Official Nvidia Maxwell slides

[GoldenTiger]

The gt 640 has zip to do with the slide I posted....

Third slide in the OP, man...

The slide I posted. As in me. Not the OP. This one:

No, you are extrapolating performance with those undefined slides.

135% performance per core = 35% increase in SMX performance.

2x performance per watt = undefined. At what level? Shader? Core/SMX? ASIC? Architecture?

Sigh... nvidia is saying the numbers in plain English: 35% increase in performance per core is pretty plain, per cuda core. 2x performance per watt means, obviously, twice the performance per watt used of the GPU. You're reading way too deeply into it for things that they haven't said, nor I. Hence why I said "2.7x" per nvidia as a base factor. I would expect no less than double the performance of the 780, from the GTX 880 or whatever the Big Maxwell is, with these two factors known currently, and possibly more. Add in that this is first-gen Maxwell and GM200/204 will be "second generation" Maxwell, and things are looking rosy.

 

[xoleras]

Just a funny side observation. When any NV user posts on the AMD forum anything that isn't completely and utterly 100% positive and praising the feet of AMD, AMD users generally complain about it. As if we can't voice our opinions. How dare we say anything that isn't completely positive about AMD.

And here we are. People downplaying the Maxwell. Personally, I take it all in stride. It's a good discussion. Don't really care if people want to be negative about Maxwell, it's whatever. Like I said. Just a funny observation - lots of AMD fans seem to have an emotional complex when anything that isn't completely 100% positive is stated about AMD. Even just criticism. Such as their alpha ++ shit the bed Mantle driver. But here we are. I can find tons of negative posts in all of the g-sync, maxwell, and physx threads, quite the double standard. Can't say anything non positive about AMD.

Anyway, that tangent aside, as far as when GM200 or GM204 will be released? Well, HPC is a super profitable segment for nvidia and I would think that they would release it as soon as fucking possible - this is big money for NV. They will release big Maxwell as soon as 20nm is ready, and rumors seem to indicate that volume production is ramping up right now. Now I don't know a precise date, no one here does. But I suspect it will be the earlier part of 2H 2014. As soon as 20nm is ready, NV will *want* Maxwell to hit shelves ASAP in some form or the other, this is in their best interest. Like GT said, this is all just speculation on my part, obviously no one here knows. But like I said, HPC is a profitable segment for NV and they will want to get it out ASAP. They do not want to just simply sit on it.

 

[evilsofa]

The slide I posted. As in me. Not the OP. This one:

What I posted had nothing to do with your slide. It's not all about you.

 

[GoldenTiger]

Just a funny side observation. When any NV user posts on the AMD forum anything that isn't completely and utterly 100% positive and praising the feet of AMD, AMD users generally complain about it. As if we can't voice our opinions. How dare we say anything that isn't completely positive about AMD.

And here we are. People downplaying the Maxwell. Personally, I take it all in stride. It's a good discussion. Don't really care if people want to be negative about Maxwell, it's whatever. Like I said. Just a funny observation - lots of AMD fans seem to have an emotional complex when anything that isn't completely 100% positive is stated about AMD. Even just criticism. Such as their alpha ++ shit the bed Mantle driver. But here we are. I can find tons of negative posts in all of the g-sync, maxwell, and physx threads, quite the double standard. Can't say anything non positive about AMD.

Anyway, that tangent aside, as far as when GM200 or GM204 will be released? Well, HPC is a super profitable segment for nvidia and I would think that they would release it as soon as fucking possible - this is big money for NV. They will release big Maxwell as soon as 20nm is ready, and rumors seem to indicate that volume production is ramping up right now. Now I don't know a precise date, no one here does. But I suspect it will be the earlier part of 2H 2014. As soon as 20nm is ready, NV will *want* Maxwell to hit shelves ASAP in some form or the other, this is in their best interest. Like GT said, this is all just speculation on my part, obviously no one here knows. But like I said, HPC is a profitable segment for NV and they will want to get it out ASAP. They do not want to just simply sit on it.

A++++++++ post, would read again, and indeed it is pretty funny to see. Don't forget ShadowPlay being deemed "worthless" by them too! As you said it's speculation about launch dates and all that right now for the most part but it's a good discussion.

