Likely RTX 2080 3DMark Time Spy Score

And the 1080 ti looks to not be OCed to the (fairly standard) 2GHz.

/yawn

Looks like Nvidia didn't bring lube this time. At least the 10 series has the 1070 trade blows with the 980 ti....for less.
 
From reddit:


"So it takes a 300Mhz overclock to actually beat a 1080ti if this is real?

Why is it an image an not a link? Really not good for Turing if this is true.

EDIT: TimeSpy Result: https://www.3dmark.com/spy/4293425

EDIT: FireStrike Ultra Result: https://www.3dmark.com/fs/16227476

EDIT: Ealier FSU Result: https://www.3dmark.com/fs/16166250

EDIT: FSU Score is comparable to that of a GTX 1050 or a GTX 1050 Ti. I do not believe these results are legit."

The 3DMark user is r_d_yang - several Yangs working at NVIDIA including a Sr ASIC Engineer (although it is a very common last name):

https://www.linkedin.com/search/results/all/?keywords=yang nvidia&origin=GLOBAL_SEARCH_HEADER
 
Last edited:
A 300 MHz overclock to make the 2080 beat a 1080Ti would align perfectly with the shader math assuming zero IPC improvement over Pascal.

1080Ti at stock boost clocks = 11.3 TFLOPS
2080 at FE stock boost clocks = 10.6 TFLOPS
2080 at 2025 MHz = 11.9 TFLOPS

That would align perfectly with the Time Spy result and clock speed.

I'm not saying it's real, I'm just saying what the arithmetic equates to.
 
Isn’t that pretty much where the nVidia charts put it? So 2080ti should be a little more than 50% above a 1080ti. That’s awesome.

Do I want to order one or two... I have to keep reminding myself multiGPU sucks.
 
The score is just going to make the 1080ti look good until it runs out of stock if that ever happens.
If the gfx score was 11,000-12,000 I would consider getting a 2080 but not for that slight bump.
 
The score is just going to make the 1080ti look good until it runs out of stock if that ever happens.
If the gfx score was 11,000-12,000 I would consider getting a 2080 but not for that slight bump.

Yeah, I mean - already today if you look at the prices it is about a $200 jump to go from 1080 Ti to 2080 - so people have choice (and the used market) if they don't want the "whole package" the RTX 2080 brings versus the 1080 Ti.
 
The score is just going to make the 1080ti look good until it runs out of stock if that ever happens.
If the gfx score was 11,000-12,000 I would consider getting a 2080 but not for that slight bump.

The 1080 was the same way compared to the 980ti. The 2080 is probably better off tbh.

This and the nVidia charts all suggest the 2080ti being 50% over the 1080ti in traditonal games ignoring the new features. That’s better than I think anyone imagined. These things will sell like hotcakes if the reviews match up.
 
The 1080 was the same way compared to the 980ti. The 2080 is probably better off tbh.

This and the nVidia charts all suggest the 2080ti being 50% over the 1080ti in traditonal games ignoring the new features. That’s better than I think anyone imagined. These things will sell like hotcakes if the reviews match up.

What nvidia charts? The only chart I saw was the non-ti 1080 vs. non-ti 2080.
 
RTX series are a flop.


How is the also yet not shipping new AMD Vega RX, is it a flop too, or just the unshipped/fully reviewed RTX line?

I'm wondering where the current Vega 64 will match up (2060 or 2070 or ???) to the RTX line myself, and if AMD will drop pricing to be competitive to it - also, there was talk of RTX Titan to follow at $3k price point and being considerably stronger than the Ti this time, will it matter even if that's so at that price.

On the topic of pricing, I believe nVidia looked at the Vega 64, priced the nearest RTX in-line with the Vega 64, and then market bumped each series up from there using their nearest competition as a pricing benchmark realizing they can (for now) charge more based upon the competitive products and their pricing.
 
What nvidia charts? The only chart I saw was the non-ti 1080 vs. non-ti 2080.

Those charts. We know the specs of both cards it’s straightforward math if we have a datapoint from each architecture.

We still don’t have a solid non biased datapoint for Turing though. But from the charts and this both suggest the 2080ti will be about 50% faster than a 1080ti.
 
What if that's a 2070 ?
What I was thinking. People will lose their minds if that's true. No where does it state definitively that this is a 2080. Granted, stock Boost clock is higher on the 2080, but we've been told that Turing has more overclocking headroom than Pascal.
 
