[WCCF] AMD Radeon R9 390X Pictured

[tential]

Is it really a foregone conclusion that gamers won't make an informed decision to pay well for an AMD GPU? The info is going to be out there and easy to find once these puppies are released. I paid pretty dearly for my Sapphire 290 when it was new, and it has stood the test of time fairly well.

Average consumers rarely make informed decisions.

 

[RussianSensation]

Anyway I am done arguing with you as you have a certain attitude about underestimating AMD and overestimating Nvidia. I remember you arguing for months together on these very same forums that AMD could not beat the GTX Titan in 2013. Heck you even said beating GTX 780 would be difficult. We saw how that turned out. I am predicting AMD will again prove you wrong.
:

It was hard to predict that Kepler would bomb that badly over time. At launch, 780Ti did beat 290X and overclocked 780Ti beat 290X OC easily. So really in the first half of that generational comparison, NV did beat AMD. How was I supposed to know that 780/780Ti would tank so hard in late 2014- Now?

As far as my estimates, they fall in the middle of this forum, between your 15-20% faster than Titan X and others who claim AMD will only improve performance 10-15% from 290X, with most of the benefits being power usage.

If your prediction goes through, that would be great but I think this time you really are overestimating it. I honestly don't see how 380X / 390X will beat 980 / Titan X by 15-20% as you keep saying. I guess 1 more months to wait.

 

[crashtech]

Average consumers rarely make informed decisions.

Perhaps it is wrong to assume that high-end GPU customers are "other than average," but I would hope most in that category would be more knowledgeable than average about GPUs, at least enough so to ask some questions before buying.

 

[tential]

Perhaps it is wrong to assume that high-end GPU customers are "other than average," but I would hope most in that category would be more knowledgeable than average about GPUs, at least enough so to ask some questions before buying.

You post here too much. I post on a lot of more casual forums. A lot of people spending 600-1000 in gpus literally know close to nothing about the gpu market. They make a decision based on what the fastest gpu is nvidia has at the time and so do the people advising them.
Take a look outside of hardware oriented forums and try posting a graph to substantiate and argument. No one will care. People here are recommending the r9 290x to users right now sometimes. The gtx 970 is the only card recommended I've seen outside of hardware forums on more casual forums. I've tried to help educate people, they don't care about facts, the nvidia cards are what people buy because everyone has them and they're perceived as better and you're a cheap lower when likes subpar performance for using amd. That's how it works on other places, sorry if you haven't seen it and aren't aware of it but people make very uninformed decisions all the time.

 

[Alatar]

Just gonna leave this here:

http://videocardz.com/55499/amds-hawaii-gpu-to-return-with-radeon-300-series

no proof obviously but article says most of the table is confirmed aside from some clocks which may vary.

pretty close to my estimate I guess. Fiji is still an unknown though.

 

[chimaxi83]

Brb, going to go make my own chart showing Fiji is in fact powered by nanobots from the future, powered by oxygen. I'll say that its true, that way people will know it's real.

 

[Glo.]

R9 380 is supposed to be Pitcairn, but... the chip that was in R9 270X is NOT the one from 380. 270X had 1280 GCN cores, 380 has 1024.

Also, 384 GB/s on Hawaii? Didn't Hawaii had much higher bandwith? Like 512 GB/s?

Im not saying this cannot be true, with like HBM memory, 3 or 6 GB pool would bring 384 GB/s at 1 GHz clock rate. 640GB/s is with 4/8 GB at 1.25 GHz.

 

[Cloudfire777]

Hawaii should be R9 380X.
Tonga should be R9 370X.

I`ve said that a long time now.
I dont think there are room for another Hawaii.

 

[jpiniero]

Just gonna leave this here:

http://videocardz.com/55499/amds-hawaii-gpu-to-return-with-radeon-300-series

no proof obviously but article says most of the table is confirmed aside from some clocks which may vary.

pretty close to my estimate I guess. Fiji is still an unknown though.

This chart is wrong because the 370 is a 265 rebrand and not a 270X and the 360's memory bandwidth is wrong. Putting 8 GB on Hawaii would only increase the already way too high power consumption too.

 

[Alatar]

R9 380 is supposed to be Pitcairn, but... the chip that was in R9 270X is NOT the one from 380. 270X had 1280 GCN cores, 380 has 1024.

Also, 384 GB/s on Hawaii? Didn't Hawaii had much higher bandwith? Like 512 GB/s?

Im not saying this cannot be true, with like HBM memory, 3 or 6 GB pool would bring 384 GB/s at 1 GHz clock rate. 640GB/s is with 4/8 GB at 1.25 GHz.