What I posted had nothing to do with your slide. It's not all about you.

Yes, and I was replying to you to tell you that we have a better measure of what it actually means, with the slide I then posted. I guess things need to be spelled out completely since inference is dead it would seem. (insert snippy sarcastic roll eyes smiley here with some random insulting quip like you did on the end)

 

[Deleted whining member 223597]

Just a funny side observation. When any NV user posts on the AMD forum anything that isn't completely and utterly 100% positive and praising the feet of AMD, AMD users generally complain about it. As if we can't voice our opinions. How dare we say anything that isn't completely positive about AMD.

And here we are. People downplaying the Maxwell. Personally, I take it all in stride. It's a good discussion. Don't really care if people want to be negative about Maxwell, it's whatever. Like I said. Just a funny observation - lots of AMD fans seem to have an emotional complex when anything that isn't completely 100% positive is stated about AMD. Even just criticism. Such as their alpha ++ shit the bed Mantle driver. But here we are. I can find tons of negative posts in all of the g-sync, maxwell, and physx threads, quite the double standard. Can't say anything non positive about AMD.

Seriously, this must be a joke. I didn't see much trolling in either Nvidia/AMD section until a few months ago, and most of it started with Mantle. Your own posts were quite irrational at times. The only thing we are saying is to not get too excited about Nvidia marketing, because it's always overblown same as AMD marketing when it comes to benchmarks. Stick with reasonable expectations and you won't be disappointed. If you think there is going to be a 170% increase in performance at the top end, prepare to be disappointed.

 

[xoleras]

Well, I admit, you're definitely correct. PC DIY stuff has been my hobby forever and sometimes I get a little wound up about it, but I'm trying NOW to not be that guy. Those guys, of which I have definitely been a part of - are around on both sides, but I have noted that a lot of AMD users get super offended even at reasonable criticism. I'm not saying i'm perfect here. Far from it. You are correct, that i've been a little over the top at times - but I will say, that I've owned AMD for a long time and a lot of that was due to very real frustrations I've had with their software. In fact in the past 7 years that i've easily owned AMD more often than NV. That is off topic and I won't discuss that though, yet I will say that AMD improved a lot. My main point is that a lot of AMD users just get SUPER SUPER offended at anything that isn't completely and utterly positive, and I just don't really understand that, you know. Anyone can say whatever they want about NV and I think it's cool. It's whatever.

Maybe i'm wrong on some users getting super offended. Maybe some NV guys are like that. But let's just move on - You are correct with regard to marketing. It is my opinion that AMD is a little more over the top in this respect, with some pretty "out there" claims on their slides about Kaveri, bulldozer, Mantle, etc. But I do think Maxwell GM107 should be a great indication of how well Maxwell scales once GM200 hits. That's all. I'm interested in seeing the perf/watt of GM107 as this will definitely have a correlation on how well GM200/GM204 does. We will see, I guess, once benchmarks hit - marketing can lie, but objective benchmarks generally cannot.

 

[Deleted whining member 223597]

I think both sides have gotten really bad lately, neither is better than the other. Both fling shit at each other

I am interested in Maxwell because I do want to go back to Nvidia, but I just don't want to have my expectations be too high and be disappointed. Let's see how this 750TI performs(Thought I saw a post where it says it's going on sale on the 22nd) and try to judge if the new improvements are gonna be great or just small gains like the past few cards.

 

[Spazturtle]

Some of you seam confused:

That means a 60w chip will be 35% faster then a 60w chip from last gen.

But to get the same performance as a last gen 60w chip you only need a 30w chip this time.

 

[xoleras]

Uhm, the 60W TDP 750ti outperformed the 65W TDP GT640 by a factor of two. Now that's a marketing slide, but it seems that the performance per watt has doubled if the slide is true. The 3dmark and firestrike scores over at videocardz/wccftech seem to collaborate this as well - performance in the same TDP window jumped dramatically with Maxwell. But these are leaks of course. We don't know if they have a complete basis in reality, I admit this.

