[JayzTwoCents] R9 390 vs GTX970 - Claims 300 series is NOT a rebrand - R9 390 new mid tier king?

[ivan134]

The outcry is because the whole 300 series (which there was a lot of hype before the release) turned out to be all rebrands/refreshes. Nobody cares if nvidia rebrands the low end gpus because most of the people that buy them don't even have an igpu. The gtx 770 is nvidia's only high end rebrand in the past couple of years. Meanwhile amd have been reusing most of their 2012 gpus (hd 7850,7870,7950 and 7970). Although might I say they have been doing an excellent job keeping up.

No, they haven't been reusing all those cards. 7850 is the only to make it into the 300 series.

[Curufinwe_wins]

Irrelevant numbers that mean absolutely nothing is AMD shitting the bed? There is almost enough of you people now so we can have a straw grasping olympics.

 

Sorry but your opinion does not make it fact. 290, 390, 970 and 780 are all 1440p cards. Your opinion on what they should be used for is not going to change that.

 

EDIT: Even if they were 1080p cards, how the fuck does that make them low end?

Irrelevant numbers? Like Amd's less than 20% market share? Like them taking 7 years to develop a GPU that doesn't even beat the one Nvidia released 3 months earlier? Like them in 20 months developing a grand total of one (two if you count tonga) new gpu cores in the same time nvidia released 5 (GM 107/108/200/204/206) and Pascal is likely only 6-12 months away, do you honestly think AMD will bring anything new to the table before then? They are falling to the wayside just like AMD has these last 5 years against Intel, hell if you include mobo costs there is basically no case where a AMD cpu makes sense let alone in a power user situation.

 

Lol 1440p on high sub 50 fps... If that is really acceptable, then sure you can count them as 1440p cards (hell graphics settings on ultra, you have issues pushing 40 fps on Witcher 3 1440p with anything lower than a 980. And newer releases don't get less demanding over time... EDIT: Yes I know hairworks blah blah blah, but the point of newer releases being more and more demanding is still valid.). 1080p is the currently the lowest resolution most of the pc gaming community finds acceptable. Which is why something like 97% of all steam games are run on 1080p. I think it's extremely fair to say that if the card is only capable of running contemporary titles, with acceptable settings and acceptable framerates, at the lowest common denominator resolution then the card IS low-end, regardless of price.

 

And yes by my own defintions no card is currently a 4k card, I think many if not most of the pc gaming community would agree that the marketing of any current card as 4k-ready is total stretch if not an outright lie.

[Notional]

Irrelevant numbers? Like Amd's less than 20% market share? Like them taking 7 years to develop a GPU that doesn't even beat the one Nvidia released 3 months earlier? Like them in 20 months developing a grand total of one (two if you count tonga) new gpu cores in the same time nvidia released 5 (GM 107/108/200/204/206) and Pascal is likely only 6-12 months away, do you honestly think AMD will bring anything new to the table before then? They are falling to the wayside just like AMD has these last 5 years against Intel, hell if you include mobo costs there is basically no case where a AMD cpu makes sense let alone in a power user situation.

 

I don't think you know what market share is, if you think AMD only has 20% in all the market. Also remember AMD has 100% of the console market. People seem to not remember that.

 

As for Fiji, it did not take 7 years to develop. That was the HBM, which NVidia gets to enjoy next year with Pascal, thanks to AMD in part. Speaking of Pascal, there is no way that it will release in 6 minths. Right now it's even doubtful it will release in 12 months. NVidia still hasn't signed a contract with Global Foundries, that already makes 14nm FinFet (and will make AMD A/C/GPU's for AMD next year in 14nm FinFet).

That means NVidia might be forced to use TSMC's 16nm FinFet+, which is not even ready yet. Knowing TSMC's latest history, they will be delayed here too, which means there is a very high posibility of AMD's 14nm FinFet HBM2 graphicscard, called Greenland, will launch before Pascal.

 

As for Maxwell, those 5, are all Maxwell GPU's cut and sized differently. They are not 5 unique GPU architectures, which means that AMD has launched just as many new architectures as NVidia.

