Nvidia ,Rtx2080ti,2080,2070, information thread. Reviews and prices September 14.

[psolord]

Actually, that's what DLSS is for. It lets you play at 480p and still have it look halfway decent on your 4k monitor. Too bad BFV doesn't support it.

That sounds great.

Shouldn't Nvidia had implemented this in their drivers or is it something that cannot be force selected by the user?

 

[LTC8K6]

These are just people saying what they saw in 5 games. Or what they heard.

where are the benchmarks?

where are the reviews?

no one has posted anything beyond some person in some article saying what they saw (or, really, what they heard.).

Lots of people claimed the 100fps was with 2 GPUs in SLI, the author confirmed it was a single GPU.
Unless some tech writers were fooled, Turin is much faster than Pascal all around, not just in RT.

 

[LTC8K6]

Yes it is.

Pascal went backwards compared to Maxwell. Just look at GP104 compared to GM104. GP104 has 25% more shaders and clocks roughly 45% higher, which results in 80% more raw shading performance. However the actual performance improvement in games was only 60-70%.

Of course this is not caused by the performance of individual shaders in GP104 regressing, but rather by the fact that performance never scales linearly due to Amdahl's law (whether it be due to memory bottlenecks, issues with extracting sufficient parallelism or something else).



The 2080 Ti FE has 21% more shaders than the 1080 TI, and a 0% increase in clock rate (the 1080 Ti boosts to an average of 1636MHz in real life usage). The 2080 Ti FE may of course boost higher than it's advertised boost, but we don't know that yet, and besides it's irrelevant for estimating the baseline improvement.

Based upon what we saw with Pascal and it's less than linear scaling, the baseline is thus somewhere around 15-20% (the same number also applies to the 2080 and 2070). It's of course perfectly possible that Turing ends up much higher than 15-20% faster, but 15-20% is just the baseline.

And that's not guessing, that's based on historical precedent.

GPU BOOST 4 may allow Turin to boost better than Pascal.

 

[antihelten]

Lots of people claimed the 100fps was with 2 GPUs in SLI, the author confirmed it was a single GPU.
Unless some tech writers were fooled, Turin is much faster than Pascal all around, not just in RT.

It doesn't really prove anything, until we know which games they used.

There are plenty of games where a 1080 Ti can manage 80+ FPS at 4K, so if those where games used, that would make a 2080 Ti running at 100 FPS 20-25% faster (which I wouldn't really consider "much faster").

GPU BOOST 4 may allow Turin to boost better than Pascal.

Absolutely, anythings possible, but we're talking about a baseline improvement here. Any number of things could improve Turing above this baseline, but given the lack of info from Nvidia we just don't know yet.

 

[LTC8K6]

It doesn't really prove anything, until we know which games they used.

There are plenty of games where a 1080 Ti can manage 80+ FPS at 4K, so if those where games used, that would make a 2080 Ti running at 100 FPS 20-25% faster (which I wouldn't really consider "much faster").



Absolutely, anythings possible, but we're talking about a baseline improvement here. Any number of things could improve Turing above this baseline, but given the lack of info from Nvidia we just don't know yet.

It was posted to answer the question posters had of 2 gpus vs 1 for the quote, not to prove anything.

And the settings were Ultra at 4K, not just 4K.

"in excess of 100 fps at 4K and Ultra settings" with a single GPU.

 

[LTC8K6]

If we don't see "in excess of 100 fps at 4K and Ultra settings" in reviews, I can follow up with the author and ask what happened.

 

[Malogeek]

Lots of people claimed the 100fps was with 2 GPUs in SLI, the author confirmed it was a single GPU.
Unless some tech writers were fooled, Turin is much faster than Pascal all around, not just in RT.

Facebook click bait ads must love you.

There's no basis of comparison, multiple games to compare. There's just not nearly enough information to go on, that's what reviews are for.

What game was it? What settings were used beyond default to compare identical to Pascal? Is this over a benchmark run or walking around in the game? Was he starting at a sky? Simply going by a statement like that is just useless. It's just click bait crap.

 

[Malogeek]

If we don't see "in excess of 100 fps at 4K and Ultra settings" in reviews, I can follow up with the author and ask what happened.

Unless you're going to use this severe lack of information to make a purchase decision, why bother? Just wait for reviews

 

[LTC8K6]

Facebook click bait ads must love you.

