News Nvidia's GeForce RTX 3070 Beats 2080 Ti in First Performance Results

[TheHughMan]

Just waiting for GeForce RTX 3070 TI SUPER OC! Maybe by Christmas. Will be +/- 3% of the RTX 3090 if issues persists.

 

[TheHughMan]

Just waiting for GeForce RTX 3070 TI SUPER OC! Maybe by Christmas. Will be +/- 3% of the RTX 3090 if issues persists.

 

[KevBacon]

Yeah... gonna wait for actual reviews on this one.
They lied about the 3080, this won't be any different.
Those people who panic-sold their 2080Tis...

The performance to price ratio is pretty good as everyone knows but i question some of these claims also. Benchmarks with my exact setup only achieved 5 to 20 more fps when switching from the 2080Ti to the 3080. This along with the software in the chart and not more information on actual gaming performance certainly seems fishy. Even if it is equal to the 2080Ti in performance you really just waited two years to get a 2080Ti for cheap. I’ve had mine for almost two years now and I’ll definitely be keeping it as it still maxes out most of my games in 3440x1440 at 120FPS. Paying $1,000 more and not waiting two years to experience that performance is worth it to me. I think it’s great that the prices make this performance attainable for a lot more people but making it seem as though the 30 series blows the 2080Ti away in performance seems a bit much to me.

 

[vinay2070]

What a mess toms layout is. Cant even zoom the graph. And even if you maginify the page, you cant see the details. All side real estate dedicated to ads. And on top of that, that irritating video window keeps scrolling. If you open it on ipad, the scrolling video window sometimes casues a black bar making the text above unreadable. Welcome to 1996. Oh wait, tomshardware was very much readable back then with fewer ads and better layout.

 

[Flayed]

Adding a zoom type feature would really help those of us with eyesight issues

 

[vinay2070]

Adding a zoom type feature would really help those of us with eyesight issues

Its not about eyesight issues. Even with a good eyesight, the last line of the graph, you literally have to get close to the monitor and struggle to read. Look at the same on videocardz! The default graph display size is so big, that you dont even need to zoom. https://videocardz.com/newz/nvidia-shares-official-geforce-rtx-3070-performance-figures

Thats how it should be. The layout should be reader friendly first, and ad friendly next. Toms - ads only friendly. My best prediction, they will NEVER improve. If they wanted to, they would have.

 

[rwood19]

I'm more interested in the differences between the 3070 and 3080. Remember the 2070 Super and 2080 Super?
I doubt it will be the same. But the 3080 needs to be at least 29% faster than the 3070 to get more bang for each buck (USD 499/699). And that might be the case. Especially with European prices for the 3070/3080. We'll see...

 

[HyperMatrix]

I'm more interested in the differences between the 3070 and 3080. Remember the 2070 Super and 2080 Super?
I doubt it will be the same. But the 3080 needs to be at least 29% faster than the 3070 to get more bang for each buck (USD 499/699). And that might be the case. Especially with European prices for the 3070/3080. We'll see...

As I wrote earlier, the 3080 has 50% more cores and 60% more memory bandwidth. It will therefore be at least 50% faster and have less issue with vram limitations. 3080 is a far better value than the 3070 which is hot garbage. It should be priced at $399.

 

[Flayed]

I'm hoping there will be something in AMDs lineup with similar performance to the 3070 for a little less money

 

[Friesiansam]

This is just Nvidia hype and, don't forget neither 3090 or 3080 have lived up to the claims of the Nvidia marketing department.

 

[rwood19]

As I wrote earlier, the 3080 has 50% more cores and 60% more memory bandwidth. It will therefore be at least 50% faster and have less issue with vram limitations. 3080 is a far better value than the 3070 which is hot garbage. It should be priced at $399.

At least 50%? ...which would make the 3070 slower or roughly the same than the 2080 Super, and not faster than the 2080 Ti. According to UserBenchmark data the 3080 is 41% faster than a 2080 Super. I have seen 50% increase averages with current reviews.

But I agree with you that the 3080 looks to be better value than the 3070.

 

[Shadowclash10]

As I wrote earlier, the 3080 has 50% more cores and 60% more memory bandwidth. It will therefore be at least 50% faster and have less issue with vram limitations. 3080 is a far better value than the 3070 which is hot garbage. It should be priced at $399.

Riiiiiight... Sorry, no way the 3080 is 50% faster. There is never a direct correlation. Now, it might be a better value, but that does not make the 3070 "hot garbage". You sound exactly like the people who think their 2080 Tis are now garbage. Let's imagine for a moment that we have the 3070 released, and it's roughly equal to a 2080 Ti. Imagine the 3080 is a better value. So what? The 3080 is $200 more. That's the big difference here. Also, why should it have been priced at $400?

