• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Anybody else unimpressed with new midrange Nvidia GPUs, and much higher MSRP?

Page 29 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
It's generally exciting to see new cards but I'm kinda meh over the 1070 pricing that's up another $50. This is basically a 400 dollar "mainstream" card and you have to think the xx60 class card will now be 300-320.

I guess I'll be waiting for the next price😛erformance champion from AMD...Their 7950/290/390 have all been easy choices in their respective price brackets IMO especially after typical price drops.
 
atticus14 said:
It's generally exciting to see new cards but I'm kinda meh over the 1070 pricing that's up another $50. This is basically a 400 dollar "mainstream" card and you have to think the xx60 class card will now be 300-320.

I guess I'll be waiting for the next price😛erformance champion from AMD...Their 7950/290/390 have all been easy choices in their respective price brackets IMO especially after typical price drops.
Click to expand...

Especially with their well known longevity after the fact.

Hopefully nVidia allows Maxwell owners to at least have good performance for through this fall and winter as GP104 supplies are undoubtedly going to be tight.
 
I am going to make some reasonable predictions for 1080.

25% faster than a reference 980Ti with ~20-22% overclocking headroom on air (2114mhz / 1733mhz)

For anyone who isn't on 980Ti OC/Titan X OC, this would annihilate every card out there.

1080 stock = 104%
1080 w/ 2.1-2.15Ghz OC = 125% (close to 60% faster than a stock Fury X)

perfrel_2560_1440.png


1080 OC will be 2X faster than base 290X/390, while later versions such as MSI Lightning/Gigabyte Xtreme and EVGA Classified could raise it to 2X faster than a stock 980.

MSI Lightning 1080 w/ 2.3Ghz OC = 132-138% (2X faster than a reference 980)

----

While 1080 OC won't be a big upgrade for 980Ti owners, for everyone else it will be a massive improvement when comparing AIB 1080s with a 20% manual OC from base 1733mhz.
 
Last edited:
RussianSensation said:
I am going to make some reasonable predictions for 1080.

25% faster than a reference 980Ti with ~20-22% overclocking headroom on air (2114mhz / 1733mhz)

For anyone who isn't on 980Ti OC/Titan X OC, this would annihilate every card out there.

1080 stock = 104%
1080 w/ 2.1-2.15Ghz OC = 125% (close to 60% faster than a stock Fury X)

perfrel_2560_1440.png


1080 OC will be 2X faster than base 290X/390, while later versions such as MSI Lightning/Gigabyte Xtreme and EVGA Classified could raise it to 2X faster than a stock 980.

MSI Lightning 1080 w/ 2.3Ghz OC = 132-138% (2X faster than a reference 980)

----

While 1080 OC won't be a big upgrade for 980Ti owners, for everyone else it will be a massive improvement when comparing AIB 1080s with a 20% manual OC from base 1733mhz.
Click to expand...

I don't think it will annihilate even previous generation cards. When you add up the percentages in actual fps compared to previous generation cards and current generation cards you see that the difference becomes smaller than the percentages show. For example the 780Ti is about equal to a 970 (if not slightly faster) so even owners who bought the GTX 700 series should still be able to compete somewhat with lower-mid range Pascal series. Either way it's too early to tell seeing as how the NDA hasn't been lifted yet.
 
AtenRa said:
NVIDIA GeForce GTX 1080 PCB pictured

http://videocardz.com/59818/nvidia-geforce-gtx-1080-taken-apart

NVIDIA-GeForce-GTX-1080-PCB-4-900x600.jpg


NVIDIA-GeForce-GTX-1080-PCB-2.jpg
Click to expand...

This ref board picture is making me wonder if the EVGA Hybrid would work on it? Anyone?
980 Ti Ref board

evga_gtx_980_ti_sc_pcb1.jpg



And only 1 8pin connector? Interesting...Even the 680 had 2x6pin connectors.

EDIT: I think it might if the mount screws are the same distance apart. The EVGA hybrid kit actually mounts using the stock heat spreader, so it has no "clearance" issues with any of the capacitors, VRAM modules, etc etc. It sits on top of the GPU.

I wonder if I should buy one while they're going for <$50 and just hold on to it. Hmmmmm....

EDIT:
Mounted EVGA kit
GTX-980-TI-HYBRID-9.JPG
 
Last edited:
flopper said:
3ghz Oc Pascal seems managable with LN2.
question what they can do on water
Click to expand...

I hope you're right, but it still has nothing to do how well OC scaling will be. GM200 has some of the best OC scaling I've seen, generational increase in performance in some cases.

Pascal clock per clock is slower than Big Maxwell. How much slower? That we don't know.
 
Last edited:
Face2Face said:
Pascal clock per clock is slower than Big Maxwell. How much slower, that we don't know.
Click to expand...

I don't think it matters much when you can push the clock speeds as much as they're doing. It would take a pretty substantial IPC hit to offset the speed gains that they're seeing.
 
RussianSensation said:
I am going to make some reasonable predictions for 1080.

25% faster than a reference 980Ti with ~20-22% overclocking headroom on air (2114mhz / 1733mhz)

For anyone who isn't on 980Ti OC/Titan X OC, this would annihilate every card out there.

1080 stock = 104%
1080 w/ 2.1-2.15Ghz OC = 125% (close to 60% faster than a stock Fury X)

perfrel_2560_1440.png


1080 OC will be 2X faster than base 290X/390, while later versions such as MSI Lightning/Gigabyte Xtreme and EVGA Classified could raise it to 2X faster than a stock 980.

MSI Lightning 1080 w/ 2.3Ghz OC = 132-138% (2X faster than a reference 980)

----

While 1080 OC won't be a big upgrade for 980Ti owners, for everyone else it will be a massive improvement when comparing AIB 1080s with a 20% manual OC from base 1733mhz.
Click to expand...

To add on to this post, this chart from RS is this 980 Ti sample's performance at factory OC/boost. This chart is that same card compared to a manual OC. A 11% gain over aftermarket stock and about 37% over reference stock...

BF3 is the only game though...

perf_oc.png
 
Fallen Kell said:
Am I the only one that thinks that photo has been shopped? A chip manufacturer wouldn't use 2 different silk-screening/etching processes on the same chip to get 2 different color fonts like that.
Click to expand...

I was thinking the same thing. Its very different from anything done previously.
 
Fallen Kell said:
Am I the only one that thinks that photo has been shopped? A chip manufacturer wouldn't use 2 different silk-screening/etching processes on the same chip to get 2 different color fonts like that.
Click to expand...

What am I missing? This looks like the GPUs that came before it.

21305_nvidia_gtx_780_core.jpg
 
Mopetar said:
I don't think it matters much when you can push the clock speeds as much as they're doing. It would take a pretty substantial IPC hit to offset the speed gains that they're seeing.
Click to expand...

The effect of needing a greater absolute increase in frequency to see the same percentage increase over a higher base clock and a lower IPC are potentially problematic for hopes of Pascal OCing like Maxwell.
 
Pneumothorax said:
That GP104 looks so wimpy compared to the GM200. You're paying quite of bit of dough for that tiny chip.
Click to expand...

It looks physically small because it is! Its still the fastest consumer grade GPU out there, and almost certainly the fastest that can be produced right now at a halfway sane price.

I'm not sure why people get emotive about this - NV are really stunningly predictable just now. ~30(?)% cumulative increase in top performance year on year. The top chip then gets a premium, as reasonably befits the fastest card in the world.

In all honesty, you can see them keeping it up for years to come too - Big pascal, then medium Volta (HBM2, much more mature process so bigger chips, architectural gains) then big Volta, by which time there will likely be a usable die shrink and so on.

Very easy to deal with as a consumer. Pick what performance (increase) you want and you can almost preorder the card 🙂

It might seem a bit calculated, but I suspect we do quite well out of it over time. The alternative 'moonshot' style of approach - 980ti right off that bat at 28nm or something insane?! - would have the odd triumph but also the odd total fiasco.
 
xthetenth said:
The effect of needing a greater absolute increase in frequency to see the same percentage increase over a higher base clock and a lower IPC are potentially problematic for hopes of Pascal OCing like Maxwell.
Click to expand...

Given the rumors of third party parts clocking at 2.3+ GHz, I think it's fair to say that the chip over clocks well. While a lower IPC means you need more of an OC for the same relative performance gains, I don't think the performance hit will be big enough for this to be a worry.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top