• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Polaris architecture – GCN 4.0

Must admit, i do like watching the reviews of cards on YT. :)

I'm the opposite. I'll watch a video review if it's the best source of information, or I'll watch Linus because I kind of like him and he talks very fast (this is important). But I far prefer written word as a medium because I can speed-read through it, the medium affords more depth and I can easily back-reference information to help me retain it while I read.

In the time it takes to watch a 15 minute review of a piece of technology, I can have read an eight-page article about the same on Anandtech and come away with many times the information and it will stick better.

Plus there's no chance of me accidentally reading YouTube comments. It always takes me several days to fully recover my faith in humanity after an exposure to them.
 
AMD have simply got to get beyond Hawaii performance with Polaris or it's underwhelming as hell. We already saw brand new Hawaii-based cards for £200 ages ago.

edit: <--- quads check em

Definitely above Hawaii if the AotS benchmarks are anything to go by, and under half the power consumption. Just hope we'll see custom RX 480s that go to ~1500Mhz (would probably outperform even the GTX 980 Ti in DX12 if async is enabled).

If we take the average of the GPU Utilization for each of the different benchmarks we get (0.51 + 0.719 + 0.923)/3 = ~0.717

Then we take the average score of 62.5 fps and divide it with the average utilization, so 62.5/0.717 = ~87.2, and then we lastly divide that with the scaling, which according to AMD was 1.83. So 87.2/1.83 = 47.65 fps (1080p). So one card would theoretically speaking score 47.65 fps at 100% utilization.
 
Last edited:
I'm the opposite. I'll watch a video review if it's the best source of information, or I'll watch Linus because I kind of like him and he talks very fast (this is important). But I far prefer written word as a medium because I can speed-read through it, the medium affords more depth and I can easily back-reference information to help me retain it while I read.

In the time it takes to watch a 15 minute review of a piece of technology, I can have read an eight-page article about the same on Anandtech and come away with many times the information and it will stick better.

Plus there's no chance of me accidentally reading YouTube comments. It always takes me several days to fully recover my faith in humanity after an exposure to them.

I have followed linus since before his "560ti 1ghz review in a closet" days and always liked what he put out but as of late im getting sick and tired of that channel. Its more about uploading a video than actually providing thoroughly tested results. I still think he is a cool guy including a few from his team, but after i got burned hard on the best gaming monitor yet twice with the asus rog swift and the acer XB270HU, not to mention a lot of smaller issues with how they obtain results from time to time, im done listening to that channel and only watch it for entertainment and not for accurate information.
 
It is actually far more than that:

http://tpucdn.com/reviews/NVIDIA/GeForce_GTX_1080/images/perfrel_2560_1440.png

JDH2uQM.png


An R9 390 is 60% faster than an R9 380 and an R9 390X is 74% faster than an R9 380. An R9 390 is 81% faster than a GTX960 and an R9 390X is 97% faster than a GTX960.

TPU places the R9 280X and R9 380X as having the same performance in recent reviews:

http://tpucdn.com/reviews/ASUS/GTX_950/images/perfrel_2560_1440.png

So,that makes an R9 390, 37% faster than an R9 380X and 49% faster than an R9 380X.

So an RX480 having either R9 390 or R9 390X level performance will be a massive upgrade in the segment with no price increase in the UK over the cards it is replacing. The GTX1070 is not comparable since you are looking at £320 at least for the cheapest model which is a good £70 increase in price over the £250 GTX970 which it is replacing.

Even if you take dollar amounts into consideration,it does not look good for the GTX1070 - the performance increase still comes at a price increase.

Also since the GTX1070 is meant to be around Titan X level,it means it is around 50% faster than a GTX970 looking at the TPU figures.

The numbers you show are for a 2GB 380 which is ~10% + slower at 1440p.

Here is a nine game average with R9 380 4GB coming around 40%-45% slower than R9 390 at 1440p.

http://www.pcgamer.com/sapphire-radeon-r9-380-4gb-review/

Most games in this review show ~40% or less improvement from R9 380 4GB to R9 390 8GB.
http://www.gamersnexus.net/hwreviews/1984-amd-r9-390-380-benchmark-review/Page-2

Don't get me wrong, RX 480/X could still be higher performance than R9 390 and if they are then great because given the specs that's right where they should be IMHO.
 