These results are expected considering the public specs, nothing surpising. Real performance increase will come with 2080 Ti, but with a huge cost.
 
Interesting. Don't cancel that pre-order just yet, but keep your finger on the button.

Yeah, the only reason I pre-ordered a 2080 was because I got an EVGA so I could step up to the 2080Ti within 90 days since they aren't in stock right now. I wouldn't buy another brand as an upgrade for my 1080Ti.
 
A 300 MHz overclock to make the 2080 beat a 1080Ti would align perfectly with the shader math assuming zero IPC improvement over Pascal.

1080Ti at stock boost clocks = 11.3 TFLOPS
2080 at FE stock boost clocks = 10.6 TFLOPS
2080 at 2025 MHz = 11.9 TFLOPS

That would align perfectly with the Time Spy result and clock speed.

I'm not saying it's real, I'm just saying what the arithmetic equates to.

FWIW, here's my 1080 Ti at 2025 MHz core and 12GHz on the GDDR5X side by side with that timespy result (my nvidia driver was from back in April, to be fair):

https://www.3dmark.com/compare/spy/4293425/spy/3628077
 
The 1080 was the same way compared to the 980ti. The 2080 is probably better off tbh.

Show me a stock 1080 only posting a 5% better timespy graphics score than a 980 ti. In games, the 1070 went neck-and-neck with the 980 ti, the 1080 blew it out of the water. [H]'s recent mutli-generational testing showed the 1080 sitting with an over 30% average increase of the 980 ti. So, no, the 1080 was not "the same way" compared to the 980 ti.
 
Show me a stock 1080 only posting a 5% better timespy graphics score than a 980 ti. In games, the 1070 went neck-and-neck with the 980 ti, the 1080 blew it out of the water. [H]'s recent mutli-generational testing showed the 1080 sitting with an over 30% average increase of the 980 ti. So, no, the 1080 was not "the same way" compared to the 980 ti.

980ti OC’d way better. In some games only a few points apart.



I remember all the people complaining then, too. It’s like ground hog’s day.
 
Different architectures to get to the same point. RTX is new tech, it brings it's focus on Ray Tracing. 3dmark isn't even that relevant anymore to be honest ...I wish it was (I miss me some 2001 SE)...

6 months or one year from now a killer app may show a much more distinguishable difference.
 
980ti OC’d way better. In some games only a few points apart.



I remember all the people complaining then, too. It’s like ground hog’s day.


that's not what he asked. Stock Vs Stock.

And people were complaining because the 1080 FE was a ridiculous price and people could see that it was only going to lead to further price rises in the future.
 
that's not what he asked. Stock Vs Stock.

And people were complaining because the 1080 FE was a ridiculous price and people could see that it was only going to lead to further price rises in the future.

nVidia is priced (sadly) in-line (or better than) with AMD... and I'd bet that they have the mid model RTX priced in-line with the Vega 64 since now nVidia's mid-line aligns performance with Vega 64 - although it'd be awesome if nVidia priced they top of the line to AMD's top gaming card, its not going happen, they are simply pricing based off their competitors in the market place.

There's only one way to proof them wrong right now, and that's by limited sales of their products, we all know that's not going to happen.
 
that's not what he asked. Stock Vs Stock.

And people were complaining because the 1080 FE was a ridiculous price and people could see that it was only going to lead to further price rises in the future.

This is [H].... we are men around here and compare OC vs OC. Overclocking is in the site name.

I was talking more about most didn’t upgrade from the 980ti to the “new king” 1080.
 
well, the scores aren't to far off the Tflops ratings..

a 1080 gtx is 9 TFLOPs
a 1080 ti is 11.3 TFLOPs
a 2080 is 10.1 TFLOPs (according to anandtech)
a 2080 ti is 13.4 TFLOPs (according to anandtech)

so.. i would guess that a 2080ti is approx 20-25% faster in old non-RT games.. i really hoped for more - so please surprise me nVidia!!!

but it sure would explain why Jensen thinks that we need to look at performance in a new way..
 
well, the scores aren't to far off the Tflops ratings..

a 1080 gtx is 9 TFLOPs
a 1080 ti is 11.3 TFLOPs
a 2080 is 10.1 TFLOPs (according to anandtech)
a 2080 ti is 13.4 TFLOPs (according to anandtech)

so.. i would guess that a 2080ti is approx 20-25% faster in old non-RT games.. i really hoped for more - so please surprise me nVidia!!!

but it sure would explain why Jensen thinks that we need to look at performance in a new way..