Hawaii had 320GB/s. 20% clock speed bump for memory brings it up to 384GB/s.

Also on the original article the Fiji memory speed was already fixed

 

[caswow]

Hawaii had 320GB/s. 20% clock speed bump for memory brings it up to 384GB/s.

Also on the original article the Fiji memory speed was already fixed

everyone should know that you cant simply oc hawaii without higher powerconsumption. they have to tweak something so i dont think its a simple rebrand.

 

[SimianR]

I'm not ruling out a R9 290/290x rebrand, but would it really make sense? The 290/290x's aren't selling - largely because they are seen as hot and power hungry. If you bump the clockspeed and double the VRAM as the chart shows it just makes the situation even worse. They aren't cheap cards to make either, so I'm not sure how much profit they would be making by rebranding them and potentially selling them for even less going forward.

The only reason why I wouldn't rule it out completely is the announcement of the OEM cards as rebrands and the fact that they rebranded the 7950 and 7970 into the 280/280x and those cards weren't dirt cheap to make either.

if that chart is even somewhat accurate, it is nice to see all the high end cards with 8GB minimum.

 

[Glo.]

Ok, then I messed up. What worries me is that 8 GB of 512 bit RAM, at 6k MHz would eat a LOT of energy.

But I don't believe they will go GDDR5 route again. HBM, with Tonga technology inside the core.

 

[JDG1980]

no proof obviously but article says most of the table is confirmed aside from some clocks which may vary.

pretty close to my estimate I guess. Fiji is still an unknown though.

Anyone can mock up a table and claim to have inside info. I just can't see AMD bringing back Hawaii with absolutely nothing changed except for clocks. It's not selling now, why would anyone think just giving it a new name would change that? While we don't know the exact production costs, they can't be making any kind of substantial profit at a $250-$320 retail price point, and they won't sell above that.

 

[Cloudfire777]

everyone should know that you cant simply oc hawaii without higher powerconsumption. they have to tweak something so i dont think its a simple rebrand.

28nm SHP and more dense design would be a way to increase clocks without higher power consumption
I assume Fiji got this to reduce TDP

From their internal slides:

 

[3DVagabond]

Your response tells me nothing about the errors in my calculation. AMD does not sell the cards/components to the market, but to AIBs. I have already accounted for AMD's margin. From the point the mark-up goes to AIBs and we get MSRP.

Also, it looks like you didn't look at historical GPU prices for AMD/NV. As others already pointed out, my calculation assumes horrible 40% yields too, no die harvesting for 2nd or 3rd tier 390 cards. Once those are taken into account, it's possible AMD could afford to finally manufacture a 500-550mm2 die.

The point of my analysis wasn't to get exact profit margins for AMD, but to have a ballpark idea if it's possible for them to afford a 500-550mm2 GPU. I think it is. If you disagree, provide an explanation instead of a "facepalm".

What he means by "margin is not markup" I markup is what you did. You add 35% to the cost. Margin is 35% of the selling price. To end up with 35% margin on something that cost $133 you would need to sell it for ~$205. $293 cost would be ~$450. And that's gross margin of course, not net.

If you divide the cost by .65 you get 35% margin. .60 = 40%, etc... Remember margin is the % of the selling price that was profit.

 

[goa604]

Just gonna leave this here:

http://videocardz.com/55499/amds-hawaii-gpu-to-return-with-radeon-300-series

no proof obviously but article says most of the table is confirmed aside from some clocks which may vary.

pretty close to my estimate I guess. Fiji is still an unknown though.

I would believe more if someone said that next generation of AMD cards is made by Nvidia than this silly silly graph.

 

[Valantar]

Perhaps it is wrong to assume that high-end GPU customers are "other than average," but I would hope most in that category would be more knowledgeable than average about GPUs, at least enough so to ask some questions before buying.

If that was true, the GPU market would swing back and forth according to who currently has the best GPUs. It doesn't. Even when AMD had had a CLEAR performance lead, nVidia has pretty much always had >50% of the market. Most consumers, regardless of what they're buying, go for known brands first, facts second.

 

[twjr]

Just because AMD may use a chip with the same paper specs as Hawaii that makes it a rebrand? If its on a new process at a different foundry and perhaps incorporates some of the improvements from Tonga is it really still a rebrand?

Also I don't recall there being such an outcry when AMD more-or-less rebranded the entire 7000 series below Hawaii. Maybe I have a short memory but I don't think it was this partisan 18 months ago.