Last edited:
Nice rewrite of history there - ATI/AMD dropped the ball on tessellation it was nVidia who pushed it into actual usage (when the time was right) even if subsequently they may have used it politically.

nVidia are responsible for the widespread adoption of tessellation? really?
 
AMD brought tessellation to games before Nvidia, the xbox360 had tesselation to a degree. Even with it being DX9.

I still believe they have something up their sleeve, still some things they have not fully clarified and wont till the NDA lifts. The average TDP is one of them considering they just showed max board power and Robert hallock said the info was still under NDA when questioned on reddit.

I think they were just giving us a taster to break some of nvidias hype over the 1070/1080 by doing a part unveil and showing the price and base specs of the card. Considering they said Tflops >5 they could have a fair boost clock hiding up their sleeves and showed just the base performance. But its all speculation till the NDA lifts, hallock even said that he could not confirm the NDA lifting date since it breaks the NDA.
 
It isn't that simple as the GPU Uilization was 51% for the single batch, 71.9% for the medium batch, and 92.3% for the heavy batch. Source

This is closer to the truth, but that was run on an older version so performance could have been improved in the newest version. This would put it between a Titan X and Fury X.

That isn't maxed out though. (Or were you comparing to Titan X's and FuryX X's at those settings?)
 
So does the fact that the EU is removing import tax from electronics, does it means we're going to finally pay very close to to a 1-1 parity for GPUs here? I wouldn't mind the 8GB 480 coming in at under £199

https://www.meijburg.com/news/ita-e...ctronic-products-eliminated-as-of-july-1-2016

Dont see any mention of computer components there :( closest thing is Games Consoles and Games..

Also if you read further upto 50% of that list will get the deduction straight away the remainder will be phased out over a period of years, 3 for most, upto 7 for others.

Does anyone know how much import duty there is currently on PC components?
 
So does the fact that the EU is removing import tax from electronics, does it means we're going to finally pay very close to to a 1-1 parity for GPUs here? I wouldn't mind the 8GB 480 coming in at under £199

https://www.meijburg.com/news/ita-e...ctronic-products-eliminated-as-of-july-1-2016

It doesn't say things like GPUs, monitors, CPUs, so I expect to still pay those duties. I got stung for a Sammy 4K monitor from Korea, which was a shame, as it wasn't even out in this country and cheaper than when it was launched but paying the Import tax and handling fee's, it made it dearer and of course a problem if there was any issues down the line and needs RMA.
 
Definitely above Hawaii if the AotS benchmarks are anything to go by, and under half the power consumption. Just hope we'll see custom RX 480s that go to ~1500Mhz (would probably outperform even the GTX 980 Ti in DX12 if async is enabled).

If we take the average of the GPU Utilization for each of the different benchmarks we get (0.51 + 0.719 + 0.923)/3 = ~0.717

Then we take the average score of 62.5 fps and divide it with the average utilization, so 62.5/0.717 = ~87.2, and then we lastly divide that with the scaling, which according to AMD was 1.83. So 87.2/1.83 = 47.65 fps (1080p). So one card would theoretically speaking score 47.65 fps at 100% utilization.

Oh my god. The AoTS does NOT tell you what the utilisation is, it only tells you how much of the time its at 100% utilisation. Your calculation assumes that any time its at less than 100% its at 0%, which is not the case at all.

It could be at 70 or 80 or even 90% utilisation.

Also, robert hallock on reddit said the scaling was 151%, not 1.83 times (where did that figure even come from).

In any case, if you already know the scaling, you dont need to take utilisation in to account, the scaling figure tells you what result they got with a single card already.
You would use the utilisation figure if you wanted to calculate a better dual card result at full utilisation, not for single card.

Your 1.83 scaling puts a single card at 33fps. Robert's 1.51 scaling figure puts a single card at 41fps.
 
