AMD Radeon Fiji XT Spotted in Shipping

[HalGameGuru]

They're the first concerns in newly built large data centers specifically because they're going to be around a LOOOONG time generating electrical costs. Now with dataceneters going modular, all the cooling is built into a storage container which houses a chunk of a datacenter with quick disconnects for piping, electricity, and such.

 

Also, yes, the GPUs are binned more, but it's the same cores on a different PCB (to support ECC) and slightly different drivers. Also, no one has a legitimate reason to use Hawaii over Kepler. It doesn't have even remotely comparable electrical or thermal performance, despite the raw compute, but if that's your argument, you may as well get a Xeon Phi for the same price and 2 TFlops more, rendering it the perf/BTU and perf/Watt king of accelerators.

 

I can't make any comment on ALL data centers, but a lot of the modular stuff I am seeing coming out of HP down here currently is bespoke for the most part. These newer centers are being designed from the floor up, they aren't using off the shelf, and if they source chips they will take care of their needs based on execution. I haven't seen a GPU in enterprise modules or trailers going out, but thats a very small cross section. But, I get the feeling, from these new designs I am seeing, that AMD's push for HSA and the new K12/Zen architecture is going to anchor their push into that modular future-space. Those HP modular pieces I have seen are screaming for a robust multi-facet APU. Not discrete compute.

 

But, once again that is a TINY cross section from one plant, and one specific new design for Data Centers. It's interesting stuff, if I can find the videos they were showing at their job fair for the data center assembly plant I'll share. It's interesting stuff, but all of it was straight up HP tech with custom housed chipsets.

[patrickjp93]

I can't make any comment on ALL data centers, but a lot of the modular stuff I am seeing coming out of HP down here currently is bespoke for the most part. These newer centers are being designed from the floor up, they aren't using off the shelf, and if they source chips they will take care of their needs based on execution. I haven't seen a GPU in enterprise modules or trailers going out, but thats a very small cross section. But, I get the feeling, from these new designs I am seeing, that AMD's push for HSA and the new K12/Zen architecture is going to anchor their push into that modular future-space. Those HP modular pieces I have seen are screaming for a robust multi-facet APU. Not discrete compute.

 

But, once again that is a TINY cross section from one plant, and one specific new design for Data Centers. It's interesting stuff, if I can find the videos they were showing at their job fair for the data center assembly plant I'll share. It's interesting stuff, but all of it was straight up HP tech with custom housed chipsets.

You should see Google's, Amazon's, and Facebook's videos regarding their datacenter designs. It's amazing what's being done now.

[MEC-777]

Look at the engineering and benchmark leaks of the Titan 2. AMD is fried. 3072 cores at 1.4GHz with 12GB of VRAM on a 384-bit bus delivering 8.84TFlops. Also, we know for a fact it's liquid cooled. Asetek doesn't design air coolers.

Also, I've been studying heterogeneous computing all this semester and optimizing game physics with OpenCL. Almost no game will benefit from HBM for the next 4 years. AMD's card prices will go up while Nvidia's have fallen and will still have to rely on thermal solutions most consumers cannot tolerate.

Also, the performance gain of the 970 and 980 is huge. We don't have the software to properly bench them. In OpenCL tasks it's a 30% jump over the Titan Black, both cards having been thoroughly tested by our class and our nerdy professor. It's like Intel providing so much computational power despite software not taking advantage, making AMD look good when it shouldn't at only 2 ALUs per core and a 4 cycle disadvantage against Haswell for floating point mathematics.

Also, AMD has not really made their chips more efficient architecturally. The 8370s are just highly-binned 8350s with some microcode revisions for lower idle clocks and snappier movement to idle. Their GPUs haven't gotten more efficient either. Simply see the 295x2.

 

I understand you have more GPU/CPU knowledge than I do, so let me just posit this to you from a different perspective. 