 

[crashtech]

If that was true, the GPU market would swing back and forth according to who currently has the best GPUs. It doesn't. Even when AMD had had a CLEAR performance lead, nVidia has pretty much always had >50% of the market. Most consumers, regardless of what they're buying, go for known brands first, facts second.

Radeon is a well known brand. AMD fans are just stuck in a rut of playing the victim, which is like a self fulfilling prophesy. All the whining about how everyone unthinkingly picks Nvidia is NOT helping AMD sales. I doubt they appreciate being portrayed as victims.

 

[Azix]

Just because AMD may use a chip with the same paper specs as Hawaii that makes it a rebrand? If its on a new process at a different foundry and perhaps incorporates some of the improvements from Tonga is it really still a rebrand?

Also I don't recall there being such an outcry when AMD more-or-less rebranded the entire 7000 series below Hawaii. Maybe I have a short memory but I don't think it was this partisan 18 months ago.

My guess is that they are using "rebrand" when it should be "replaces". So pretty much everything is a rebrand if there was a card in its slot before.

Radeon is a well known brand. AMD fans are just stuck in a rut of playing the victim, which is like a self fulfilling prophesy. All the whining about how everyone unthinkingly picks Nvidia is NOT helping AMD sales. I doubt they appreciate being portrayed as victims.

They do unthinkingly pick nvidia. AMD needs better PR. heard some dumb reasons for going with nvidia.

 

[raghu78]

I would believe more if someone said that next generation of AMD cards is made by Nvidia than this silly silly graph.

well said. :thumbsup: wccftech is predicting with no actual information about the GPUs except R9 390X and R9 390.

It was hard to predict that Kepler would bomb that badly over time. At launch, 780Ti did beat 290X and overclocked 780Ti beat 290X OC easily. So really in the first half of that generational comparison, NV did beat AMD. How was I supposed to know that 780/780Ti would tank so hard in late 2014- Now?

As far as my estimates, they fall in the middle of this forum, between your 15-20% faster than Titan X and others who claim AMD will only improve performance 10-15% from 290X, with most of the benefits being power usage.

If your prediction goes through, that would be great but I think this time you really are overestimating it. I honestly don't see how 380X / 390X will beat 980 / Titan X by 15-20% as you keep saying. I guess 1 more months to wait.

I am specifically talking about the GTX Titan and the time before R9 290X launch. I vividly remember how you spent months arguing that beating Titan was impossible. I agree that 780 Ti started out faster. But as you saw the 4GB limitation started showing up quite soon with Watchdogs in mid-2014 being unable to run at Ultra texture settings on 780 Ti. Then after Maxwell launch we saw R9 290X close the gap and match 780 Ti as newer games performed better/on par with R9 290X.

I still say my expectation is R9 390X is 10% faster than GTX Titan. Anything higher I would be pleasantly surprised. It was silverforce who said he expects R9 390X would be 15% faster than Titan-X. Anyway I think its best that we wait and see how this plays out on June 24th.

 

[LTC8K6]

So we can buy an XFX Black Edition R9 285 and we actually have an R9 380?

 

[LTC8K6]

Just because AMD may use a chip with the same paper specs as Hawaii that makes it a rebrand? If its on a new process at a different foundry and perhaps incorporates some of the improvements from Tonga is it really still a rebrand?

Also I don't recall there being such an outcry when AMD more-or-less rebranded the entire 7000 series below Hawaii. Maybe I have a short memory but I don't think it was this partisan 18 months ago.

Rebranding gets tiresome.

People might accept it once or twice, with little grumbling, and then be really upset when it continues.

 

[JDG1980]

Just because AMD may use a chip with the same paper specs as Hawaii that makes it a rebrand? If its on a new process at a different foundry and perhaps incorporates some of the improvements from Tonga is it really still a rebrand?

I would argue that no, it is not. Once you start making changes in the chip that go beyond minor tweaking (for example, updating the shader model or the UVD block) then it's no longer a pure rebrand; it's a new chip.

Also I don't recall there being such an outcry when AMD more-or-less rebranded the entire 7000 series below Hawaii. Maybe I have a short memory but I don't think it was this partisan 18 months ago.

You said it yourself: 18 months ago. It's now been three full years since the release of the original GCN chips, and we're tired of seeing the same outdated silicon come back over and over again.

Back in October 2013, AMD's existing chips were reasonably competitive in terms of performance, perf/$, and perf/watt against Nvidia's lineup. Maxwell changed the game, and that's no longer the case. Ever since GM107 dropped in February 2014, we've been waiting for AMD's response. A full lineup of pure rebrands isn't going to cut it any more.