Yeah from the nVidia charts and the questionable TimeSpy leaks they pulled ~15% from someplace, which would be great because the 2080ti would be about 50% above the 1080ti. That’s a lot more palatable to me than 30%. If it’s only in the 30% range I am still not certain which card to buy.... if any.
 
980ti OC’d way better. In some games only a few points apart.



I remember all the people complaining then, too. It’s like ground hog’s day.


Not going to do the math on everything but from rough estimates it looks the 980 ti OC's is under 5% from a stock 1080. With both OC'd that brings the difference up to around 15%. With custom coolers I imagine the 1080 ends up being a little closer to 20% better and the 1070 probably catches up to the 980 ti (or, at least, gets close). In games that make heavy use of VRAM the 8GB on both cards will put them over the 980 ti. The OC to OC difference is still 200% more than the 5% difference shown by these. That said, these could every well be on non-final drivers so that difference could change.

And yes, people complained. Because prices went up and people worried that it would mean they would go up more. Guess what, they went up more and people are not happy. Unless these cards show Pascal level improvements (highly unlikely) then people are going to be, justifiably, upset at the price increases.
 
Not going to do the math on everything but from rough estimates it looks the 980 ti OC's is under 5% from a stock 1080. With both OC'd that brings the difference up to around 15%. With custom coolers I imagine the 1080 ends up being a little closer to 20% better and the 1070 probably catches up to the 980 ti (or, at least, gets close). In games that make heavy use of VRAM the 8GB on both cards will put them over the 980 ti. The OC to OC difference is still 200% more than the 5% difference shown by these. That said, these could every well be on non-final drivers so that difference could change.

And yes, people complained. Because prices went up and people worried that it would mean they would go up more. Guess what, they went up more and people are not happy. Unless these cards show Pascal level improvements (highly unlikely) then people are going to be, justifiably, upset at the price increases.

Even if they do show Pascal level improvements people are going to be upset at the price increase because their pricing is absurd.
 
  • Like
Reactions: Savoy
like this
I kinda wonder if Nvidia is sandbagging a bit till Amd, or other alternative company, puts something out to challenge, then bam! Why go all out when you can dole it out piecemeal to stay on top? Plus we have all seen improvements on the driver front.

I think it’s part of the marketing. You know, like the crack cocains. Once you hooked.... Just a hypothesis. Surely major tech companies wouldn’t take advantage of us scmucks and squeeze us for what it’s worth.
 
I kinda wonder if Nvidia is sandbagging a bit till Amd, or other alternative company, puts something out to challenge, then bam! Why go all out when you can dole it out piecemeal to stay on top? Plus we have all seen improvements on the driver front.

I think it’s part of the marketing. You know, like the crack cocains. Once you hooked.... Just a hypothesis. Surely major tech companies wouldn’t take advantage of us scmucks and squeeze us for what it’s worth.

Naw at this point, Nvidia is going against themselves. AMD brings nothing against Nvidia right now. Even when you take pricing into account AMD's Vega 64 is overpriced so badly. You can get a new 1080 GTX for about $200 cheaper then Vega 64, for the same performance.
 
Different architectures to get to the same point. RTX is new tech, it brings it's focus on Ray Tracing. 3dmark isn't even that relevant anymore to be honest ...I wish it was (I miss me some 2001 SE)...

6 months or one year from now a killer app may show a much more distinguishable difference.

They said that with bulldozer and vega.
 
They said that with bulldozer and vega.

And 290x... and the Fury X.... and #waitfornavi

The difference is nVidia does their homework and gave up a 50% boost in CUDA cores to go with this tech. It’s such an interesting launch. Normally it’s pretty boring once you know the specs.
 
We've been through this before. The cards may end up being similiar in fps and performance due to different ways of getting there and doing it, but this one has Ray Tracing.
 
We've been through this before. The cards may end up being similiar in fps and performance due to different ways of getting there and doing it, but this one has Ray Tracing.

I love ray tracing tech, but it really isn't a thing for this card. It's not going to be powerful enough to really push RT stuff and it'll be 4-5 years (at least) before it's a big thing across the board for games. By the time ray tracing tech really matters the 2000 line will be woefully underpowered. I like that Nvidia is getting the ball rolling, but don't buy into the hype to heavily just yet.
 
I haven't bought in it's a great start but will take time to grow. As I've said, we've been down similiar roads with past cards.
 
Back
Top