[Vozella]

I first thought a rebrand was the same thing as a refresh. That's why I called it either way.

[Curufinwe_wins]

I don't think you know what market share is, if you think AMD only has 20% in all the market. Also remember AMD has 100% of the console market. People seem to not remember that.

 

As for Fiji, it did not take 7 years to develop. That was the HBM, which NVidia gets to enjoy next year with Pascal, thanks to AMD in part. Speaking of Pascal, there is no way that it will release in 6 minths. Right now it's even doubtful it will release in 12 months. NVidia still hasn't signed a contract with Global Foundries, that already makes 14nm FinFet (and will make AMD A/C/GPU's for AMD next year in 14nm FinFet).

That means NVidia might be forced to use TSMC's 16nm FinFet+, which is not even ready yet. Knowing TSMC's latest history, they will be delayed here too, which means there is a very high posibility of AMD's 14nm FinFet HBM2 graphicscard, called Greenland, will launch before Pascal.

 

As for Maxwell, those 5, are all Maxwell GPU's cut and sized differently. They are not 5 unique GPU architectures, which means that AMD has launched just as many new architectures as NVidia.

I had a long reply vacating basically everyone of those points except TSMC (still k12 in 2017 lol), but I hit backspace and it's really too much effort to rewrite that when no matter what I say (or you say for that matter)neither of us is going to change either of our opinions.

[Echodamus]

You can call it a cat for all I care, it doesn't change what it is...

[Dabombinable]

I had a long reply vacating basically everyone of those points except TSMC (still k12 in 2017 lol), but I hit backspace and it's really too much effort to rewrite that when no matter what I say (or you say for that matter)neither of us is going to change either of our opinions.

That's Notional for you.

[Notional]

I had a long reply vacating basically everyone of those points except TSMC (still k12 in 2017 lol), but I hit backspace and it's really too much effort to rewrite that when no matter what I say (or you say for that matter)neither of us is going to change either of our opinions.

 

Too bad, would have liked to hear your views on things. As for k12, do anyone in here really care? It's not an x86 part, so what would any of us need it for next year?

 

As for Global Foundries their 14nm FinFet LPE node us fully operational with mass production yields (whatever that means in percentage). LPP will be certified later this year. Maybe in time to be used for Zen and Greenland. It would be interesting to see AMD and NVidia on different manufacturing foundries (and nodes).

 

That's Notional for you.

 

Yes, how DARE I stick to facts, and state things there are legimate sources supporting!

[Kobz360]

Irrelevant numbers? Like Amd's less than 20% market share? Like them taking 7 years to develop a GPU that doesn't even beat the one Nvidia released 3 months earlier? Like them in 20 months developing a grand total of one (two if you count tonga) new gpu cores in the same time nvidia released 5 (GM 107/108/200/204/206) and Pascal is likely only 6-12 months away, do you honestly think AMD will bring anything new to the table before then? They are falling to the wayside just like AMD has these last 5 years against Intel, hell if you include mobo costs there is basically no case where a AMD cpu makes sense let alone in a power user situation.

 

Lol 1440p on high sub 50 fps... If that is really acceptable, then sure you can count them as 1440p cards (hell graphics settings on ultra, you have issues pushing 40 fps on Witcher 3 1440p with anything lower than a 980. And newer releases don't get less demanding over time... EDIT: Yes I know hairworks blah blah blah, but the point of newer releases being more and more demanding is still valid.). 1080p is the currently the lowest resolution most of the pc gaming community finds acceptable. Which is why something like 97% of all steam games are run on 1080p. I think it's extremely fair to say that if the card is only capable of running contemporary titles, with acceptable settings and acceptable framerates, at the lowest common denominator resolution then the card IS low-end, regardless of price.

 

And yes by my own defintions no card is currently a 4k card, I think many if not most of the pc gaming community would agree that the marketing of any current card as 4k-ready is total stretch if not an outright lie.