 

I'm a gamer and PC enthusiast. The #1 thing I look at is gaming performance. Frame rates, frame timing etc. Other features and aspects like power efficiency, noise and heat are also of high importance, but within reason. Yeah, my 290 runs hotter and pulls more power than the 970 I could have bought, but it still runs cool and quiet enough to make it well worth it (to me) to spend $100 less than what a 970 would have cost me. I'm a PC gamer and PC enthusiast. I'm not so much interested in compute capabilities. That being said, looking at pure gaming performance, the 970/980 were not THAT much of a jump over the 290/X and 780/Ti. That's the way I'm looking at it. These are gaming cards and should be judged/compared within that context.  

 

Yes, AMD have been making their chips more efficient, architecturally. Look at Kaveri vs. Richland. Not earth-shattering, but Kaveri is more efficient and that was one of their main goals with that chip. My point with this is that AMD is making an effort, and you can bet they're very much aware of Maxwell's efficiency and will make an effort to address that. Also, look at Tonga (R9-285) quite a bit more efficient vs. the 280 it replaces and also uses a more narrow bus of 256-bit along side other architectural optimizations just like Maxwell. 

[HalGameGuru]

You should see Goodle's, Amazon's, and Facebook's videos regarding their datacenter designs. It's amazing what's being done now.

 

I don't doubt it, I've just seen a larger push for bespoke design and smaller chunks of capacity stacked rather than big units. I mean these new ones from HP are gorgeous. If so much of it wasn't so far beyond my practical experience I would want to jump in feet first, I just don't like devoting that much time to things I try to justify as work but are really just technological masturbation. I luvs me some sexy tech infrastructure

[patrickjp93]

I understand you have more GPU/CPU knowledge than I do, so let me just posit this to you from a different perspective. 

 

I'm a gamer and PC enthusiast. The #1 thing I look at is gaming performance. Frame rates, frame timing etc. Other features and aspects like power efficiency, noise and heat are also of high importance, but within reason. Yeah, my 290 runs hotter and pulls more power than the 970 I could have bought, but it still runs cool and quiet enough to make it well worth it (to me) to spend $100 less than what a 970 would have cost me. I'm a PC gamer and PC enthusiast. I'm not so much interested in compute capabilities. That being said, looking at pure gaming performance, the 970/980 were not THAT much of a jump over the 290/X and 780/Ti. That's the way I'm looking at it. These are gaming cards and should be judged/compared within that context.  

 

Yes, AMD have been making their chips more efficient, architecturally. Look at Kaveri vs. Richland. Not earth-shattering, but Kaveri is more efficient and that was one of their main goals with that chip. My point with this is that AMD is making an effort, and you can bet they're very much aware of Maxwell's efficiency and will make an effort to address that. Also, look at Tonga (R9-285) quite a bit more efficient vs. the 280 it replaces and also uses a more narrow bus of 256-bit along side other architectural optimizations just like Maxwell. 

You have between 1 and 4 GPUs in your system. supercomputers have hundreds to tens of thousands. Those extra 35 watts per chip add up, and then it's minimum 1.28 * 35 x card count in electricity to keep the systems cool. These are also 24-hour running systems which go on running for 8+ years. By comparison, you're a flea on an elephant.

 

The 980 is not much better in games currently, but you have to look at what the card actually is. It's 4GB of global memory, a 12x increase in local memory per work group, a shrinking of the core count (decreased granularity for thread management), and a butt ton of cache per core with a stupid-high clock rate. If you build something optimized for that, the 780TI cannot possibly keep up. That 5-6 TFlop rating isn't for show. In scientific computing, I'd choose the 980 over a Titan Black no contest.

 

Mind you the 980 and 970 are this good on release drivers and without software updates to optimize for preferred working group and block sizes (every accelerator is optimized for certain block sizes). Give it a couple months and we'll revisit gaming performance. In scientific computing (the matter of discussion) the 980 rips the 290x apart.

 

Kaveri is efficient? It can't maintain a 1GHz GPU clock rate whereas Intel can maintain 1.3GHz on their top SKU, and that's a 95-Watt AMD part against a 65-watt Intel part.

 

Tonga also dropped 1GB of VRAM. I call shenanigans on calling that actually efficient.

