NVIDIA GeForce RTX 2060 Founders Edition 6 GB

[M2B]

They changed up the core so the core count doesn't really say much does it The perf/watt of a GTX 1080 is 1% better than that of the 2060. Turing is just Pascal with RT and some minor shader changes to make it fit.

So in that sense, AMD is looking at the same situation as they did with Pascal.

The main difference between turing and pascal is caching. seems like modern games do benefit more from better caching and lower latencies than older games, I think. I'm not sure about that but seeing the RTX 2080 performing better against 1080Ti in newer games (which use way more post processing effects) might be a proof.

 

[Vayra86]

This is pretty odd @W1zzard

And there is a trend, too. On Anandtech I see the 1080 and sometimes the Vega64 put a 10% perf gap between the 2060 FE. And that is even with the 2060 FE doing 93.5 versus 89.1 on TPU.

One other point to make; this is why I feel TPU needs to be adding minimum FPS / 99th percentile results. The 2060 falls apart at 4K, this is where you see the lacking VRAM department at work. TPU's review completely misses out on that info.

The the exception that the 2080ti is a little further ahead of the 1080ti.

I honestly can't consider a 1200 dollar MSRP card a relevant part of the product stack. Its priced well out of the comfort zone.

 

[moproblems99]

So in that sense, AMD is looking at the same situation as they did with Pascal.

The the exception that the 2080ti is a little further ahead of the 1080ti.

 

[jabbadap]

They changed up the core so the core count doesn't really say much does it The perf/watt of a GTX 1080 is 1% better than that of the 2060. Turing is just Pascal with RT and some minor shader changes to make it fit.

So in that sense, AMD is looking at the same situation as they did with Pascal.

Well in sense of fp32 shader performance improvements one could use perf/Tflops metrics. I.E.
RTX 2060: 100%/(2*1920*1.865) ~ 0.01396 -> 100%
GTX 1080: 102%/(2*2560*1.783) ~ 0.01117 -> 100% * 0.01117/0.01396 = 80%

Of course it not all from shader, but between gtx 1080 and RTX 2060 impact of different memory configurations are quite minor.

 

[moproblems99]

I honestly can't consider a 1200 dollar MSRP card a relevant part of the product stack. Its priced well out of the comfort zone.

I don't disagree with you but for some, if you aren't the best your not relevant.

 

[jabbadap]

This is pretty odd @W1zzard

And there is a trend, too. On Anandtech I see the 1080 and sometimes the Vega64 put a 10% perf gap between the 2060 FE. And that is even with the 2060 FE doing 93.5 versus 89.1 on TPU.

One other point to make; this is why I feel TPU needs to be adding minimum FPS / 99th percentile results. The 2060 falls apart at 4K, this is where you see the lacking VRAM department at work. TPU's review completely misses out on that info.

I honestly can't consider a 1200 dollar MSRP card a relevant part of the product stack. Its priced well out of the comfort zone.

I would say anandtech has some odd numbers there on RTX 2060 at shadow of war. Gap between RTX 2060 and RTX 2070 is really huge. On Pcper's review is quite in line with tpu, but obviously fps differs a lot between review to review.

 

[bug]

Sure it has, but in this case there is no competition at all. i hope we dont see an era similar to the cpus prior-to-ryzen release era

Sure there is. It's not in the same league, but that doesn't mean there's nothing to compare against

 

[Vayra86]

I would say anandtech has some odd numbers there on RTX 2060 at shadow of war. Gap between RTX 2060 and RTX 2070 is really huge. On Pcper's review is quite in line with tpu, but obviously fps differs a lot between review to review.

Yes, the numbers are different across the whole thing (slightly higher FPS across the board @ Anandtech), but the % gaps are not. The 2060 is very close to TPU's results (only 4 FPS more), while Vega 64 scores 13 FPS over the TPU result. And the 1080 is also higher up the tree than it is on TPUs results - consistently - across all games tested. I mostly looked at the cards the 2060 competes against - because the whole argument for this card is that it pushes the price of that performance level down. But Anandtech's review actually shows a very different picture, with a notable performance gap that doesn't favor the 2060. Additionally, going off the TPU 4K numbers you might think the card is very consistent at pushing 4K compared to 1440p and 1080p, but the polar opposite is true - its min FPS can only compete with a 6GB 980ti and a GDDR5 driven 1070 - and that is when considering the slightly higher numbers in the Anandtech review.

