Your logic is great but I would be a little reserved.The only reason Vega is using HBM2 is to get more performance (high core clockspeeds) under a 300W envelope, not for cost reasons because a bigger die with GDDR5X is actually the cheaper solution.
You mean 500USD FULL GP102(alias GTX285/GTX480/GTX580) and not 1200USD cutdown TITANX, 250USD GTX1080(alias GTX560TI), 200USD GTX1070 (alias GTX560) and 100USD GTX1060 alias whatever low-end it was back in those nice days..Imagine a lineup of $649 GTX1080Ti, $399 GTX1080, $299 GTX1070 and $199 GTX1060.. something similar to the old school GPU pricing.
This card is nearly 2x the price of 1080. In terms of FPS/$, new Titan X is worse than 1080 by a lot. Not sure why you'd diss 1080 because of Titan X.
This is the thing with stable competition. At least the current gpu market has thought us that we can't even quantify the level intel is screwing us price-wise.I will be waiting for the inevitable 1080ti. AMD hurry up with Vega so the 1080ti can come out!
Using hbm should help. A fury gets 60% more performance/watt than a 390 according to techpowerup.
I hear hbm also helps with die size, since you don't need the on die stuff to drive the signal through the pcb as with gddr.
This is the thing with stable competition. At least the current gpu market has thought us that we can't even quantify the level intel is screwing us price-wise.
That's a non sequitur, underutilized shaders cannot increase perf/watt.Also, the Fury had good perf/watt partially due to the shaders being choked out by the front-end and keeping it way under-utilized.
If that metric means nothing then the 1080 was not a bad product either, so any way you cut it Arachnotronic's point stands: there's no reason to diss 1080 because of Titan X.
AMD can't compete with a GP102. They're going to have to use HBM2 just to compete with the 1080. and only because it would otherwise have too high of a TDP. Pascal is only in Maxwell territory perf/watt wise. That is not good.
Nvidia has gotten advatage in almost every front now- perfomance, power consumtion, getting the chips out first etc. Unless AMD gets to improve their architecture and/or process tech for Vega considerably (which is unlikely) i see top Vega being maybe 10-15% faster than 1080 at best. And even then most people who did want perfomance in that territory have bought Nvidia card by the time Vega gets out+ Volta would be really close by then. Sad times really.
GPU wise is almost a tie. AMD can back fire anytime.AMD can't compete with a GP102. They're going to have to use HBM2 just to compete with the 1080. and only because it would otherwise have too high of a TDP. Pascal is only in Maxwell territory perf/watt wise. That is not good.
GPU wise is almost a tie. AMD can back fire anytime.
CPU wise, the gap between AMD and Intel is humongous.
Gonna lay low for a while since I've gone off topic a lot.
It's damn hard to talk about Titan X without bringing AMD/Vega up.
Nvidia has gotten advatage in almost every front now- perfomance, power consumtion, getting the chips out first etc. Unless AMD gets to improve their architecture and/or process tech for Vega considerably (which is unlikely) i see top Vega being maybe 10-15% faster than 1080 at best. And even then most people who did want perfomance in that territory have bought Nvidia card by the time Vega gets out+ Volta would be really close by then. Sad times really.
Why is it sad times? NVIDIA has product out now that hits these performance levels.
Anyway, AMD doesn't really sell many high end GPUs (see Steam Hardware Survey), high end buyers overwhelmingly choose NVIDIA, so it's probably smart for them to focus on stuff like Polaris that sells into lower price points and OEM systems.
100% agree with you. Overclock that 1080 and you suddenly become within 10% of a stock Titan X. Yeah, we know a Titan X can overclock but overclocked vs overclocked will only be a ~ 25-30% performance difference for double the price (Non founders edition pricing wise)
I am so glad I picked up my 1080 and even got it for $620. Works great on 1440p @ 144hz.
I'm not sure I agree with that. There seems to be a correlation between sales and who has the fastest single GPU.
Why is it sad times? NVIDIA has product out now that hits these performance levels.
That would be a terrible idea. Nvidia would have no reason to keep prices in the sane world (case in point with Titan X). AMD still has a ~20% market share so getting them back to parity with Nvidia should be on everyone's wish list. That would help drive prices down for team green and team red.Anyway, AMD doesn't really sell many high end GPUs (see Steam Hardware Survey), high end buyers overwhelmingly choose NVIDIA, so it's probably smart for them to focus on stuff like Polaris that sells into lower price points and OEM systems.
Do you mean ever or just with their current line up? Titan X is certainly well, well out of reach of the 480. But if you're saying Titan X will face no competition from AMD at all that's pretty absurd. They've already announced Vega... nVidia just has a huge TTM advantage here. This is like when 58xx series dropped months before Fermi.AMD can't compete with a GP102. They're going to have to use HBM2 just to compete with the 1080. and only because it would otherwise have too high of a TDP. Pascal is only in Maxwell territory perf/watt wise. That is not good.
Because we as consumers now get to pay $1200 (#$#$&$&*&!!!) for the top end. I'm not sure how you don't see that as a negative.
That would be a terrible idea. Nvidia would have no reason to keep prices in the sane world (case in point with Titan X). AMD still has a ~20% market share so getting them back to parity with Nvidia should be on everyone's wish list. That would help drive prices down for team green and team red.
The price range has expanded upwards, but for the same money as you paid in previous generations you're still getting a ton more performance. GTX 1070 at $379-$449 packs a lot of punch compared to what you could get for that money prior gen. GTX 1060 for $250 (yes, you can buy them for $250) is also a great deal.
Just because NVIDIA is introducing products at higher price points than previous doesn't mean that gamers are "getting screwed." It just means that for those willing to pay more for even more performance, the option is there.
AMD has 30% share of the total discrete GPU market, but around 22% share of the desktop dGPU market by units. However, from a business perspective, AMD's ultra high end GPUs have not fared well in the marketplace.
Look at the Steam Hardware Survey numbers:
http://store.steampowered.com/hwsurvey/videocard/
GTX 1080 is already 0.42% of the Steam DX12-capable user base and that card is what, 2 months old and constantly out of stock? R9 290-series (that's 290/290X) is 0.99% of that same base and R9 390-series is just 0.73%, and these have been in the market for quite a while at much lower prices.
GTX 970 is 7.36% of that same population, GTX 980 Ti is 1.39%, and GTX 980 is 1.44%.
If I were in charge of making business decisions at AMD, I would take a very long, hard look at whether it makes sense to continue investing in ultra high end dGPUs. They just don't sell that well, so it's a whole lot of R&D and marketing dollars spent for what might not even be a positive economic return.
During those times, their share lead shrunk. Since the 5870, where AMD had its largest recent lead, they've steadily lost market share. The ATI 980 pro days were also good for them.Too many confounding variables for such a correlation to be meaningful.
Having the halo product crown helps, but you have to understand that a large part of the reason that NVIDIA has been successful has been that it has invested pretty successfully in the marketing side of things. It has built a brand and presence that's really on a whole different level than what AMD has managed to do.
That success has allowed it to invest more in R&D to gain what today looks like an unprecedented product advantage in terms of raw performance at the top (1070/1080/Titan X) and efficiency at iso performance levels (i.e. GTX 1060).
So, I think from here on out we will see NVIDIA really start to pull ahead from AMD technologically, but historically the technological gap couldn't totally explain the market share lead that NVIDIA has had. There have even been times where NVIDIA was behind on technology but it still managed to have the share lead.