There's no basis of comparison, multiple games to compare. There's just not nearly enough information to go on, that's what reviews are for.

What game was it? What settings were used beyond default to compare identical to Pascal? Is this over a benchmark run or walking around in the game? Was he starting at a sky? Simply going by a statement like that is just useless. It's just click bait crap.

These cards are not for me. Maybe the 2070, but I doubt it. I'll be looking lower on the ladder.

It was multiple games at 4K Ultra according to the author...“We also played a variety of other PC games that shall not be named, and saw performance run in excess of 100 fps at 4K and Ultra settings."
There's no reason yet to believe he was tricked, is there?

 

[rainy]

There's no reason yet to believe he was tricked, is there?

Why not, maybe it was beneficial for him to said that?

 

[Malogeek]

There's no reason yet to believe he was tricked, is there?

I'm saying people should wait for studied empirical evidence of performance to make a judgement for purchase. ie. reviews.

 

[LTC8K6]

I'm saying people should wait for studied empirical evidence of performance to make a judgement for purchase. ie. reviews.

Of course they should. No one said they shouldn't.
You can't purchase anything for a while anyway.
And the 2070 can't be purchased until October.

But you can't claim it was SLI at 4K Ultra when the author says it wasn't, and you can't claim it was one game when the author says it was a variety of games, etc.

The author was there and presumably has better info.

Those claims and others, contrary to the article, and to other articles, deserve to be countered, imo.

If reality fails to live up to the what the reporters said, then they will look foolish and need to answer.

 

[Guru]

Seems like they were running a mix of medium and high settings at 4k to achieve 60-100fps.

 

[crisium]

Are you referring to Tech Radar?

"We also played a variety of other PC games that shall not be named, and saw performance run in excess of 100 fps at 4K and Ultra settings"

Clearly says Ultra and 100+, so not medium-high 60-100.

 

[LTC8K6]

Seems like they were running a mix of medium and high settings at 4k to achieve 60-100fps.

No indication of that as far as I can tell in the non-RT testing.

 

[sze5003]

Saw something on reddit about some press conference today for Nvidia. So I wonder if they will answer any questions regarding the performance. Either way we have a while before AIB cards are actually available so nothing to do but sit and wait anyway for reviews.

I just wish they wouldn't keep the nda until these things ship. Of course they don't want people cancelling their pre orders.

 

[TeknoBug]

The price range is super ridiculous, and even then they'll still fly off the shelves. I'm glad i waited it out and just grabbed a 1070 Ti when the price dropped a few days ago.

 

[Thor86]

Wow, my local seller has the preorders listed, and the 2080Tis are now on par with Titan pricing. Ugh.

 

[LTC8K6]

Saw something on reddit about some press conference today for Nvidia. So I wonder if they will answer any questions regarding the performance. Either way we have a while before AIB cards are actually available so nothing to do but sit and wait anyway for reviews.

I just wish they wouldn't keep the nda until these things ship. Of course they don't want people cancelling their pre orders.

Keeping the pre-orders will be easy. Just show the 2080ti demolishing the 1080ti, without RT involved, which it's supposedly able to do.

 

[sze5003]

This is pretty funny looking : https://www.theverge.com/circuitbreaker/2018/8/22/17767650/galax-geforce-rtx-2080-ti-hof-announced

 

[LTC8K6]

This is pretty funny looking : https://www.theverge.com/circuitbreaker/2018/8/22/17767650/galax-geforce-rtx-2080-ti-hof-announced

Burger King edition...

 

[realibrad]

It would not be so bad if not for that crown. I wonder if they ever run these designs through American firms to get feedback. I can't imagine they do.

 

[Malogeek]

This is pretty funny looking : https://www.theverge.com/circuitbreaker/2018/8/22/17767650/galax-geforce-rtx-2080-ti-hof-announced

That's terrible, who approves these designs?

 

[LTC8K6]

WCCF:

Word has reached us that reviewers returning from Gamescom are going to be able to start testing the new toys straight away. NVIDIA’s reviewer’s briefing is taking place tomorrow where the company is expected to hand over a working driver to reviewers, so it’s likely they could begin running benchmarks no later than this weekend.

 

[crisium]

Hopefully someone has loose lips.