Geforce GTX1180/2080 Speculation thread

EastCoastHandle · 23 Aug 2018 at 00:29

Something is not making sense.
Is it 64 CC per cluster or 128 CC per cluster?

Based on what's leaked so far:
Because 72 SM is a pretty high number for just 4608 CC's.
RTX Titan = 4608 CC and 72 SM = 4352/72 = 64 CC per cluster
RTX Titan = 4608 CC and 36 SM = 4352/36 = 128 CC per cluster <---that's if its 36 SM and not 72 SM

While the 2080 and 2070 seem to fix the traditional 128 CC per cluster

2080 = 2944 CC and 23 SM = 2944/23 = 128 CC per cluster
2070 = 2304 CC and 18 SM = 2304/18 = 128 CC per cluster

AthlonXP1800 · 23 Aug 2018 at 00:31

Interesting comment from HardwareCanucks.

All people posted negative posts on OCUK forum about Turing will eat their negative posts on 14 Sept.

Kelt · 23 Aug 2018 at 00:33

AthlonXP1800 said:
Interesting comment from HardwareCanucks.

All people posted negative posts on OCUK forum about Turing will eat their negative posts on 14 Sept.

Only thing I've commented on negatively about the new Nvidia range is the extreme pricing, and I don't think I'll be eating my words anytime soon on that.

EastCoastHandle · 23 Aug 2018 at 00:36

I would take his statement with a pinch of salt. Besides, his post would have to be approved by nvidia if he's under NDA. Typically, you wouldn't be allowed to say that much. So again, pinch of salt.

Lokken86 · 23 Aug 2018 at 00:37

HACO said:
Remember this. At the time of release, a 980 Ti and 1080 were almost identical in games.

Um no. The 1080 was miles faster.

ltron · 23 Aug 2018 at 00:41

I think people have misunderstood DLSS in those graphs, the 1080 is not doing it and it's not capable of it as it doesn't have tensor cores; that is not why the 2080 gets a speed boost. DLSS is using AI to teach the GPU to render more efficiently with improving anti aliasing being the icing on the cake. This is revolutionary and my favourite thing about RTX, see: https://blogs.nvidia.com/blog/2018/08/22/geforce-rtx-60-fps-4k-hdr-games/?linkId=100000003301637

CAT-THE-FIFTH · 23 Aug 2018 at 00:44

Kelt said:
Only thing I've commented on negatively about the new Nvidia range is the extreme pricing, and I don't think I'll be eating my words anytime soon on that.

So let me get this straight a reviewer is threatening people,ie,gamers who are annoyed about the price increase - even if its 50% faster for 50% more money price/performance has stagnated.

Edit!!

Oh wait he made this video:

https://www.youtube.com/watch?v=162DeB3xbqk

:rolleyes:

tommybhoy · 23 Aug 2018 at 00:53

AthlonXP1800 said:
Interesting comment from HardwareCanucks.

melmac · 23 Aug 2018 at 00:55

AthlonXP1800 said:
Interesting comment from HardwareCanucks.

All people posted negative posts on OCUK forum about Turing will eat their negative posts on 14 Sept.

Actually, no they won't. There will be some of course, but, the problem most people have with the new cards is the price. I know I won't be eating my words, doesn't matter what the performance is like, the price jump from Pascal is a joke.

HACO · 23 Aug 2018 at 00:56

Lokken86 said:
Um no. The 1080 was miles faster.

It wasn't. In fact it was slower in many games and only faster in a handful of them. After 3 months of driver enhancements, it became consistently faster. Just go back to forum posts and YouTube comparisons from the time of release to see this.

EastCoastHandle · 23 Aug 2018 at 00:57

Looks like we have a nvidia cheerleader.

Lokken86 · 23 Aug 2018 at 00:58

tommybhoy said:

Sure it will be.

Lokken86 · 23 Aug 2018 at 00:59

melmac said:
Actually, no they won't. There will be some of course, but, the problem most people have with the new cards is the price. I know I won't be eating my words, doesn't matter what the performance is like, the price jump from Pascal is a joke.

Links to any reviews of 1080 being slower than 980ti?

CAT-THE-FIFTH · 23 Aug 2018 at 01:00

Wait he saw a briefing from Nvidia and didn't test the cards??

So how is he allowed to break NDA unless Nvidia said OK?

CAT-THE-FIFTH · 23 Aug 2018 at 01:04

Lokken86 said:
Links to any reviews of 1080 being slower than 980ti?

Guy is talking about the RTX2070 not the RTX2080.

Also how does that tie in with the oficial Nvidia figures:

https://hexus.net/media/uploaded/2018/8/f4bc9c91-8e4c-4007-8dc6-882852031513.png

The RTX2080 is meant to be upto 50% faster overall than a GTX1080. The GTX1080TI is 30% to 40% faster than a GTX1080. So the 10% leaked figure past a GTX1080TI fits what you say.

Now apparently the RTX2070 is also faster,which means the RTX2080 is barely faster than a RTX2070??

Does that sound right to you??

ltron · 23 Aug 2018 at 01:09

Lokken86 said:
Links to any reviews of 1080 being slower than 980ti?

I think he means the 1070.

tommybhoy · 23 Aug 2018 at 01:11

CAT-THE-FIFTH said:
So how is he allowed to break NDA unless Nvidia said OK?

Should clarify that screenshot was taken a few hours after the presentation the other day which I imagine precedes Athlons HWC other tweet above.

FoxEye · 23 Aug 2018 at 01:17

CAT-THE-FIFTH said:
Now apparently the RTX2070 is also faster,which means the RTX2080 is barely faster than a RTX2070??

Does that sound right to you??

I don't think anyone should be expecting 2070 to be faster than the 1080 Ti.

Going to be a long three weeks waiting for benchmarks.

Guess nV are using that time to find out how many customers will throw money at them regardless of performance

Wrinkly · 23 Aug 2018 at 01:18

HACO said:
Remember this. At the time of release, a 980 Ti and 1080 were almost identical in games. Now two years later, 1080 is often a good 25-30% ahead. This is always the case. The gap between current and last gen keeps on widening as time goes by and newer drives and titles come.

I thought an overclocked 980Ti was around the same performance as an overclocked 1070 on release. Too lazy to check, but I think the same still stands.

EastCoastHandle · 23 Aug 2018 at 01:19

Here is something to help us keep tabs on the mArch for the past 2 Generation including this upcoming release Turing...