Something super is coming...

[Rofflay]

They make a good difference to the 1% lows i noticed Bit tech doing the reviews 1080 and 2080s at 4k were like 32min where as the 2080ti was around 45 up to 60 sometimes almost twice the 1080 and always a good bit ahead of the super.

Especially the 1% low at 4k there they show most just saying.

 

[Spyhop]

How does the 2560 stream processors of the 2070S outpace/match the 3584 of the 1080Ti? That's 1024 extra processors.

Loads of big brains being paid six-figure salaries by leather jacket man to make it happen

 

[willhub]

Loads of big brains being paid six-figure salaries by leather jacket man to make it happen

Are you saying they're gimping the 1024 processors?

 

[TNA]

Are you saying they're gimping the 1024 processors?

He is saying they are getting people to part with their money and upgrade from a 1080Ti to a 2070S. Lol

 

[Rroff]

How does the 2560 stream processors of the 2070S outpace/match the 3584 of the 1080Ti? That's 1024 extra processors.

Just because they are all CUDA cores doesn't mean they are architecturally identical - the SMs for Turing are quite a bit different to Pascal able to do concurrent execution in ways that would have become serial operations on Pascal, upto 50% improvement in shading throughput, etc. etc.

 

[willhub]

Just because they are all CUDA cores doesn't mean they are architecturally identical - the SMs for Turing are quite a bit different to Pascal able to do concurrent execution in ways that would have become serial operations on Pascal, upto 50% improvement in shading throughput, etc. etc.

So 2560 Turing = 3840 Pascal, which is why the 2070S is equal/faster than a 1080Ti?

He is saying they are getting people to part with their money and upgrade from a 1080Ti to a 2070S. Lol

No regrets lol, my rig is a babe now.

 

[Rroff]

So 2560 Turing = 3840 Pascal, which is why the 2070S is equal/faster than a 1080Ti?

It isn't really as simple as that - Turing at a 1:1 level will be faster than Pascal for some things and slower than others but overall there are lots of little architecture improvements in Turing - like Pascal was from Maxwell they've significantly improved the memory implementation with better cache efficiency and throughput which helps as well.