Steam Hardware Survey - 34% of users on 1080p, 26.89% on 1366x768 and 7.41% on 900p..... There's barely 3% of total users running anything above 1080p... We are the niche in the PC gaming market man, too many of you fail to remember that. Yes, 1080p is seen as the standard (barely, as shown by statistics), but just because a minority of games are harder to run than others, doesn't mean that it makes specific GPU's are 'low-end', simply because it doesn't run the most demanding games to your standards. I wouldn't be surprised if a 270X could run the majority of games on the Steam market, max settings at 1080p. 

 

It is ridiculous to argue that just because the 390/970 can't run Witcher strongly at a resolution which a strong minority of people actually have (1440p), they become 'low-end'. What I would like to know is what your mid-end and high-end is? Because you're shaping up a pretty narrow market. 

[Curufinwe_wins]

Too bad, would have liked to hear your views on things. As for k12, do anyone in here really care? It's not an x86 part, so what would any of us need it for next year?

 

As for Global Foundries their 14nm FinFet LPE node us fully operational with mass production yields (whatever that means in percentage). LPP will be certified later this year. Maybe in time to be used for Zen and Greenland. It would be interesting to see AMD and NVidia on different manufacturing foundries (and nodes).

Short summary. If AMD gets console, intel gets everything with discrete. GCN 1.2 blah, poor gaming to SP compute ratios still indicate 'relatively' piss poor architecture. Hopefully TSMC can finally get shit together (and this is no more than a Broadwell sized hiccup).

[Dabombinable]

Too bad, would have liked to hear your views on things. As for k12, do anyone in here really care? It's not an x86 part, so what would any of us need it for next year?

 

As for Global Foundries their 14nm FinFet LPE node us fully operational with mass production yields (whatever that means in percentage). LPP will be certified later this year. Maybe in time to be used for Zen and Greenland. It would be interesting to see AMD and NVidia on different manufacturing foundries (and nodes).

 

 

Yes, how DARE I stick to facts, and state things there are legimate sources supporting!

Gameworks is cuda accelerated.

[othertomperson]

I first thought a rebrand was the same thing as a refresh. That's why I called it either way.

 

What Intel did with Devil's Canyon was a refresh.

 

What AMD have done here is the equivalent of changing the TIM, improving the power delivery, calling it Broadwell, renaming the resulting chip the 5770k, increasing the price by $150 and then consistently denying what they had done.

 

They did not "refresh" the 290 and 290X, they released what they claimed was an all new flagship. They did not call it "200 series refresh" in line with what Intel did last year, they released the 300 series, and bumped prices up significantly.

 

There are ways they could have done this honestly. They chose not to.

[Dabombinable]

What Intel did with Devil's Canyon was a refresh.

 

What AMD have done here is the equivalent of changing the TIM, improving the power delivery, calling it Broadwell, renaming the resulting chip the 5770k, increasing the price by $150 and then consistently denying what they had done.

 

They did not "refresh" the 290 and 290X, they released what they claimed was an all new flagship. They did not call it "200 series refresh" in line with what Intel did last year, they released the 300 series, and bumped prices up significantly.

 

There are ways they could have done this honestly. They chose not to.

They did the same thing with the FX 8*** as well, as soon as AMD saw that people had them running fine at 5Ghz+, they decided to start binning them and calling them FX 9590.

[Notional]

Short summary. If AMD gets console, intel gets everything with discrete. GCN 1.2 blah, poor gaming to SP compute ratios still indicate 'relatively' piss poor architecture. Hopefully TSMC can finally get shit together (and this is no more than a Broadwell sized hiccup).

 

Not sure I follow, how would Intel get everything descrete? Sure Intel is the largest GPU seller, with their integrated GPU's, but that is about it. I doubt they will ever gain any control of gaming, as that requires equal parts software (drivers).

As for GCN, I believe Greenland will be GCN 2.0, but that of course is not official yet. I read a little more, looks like TSMC won't put a lot of effort into 16nm FinFet+, but focus on getting 10nm done for 2017.

 

Gameworks is cuda accelerated.

 

Only APEX based GameWorks effects are CUDA based:

https://developer.nvidia.com/content/introducing-nvidia-gameworks

 

As you can see, almost all VisualFX parts of GameWorks, are NOT APEX based, so it has nothing to do with CUDA. Also what does that have to do with the post I made in answer to Curufinwe?