I suspect it will be around that or maybe slightly lower depending on various factors such as resolution, etc. But if the slide is reasonably close to reality, the performance slide specifically - then it does seem to be better than a 35% perf/watt jump. Also, the SMX cores were completely overhauled for Maxwell; you can't do a direct per core comparison from what I understand. Anyway, we'll see in 3 days, or later this week how close to reality that claim is. If it's double the performance at a 5W lower TDP, then that's pretty impressive.

I admit i'm being somewhat optimistic, and I may be wrong, but the leaked performance benchmarks i've seen for the 750ti are downright great for a 60W TDP.

 

[{NG}Fidel]

Just ignore AMD and Nvidia Marketing benchmarks. Not saying this isn't faster or an improvement, just saying of course the two companies are going to cherry pick benchmarks that make things shine.

 

[LordEC911]

Uhm, the 60W TDP 750ti outperformed the 65W TDP GT640 by a factor of two. Now that's a marketing slide, but it seems that the performance per watt has doubled if the slide is true. The 3dmark and firestrike scores over at videocardz/wccftech seem to collaborate this as well - performance in the same TDP window jumped dramatically with Maxwell. But these are leaks of course. We don't know if they have a complete basis in reality, I admit this.

I suspect it will be around that or maybe slightly lower depending on various factors such as resolution, etc. But if the slide is reasonably close to reality, the performance slide specifically - then it does seem to be better than a 35% perf/watt jump. Also, the SMX cores were completely overhauled for Maxwell; you can't do a direct per core comparison from what I understand. Anyway, we'll see in 3 days, or later this week how close to reality that claim is. If it's double the performance at a 5W lower TDP, then that's pretty impressive.

I admit i'm being somewhat optimistic, and I may be wrong, but the leaked performance benchmarks i've seen for the 750ti are downright great for a 60W TDP.

They are comparing a 65w GT 640 using DDR3 to a 750Ti using 5.4Gbps GDDR5... I hope it can beat it by a factor of 2.

Sigh... nvidia is saying the numbers in plain English: 35% increase in performance per core is pretty plain, per cuda core. 2x performance per watt means, obviously, twice the performance per watt used of the GPU. You're reading way too deeply into it for things that they haven't said, nor I. Hence why I said "2.7x" per nvidia as a base factor. I would expect no less than double the performance of the 780, from the GTX 880 or whatever the Big Maxwell is, with these two factors known currently, and possibly more. Add in that this is first-gen Maxwell and GM200/204 will be "second generation" Maxwell, and things are looking rosy.

Nvidia has never been vague with TDP numbers before... In the last year and a half, they have purposefully misrepresented their TDP numbers on at least 3 future products.

 

[Spazturtle]

Uhm, the 60W TDP 750ti outperformed the 65W TDP GT640 by a factor of two. Now that's a marketing slide, but it seems that the performance per watt has doubled if the slide is true. The 3dmark and firestrike scores over at videocardz/wccftech seem to collaborate this as well - performance in the same TDP window jumped dramatically with Maxwell. But these are leaks of course. We don't know if they have a complete basis in reality, I admit this.

I suspect it will be around that or maybe slightly lower depending on various factors such as resolution, etc. But if the slide is reasonably close to reality, the performance slide specifically - then it does seem to be better than a 35% perf/watt jump. Also, the SMX cores were completely overhauled for Maxwell; you can't do a direct per core comparison from what I understand. Anyway, we'll see in 3 days, or later this week how close to reality that claim is. If it's double the performance at a 5W lower TDP, then that's pretty impressive.

I admit i'm being somewhat optimistic, and I may be wrong, but the leaked performance benchmarks i've seen for the 750ti are downright great for a 60W TDP.

The 750ti has more improvements then just a new GPU, it has faster memory and an improved bus.

The GPU by its self is only 35% faster, but the other improvements also cause the performance of the card to improve.

But then again this is all marketing stuff so who knows.

 

[Sodapopjones]

Seriously, this must be a joke. I didn't see much trolling in either Nvidia/AMD section until a few months ago, and most of it started with Mantle. Your own posts were quite irrational at times. The only thing we are saying is to not get too excited about Nvidia marketing, because it's always overblown same as AMD marketing when it comes to benchmarks. Stick with reasonable expectations and you won't be disappointed. If you think there is going to be a 170% increase in performance at the top end, prepare to be disappointed.

