I will bet my tag that maxwell GM200 will release with a 384bit bus or better
I'd be disappointed if GM200 doesn't have a 512bit bus.
I will bet my tag that maxwell GM200 will release with a 384bit bus or better
Say what?Another Titan re-branded card, me pass.
Well, now you can play DOTA on 3 4k monitors.All these new cards
And all I play is dota
That's based on one 750Ti part. Do we know if it scales well from 640 cores up to high 2000s? You need to wait till their cards arrive at Toms/Anand to see.Why are you purposely ignoring the obvious large gains they have made at 28nm with maxwell? Your moore's law has hit a brick wall excuse holds no water
We need to be into chip design to know for sure. Here people are taking laymen terms like TDP and 28nm, and project to everything including sinister predator pricing. Anand did mention the design turnaround in their 680 review, including the refocusing on reducing the high TDPs of 580. Maybe it was for mobility, since laptops are the biggest beneficiaries (correct me if wrong).Again, nvidia went down this same path with the move from 40 to 28nm , what's your excuse for that?
Fixt.That's based on one 750Ti part. Do we know if it scales well from 640 cores up to high 2000s? You need to wait till their cards arrive at Tech Report/PCPer to see.
No I'm just commenting on you acting like console games being designed for amd hardware with multiplatform games will somehow make up for amd getting really far behind performance wise (once gm200 is out) and being far behind performance/watt wise.
Amd cards won't be irrelevant if they don't have a proper successor to catch up to maxwell no, it'll be like the cpu market where they can get the crumbs at the low end (similarly with underpowered yet power hungry hardware) while the consumer gets fucked up the ass due to there being no competition at the midrange or high end anymore for nvidia.
The cpu market is an ugly thing right now, I'd hate to see the gpu market fully barrel down that abyss.
This is just a what if scenario that you are proposing (preemptive damage control with that amd gpu optimisation nonsense, hence why I replied), I don't know what amd have coming next. (but they better have something good up their sleeves for OUR sake.)
As for the midrange 960 ti, that is what the 980 already is... it's the successor to the 560 ti
Anything nvidia would call 960 ti right now would probably be some low end 196 bit bus abomination like the 2GB 660 ti was and would be no faster than a gtx680.
Anyhow that's enough down to earth realistic negativity from me for today, here is to hoping nvidia sell the 980 for 400 or less and that the 970 can become the next 560ti that we all want.
(so you're saying there's a chance.gif)
Anyhow that's enough down to earth realistic negativity from me for today, here is to hoping nvidia sell the 980 for 400 or less and that the 970 can become the next 560ti that we all want.
That's based on one 750Ti part. Do we know if it scales well from 640 cores up to high 2000s? You need to wait till their cards arrive at Toms/Anand to see.
We need to be into chip design to know for sure. Here people are taking laymen terms like TDP and 28nm, and project to everything including sinister predator pricing. Anand did mention the design turnaround in their 680 review, including the refocusing on reducing the high TDPs of 580. Maybe it was for mobility, since laptops are the biggest beneficiaries (correct me if wrong).
Well my problem is I don't have a 4K monitor or in the plans to get one any time soon.
Although 120fps gaming does interest me.
So if whatever card I get next to surpass my HD7950 can at least feasibly accommodate 1080p/120fps
snip.
That's based purely on die area and TDP, again we are using two layman measurements to project the world on these parts. We're not accounting for actual complexity of chip and the associated defect rates if they decide to pack more into a larger die. Sure Maxwell looks extremely promising now, but how realistic is it to expect Nvidia to reach that promised land without the learning curve of Kepler + first stab at larger die 28nm back at 2012?They didn't refocus shit, they just released a small die midrange gpu (680) and 6 months later released the high end one.
That's based purely on die area and TDP, again we are using two layman measurements to project the world on these parts. We're not accounting for actual complexity of chip and the associated defect rates if they decide to pack more into a larger die. Sure Maxwell looks extremely promising now, but how realistic is it to expect Nvidia to reach that promised land without the learning curve of Kepler + first stab at larger die 28nm back at 2012?
So basically, even though I could get one in January, I should save up and wait for GM200?
What's the point of this though? Were you guys actually expecting Moore's Law to be eternal? This is like shouting at the clouds.