[Curufinwe_wins]

Steam Hardware Survey - 34% of users on 1080p, 26.89% on 1366x768 and 7.41% on 900p..... There's barely 3% of total users running anything above 1080p... We are the niche in the PC gaming market man, too many of you fail to remember that. Yes, 1080p is seen as the standard (barely, as shown by statistics), but just because a minority of games are harder to run than others, doesn't mean that it makes specific GPU's are 'low-end', simply because it doesn't run the most demanding games to your standards. I wouldn't be surprised if a 270X could run the majority of games on the Steam market, max settings at 1080p. 

 

It is ridiculous to argue that just because the 390/970 can't run Witcher strongly at a resolution which a strong minority of people actually have (1440p), they become 'low-end'. What I would like to know is what your mid-end and high-end is? Because you're shaping up a pretty narrow market. 

Anything can run CS:GO on max with insane fps. That comment is completely irrelevant. Modern AAA titles are more and more demanding as a rule. If the "console experience" isn't in your opinion low end, then the exact opposite issue of mine is the case. The other issue with looking at steam user sets is it reflects adoption rates more than current standard experiences.  In 2015, 1080p is definitively the lowest common denominator in pc gaming.

 

Medium end and high end isn't a particularly well defined barrier imho, because the return on investment becomes so insanely small as you climb up the ladder. The difference in real performance and your ability to break down well accepted standard barriers with increasingly expensive items on the same architecture is rather underwhelming. (The difference between say a 5820k and a 5930k on the med/high end edge is a perfect example of this).

 

Perhaps this is a reflection of this particular situation we are in now. Perhaps in the next generation 1080p will still be the same lowest common denominator, while 1440p and 4k resolutions will become more accessible with modern titles in a wide range of gpus.

 

This trend is very much mirrored in the tv entertainment world. 1080p is such a widely accepted format, but such an insanely marginal number of media is capable of higher resolutions that while almost anything less than 1080p is considered low-end for tv's (albeit its amusing that dvd's are still fairly common, but up-sampling in videos is much more effective than games ofc.) while 1440p is extremely rare, and 4k is more common, but generally seen as beyond excessive.

[Curufinwe_wins]

Not sure I follow, how would Intel get everything descrete? Sure Intel is the largest GPU seller, with their integrated GPU's, but that is about it. I doubt they will ever gain any control of gaming, as that requires equal parts software (drivers).

As for GCN, I believe Greenland will be GCN 2.0, but that of course is not official yet. I read a little more, looks like TSMC won't put a lot of effort into 16nm FinFet+, but focus on getting 10nm done for 2017.

 

 

Only APEX based GameWorks effects are CUDA based:

https://developer.nvidia.com/content/introducing-nvidia-gameworks

 

As you can see, almost all VisualFX parts of GameWorks, are NOT APEX based, so it has nothing to do with CUDA. Also what does that have to do with the post I made in answer to Curufinwe?

I messed up writing mine really quick. I meant ofc that if you throw consoles in there then you really have to throw intel's igpu in there and then AMD's market (and nvidia's for that matter) becomes laughably small.

 

I have no idea what the CUDA comments were about. 

 

I will straight out say this. IFF GCN 2.0 can come close (+/-5%) to bringing the extremely efficient compute to gaming performance that Maxwell brings, then AMD should have no issues whatsoever keeping up with Nvidia in the near future. I worry about that possibility because thus far (yes its ATI vs AMD and there is a huge difference in the types) its cpu market has had a similar issue getting efficient architectures in place.

[ivan134]

Irrelevant numbers? Like Amd's less than 20% market share? Like them taking 7 years to develop a GPU that doesn't even beat the one Nvidia released 3 months earlier? Like them in 20 months developing a grand total of one (two if you count tonga) new gpu cores in the same time nvidia released 5 (GM 107/108/200/204/206) and Pascal is likely only 6-12 months away, do you honestly think AMD will bring anything new to the table before then? They are falling to the wayside just like AMD has these last 5 years against Intel, hell if you include mobo costs there is basically no case where a AMD cpu makes sense let alone in a power user situation.