[patrickjp93]

I don't doubt it, I've just seen a larger push for bespoke design and smaller chunks of capacity stacked rather than big units. I mean these new ones from HP are gorgeous. If so much of it wasn't so far beyond my practical experience I would want to jump in feet first, I just don't like devoting that much time to things I try to justify as work but are really just technological masturbation. I luvs me some sexy tech infrastructure

Oh yeah, horizontal scaling is taking off, but the factor of your horizontal scaling depends on the vertical scale of each unit.

[worldwin]

You have between 1 and 4 GPUs in your system. supercomputers have hundreds to tens of thousands. Those extra 35 watts per chip add up, and then it's minimum 1.28 * 35 x card count in electricity to keep the systems cool. These are also 24-hour running systems which go on running for 8+ years. By comparison, you're a flea on an elephant.

 

The 980 is not much better in games currently, but you have to look at what the card actually is. It's 4GB of global memory, a 12x increase in local memory per work group, a shrinking of the core count (decreased granularity for thread management), and a butt ton of cache per core with a stupid-high clock rate. If you build something optimized for that, the 780TI cannot possibly keep up. That 5-6 TFlop rating isn't for show. In scientific computing, I'd choose the 980 over a Titan Black no contest.

 

Mind you the 980 and 970 are this good on release drivers and without software updates to optimize for preferred working group and block sizes (every accelerator is optimized for certain block sizes). Give it a couple months and we'll revisit gaming performance. In scientific computing (the matter of discussion) the 980 rips the 290x apart.

 

Kaveri is efficient? It can't maintain a 1GHz GPU clock rate whereas Intel can maintain 1.3GHz on their top SKU, and that's a 95-Watt AMD part against a 65-watt Intel part.

 

Tonga also dropped 1GB of VRAM. I call shenanigans on calling that actually efficient.

The person you quoted never discussed servers so the point is moot. From a gamer perspective (which is what MEC-777 was coming from) that 35W is not that big. If you are willing to spend a few hundred dollars on GPU(s) then you should be able to afford the power required to run it. Also Kaveri is more efficient compared to Richland which is what it was being compared to. Don't take them out of context. "Look at Kaveri vs. Richland."~MEC-777

[patrickjp93]

The person you quoted never discussed servers so the point is moot. From a gamer perspective (which is what MEC-777 was coming from) that 35W is not that big. If you are willing to spend a few hundred dollars on GPU(s) then you should be able to afford the power required to run it. Also Kaveri is more efficient compared to Richland which is what it was being compared to. Don't take them out of context. "Look at Kaveri vs. Richland."~MEC-777

The person I quoted, who quoted me in a conversation about AMD's financial future depending on breaking into the HPC market, will find the information relevant. 

 

I didn't take it out of context. AMD has huge chasms it needs to close regarding efficiency. Richland to Kaveri isn't that much of an improvement. It's increasing throughput without touching the power and heat envelopes. That's not good enough.

 

As per costs, that's easy for you to say from a gaming perspective. Try being a CFO in charge of a datacenter. You're talking hundreds of thousands of dollars per month in electricity. Watts matter.

[mr moose]

Look at just the top 500 supercomputers list, pull down the specs from each, multiply card counts by their current prices (lower than original buying prices in their respective years, going back to 2007 at the earliest). Compare that sum of cards*respective prices against 2 years of consumer GPU purchases from AMD and Nvidia which you can find pretty trivially (not perfect estimate, but we all know there are many smaller clusters and supercomputers built each year by the hundreds and thousands). 

 

It's about 15794:1.

 

It appears only 62 of the top 500 use a GPU/co processor.  At any rate, if we take the 2nd biggest which has 19,000 Nvidia K20's in it, and multiply that by the  62 computers that use accelerators and we have a total of 1.178M gpus. Which is a very long shot from the 1600:1 you claimed given that according john peddy 438Million GPUs where shipped in quater2 alone.

 

 

A total of 62 systems on the list are using accelerator/co-processor technology, up from 53 from November 2013. Forty-four of these use NVIDIA chips, two use ATI Radeon, and there are now 17 systems with Intel MIC technology (Xeon Phi). The average number of accelerator cores for these 62 systems is 78,127 cores/system.