In all honesty, I was like meh when I saw the Mantle promotion, it wasn't till it was actually put into action was I impressed (Both promotions were for parts most enthusiasts don't use). I'm sure Big Maxwell will be pretty rad, but I'm not getting excited till I see real life data. I also now own AMD and nVidia, each have their own pros and cons.

 

[Sunin]

So no math put to this yet? At a minimum, not assuming gains in cuda cores from going 20nm and efficiency gains that will bring from a wattage side. Here is the minimum we will see at 28nm:
3200 cuda cores (640@60w projected to a 300w spec)
That is 11.11% gain in cores! that are 35% faster or at a min we will see 50% increase in speed at 28nm. I also would expect if nvidia is smart they will put out the first 880 at 28nm unless amd comes in strong

Carry this logic to 20nm, and depending on wattage drops and such I would expect we will see 6x that of the 750, thus 3840 cuda cores or a peak gain of 33.33% in cuda cores and about 80% increase in performance this will be their 880ti refresh, they may even cripple it some so they can do a 990 refresh and offer up another 10-20% performance if again amd doesn't come in strong

Just my magic ball prediction

 

[GoldenTiger]

Some of you seam confused:

[*IMG]http://imageshack.com/a/img203/1222/8v51.jpg[/IMG]

That means a 60w chip will be 35% faster then a 60w chip from last gen.

But to get the same performance as a last gen 60w chip you only need a 30w chip this time.

Yeah, I posted that already twice. And you're contradicting your first line with the second one... . The actual case is that for the same performance level we should be seeing half the wattage used even on 28nm, and that per-core we are going to see a 35% gain (by their claims, at this point of course). So a card that would take Kepler 300w can be done in a 150w or less envelope is their claim, on 28nm even. That leaves a huge amount of extra power to continue stuffing in more cores and increasing clocks. Now add in that the Big Maxwell is going to be (GM200/204) on 20nm, and that improves power efficiency and die size usable even further. Now add in that this is Maxwell FIRST generation, for the GM107, and that GM200/204 are SECOND generation by all known info.... and we have a recipe for easily doubling the performance at least, if their claims are anywhere near true.

 

[jwcalla]

It'd be nice if Intel licensed the Maxwell architecture and released a really awesome Steam Machine this fall.

 

[Lorien]

So BEAST mode means 1080P at medium setting now?

This is so underwhelming all AMD had to do is re-release the 2+ year old 7850 with a slight overclock to beat it. Nvidia's architectures have never scaled down well, this simply continues the trend. I predict we'll see reviews clinging to the tree hugger argument.

I want to see a Kaveri system with a cheap 250 or 260X running dual graphics against Nvidia's Mini-ITX 300W system.

 

[evilsofa]

This is so underwhelming all AMD had to do is re-release the 2+ year old 7850 with a slight overclock to beat it.

The 750 Ti is to be released for $130-$140. The R7 260X for $120-$135 is slower than the 7850. The R9 270 is faster than the 7850 and is supposed to cost $180, but the bitcoin craze is doing crazy things to its price and the 270X at $240 is cheaper than the 270.

 

[Sunin]

If those efficiency #'s are correct I expect the crypto miners will be all over this card, first profitable nvidia card?

 

[Unknown-One]

Nvidia has never been vague with TDP numbers before... In the last year and a half, they have purposefully misrepresented their TDP numbers on at least 3 future products.

We're talking about cards with no external power connector here. It's not possible to misrepresent TDP by that much.

Maximum amount of power you're allowed to draw through the PCIe slot is 75w. Draw more than that and you risk the motherboard dropping the slot to PCIe 1.1 mode and the graphics card being forced into a lower performance state.

In short, we know it draws (at most) 75w.

 

[Spazturtle]

We're talking about cards with no external power connector here. It's not possible to misrepresent TDP by that much.

Maximum amount of power you're allowed to draw through the PCIe slot is 75w. Draw more than that and you risk the motherboard dropping the slot to PCIe 1.1 mode and the graphics card being forced into a lower performance state.

In short, we know it draws (at most) 75w.