 

Lol 1440p on high sub 50 fps... If that is really acceptable, then sure you can count them as 1440p cards (hell graphics settings on ultra, you have issues pushing 40 fps on Witcher 3 1440p with anything lower than a 980. And newer releases don't get less demanding over time... EDIT: Yes I know hairworks blah blah blah, but the point of newer releases being more and more demanding is still valid.). 1080p is the currently the lowest resolution most of the pc gaming community finds acceptable. Which is why something like 97% of all steam games are run on 1080p. I think it's extremely fair to say that if the card is only capable of running contemporary titles, with acceptable settings and acceptable framerates, at the lowest common denominator resolution then the card IS low-end, regardless of price.

 

And yes by my own defintions no card is currently a 4k card, I think many if not most of the pc gaming community would agree that the marketing of any current card as 4k-ready is total stretch if not an outright lie.

Nice rant about something I wasn't even talking about. The irrelevant numbers I was referring to was the naming scheme you were complaining about. Like I said to the other guy, the name is fucking irrelevant. Would calling the Fury an R5 350 stop you from buying it regardless of performance? If so, then you have serious problems. Also the 390x and 390 did drop a tier. The 290x was AMD's former single GPU flagship. The Fury X is now their current flagship.

 

I'm not going to respond to the rest of your nonsense that seems to comes from a few enthusiasts who buy very high level hardware.

[ivan134]

What Intel did with Devil's Canyon was a refresh.

 

What AMD have done here is the equivalent of changing the TIM, improving the power delivery, calling it Broadwell, renaming the resulting chip the 5770k, increasing the price by $150 and then consistently denying what they had done.

 

They did not "refresh" the 290 and 290X, they released what they claimed was an all new flagship. They did not call it "200 series refresh" in line with what Intel did last year, they released the 300 series, and bumped prices up significantly.

 

There are ways they could have done this honestly. They chose not to.

Wow, you people just keep pouring in. This is my response to someone else who like you, who didn't seem to understand how retail works and keeps making ridiculous assertions.

 

I feel like you're just tunnel visioning in on your preconceived opinion and not actually listening to any facts being presented.

 

The price is justified because the manufacturing process was refined which led to better thermals, the card overclocks better and it has an extra 4GB of VRAM. There is no price hike. The 290 launched at $399. The 390 launched at $329. The 290 went on sale for as low as $220 sometimes, because it's been on the market for a while. I don't how much experience you have with retail of ANY product, but that's what happens. You don't price your new (or in this case refreshed) product based on the sale price of your older product that has been on the market for a while.

 

I think your confusion is who actually sets the price, and for the most part it's retailers. AMD just sets the MSRP, but it's up to the retailers to price a product according to what they think the market will bear. You would think people would understand this by now, especially after the mining craze that plagued the r9 200 series in the beginning and everyone blaming AMD for the high prices when they had nothing to do with it. That was just supply and demand doing what it does. AMD isn't duping anyone with the 300 series.

 

 

You can believe whatever you want, but it's still not going to make it true. There is no price hike, period. 390 launched at $329 while the 290 launched at $399. 390x launched at $429 while 290x launched at $549. I've already explained to you the way retail products are priced and you're choosing to ignore because you prefer your tantrum induced reasoning.

 

Were you this angry when Haswell was refreshed for exactly the same price?

[Notional]

I messed up writing mine really quick. I meant ofc that if you throw consoles in there then you really have to throw intel's igpu in there and then AMD's market (and nvidia's for that matter) becomes laughably small.

 

I have no idea what the CUDA comments were about. 

 

I will straight out say this. IFF GCN 2.0 can come close (+/-5%) to bringing the extremely efficient compute to gaming performance that Maxwell brings, then AMD should have no issues whatsoever keeping up with Nvidia in the near future. I worry about that possibility because thus far (yes its ATI vs AMD and there is a huge difference in the types) its cpu market has had a similar issue getting efficient architectures in place.