The 2070 then - compare it to Vega 64. On TPU: 91.6 > 104 (12,4 FPS) On Anandtech: 104 > 116.9 (12.9 FPS). Seems legit in terms of relative performance. As does the gap between 2070 and 2080.

Something is wrong with those 2060 numbers and I am inclined to believe Anandtech for accuracy here. And only because of its consistency.

 

[Easy Rhino]

I will probably upgrade from the gtx970 SC next Spring to this card. Although I don't like paying more than $300 for a GPU. Afterall most of the time I am playing games like Rimworld, Prison Architect, Astroneer, etc.

 

[W1zzard]

On Anandtech

Do they start their benches with cold cards (causing some to boost high for 30 seconds and then drop clocks) ? Or maybe using a non-reference Vega?

 

[cucker tarlson]

Yes, the numbers are different across the whole thing (slightly higher FPS across the board @ Anandtech), but the % gaps are not. The 2060 is very close to TPU's results (only 4 FPS more), while Vega 64 scores 13 FPS over the TPU result. And the 1080 is also higher up the tree than it is on TPUs results - consistently - across all games tested. I mostly looked at the cards the 2060 competes against - because the whole argument for this card is that it pushes the price of that performance level down. But Anandtech's review actually shows a very different picture, with a notable performance gap that doesn't favor the 2060. Additionally, going off the TPU 4K numbers you might think the card is very consistent at pushing 4K compared to 1440p and 1080p, but the polar opposite is true - its min FPS can only compete with a 6GB 980ti and a GDDR5 driven 1070 - and that is when considering the slightly higher numbers in the Anandtech review.

The 2070 then - compare it to Vega 64. On TPU: 91.6 > 104 (12,4 FPS) On Anandtech: 104 > 116.9 (12.9 FPS). Seems legit in terms of relative performance. As does the gap between 2070 and 2080.

Something is wrong with those 2060 numbers and I am inclined to believe Anandtech for accuracy here. And only because of its consistency.

it may be about other cards. what vega are they using ? and what fan profile ?

here's a ppc.pl review,they're a trustworthy source

https://www.purepc.pl/karty_graficzne/test_geforce_rtx_2060_nastepca_gtx_1060_dogania_gtx_1080

@1440p 2060 beats 1080 narrowly in 6 games,matches in 1 and in 4 1080 beats it narrowly. 2060 is a faster card,though only by a whisker.
This is not something we just imagined, the smallest rtx card kicks ass.

I will probably upgrade from the gtx970 SC next Spring to this card. Although I don't like paying more than $300 for a GPU. Afterall most of the time I am playing games like Rimworld, Prison Architect, Astroneer, etc.

maybe there's 1160 coming,who knows.

 

[eltano06]

And get lower performance. Crazy right?

18% faster, 50+% more expensive.

 

[Vayra86]

Do they start their benches with cold cards (causing some to boost high for 30 seconds and then drop clocks) ? Or maybe using a non-reference Vega?

I don't know, honestly, but we do know you have older results in your data. You also have a faster (CPU side) test setup with a whopping 500mhz gap and marginally faster RAM - while your results are generally lower FPS (and these were 1080p results). The combination of things doesn't really make a whole lot of sense to explain this difference. Pascal being warm versus cold also wouldn't really explain this amount of a difference - you're talking about three to five boost bins and only briefly in that sense. Driver wise the situation is near identical, at least on the Nvidia side. And I really doubt quad channel RAM has this impact.

Don't get me wrong, not questioning your being faithful about the results, just curious what could cause such a gap.

 

[N3M3515]

The legend says there was a time you could buy an new gen x60 card at $200 -$250 and get previous gen x80 performance...

 

[Fluffmeister]

The legend says there was a time you could buy an new gen x60 card at $200 -$250 and get previous gen x80 performance...