Don't expect cards to be fully within their power limits.

Gaming won't load a card nearly hard enough to go close to the power limits, but something like litecoin mining or furmark will.

And drivers already have to underclock cards when running furmark to keep them within power limits.

Without the restrictions both furmark and litecoin mining push cards well over their rated power limits.

They can only draw 75W on the bus yes, but I expect the card to have a max power limit of around 85-95W and if they get too high it will simply throttle down.

 

[LordEC911]

We're talking about cards with no external power connector here. It's not possible to misrepresent TDP by that much.

Maximum amount of power you're allowed to draw through the PCIe slot is 75w. Draw more than that and you risk the motherboard dropping the slot to PCIe 1.1 mode and the graphics card being forced into a lower performance state.

In short, we know it draws (at most) 75w.

Correct, if 25% isn't "that much" to you...
There has been motherboards in the past that have allowed up to 150w through the PCI-e slot though.
As long as the "typical" power is less than 75w it is possible to still meet PCI-e spec.

 

[yanmeng]

I am very impressed with these numbers, especially if they are achieved with 28nm.

 

[xoleras]

We're talking about cards with no external power connector here. It's not possible to misrepresent TDP by that much.

Maximum amount of power you're allowed to draw through the PCIe slot is 75w. Draw more than that and you risk the motherboard dropping the slot to PCIe 1.1 mode and the graphics card being forced into a lower performance state.

In short, we know it draws (at most) 75w.

Basically, from what I've seen, the reference design is a 60W TDP as specified per nvidia. So if they're saying a 60W TDP i'd expect that to be pretty close regardless of what the slot provides; although I do think you're correct about going over 75W. Now, the thing is, the 6 pin connector is optional, so it probably isn't necessary. If an AIB wants their card to be overclocked, they'll just opt for a 6 pin.. I'm guessing that nearly all factory overclocked models, will probably have it to facilitate overclocking above the stock reference design.

 

[x3sphere]

So the 750 Ti is Maxwell-based? I wonder how it will fare for mining then. If they've tweaked the architecture so integer GPGPU performance is comparable to AMD this card could be amazing considering the low TDP.

 

[xoleras]

I hope it's downright horrible for mining. All of the Maxwell cards, even GM200, here's hoping it's godawful at mining.

Would just be cool if they were more favored by actual PC gaming enthusiasts instead. Which is where I see NV now. Just IMO. AMD has entered strange territory in this respect, because miners aren't the loyal customers that a PC gamer heavily invested in the ecosystem would be. I mean, I can't see any of the miners who don't game being long term customers if China and corporations start implementing data centers with 1000000's of scrypt ASICs. I mean, a few miners do game - but there are many more miners who don't care about gaming at all. And those guys won't come back if anything happens, and I do think it will at some point; whether it's 2 months from now, 6 months, or a year, who knows. We do know that scrypt ASICs are being developed, and once corporations can power that stuff with hydro or solar power GPU mining won't be ideal. That's essentially what happened to BTC which caused it to go bust for GPUs. China and corporate data centers with a billion ASICs.

So I guess what i'm saying. I personally wouldn't be interested in having those guys fluck with NV pricing. Even if scrypt mining is still around when GM200 hits.

 

[Spazturtle]

So the 750 Ti is Maxwell-based? I wonder how it will fare for mining then. If they've tweaked the architecture so integer GPGPU performance is comparable to AMD this card could be amazing considering the low TDP.

Still got the fact that NVIDIA drivers are pretty broken and have been for a few years now for GPGPU to deal with. Heck they use a whoule CPU thread when running OpenCL code.

xeloras: So you hope its a shit GPU?

 

[xoleras]

Still got the fact that NVIDIA drivers are pretty broken and have been for a few years now for GPGPU to deal with. Heck they use a whoule CPU thread when running OpenCL code.

xeloras: So you hope its a shit GPU?

No, I hope it's a great gaming GPU. Which is what these cards SHOULD be. Radeon is gaming? Where have I seen that on AMD's website. I never saw Radeon is mining. I don't give a fuck about mining.