 

Yeah of course, Intel has a huge market share, but most of it is just for desktops and laptops 2D applications. In the perspective of gaming, it's less relevant. My 2011 i3 laptop, cannot even run telltales games like The Walking Dead (I know a lot has happened since then, but most people don't have the newest in iGPU's anyways).

 

The question I'm left with, is it it is necessary? NVidia are only marginally faster in the high end graphics market, but has chips, that are much larger (although Fiji is pretty damn close to Titanx in chip size). Either way, I cannot wait for for Greenland versus Pascal next year. Fiji was revolutionary on the memory side, but Greenland and Pascal will be on new nodes, and both have HBM2. So maybe higher hz, maybe smaller dies, but definitely less power hungry.

[Curufinwe_wins]

Nice rant about something I wasn't even talking about. The irrelevant numbers I was referring to was the naming scheme you were complaining about. Like I said to the other guy, the name is fucking irrelevant. Would calling the Fury an R5 350 stop you from buying it regardless of performance? If so, then you have serious problems. Also the 390x and 390 did drop a tier. The 290x was AMD's former single GPU flagship. The Fury X is now their current flagship.

 

I'm not going to respond to the rest of your nonsense that seems to comes from a few enthusiasts who buy very high level hardware.

 

The naming scheme doesn't particularly matter to people like us, but it does to the general public. It is an intentional attempt to capitalize on the general public's ignorance of the pc world. Ofc we should denounce it. Calling the Fury X a 350 wouldn't stop me from buying it but it would massively hurt overall sales (not that I would buy it in the first place with it's rather poor maximum performance at the moment, perhaps when people bios hack it, it may show more potential).

 

As to the rest of my 'nonsense', I do not own any components that classify on my own personal hardware that would count as med-high end. My primary (non-gaming) driver is a y580 rocking a massively overclocked 660m. So feel free to call me elitist, but I do not belong to my own 'elite' characterization. Although I do/have built and ran non-low end builds (by my standards), for friends and family, because as many of us here are I am within my person circle the go-to tech person.

[Curufinwe_wins]

Yeah of course, Intel has a huge market share, but most of it is just for desktops and laptops 2D applications. In the perspective of gaming, it's less relevant. My 2011 i3 laptop, cannot even run telltales games like The Walking Dead (I know a lot has happened since then, but most people don't have the newest in iGPU's anyways).

 

The question I'm left with, is it it is necessary? NVidia are only marginally faster in the high end graphics market, but has chips, that are much larger (although Fiji is pretty damn close to Titanx in chip size). Either way, I cannot wait for for Greenland versus Pascal next year. Fiji was revolutionary on the memory side, but Greenland and Pascal will be on new nodes, and both have HBM2. So maybe higher hz, maybe smaller dies, but definitely less power hungry.

Finally moving away from the 28 nm node should be a huge improvement. It is an exciting time. Imagine if intel was still on 28 nm. EWWWW.

[Kobz360]

Anything can run CS:GO on max with insane fps. That comment is completely irrelevant. Modern AAA titles are more and more demanding as a rule. If the "console experience" isn't in your opinion low end, then the exact opposite issue of mine is the case. The other issue with looking at steam user sets is it reflects adoption rates more than current standard experiences.  In 2015, 1080p is definitively the lowest common denominator in pc gaming.

 

Medium end and high end isn't a particularly well defined barrier imho, because the return on investment becomes so insanely small as you climb up the ladder. The difference in real performance and your ability to break down well accepted standard barriers with increasingly expensive items on the same architecture is rather underwhelming. (The difference between say a 5820k and a 5930k on the med/high end edge is a perfect example of this).

 

Perhaps this is a reflection of this particular situation we are in now. Perhaps in the next generation 1080p will still be the same lowest common denominator, while 1440p and 4k resolutions will become more accessible with modern titles in a wide range of gpus.

 

This trend is very much mirrored in the tv entertainment world. 1080p is such a widely accepted format, but such an insanely marginal number of media is capable of higher resolutions that while almost anything less than 1080p is considered low-end for tv's (albeit its amusing that dvd's are still fairly common, but up-sampling in videos is much more effective than games ofc.) while 1440p is extremely rare, and 4k is more common, but generally seen as beyond excessive.

