070 will have the same power consumption? That's the only context of 1070 discussion here. That's still over 150W. So you think the 1x6pin 480 will typically use over 150W?
 

Sweepr

Diamond Member
May 12, 2006
5,148
1,142
131
So are you still saying 480 and 1070 will have the same power consumption? That's the only context of 1070 discussion here.

Probably not (for their sake), but I don't think AMD picked this number for no reason, max power consumption for the 8GB model could be close.

That's still over 150W.

For a factory OCed custom model, yes, as expected. Barely above 150W (stress testing) for FE according to TPU.
 
Last edited:

poofyhairguy

Lifer
Nov 20, 2005
14,612
318
126
So someone tell it to me straight- will the 8GB 480 be an upgrade over my 390x at 1080p? Because if not I might just buy the 4GB model for a secondary rig.
 

zinfamous

No Lifer
Jul 12, 2006
110,810
29,564
146
So someone tell it to me straight- will the 8GB 480 be an upgrade over my 390x at 1080p? Because if not I might just buy the 4GB model for a secondary rig.

What my uninformed self is gathering from all of this discussion and numbers, is that it will probably be a slight upgrade: slight performance, significant power, and not-yet disclosed % of upgrade for DX12.

It is looking like the big difference here is cost/performance and DX12, which NVidia seems to have screwed the pooch over for the next 2-3 years with their new architecture. Updated and optimized drivers .5-1 year out might just show that Polaris is a monster for DX12 at budget cost.
 

HannooFX

Member
Jun 6, 2016
56
22
41
So someone tell it to me straight- will the 8GB 480 be an upgrade over my 390x at 1080p? Because if not I might just buy the 4GB model for a secondary rig.

I don't believe so, from what was seen so far performance is on par with 390, or slightly higher.

But we should wait and see on the 29th
 
Status
Not open for further replies.
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |