Which 1070 to buy? Or 1080? Or wait for Price drop in 2017??

>>58260886


gtx 1080 is a massive upgrade over your current card. get the 1070 if you don't need that.

I know that the 1080 is a massive upgrade... but I don't know if the 1070 is good enough or if there are better options or reasons...

From a personal experience I'd recommend a Gainward/Palit for great cooling and the virtually no noise, but as long as you don't get an EVGA housefire, they're all good.

I'd either save and wait for the 1080Ti or get the 1070 now.
1080 is too small of a performance increase over the 1070 for the price it's going for.
Besides OC'd 1070 can hit stock 1080 levels of performance anyways.

I never heard about a Gainward / Palid here in germany...

File: EVGA-FTW-GTX-1080[1].jpg (47KB, 640x360px) Image search: [Google]

47KB, 640x360px

>>58260886
You're in germany which means you have the advantage of gainward and palit, brands we can't get with out importing here in the states. All 1070s will run at stock just fine, the difference is overclocking. All cards can overclock to the default 10%, every nvidia card seems to do this for some reason. However a few manufacturers make cards that can go past that and are only limited by thermals. The white galax 1060 is one example. I believe that the gainward phoenix has a very good power delivery system that will allow you to go past the 10% overclock but the 1080 hits a wall at a point. I think the 1070 is the best value right now.

This is what I'm basing my opinion off of http://videocardz.com/60547/comparison-of-custom-geforce-gtx-1080-pcbs

Maybe someone else with more experience overclocking GPUs can chime in.

>>58261282
No it can't this isn't maxwell

the one that is a 480

>>58261759
Let me also add that here you can see
http://wccftech.com/nvidia-pascal-gpu-frequency-world-record-3-ghz/
that the 1060 was able to go as far as they wanted. I water cooled my 1060 and hit a power or voltage wall at 2130mhz core even though the GPU was at only 45c. There was absolutely nothing I could do to make it go further, this same overclock was stable on air, and my 1060s air cooler is shittier than the intel stock one. I am completely convinced nvidia leaves that extra 10% there as sort of an appeasement to "enthusiasts". So this same situation would happen if I had a 1070 or a 1080, no matter what cooling I use I'm only going to get that 10%.

The next card I buy will be something like those galax cards that don't have this bullshit jewery and let you go as far as your cooling is able. Most RX 480s seem to have very overbuilt power delivery systems too. If I water cooled one of them I'd be able to take it to 1500mhz which would be superior to the 1060 at 2130.

>>reddit fucking newfag, you clearly don't belong here

>>58260886
Gigabyte makes some abysmally shitty cards (sans the 1060, heard they did a good job there).

Go with MSI (gaming only), EVGA (for dat warranty), ASUS (strix only), or Palit.

Alternatively, XFX GTR for AMD if you choose to stick to red.

>>58261715

That's what google is for anon.

http://www.mindfactory.de/product_info.php/8GB-Palit-GeForce-GTX-1070-JetStream-Aktiv-PCIe-3-0-x16--Retail-_1094585.html

https://www.compuland.de/product_info.php/8GB-Gainward-GeForce-GTX-1070-Phoenix-Aktiv-PCIe-3-0-x16--Retail-_1094868.html

>>58261777
If you get Samsung memory in your 1070, it will hit almost exactly the same speeds as stock 1080.
The difference is like 5% at best, that's close enough.
http://www.overclockersclub.com/reviews/nvidia_geforcegtx_1070_overclocking/6.htm
If you get Micron crap memory, you'll get a slightly worse OC out of the card.

keep R9 390, it's still a good GPU

>>58261839
Impressive actually
I wonder why pascal scales so shit in 4k though vs maxwell
Look at the oc gains of the 980ti and the 1070
Are pascal stock clocks just set to be more aggressive so there's a lower ceiling?

"keep R9 390, it's still a good GPU"

OF course it is still good! But not for my Monitor!!! Read my post first!

>>58261919

I haven't looked that deep into the subject, but they seem to have restricted the voltage on Pascal cards on the hardware level.
https://xdevs.com/guide/pascal_oc/#voltsc
And according to these guys the cards don't like higher voltages, even if you're able to pump more of it there.
Guess you could say that the ceiling is a lot lower with this generation.

>>58262154
I'm not too /g/esque but with all the lurking I've heard that node shrinks result in higher voltages being more likely to damage transistors
At least in CPUs
Correct me if I'm wrong but this could be a potential factor. Is pascal all that much of a different arch? I thought the biggest deal was bringing it to 16nm from 32