The legend also said AMD were once competitive too. Mythical times of horses and knights, King Vega was going to rule, but he died a horrible painful death.

 

[moproblems99]

The legend also said AMD were once competitive too. Mythical times of horses and knights, King Vega was going to rule, but he died a horrible painful death.

Why can't your posts always be like this?

 

[Fluffmeister]

Why can't your posts always be like this?

I'm drinking tonight (long story), so I always get more whimsical.

 

[moproblems99]

I'm drinking tonight (long story), so I always get more whimsical.

Hopefully your drinking for positive reasons - and not alone!

 

[bug]

I don't know, honestly, but we do know you have older results in your data. You also have a faster (CPU side) test setup with a whopping 500mhz gap and marginally faster RAM - while your results are generally lower FPS (and these were 1080p results). The combination of things doesn't really make a whole lot of sense to explain this difference. Pascal being warm versus cold also wouldn't really explain this amount of a difference - you're talking about three to five boost bins and only briefly in that sense. Driver wise the situation is near identical, at least on the Nvidia side. And I really doubt quad channel RAM has this impact.

Don't get me wrong, not questioning your being faithful about the results, just curious what could cause such a gap.

Are we talking standard benchmarks here? Because if each/one site benchmarks their own game play session, that would explain a lot.

 

[Palladium]

Man, I can't remember the last time I was this bored with PC hardware. The CPU core wars are nice and all but both camps are now already deep in the overkill territory for games, and this new 2060 is barely better than my 2.5 year old 1070 with the only real perk that it costs $100 less than the inflated launch price I paid for my 1070 at.

 

[moproblems99]

this new 2060 is barely better than my 2.5 year old 1070 with the only real perk that it costs $100 less than the inflated launch price I paid for my 1070

Luckily, the 2070 is the natural successor and not the 2060.

 

[chr0nos]

This is pretty odd @W1zzard

And there is a trend, too. On Anandtech I see the 1080 and sometimes the Vega64 put a 10% perf gap between the 2060 FE. And that is even with the 2060 FE doing 93.5 versus 89.1 on TPU.

One other point to make; this is why I feel TPU needs to be adding minimum FPS / 99th percentile results. The 2060 falls apart at 4K, this is where you see the lacking VRAM department at work. TPU's review completely misses out on that info.

I honestly can't consider a 1200 dollar MSRP card a relevant part of the product stack. Its priced well out of the comfort zone.

Drivers on AMD side: Catalyst 18.8.2 WHQL are just plain OLD and it was brought up some time ago but dismissed as nonsense, Anand uses latest it seems.

 

[bajs11]

it seems a lot of people have bad memory and totally forgot that the GTX 1060 with its 250 usd msrp outperformed the GTX 980:
we are talking about a new generation of cards released two and a half years later and costs 40% more than its predecessor

 

[Shatun_Bear]

The main difference between turing and pascal is caching. seems like modern games do benefit more from better caching and lower latencies than older games, I think. I'm not sure about that but seeing the RTX 2080 performing better against 1080Ti in newer games (which use way more post processing effects) might be a proof.

I don't think that has anything to do with it. Simply, Nvidia is prioritising game optimisations for their latest gen of cards like they have always done. Same thing happened between Maxwell and Pascal (I recall the gap between the 1070 and 980 Ti was minimal with older games but got larger as newer games were released and were more optimised for Pascal).

 

[londiste]

it seems a lot of people have bad memory and totally forgot that the GTX 1060 with its 250 usd msrp outperformed the GTX 980:
we are talking about a new generation of cards released two and a half years later and costs 40% more than its predecessor

View attachment 114296

You are quibbling about a couple %? RTX2060 is 1-2% slower than GTX1080.
https://www.techpowerup.com/reviews/NVIDIA/GeForce_RTX_2060_Founders_Edition/33.html

I don't think that has anything to do with it. Simply, Nvidia is prioritising game optimisations for their latest gen of cards like they have always done. Same thing happened between Maxwell and Pascal (I recall the gap between the 1070 and 980 Ti was minimal with older games but got larger as newer games were released and were more optimised for Pascal).

Not so much Nvidia but games themselves are increasingly optimized for newer features as time passes.