Benchmark leak puts RTX 2070 Max-Q ahead of desktop GTX 1080, Vega 64

[midian182]

Something to look forward to: With CES drawing ever closer, we’re looking forward to the slew of tech products that will be on show. One of these could be the laptop versions of Nvidia’s RTX cards, and according to a new leak, they look pretty impressive, with the Max-Q version of the RTX 2070 offering more performance than a GTX 1080 and the Radeon Vega 64.

The leak comes from the Final Fantasy XV benchmark database, which has contained results from unreleased graphics cards in the past, including the Radeon 590 and Nvidia’s upcoming RTX 2060.

Leaker TUM APISAK spotted the entry. Assuming it’s legitimate, the RTX 2070 Max-Q’s score of 3,080 puts it ahead of the desktop version of the GTX 1080 (2,892) and AMD’s Radeon Vega 64 (2,895). The RTX 2060, for comparison, scored 2,589.

Nvidia revealed its Max-Q initiative last year. The design philosophy focuses on efficiency, allowing the creation of ultra-thin gaming laptops. It’ll certainly be interesting to see if the mobile RTX 2070 does make an appearance in a few weeks, though the RTX 2080 version is rumored to get its unveiling later in 2019. Both these cards could offer the kind of performance never seen before in thin laptops.

In other Nvidia news, the company recently showed off its mesh shading technology for Turing graphics cards in a new video for developers. The firm says it can improve performance when rendering a large number of objects in games. Check it out in the clip below.

Permalink to story.

https://www.techspot.com/news/77997-benchmark-leak-puts-rtx-2070-max-q-ahead.html

 

[gigantor21]

...the Max-Q variant is just as fast as the full fat desktop version? Considering the performance delta between the Max-Q and AIB versions of Pascal, that seems highly dubious.

 

[QuantumPhysics]

As if there was any doubt that the 2000 series was gonna walk all over AMD's best?

And if Nvidia returns next year with a 3000 series, AMD will be so far behind they'll need to create another cryptocurrency craze.

 

[Namtrooper]

We are so blessed.

 

[Evernessince]

As if there was any doubt that the 2000 series was gonna walk all over AMD's best?

And if Nvidia returns next year with a 3000 series, AMD will be so far behind they'll need to create another cryptocurrency craze.

How about ya take a step back there and look at current max-q products on the market before jumping to conclusions. Given that Max-Q products are designed for thin form factors and that Turning isn't particularly good power consumption wise, I'd be surprised if the 2070 was more then 15% faster then the 1070.

This is only a single benchmark as well and it's one of the worst one's at that.

 

[joe nodden]

As if there was any doubt that the 2000 series was gonna walk all over AMD's best?

And if Nvidia returns next year with a 3000 series, AMD will be so far behind they'll need to create another cryptocurrency craze.

Navi.

 

[JaredTheDragon]

I don't understand the video about mesh shading. We already had all of that (dynamic LOD) like ten years ago, even longer ago in Maya, and those asteroids don't even look as good as Everspace already did two years ago. Evidently nobody at Nvidia got the note, you know, that Everspace already existed. That tech demo looks like crap to me.

 

[BigBoomBoom]

...the Max-Q variant is just as fast as the full fat desktop version? Considering the performance delta between the Max-Q and AIB versions of Pascal, that seems highly dubious.

GTX 1070 Max-Q and 1060 Max-Q were not that different from desktop AIB. It's the 1080 Max-Q that was disappointed (GTX 1070 level) due to the high thermal of the card when forced into lightweight, super thin laptop.

RTX 2070 however is more energy efficient so it can be much easier to reduce voltage and still keep performance.

 

[Manuel Diego]

"The RTX 2060, for comparison, scored 2,589." and the 1070 2750: it had been a while since an x60 couldn't outdo an (x-1)70. Is RTX 2xxx series Nvidia's biggest blunder? Should they have waited to introduce the gigarays?

 

[gigantor21]

GTX 1070 Max-Q and 1060 Max-Q were not that different from desktop AIB. It's the 1080 Max-Q that was disappointed (GTX 1070 level) due to the high thermal of the card when forced into lightweight, super thin laptop.

RTX 2070 however is more energy efficient so it can be much easier to reduce voltage and still keep performance.

In my personal use, the difference between the mobile (non-MaxQ) 1070 and the desktop variant were significant in several games--especially when running at resolutions higher than 1080p. This would be a big improvement if these numbers held across most titles, especially given mobile systems would be running slower CPUs and memory.

 

[BigBoomBoom]

In my personal use, the difference between the mobile (non-MaxQ) 1070 and the desktop variant were significant in several games--especially when running at resolutions higher than 1080p. This would be a big improvement if these numbers held across most titles, especially given mobile systems would be running slower CPUs and memory.

Depends, non Max-Q can be in shitty laptop with shitty built that leads to downclock. Usually Max-Q are more premium and the laptops equipped with those are better built despite being thinner. Only the most expensive ones with 1080 are the one being throttled compare to normal laptop, as there's a point where you can't make 90W TDP 1080 in a super thin (20mm ish?) Laptop w orks wonder. 1060 and 1070 Max-Q are like 45-70W TDP.

Now if the RTX 2070 Max-Q are like it's desktop counter part, possibly you are looking at sub 80W TDP. Easier to handle than the 1080.

 

[cliffordcooley]

LOL - Benchmark stops at 3080. Someone is playing a joke on us.

 

[Knot Schure]

...the Max-Q variant is just as fast as the full fat desktop version? Considering the performance delta between the Max-Q and AIB versions of Pascal, that seems highly dubious.

Dubious, why?

The 1070M was 'bigger' than its Desktop brethren (2048 shaders vs desktop 1920), albeit with lower clocks...