 

 

Also interesting to note that of the 62 systems 2 are running AMD GPU's and the second largest super computer is running AMD CPU's, Which by my calculation is 3.2% of the total accelerated supercomputer market not including the CPU's.

 

http://www.top500.org/lists/2014/06/

[patrickjp93]

It appears only 62 of the top 500 use a GPU/co processor At any rate, if we take the 2nd biggest which has 19,000 Nvidia K20's in it, and multiply that by the  62 computers that use accelerators and we have a total of 1.178M gpus. Which is a very long shot from the 1600:1 you claimed given that according john peddy 438Million GPUs where shipped in quater2 alone.

 

 

Also interesting to note that of the 62 systems 2 are running AMD GPU's and the second largest super computer is running AMD CPU's, Which by my calculation is 3.2% of the total accelerated supercomputer market not including the CPU's.

 

http://www.top500.org/lists/2014/06/

19,000 x 62, x $5900 per GPU vs consumers where GPU prices top out at 1100 up until the most recent Titan Z and 295x2. Please try to keep up.

[patrickjp93]

It appears only 62 of the top 500 use a GPU/co processor.  At any rate, if we take the 2nd biggest which has 19,000 Nvidia K20's in it, and multiply that by the  62 computers that use accelerators and we have a total of 1.178M gpus. Which is a very long shot from the 1600:1 you claimed given that according john peddy 438Million GPUs where shipped in quater2 alone.

 

 

Also interesting to note that of the 62 systems 2 are running AMD GPU's and the second largest super computer is running AMD CPU's, Which by my calculation is 3.2% of the total accelerated supercomputer market not including the CPU's.

 

http://www.top500.org/lists/2014/06/

These top 500 are not the whole market, but they do use clusters (modules basically) which are building blocks for a number of smaller supercomputers. 

 

Yes, AMD Opterons are used in a few of the top supercomputers. Congrats to them for having a product cheap enough for the Department of Energy (arm cores, so duh). Also, AMD's FirePros are not at all popular in the mid-range and low-end, because the heat density is that bad. Frankly I wouldn't be surprised if the Mac Pros use more FirePros than a few HPC vendors. And again, number of computers using vs. total amount used... AMD has less than 1% of the market in accelerators last I heard in market analytics.

[mr moose]

19,000 x 62, x $5900 per GPU vs consumers where GPU prices top out at 1100 up until the most recent Titan Z and 295x2. Please try to keep up.

 

So do the math then, 19,000 x 62 x $5900 = $6.95B

 

Even if they only shipped 20M cards (which coincidentally is the total number of NG consoles sold) x 500 (average price) =  $10B

 

And you're telling me to try and keep up? 

 

I've got an Idea, actually do some research before you form an opinion let alone tell everyone about it.

[worldwin]

The person I quoted, who quoted me in a conversation about AMD's financial future depending on breaking into the HPC market, will find the information relevant. 

 

I didn't take it out of context. AMD has huge chasms it needs to close regarding efficiency. Richland to Kaveri isn't that much of an improvement. It's increasing throughput without touching the power and heat envelopes. That's not good enough.

 

As per costs, that's easy for you to say from a gaming perspective. Try being a CFO in charge of a datacenter. You're talking hundreds of thousands of dollars per month in electricity. Watts matter.

Like I said before, the person you quoted (MEC-777) was going from a gamers perspective. Energy consumption isn't a massive concern for PC gamers except when considering whether the power supply is sufficient. If you wish to discuss differing topics with different people quote them individually. What you did was quote one of them and discussed the topic for both people bringing irrelevant topics to the uninvolved party. Also Kaveri is an improvement. Once again you are ignoring the arguement. The arguement was that AMD is making more efficient chips, which you agreed "Richland to Kaveri isn't that much of an improvement." MEC-777 was not comparing AMD to intel,(which you brought in from nowhere with "AMD has huge chasms it needs to close regarding efficiency") rather AMD to AMD. You should focus on what the argument is.