Last edited:
I have followed linus since before his "560ti 1ghz review in a closet" days and always liked what he put out but as of late im getting sick and tired of that channel. Its more about uploading a video than actually providing thoroughly tested results. I still think he is a cool guy including a few from his team, but after i got burned hard on the best gaming monitor yet twice with the asus rog swift and the acer XB270HU, not to mention a lot of smaller issues with how they obtain results from time to time, im done listening to that channel and only watch it for entertainment and not for accurate information.

well he's a shill so...
 
Dont see any mention of computer components there :( closest thing is Games Consoles and Games..

Also if you read further upto 50% of that list will get the deduction straight away the remainder will be phased out over a period of years, 3 for most, upto 7 for others.

Does anyone know how much import duty there is currently on PC components?

It doesn't say things like GPUs, monitors, CPUs, so I expect to still pay those duties.

I figured computers and their parts would fall under these two.

Video games and consoles
Multi-component integrated circuits (MCOs)


It would be nice if they are included technically. If not nothing changes for our hobby here :(
 
The FX CPUs were very forward thinking, designed for software that simply didn't exist at the time and ignoring the fact that such software would take some time to be developed. Multi-threading is very hard to do right, I know, I do it professionally, I can spend days debugging the most innocuous looking code. Strangely AMD made the same mistake with Hawaii and Fiji, a hardware that in theory should be incredibly powerful but hammered by the current APIs and some frontend bottlenecks. This is something of a theme for AMD - tomorrow's technology today. Sounds great, forward thinking, but in reality always comes back to bite them. HBM, great stuff, but utterly pointless on a FuryX, Pioneering Tessellation, and the totally ignoring it when it becomes practical.

Intel made a similar error with the Itanium. X86 architecture sucks and is far to antiquated, Buick on crumbling bloated foundations. Itanium-64 is fantastic improvement in theory. In practice they never made decent compilers, making decent compilers was incredibly tough, and even when they did it was a failure because no software developers would code specifically for that architecture.

Ironically AMD by choosing the cheaper and simpler 64bit extension x86 was a fantastically successful strategic move that really kicked Intel down a beat.

yes and no, it was forward thinking, but if you remember few year prior, there was a lot of talk about DX and multi-threading, but DX10-11 came out not very far from DX9, microsoft didnt follow, and didnt seem to have plans to change that anytime soon, which prompted the birth of Mantle, but that was like 4 years late, enough time for AMD CPUs to get the reputation of being bad, which they are not if put in their element.
about GPUs to me thats not a mistake, they did it when they decided to do custom soc specific for consoles, and it paid out, so they had to sacrifice a little bit on the PC front in efficiency for that, but the upside is that they are in every console on the market, Devs are more familar with their hardware, they are in the middle of it being able to influence change from the inside, and their next step is pushing multi-gpu into console, and laveraging that on PC by making multiple chips on an interposer.
now that the API is less crappy than before, they are doing better with software using them on both fronts gpu and cpu.
to be honest AMD did really well, with so little money, and really bad luck during these last years, most companies would have gone belly up long ago, or gave up, they have a really good long term planning, so polaris/vega/zen should help them get out of the red zone definitely by the end of the year, at E3 if sony and microsoft announce their new consoles upgrade and turns out to be multi-gpu, then AMD is going to have a great 2017/2018, from market share to margins, and money will flow...hopefully
 
There is benchmarks to compare in aots. On one day he benched single and dual at the same settings. He got 40 for single and 60 for dual. Scaling here was 50%. The next day he done another dual run at the same settings and got 66. He never done a single run this day so I don't know if his score would have improved. This was done at 1440p Extreme settings.

Anyhow if we take 40 and 66 scaling here would be 65%. If he got the extra performance because single card performance was up then scaling might have been in the 50% area again.

The benches in question were all from Radeon demo.
 
Are you sure about it being 1440p? looking on the aots bench site the 1440p dual card result was 51fps
The 62.5fps figure they used in the presentation was 1080p

Yea am sure. The ones you are seeing were done at crazy settings. I posted the links a few pages back and they are 1440p extreme settings hence the higher scores. There is around 30 different runs done by him. Most are Gtx1080 though :D:D.
 
Back
Top Bottom