 

[Sausagemeat]

As if there was any doubt that the 2000 series was gonna walk all over AMD's best?

And if Nvidia returns next year with a 3000 series, AMD will be so far behind they'll need to create another cryptocurrency craze.

How about ya take a step back there and look at current max-q products on the market before jumping to conclusions. Given that Max-Q products are designed for thin form factors and that Turning isn't particularly good power consumption wise, I'd be surprised if the 2070 was more then 15% faster then the 1070.

This is only a single benchmark as well and it's one of the worst one's at that.

Current max Q isn’t that different to desktop variants with the exception of the 1080. But the 2070 is a lot more efficient and produces less heat than the 1080 so it’s not unreasonable to expect the max Q 2070 variant performing similarly to the desktop variant.

 

[Bpsti]

The 2070 beats the 1080 (by a little bit) all for pretty much the exact same price. So for the same amount of money as you spent last gen you can get.... pretty much the same performance as last gen..

Thanks Nvidia. Sliding the model numbers around didn't fool us. Just check your stock prices.

 

[Sausagemeat]

The 2070 beats the 1080 (by a little bit) all for pretty much the exact same price. So for the same amount of money as you spent last gen you can get.... pretty much the same performance as last gen..

Thanks Nvidia. Sliding the model numbers around didn't fool us. Just check your stock prices.

2070’s in the UK are selling at around £450. Cheaper than 1080’s we’re most of their life. In fact 1080’s rarely dropped below £500.

The 2070 is also more efficient and has tensor cores. We haven’t seen a big reduction in price from Nvidia but they haven’t raised prices for anyone. People should only really be upset if they felt entitled to pay less. Which with no competitor isn’t going to happen. If Nvidia had raised prices I would condemn them. But they haven’t. Unlike AMD who are charging more for their same Polaris chip just with an overclock. At $280 the 590 is worse value than the $200 480 from 2.5 years ago.

 

[urbanman2004]

AMD, make NVIDIA great again

 

[Evernessince]

The 2070 beats the 1080 (by a little bit) all for pretty much the exact same price. So for the same amount of money as you spent last gen you can get.... pretty much the same performance as last gen..

Thanks Nvidia. Sliding the model numbers around didn't fool us. Just check your stock prices.

2070’s in the UK are selling at around £450. Cheaper than 1080’s we’re most of their life. In fact 1080’s rarely dropped below £500.

The 2070 is also more efficient and has tensor cores. We haven’t seen a big reduction in price from Nvidia but they haven’t raised prices for anyone. People should only really be upset if they felt entitled to pay less. Which with no competitor isn’t going to happen. If Nvidia had raised prices I would condemn them. But they haven’t. Unlike AMD who are charging more for their same Polaris chip just with an overclock. At $280 the 590 is worse value than the $200 480 from 2.5 years ago.

The 590 is die shrink.

As if there was any doubt that the 2000 series was gonna walk all over AMD's best?

And if Nvidia returns next year with a 3000 series, AMD will be so far behind they'll need to create another cryptocurrency craze.

How about ya take a step back there and look at current max-q products on the market before jumping to conclusions. Given that Max-Q products are designed for thin form factors and that Turning isn't particularly good power consumption wise, I'd be surprised if the 2070 was more then 15% faster then the 1070.

This is only a single benchmark as well and it's one of the worst one's at that.

Current max Q isn’t that different to desktop variants with the exception of the 1080. But the 2070 is a lot more efficient and produces less heat than the 1080 so it’s not unreasonable to expect the max Q 2070 variant performing similarly to the desktop variant.

Power efficient isn't really what I'd call the 2070. It consumes 15w less during load then a 1080 for similar performance. That's not impressive by any measure and it most certainly isn't noteworthy enough to see much gains in mobile.

https://www.anandtech.com/show/13431/nvidia-geforce-rtx-2070-founders-edition-review/14

If both the 2070 and 1080 consume similar power then we can safely assume that mobile versions will won't break the mold on a pattern we saw precisely last generation.

 

[QuantumPhysics]

The 590 is die shrink.
Power efficient isn't really what I'd call the 2070. It consumes 15w less during load then a 1080 for similar performance. That's not impressive by any measure and it most certainly isn't noteworthy enough to see much gains in mobile.

https://www.anandtech.com/show/13431/nvidia-geforce-rtx-2070-founders-edition-review/14

If both the 2070 and 1080 consume similar power then we can safely assume that mobile versions will won't break the mold on a pattern we saw precisely last generation.

15 W less during load IS EXACTLY what I'd call power efficient. That 15W adds up on a monthly basis.

And if the 2070 is 15W better at load than a 1080, I already know it's even more efficient than the AMD counterpart.

 

[Gus Fring]

Where are the AMD mini cards .. I sit with my money , waiting for aMD to bring ut even a sniff of mini , it was looked upon as shocking , but now? anything goes !

 

[Evernessince]

15 W less during load IS EXACTLY what I'd call power efficient. That 15W adds up on a monthly basis.

And if the 2070 is 15W better at load than a 1080, I already know it's even more efficient than the AMD counterpart.

15 out of 324w. And no, getting 15w less then a last generation card in a higher tier is by no definition power efficient. You should be comparing it to the 1070, which is the equivalent model of last generation. It's ridiculous to claim a card is power efficient by making apples to oranges comparisons.

 

[QuantumPhysics]

15 out of 324w. And no, getting 15w less then a last generation card in a higher tier is by no definition power efficient. You should be comparing it to the 1070, which is the equivalent model of last generation. It's ridiculous to claim a card is power efficient by making apples to oranges comparisons.

Whatever.

As long as my 2080Ti works well I don't really care.