[MEC-777]

You have between 1 and 4 GPUs in your system. supercomputers have hundreds to tens of thousands. Those extra 35 watts per chip add up, and then it's minimum 1.28 * 35 x card count in electricity to keep the systems cool. These are also 24-hour running systems which go on running for 8+ years. By comparison, you're a flea on an elephant.

 

The 980 is not much better in games currently, but you have to look at what the card actually is. It's 4GB of global memory, a 12x increase in local memory per work group, a shrinking of the core count (decreased granularity for thread management), and a butt ton of cache per core with a stupid-high clock rate. If you build something optimized for that, the 780TI cannot possibly keep up. That 5-6 TFlop rating isn't for show. In scientific computing, I'd choose the 980 over a Titan Black no contest.

 

Mind you the 980 and 970 are this good on release drivers and without software updates to optimize for preferred working group and block sizes (every accelerator is optimized for certain block sizes). Give it a couple months and we'll revisit gaming performance. In scientific computing (the matter of discussion) the 980 rips the 290x apart.

 

Kaveri is efficient? It can't maintain a 1GHz GPU clock rate whereas Intel can maintain 1.3GHz on their top SKU, and that's a 95-Watt AMD part against a 65-watt Intel part.

 

Tonga also dropped 1GB of VRAM. I call shenanigans on calling that actually efficient.

 

I just told you how I'm looking at and comparing these GPUs - in the context of a PC enthusiast from a gaming-centric perspective. Not scientific computing. I don't care about, nor am I commenting on that, as I made that clear in my last post. So why you continue to talk about that in response to my post, I don't know. It's irrelevant to me. 

 

Will the 970/980 improve in gaming performance as newer drivers are released? Of course. So did the 290/X and so will the next AMD and Nvidia GPUs in the next generations to come. I don't deny that. 

 

As for AMD improving efficiency in their CPUs and GPUs, you said they haven't been doing so, which is completely false. I never said "Kaveri is efficient", I said; Kaveri is more efficient vs. Richland. Also, as I said, it's not earth-shattering, but it IS an improvement over it's replacement. The A8-7600 is a particular one to mention as it delivers respectable gaming performance even at just 45w that is much stronger than anything from the previous Richland line up. That's an architectural improvement in efficiency. With Tonga, the GPU itself is quite a bit more efficient over the 280 (7950) chip it replaces and the overall system architecture on that card is also more efficient - delivering equal to or better than the performance of the 280 (7950) which had more Vram and a wider 384 bit bus. You don't necessarily need more Vram nor a wider bus if it's being utilized more efficiently. Once again, that is also an architectural improvement in efficiency. So you cannot say AMD has or is not improving efficiency in their chips architecturally. 

 

The person I quoted, who quoted me in a conversation about AMD's financial future depending on breaking into the HPC market, will find the information relevant. 

 

I didn't take it out of context. AMD has huge chasms it needs to close regarding efficiency. Richland to Kaveri isn't that much of an improvement. It's increasing throughput without touching the power and heat envelopes. That's not good enough.

 

As per costs, that's easy for you to say from a gaming perspective. Try being a CFO in charge of a datacenter. You're talking hundreds of thousands of dollars per month in electricity. Watts matter.

 

Once again, I stated very clearly what my perspective is and that that is what I was commenting on, not the rest. 

 

This isn't the first time you've taken things out of context in a discussion. You need to read more carefully what people are saying. That's all fine and dandy if you're a CFO in charge of a datacenter. But you have to realize, we're talking about gaming GPUs here, which the majority of people on this forum will use for gaming and care mostly about how well they run games, not how good they are at compute or other non-gaming stuff. 

[SpaghettiCarbonara]

Cool news for the first post, Video cardz is a questionable source almost as same as wccf but the source seems real to me...

 

Keep up the good posts  B)

You did EVERYTHING WRONG! I can't believe you posted such trash!

Just kidding! :lol:

You did great mate!

 

Yeeeyyy! Look mum, I'm a reporter! I did guud

If I said I wasn't surprised at how much this thread blew up, I'd be lying. I expected a handful of replies with a lot of "this isn't news!"

[Jerakl]

Well winter is cettainly coming.

