FEATURED Nvidia 1080 launch thread

[bob4933]

Do we have an idea when the HBM2 Versions of the cards will be released? I'm hesitant on jumping into it until then, just to eliminate that as a factor.

How much does this REALLY matter? I mean how fast is "fast"? gddr5x is hella quick and far beyond anything consumer for a long time.

 

[wingman99]

I don't think the GDDR5x is the bottle neck, we will have to see how much it improves the performance alone when overclocked with the GPU, having the GPU overclocked first.

 

[bob4933]

I don't think the GDDR5x is the bottle neck, we will have to see how much it improves the performance alone when overclocked with the GPU, having the GPU overclocked first.

memory has never been the limiting factor to my experiences. I dunno why they make it such a big deal

 

[redrumy3]

So I just got my card and did clean install on drivers and decided to download 3dmark demo and run it.

Does this seem about right for stock 1080 and my i5 @ 4.6ghz

 

[Evilsizer]

guru3d.com is showing 19k for the 1080
http://www.guru3d.com/articles-pages/nvidia-geforce-gtx-1080-review,28.html

after looking looks like your physics score is lower then on guru3d.com but the first score on the left, yours is a touch higher.

 

[redrumy3]

guru3d.com is showing 19k for the 1080
http://www.guru3d.com/articles-pages/nvidia-geforce-gtx-1080-review,28.html

after looking looks like your physics score is lower then on guru3d.com but the first score on the left, yours is a touch higher.

hmm I don't know how they get such a high physics score.

I guess the 8 core cpu helps that part of the test a lot lol.

so it looks like besides physics I don't seem to be bottlenecking card

 

[Evilsizer]

yea the 8 core might help but im not sure if its set to run on the cpu only, i thought that test was for the GPU. nah you shouldnt be, anyone using a 3rd gen+ intel shouldnt be, barred IMO the cpu speed is at least 4ghz.

 

[Janus67]

I believe the physics test in 3dmark greatly benefits from additional cores, including hyperthreaded cores. an i5 vs i7 makes a difference here.

 

[EarthDog]

Indeed it does... they are using an Octo-core Intel at 4.4Ghz. You have a quad core clocked a bit higher.

memory has never been the limiting factor to my experiences. I dunno why they make it such a big deal

It is at 4K. Look at the 980Ti versus the Fury line and its HBM1. You will notice in benchmarks that as the resolution goes up, the gap the 980Ti holds over the Fury products shrinks.

 

[Woomack]

Memory bandwidth was limiting factor in 256bit graphics. In 386b+ you couldn't see performance gain so much from memory overclocking but cards like GTX980 are scalling really good with higher memory clock. GTX1080 is also 256b card so I expect about the same.

I see that performance in 3DMarks on GTX1080 is about as high as on GTX980 at about the same clock ... but there are no GTX980 which can run at stable 1700+MHz. I had to use ss/dice cooling to make my GTX980 pass 1700MHz and then it made ~16.5k in 3DM.

I'm wondering if there is any annoying coil whining on reference GTX1080. I had bad experiences with some cards from previous series but on ASUS Strix all was perfect ... at the same time I don't want to buy Strix anymore because of some other issues.

 

[EarthDog]

That is exactly what I was alluding to Bart.

 

[Kenrou]

at the same time I don't want to buy Strix anymore because of some other issues.

"other issues"

 

[Woomack]

My personal preferences. Strix required higher voltage to OC and GTX980 were just not overclocking good comparing to competitive high end cards. GTX980 also required special BIOS to pass OCP/OVP which was locking memory at some values. Other thing is that some versions of ASUS software ( mainly early ) were causing instability and one of the versions could damage cards. In GTX780 was about the same. I just decided that I will get something else in next series. Not that ASUS is bad, I just expect something more when I spend $700+ on a graphics card.
Other thing is that standard versions are not enough for me and top series cost too much so I'm not sure if I buy anything.

 

[Kenrou]

Update on Strix 1070 :

http://www.guru3d.com/news-story/asus-rog-strix-geforce-gtx-1070-spotted.html

 

[JrClocker]

Man...this thread is like crack!

I really want to wait for the 1080 Ti...but not sure if I can wait - hehe.

 

[magellan]

hmm I don't know how they get such a high physics score.

I guess the 8 core cpu helps that part of the test a lot lol.

so it looks like besides physics I don't seem to be bottlenecking card

A 4.6 Ghz i7 4670 is bottlenecking the physics score? Hard to believe.

 

[Kenrou]

hmm I don't know how they get such a high physics score.

I guess the 8 core cpu helps that part of the test a lot lol.

6700k 4.7ghz

 

[EarthDog]

A 4.6 Ghz i7 4670 is bottlenecking the physics score? Hard to believe.

It's not a bottleneck per say. The physics tests in these benchmarks respond to more cores/threads. It makes complete sense a 16 thread cpu is womping on a 4 thread.

These are synthetic benchmarks... it doesn't work in the 'real world' that way....hence my statement about it not being a bottleneck.

 

[magellan]

It's not a bottleneck per say. The physics tests in these benchmarks respond to more cores/threads. It makes complete sense a 16 thread cpu is womping on a 4 thread.

These are synthetic benchmarks... it doesn't work in the 'real world' that way....hence my statement about it not being a bottleneck.

Thanks, interesting.

At my work, they seem to be moving away from HPC builds based on more Intel cores and looking at less cores and higher clock rates. They have no plans to use deca-core Intel parts.

BTW, I read Tom's hardware review on the GTX 1070 and your prediction as to its performance being between the 980Ti and 1080 was perfect.

If you have stock market predictions, I'd like to read them.

 

[EarthDog]

Sun shines on a dogs, um, rear every once in a while, ehh?