You know what though, your post is goddamn hilarious. Nvidia drivers pretty broken. Mining correlating to a "shit GPU". LOL. Nice troll. Not even a subtle troll, complete blatant troll. 10/10. High fives all around. Good stuff. I hope no AMD fans get super offended at constructive criticism, but unfortunately a lot of them do. And then this happens. Great post. And by great, I mean, nice work trying your best to troll without even being subtle.

 

[jwcalla]

Still got the fact that NVIDIA drivers are pretty broken and have been for a few years now for GPGPU to deal with. Heck they use a whoule CPU thread when running OpenCL code.

Jeez this forum is a trainwreck.

Go ask the Blender guys why they don't support AMD and OpenCL.

 

[Meeho]

Jeez this forum is a trainwreck.

Go ask the Blender guys why they don't support AMD and OpenCL.

Well, if madVR is anything to go by, Nvidia's OpenCL is pretty buggy and not that fast when it does work.

 

[Spazturtle]

Jeez this forum is a trainwreck.

Go ask the Blender guys why they don't support AMD and OpenCL.

Go look at how NVIDIA cards work in OpenCL folding, look at how much CPU they use.

NVIDIA drivers are pretty broken and need fixing.

No, I hope it's a great gaming GPU. Which is what these cards SHOULD be. Radeon is gaming? Where have I seen that on AMD's website. I never saw Radeon is mining. I don't give a fuck about mining.

You know what though, your post is goddamn hilarious. Nvidia drivers pretty broken. Mining correlating to a "shit GPU". LOL. Nice troll. Not even a subtle troll, complete blatant troll. 10/10. High fives all around. Good stuff. I hope no AMD fans get super offended at constructive criticism, but unfortunately a lot of them do. And then this happens. Great post. And by great, I mean, nice work trying your best to troll without even being subtle.

How good a card is at mining directly relates to how good it is at GPGPU, so you hope for bad GPGPU performance. So you hope every single GPU accelerated program runs badly on Maxwell, you hope games are slow as they can't do GPGPU physics well.

GPGPU is almost the most important factor in GPUs these days and you say you hope its shit on Maxwell.

So yeah, you said you hope Maxwell is shit.

You either don't understand what a GPU does or you are a troll.

 

[jwcalla]

Well, if madVR is anything to go by, Nvidia's OpenCL is pretty buggy and not that fast when it does work.

You shouldn't use madVR.

 

[jwcalla]

Go look at how NVIDIA cards work in OpenCL folding, look at how much CPU they use.

NVIDIA drivers are pretty broken and need fixing.

How good a card is at mining directly relates to how good it is at GPGPU, so you hope for bad GPGPU performance. So you hope every single GPU accelerated program runs badly on Maxwell, you hope games are slow as they can't do GPGPU physics well.

GPGPU is almost the most important factor in GPUs these days and you say you hope its shit on Maxwell.

So yeah, you said you hope Maxwell is shit.

You either don't understand what a GPU does or you are a troll.

The idea that nvidia GPUs are "bad at GPGPU" is idiotic. If that was the case they wouldn't be in Top500 supercomputers.

Use CUDA.

OpenCL is still a turd and needs some time to cook.

And yes, it's a turd on AMD as well.

 

[Spazturtle]

The idea that nvidia GPUs are "bad at GPGPU" is idiotic. If that was the case they wouldn't be in Top500 supercomputers.

Use CUDA.

OpenCL is still a turd and needs some time to cook.

And yes, it's a turd on AMD as well.

Stanford University dropped supporting CUDA on Folding for a reason.

The new OpenCL core seams to run very well on AMD GPUs and even Intel CPUs. No CPU usage on AMD GPUs, but on NVIDIA is is slower and uses a CPU thread.

CUDA is not a solution, it is a propriety problem.

Also I didn't say NVIDIA cards were bade at GPGPU, there workstation GPUs are very good (consumers ones are artificial restricted) its just the driver has issues.

I said that by Xeloras hoping they are bad at mining he is also hoping they are bad at GPGPU as the two are linked.