Should tagged me :L

[patrickjp93]

I just told you how I'm looking at and comparing these GPUs - in the context of a PC enthusiast from a gaming-centric perspective. Not scientific computing. I don't care about, nor am I commenting on that, as I made that clear in my last post. So why you continue to talk about that in response to my post, I don't know. It's irrelevant to me.

Will the 970/980 improve in gaming performance as newer drivers are released? Of course. So did the 290/X and so will the next AMD and Nvidia GPUs in the next generations to come. I don't deny that.

As for AMD improving efficiency in their CPUs and GPUs, you said they haven't been doing so, which is completely false. I never said "Kaveri is efficient", I said; Kaveri is more efficient vs. Richland. Also, as I said, it's not earth-shattering, but it IS an improvement over it's replacement. The A8-7600 is a particular one to mention as it delivers respectable gaming performance even at just 45w that is much stronger than anything from the previous Richland line up. That's an architectural improvement in efficiency. With Tonga, the GPU itself is quite a bit more efficient over the 280 (7950) chip it replaces and the overall system architecture on that card is also more efficient - delivering equal to or better than the performance of the 280 (7950) which had more Vram and a wider 384 bit bus. You don't necessarily need more Vram nor a wider bus if it's being utilized more efficiently. Once again, that is also an architectural improvement in efficiency. So you cannot say AMD has or is not improving efficiency in their chips architecturally.

Once again, I stated very clearly what my perspective is and that that is what I was commenting on, not the rest.

This isn't the first time you've taken things out of context in a discussion. You need to read more carefully what people are saying. That's all fine and dandy if you're a CFO in charge of a datacenter. But you have to realize, we're talking about gaming GPUs here, which the majority of people on this forum will use for gaming and care mostly about how well they run games, not how good they are at compute or other non-gaming stuff.

Sigh. My point I guess is AMD really shouldn't be caring so much about gamers. There's just not enough money in the gaming community to keep AMD afloat. AMD not focusing on larger markets will eventually make it impossible for them to bring innovations to the gaming market. Without going the route Intel and Nvidia went, AMD can't compete long-term. HSA is a stopgap attempt given Intel is fully caught up to Kaveri, so I just don't understand how so many on here can't see the bigger picture. If you want better gaming performance, the last thing you should be doing is demanding more power from AMD. That's a losing atrategy.

Also, be aware you started by quoting an enterprise-centered post. I thought the idea was that somehow you expected to have AMD cater to you and your needs when it's getting eaten alive for the reasons I laid out.

Also, the only difference between a gaming card and a workstation/supercomputer one is binning, non-disabled ECC support, and a tweaked driver for double precision. Anything which applies to a tesla applies to a gtx architecturally speaking.

[Dive5885]

I have a 290x from Powercolor.

http://media.bestofmicro.com/J/B/418007/original/powercolor_pcs-r9-290x.jpg,

So I'm not about to need one, but I have to say I feel the coming AMD cars will be worth their while.

I'm waiting till the 490x to upgrade. Second gen HBM and 16nm will make those a doozy.

 

Hope you don't mind me asking, How's that Powercolor 290X treating you?

 

I might switch from my 3-way to a 290X and that card caught my eye!

[LukaP]

@patrickjp93 dont respond to this as i wont even get notifications, and dont plan on looking at it...but dont you think that when there are many respected forum members (some of which study the stuff you spit your BS about) taht disagree with you and provide you facts and points that you just keep on ignoring, that maybe JUST FUCKING MAYBE it would be nice to stop being a dick and ruining every GPU related thread you go in, just so you can feel smart because you know a few words, some of them not even used in the correct context.

 

peace, bye

[SpaghettiCarbonara]

@patrickjp93 dont respond to this as i wont even get notifications, and dont plan on looking at it...but dont you think that when there are many respected forum members (some of which study the stuff you spit your BS about) taht disagree with you and provide you facts and points that you just keep on ignoring, that maybe JUST FUCKING MAYBE it would be nice to stop being a dick and ruining every GPU related thread you go in, just so you can feel smart because you know a few words, some of them not even used in the correct context.