 

[Chung Leong]

As I wrote earlier, the 3080 has 50% more cores and 60% more memory bandwidth. It will therefore be at least 50% faster and have less issue with vram limitations. 3080 is a far better value than the 3070 which is hot garbage. It should be priced at $399.

The 3070 will offers ~70% of the 3080's performance for 70% of its price. What's so outrageous about that?

 

[Phaaze88]

Pardon my use of mulit-quotes, it's a bad habit of mine.

The performance to price ratio is pretty good as everyone knows but i question some of these claims also.

Typical marketing shenanigans. It's their job to make the product more attractive to people who don't do their 'homework'. Sales > everything else.
Everyone should take a company's performance metrics with a grain of salt, and check independent reviews instead.

Benchmarks with my exact setup only achieved 5 to 20 more fps when switching from the 2080Ti to the 3080.

5-20 more on average in 1-2 titles?
10 games?
50?
The number(%) should be lower with a larger sample size.

This along with the software in the chart and not more information on actual gaming performance certainly seems fishy.

Something I've noticed that's still being overlooked is that Ampere is not a gaming card, but is still being marketed as one. It is a compute card at its core - the same kind used in datacenters and the like, but cut down.
If I recall correctly, the same is true for Turing. It is cheaper for Nvidia to do it that way.
The tensor cores, the shader models, the AI acceleration... does not have that much utility in games, but they tried to hide that with all those rendering benchmarks.

Even if it is equal to the 2080Ti in performance you really just waited two years to get a 2080Ti for cheap. I’ve had mine for almost two years now and I’ll definitely be keeping it as it still maxes out most of my games in 3440x1440 at 120FPS.

That compute performance is the same reason that the 3080 and 3090 are so good at 4K and up, so-so at 1440p, and poor at 1080p.
Your 1440p UW is still closer to regular 1440p at ~34% more pixels. 4K is still a crazy 67% more pixels than 1440p UW... so yeah, a 3080 isn't going to be all that great for you on average; just so-so.

Paying $1,000 more and not waiting two years to experience that performance is worth it to me.

Individual values vary, kind of thing, blah, blah, blah~. Gonna leave that one alone.
Some will agree with you, some won't, etc.

...making it seem as though the 30 series blows the 2080Ti away in performance seems a bit much to me.

That's the marketing dept's job.
Knowledge is power; they can't take advantage of the well-informed.

 

[HyperMatrix]

The 3070 will offers ~70% of the 3080's performance for 70% of its price. What's so outrageous about that?

because you’re paying 71.5% of its price for 66% of its performance. And you’re also limiting yourself with 8GB vram at the same time which means bad long term prospects for next gen games.

 

[TechyInAZ]

Just wondering, roughly what difference is there on your 2060 Super at Ultrawide w/DLSS on performance vs. quality?

It's about a 20FPS difference. With performance mode, I can maintain ultra settings on everything (including maxed out RTX) and touch 60fps with 50-55fps averages. With quality that drops into the 35s and 40s.

 

[csm101]

i have a history of 70's. so its time to move on to 80. i hope i can get a EVGA hybrid once its launched. looking at these performances I m sure the <Mod Edit> scalpers will target 3070 more than 3080.

 

[Chung Leong]

because you’re paying 71.5% of its price for 66% of its performance. And you’re also limiting yourself with 8GB vram at the same time which means bad long term prospects for next gen games.

The 3080 is not going to be 50% faster than the 3070 when it only has 25% more L2 cache and memory controllers.

 

[Avro Arrow]

I can't even bring myself to bother reading this after seeing how much BS was involved in the RTX 3080 "benchmarks" from digital foundry. It's minutes of my life that I'll never get back only to find out later that it was utter BS.

I generally ignore manufacturer-provided "benchmarks" and "performance numbers" because they're always skewed one way or the other. I'll wait for the independent benchmarks this time.

 

[Avro Arrow]

View: https://www.youtube.com/watch?v=VjOnWT9U96g

How Good...or Bad...is Ampere?
um yea. the more i read about ampere, and the more that is starting to be leaked about rdna2 the more i am glad i am waiting till all the cards are on the table, or, in the PC case, i should say.

Jim really is the best hardware analyst that I've ever seen. He uses historical context to show long-term trends which are at least equally as important as what is happening right now. He shows how we got here and where we're pointed.

He really is the best investigative tech reporter/analyst since Charlie Demerjian.