Gossip on 3080 TI - Power to the X

Accessories Other Tech Talk
,
21

I’m getting the 3080ti. Gonna start saving now so I can order at the reveal.

5 Likes
22

Just mortgage the neighbor’s house. :beers::smile::vulcan_salute::sparkles:

8 Likes
23

After 11 months, my RTX2080ti started to die. I’ve recently returned it - hoping for a refund ($2400 AUD) … that I’ll put into a 3080 :slight_smile:

5 Likes
24

I think my 4790k might bottleneck this…

2 Likes
25

Fingers crossed for HDMI 2.1 support. Will probably not buy a new GPU without that feature.

4 Likes
26

And pciX 4 or x…

2 Likes
27

you’d be surprised! I have the same chip since it released and every time a new video card comes out I think, is it time to upgrade my pc? then i look at the benchmarks and realize upgrading my cpu would gain me negligible frames.

I think the recent jump to 4k rendering as ensured the cpu won’t be the bottleneck for some time, but who knows the 3080ti could be so fast it eats 4k for breakfast.

Also comparing alot of people here on the forums and youtubers with pimax, and the same card as me(2080Ti) event those with OC’ed 99900ks arern’t blowing me away with the frame increase.

I think a upgrade to ryzen 4000 series might be in my future, that might finally be the time amd takes the single- threaded throne from intel.

the only games ithink would benefit us from upgrading right now would be sims. They tend to be more heavy on the CPU. Only other thing i can think of is just having more cpu head room in general.

2 Likes
28

good luck with that if you love intel.

2 Likes
29

Done… :wink:

4 Likes
30

Pretty much ditched Intel as likely a new socket/chipset with every new release. Plus the inflated price. Lol.

Since Ryzen 2000 Amd is gaining on Intel in sales. I am curious if those experiencing studder might be in part cpu related.

2 Likes
31

Thanks! (work keeps me busy atm)

3 Likes
32

Those specs are fake…

33

Why do you say that?

1 Like
34

thats a hell of a warranty. Cutting edge gamer is starting to look better each day…lol

35

My biggest question will be:
Will this thing fit in my mobo and/or how would it play with my current i5 7600 OC 4.6ghz??

36

Tmsc 12 nm is indeed 16 nm.
Tmsc 7 nm euv is second generation 7 euv

3070 has for the leak 48 SM and 3072 cores. These are 64 cores x SM
3080 has for the leak 60 SM and 3840 cores . These are 64 cores x SM

If 3080 Ti has 72 SM then cores are 4608 the same as Titan RTX

Then there is the photo of the architecture of Ampere and it’s similar to Turing

This is strange because Nvidia would like to improve the numbers of RT cores with Ampere

Really i don’t know how so many reviewers and magazines could write post on future Nvidia products
based on rumours so bad made

37

I preface this by saying I have no reason to believe or deny this rumour. To me it is just interesting but I can hope there is some truth in it.

Perhaps I’m stupid but I thought 60 is more than 48???

I did not think there was a leak for the 3080ti, just speculation based on the current leaks.

Let’s pretend this rumour is accurate.

And let’s pretend the GPU in the VR benchmark the other day was big Navi.

What we are potentially looking at is big Navi having over 25% more performance than the 2080ti and the 3080 having possibly over 60% more performance than the 2080ti.

These rumours are way to unreliable to put any faith in but one can hope there is some truth in them because that’s a big step up in performance.

Interestingly if these rumours did turn out to be accurate it doesn’t look like AMD has gained anything against Nvidia.

38

2080 makes 60T RTX-Ops
2080 ti makes 78T RTX-Ops

3080 e 3080ti are at least doubling that performance

and this means another architecture, many more cores and RT CORES than the leak

39

Previously you said “This is strange because Nvidia would like to improve the numbers of RT cores with Ampere”

That is why I said 60 is bigger than 48 because in the picture it has 60 RT core while the 2080 only has 48 RT cores.

Unless I’m missing stuff I can’t see enough info to calculate RTX-OPS

RTX-OPS = Peak FP32 performance (in TFLOPS) x 80% + Peak INT32 performance (in TFLOPS) x 28% + Peak Ray Tracing performance (in tera-ops per second) x 40% + Tensor core performance (in TFLOPS) x 20%

I am pretty stupid with this stuff and I’m asking because I’m trying to understand it better. How are you working out performance to be able to calculate RTX-OPS?

2 Likes
40

Going from 48 to 60 + RTX optimization for Ampere you don’t double RTX-OPS.

RTX-OPS is NVIDIA benckmark for ray tracing;

As i have said they want at least double the RTX-OPS performance of their next generation cards