 

peace, bye

Patrick may not respond... BUT I WILL

[LukaP]

Patrick may not respond... BUT I WILL

 

post not really understood

[MEC-777]

Sigh. My point I guess is AMD really shouldn't be caring so much about gamers. There's just not enough money in the gaming community to keep AMD afloat. AMD not focusing on larger markets will eventually make it impossible for them to bring innovations to the gaming market. Without going the route Intel and Nvidia went, AMD can't compete long-term. HSA is a stopgap attempt given Intel is fully caught up to Kaveri, so I just don't understand how so many on here can't see the bigger picture. If you want better gaming performance, the last thing you should be doing is demanding more power from AMD. That's a losing atrategy.

Also, be aware you started by quoting an enterprise-centered post. I thought the idea was that somehow you expected to have AMD cater to you and your needs when it's getting eaten alive for the reasons I laid out.

Also, the only difference between a gaming card and a workstation/supercomputer one is binning, non-disabled ECC support, and a tweaked driver for double precision. Anything which applies to a tesla applies to a gtx architecturally speaking.

 

This is a PC enthusiast community, full of PC gamers and this is a thread about gaming GPUs. We're talking about gaming GPUs, not AMD's business structure and non-gaming GPU performance. We know from the Titan that compute capability doesn't translate to gaming prowess.

 

Really? AMD shouldn't care about gamers? So would you also agree that Nvidia shouldn't care about gamers either? Because as I recall, Nvidia has just posted tremendous profits as a direct result of the 970/980 release, that is, from the sales of gaming GPUs to [mostly] gamers. If that's any indication, AMD should certainly care about gamers.

 

I quoted your post and then specifically stated in the first few lines that I don't care about the non-gaming aspect and then clearly stated the perspective I was commenting from and why. I don't know where you're coming up with this "cater to your needs" stuff...

 

Thank you for pointing out the differences between a gaming and workstation card. Workstation cards are also vastly more expensive and are of very little to no interest to gamers. I don't really understand why you mentioned that because it's not really relevant to gaming performance.

 

 

Take a look at this: http://www.anandtech.com/bench/product/1068?vs=1060

 

See how the $299 290 slaps around the ~$1000 Titan in gaming performance? Gamers aren't going to buy the Titan just because it's "better at compute". They're going to buy the 290 (or a 780/970) or any GPU that costs less and gives them far better performance per dollar. Titan's aren't flying off the shelves. 290's and 970's are. They make up a significant portion of profit for those companies and they are far more interesting products for people like me vs. compute/workstation cards. This is my whole point and the perspective that the majority of the members here are looking at this from.

 

Just like before when you compared a much more expensive Intel APU against Kaveri (talking $100-200+ difference), you're doing the same thing here, pulling things out of context.

[LukaP]

-snp-

i would recommend you stop arguing an idiot... 

 

ive tried it many times, this guy... you wont even begin to change his opinion he is that self entitled

[MEC-777]

i would recommend you stop arguing an idiot... 

 

ive tried it many times, this guy... you wont even begin to change his opinion he is that self entitled

 

Yeah, he and I have gone at it before. I can't help but set the record straight when someone tries to skew an argument. It's in my nature.

 

Typically, if I put in the effort to argue about something, it's not so much to change the other person's opinion (rarely ever happens), but more so for others reading this, so they won't be confused or come away with the wrong idea/information.

[LukaP]

Yeah, he and I have gone at it before. I can't help but set the record straight when someone tries to skew an argument. It's in my nature.

 

Typically, if I put in the effort to argue about something, it's not so much to change the other person's opinion (rarely ever happens), but more so for others reading this, so they won't be confused or come away with the wrong idea/information.

i just post wherever i see him spewing his bullshit, warning others about it. its sad because he has enough posts to seem credible to newcomers when in reality, he knows jack shit apart from a few words (some of them he picked up from me arguing with him) which he then uses in complete wrong context (eg i was saying intel iGPU is crap at games no matter how good it is at theoretical compute because it lacks dedicated hardware for certain tasks, i mentioned nvidias polymorph engine somewhere in the middle, and from then on he kept on saying intel just needs a polymorph engine to beat every gpu on the market) -.-