 

[xoleras]

Spaz, Mining has absolutely nothing to do with GPGPU. I should add, that nvidia created GPGPU ecosystem with the Fermi, and AMD was left scrambling for some type of answer. Mining has more to do with being ALU bound which is something that AMD does 3 times per clock versus one on NV, but that is not a specific GPGPU function. GPGPU uses the GPU to do traditional CPU functions, and again, the entire GPGPU ecosystem was originally created by NV in the first place. So your assertions and trolling are downright comical. Especially since you're basically FOS. Furthermore, nvidia basically owns the roost in terms of GPGPU sales for supercomputers and professional applications, AMD has got nothing here. But you would say that all because of mining, that AMD is better at GPGPU. How comical.

If you're going to troll, maybe not be so blatant and get your facts straight. Here's another hilarious fact. That i'm sure you don't know because you're completely FOS. Double Precision was crippled on the Hawaii GPUs by a factor of 1/8th. So Tahiti has better DP performance than Hawaii. Guess what matters more for GPGPU? Not that this has JACK SHIT to do with gaming performance. Which was the original point - I want gaming performance in a Geforce Maxwell, I don't want goddamn DP, mining, or any of that shit. None of that stuff has anything to do with gaming performance and never has.

I don't give a fuck about mining. I want a good gaming GPU with a Geforce part. Not a good mining GPU. If you want a good mining GPU clearly you're in the wrong goddamn forum. That was my original point, and mining has absolutely jack shit to do with good gaming or GPGPU. Mining has never had a correlation with gaming performance - the fermi based GTX 580, the first GPGPU, decimated Cayman in DX11 gaming. Although for some reason Cayman was far better at mining. Gonna be hard to explain that one huh.... But by all means continue your blatant trolling. It's pretty hilarious at this point. Your facts are so incorrect that it's hard to take a word you type even remotely seriously.

 

[Spazturtle]

Spaz, Mining has absolutely nothing to do with GPGPU. I should add, that nvidia created GPGPU ecosystem with the Fermi, and AMD was left scrambling for some type of answer. Mining has more to do with being ALU bound which is something that AMD does 3 times per clock versus one on NV, but that is not a GPGPU function. GPGPU uses the GPU to do traditional CPU functions, and again, the entire GPGPU ecosystem was originally created by NV in the first place. So your assertions and trolling are downright comical. Especially since you're basically FOS. Furthermore, nvidia basically owns the roost in terms of GPGPU sales for supercomputers and professional applications, AMD has got nothing here. But you would say that all because of mining, that AMD is better at GPGPU. How comical.

If you're going to troll, maybe not be so blatant and get your facts straight.

I don't give a fuck about mining. I want a good gaming GPU with a Geforce part. Not a good mining GPU. If you want a good mining GPU clearly you're in the wrong goddamn forum. That was my original point, and mining has absolutely jack shit to do with good gaming or GPGPU. But by all means continue your blatant trolling. It's pretty hilarious at this point. Your facts are so incorrect that it's hard to take a word you type even remotely seriously.

Mining has nothing to do with GPGPU when mining is literally GPGPU work? Mmkay sure...

 

[xoleras]

LOL. The trolling continues. Just throw a vague random comment out there and pretend it's true. I'm not sure why anyone even bothers responding to you. It's just hilarious at this point. I'm going to just watch you continue so I can sit back and laugh at how stupidly comical your posts are.

While you're at it, maybe you can tell us about how great the 290X is at GPGPU since it has crippled DP by a factor of 1/8th. That should be a VERY enlightening discussion. But I wouldn't expect a serious non troll answer since you've proved your FOS status long ago.

 

[Spazturtle]

LOL. The trolling continues. Just throw a vague random comment out there and pretend it's true. I'm not sure why anyone even bothers responding to you. It's just hilarious at this point. I'm going to just watch you continue so I can sit back and laugh at how stupidly comical your posts are.

If Mining is not GPGPU then what is it? Gaming?

 

[Stoly]

Yes nvidia plain sucks at computing, that's why AMD GPUs dominate the supercomputer charts... no wait...

 

[Spazturtle]

Yes nvidia plain sucks at computing, that's why AMD GPUs dominate the supercomputer charts... no wait...

Do supercomputers use consumer GPUs?

Do supercomputers use windows with NVIDIA windows drivers?

if the answer to any of those is no then you point is irrelevant.