These progresses listed were accompanied with a pretty significant die shrink and clockspeed increases because they weren't hitting electrical and thermal limitations that chipmakers are currently facing now on the 22/28nm generation. You knew the writing was on the wall Intel Isreal told HQ that physics is getting tired of their shithaha
I've covered all of this earlier in the thread, so if you want to talk about it with me just read the previous pages, I don't want to type it all again![]()
One thing I forgot to add earlier: nvidia did have big kepler lined up and on track before the 7970 released, all the sources at the time about gk104 were accurate and those same sources said gk110 was on the way at the same time.
Then the 7970 happened , was shit (not shit compared to ancient 40nm and shitty 6000 series architecture, but obviously shite compared to kepler) and sold well at a very high price (again the reason why I covered before) , so nvidia said we'll have some of that ! and postponed the gk110 launch since they didn't need it to compete.
I doubt you'll find many people who don't agree on this (gk 110 release) by now.
If I don't reply anymore it's because I passed out it's 4 am![]()
Well, now you can play DOTA on 3 4k monitors.
I'd be disappointed if GM200 doesn't have a 512bit bus.
More info seems to have been leaked about the 970:
http://wccftech.com/maxwell-geforce-gtx-970-alleged-specs-leak-shots/
1644 CUDA cores (had leaked earlier)
4GB GDDR5
32 ROPs
1051+ MHz core clock (1178 boost clock)
7.0 GHz memory clock
~150W TDP
Synthetic benchmarks do seem to put it around the performance level of the 780.
Here you go:I just want a single card that beats my 690.
So I can get two
don't get your hopes up.My GTX670 is almost dead so I really, really need the 980 released as soon as possible.
It better be more powerful than a 780Ti. Hopefully 6Gb, too.
don't get your hopes up.
i'm still pretty much a layman when it comes to computer technology but it's looking to me that nvidia are setting the foundations for a product (i.e the next iteration of the big maxwell cards) that will outstrip the 780ti rather than beating it out of the gate. what we're getting is something on the same level or marginally more powerful, with slightly more vram, and lower power consumption. it's not a total waste of a card, but the price point is probably going to make it very unattractive.
Here you go:
Titan-ZRadeon R9 295X2
all we can do is remain optimistic right now.Man, I really hope it's priced decently. It better be under $600 AUD.
I was just joking. From what you wrote, I knew you wanted to get away from dual GPU on a single card.Radeon is temping. Temping, but I've had tons of ATI problems in the past... plus it's a dual card as well. Trying to get away from that.
I want to believe that the GPUs will be monsters at overclocking at least, considering the 970 has a higher boost clock than the 770 while having an 80W lower TDP (230 vs ~150).
All these people saying they NEED the 980 right now because their 600 series is outdated... I'm on 470 GTX. Feel my first world pain. I feel like a homeless drunkard to your droogs.
But as soon as EVGA or ASUS bring out their version with 6GB or more they will be singing in the rain, of my money.
really really hope nvidia stops being greedy and prices these affordably, I would love to upgrade from my 650ti
Yeah I haven't updated my 560Ti drivers since Nvidia went through that spot where every single update would either crash your system or run the card so hot it diedYeah, I guess that card is about 3 years old now. It has actually done well by me, although truth be told I have not updated the drivers in ages, due to the fact that the most recent drivers were causing it to crash.
You can always expect Asus to come out with DCUII non-reference, even their 290X was fairly discreet.I just need a variant that is quiet, I recently upgraded my case fans and my GPUs are now incredibly noisy.
I just need a variant that is quiet, I recently upgraded my case fans and my GPUs are now incredibly noisy.
Also, count me in the 560 Ti group, although I have the 448 core version.
So it is true we are getting 16nm GPU next year?
http://wccftech.com/tsmc-accelerating-16nm-process-gen-gpus-coming-sooner-expected/
28nm to 16nm should be a massive jump. It seems we been at 28nm forever.
Same. Wondering if I'll ever feel a need to upgrade.
Did I take a wrong turn and end up in some console vs pc thread, you believe in your optimisation fairytales, but in the real world the 980 in name 960 in specs is outperforming the 290x by 20 percent while using less than half the power.
Which reminds me,does anyone know what amd actually have coming next?
This tonka toy 285 can't be it...
I haven't heard anything about their next architecture
You haven't heard anything about AMDs new GPUs because you aren't really looking. Their new line of GPUs will be called Pirate Islands. The 390X is rumored to have over 4000 stream processors, 96 rops, and a 512 bit memory bus. production right now.