My issue is that I don't believe there are enough demanding games out there, which at 1080p make the 390/970 be seen as low-end cards. Games like BF, CoD:AW, GTA 5 all run on these 270X and 760 range cards quite easily at near max settings, 1080p. I'm not just talking about typical source engine Dota 2, CS:GO games. 

This whole issue also becomes less clear cut because at the end of the day, you need to consider raw graphical settings, on top of the resolution. 

 

IMO, anything worse than a GTX 650 is low end. I'd put the consoles somewhere between a low mid-end experience as the PS4 has essentially a GPU of power between a 7850 and 7870. I do, however agree with you that it's pretty hard to state a mid and high end because there's so many different generations of GPUs, it's hard to really pinpoint. Anyway, I'd pretty much start a mid-end from around a 270 but it ranges all the way up to a 380/770/280. When we get into the 390's and whatnot, I believe them to be the higher range cards with Fury X's and 980 Ti's of the world being high range/enthusiast grade gaming cards. 

[Notional]

Finally moving away from the 28 nm node should be a huge improvement. It is an exciting time. Imagine if intel was still on 28 nm. EWWWW.

 

I know right? 22nm did seem to work just fine (as my Devil's Canyon), but 14nm just seems to be that much better in every term (although Intel is struggling with yields). But seeing a highend Intel chip at 65w TDP standard, is very impressive. I really think people need to rethink putting an AIO water cooler on the CPU, and do it on GPU's instead.

 

It is however impressive that AMD could still be competitive on 32/28nm in the low and midend segment. Those high density libraries are pretty impressive.

[Curufinwe_wins]

My issue is that I don't believe there are enough demanding games out there, which at 1080p make the 390/970 be seen as low-end cards. Games like BF, CoD:AW, GTA 5 all run on these 270X and 760 range cards quite easily at near max settings, 1080p. I'm not just talking about typical source engine Dota 2, CS:GO games. 

This whole issue also becomes less clear cut because at the end of the day, you need to consider raw graphical settings, on top of the resolution. 

 

IMO, anything worse than a GTX 650 is low end. I'd put the consoles somewhere between a low mid-end experience as the PS4 has essentially a GPU of power between a 7850 and 7870. I do, however agree with you that it's pretty hard to state a mid and high end because there's so many different generations of GPUs, it's hard to really pinpoint. Anyway, I'd pretty much start a mid-end from around a 270 but it ranges all the way up to a 380/770/280. When we get into the 390's and whatnot, I believe them to be the higher range cards with Fury X's and 980 Ti's of the world being high range/enthusiast grade gaming cards. 

I mean if that's the way you want to look at it, that's perfectly valid. I would say that most things lower than a 270/consoles are in a completely different category (unplayable), because based on my own standards they are unable to meet the 'lowest common denominator' standard (btw my laptop 660m is currently in that range even after the insanely high overclock I have on it. The newest game I can play on it is DA:I and that's on min settings at 30ps). So in that way perhaps, I share a similar setup to you.

[Curufinwe_wins]

I know right? 22nm did seem to work just fine (as my Devil's Canyon), but 14nm just seems to be that much better in every term (although Intel is struggling with yields). But seeing a highend Intel chip at 65w TDP standard, is very impressive. I really think people need to rethink putting an AIO water cooler on the CPU, and do it on GPU's instead.

 

It is however impressive that AMD could still be competitive on 32/28nm in the low and midend segment. Those high density libraries are pretty impressive.

I really think people unfairly wrote off Broadwell as well. The 5775C was beating/matching the 5960x in most games with both overclocked to 4.2. That's been a major reason for me I haven't gone x99. Hell at stock the 5775C was competing with the 4790k while being massively lower clock speed. I am so excited for a significantly more powerful 14nm cpu. Don't get me wrong, it wasn't a step-up in an absolute scale, but it is a major breakthrough from a future technology perspective.

 

As to AMD it is pretty impressive, but rather depressive as well that it has been relegated to that segment.