First Alleged Benchmark Results GeForce RTX 3080 Surface

980ti is fast 1070 when you think about it. That isnt destroying is it? Nvidia stagnating for years, increasing only prices . Ampere would be this big thing but if it dissapoint now people will be pissed. Thats good for amd though.

 

They aren't very accurate for me and my trusty 2080 Ti and i7-10700k. Referring to https://www.guru3d.com/news_story/first_alleged_benchmark_results_geforce_rtx_3080_surface.html

The percentage differences are a off a fair bit for an RTX 2080 Ti! Looking at my results, in general its less than 1% to max 17% difference.

Bit like the Doom Eternal comparison nvidia did (I did the same intro to the boss at max settings 4k and was 20-30fps faster than the el cheapo underclocked 2080ti nvidia used for comparison and damn close to the 3080 side by side).

So, if those results are true for a 3080 it certainly isn't a massive change for most games. Nor would <r Leather Jacket man claiming a 3070 is the same or better. Unless you are bitcoining but for gaming. Neh!

Here are my random plucked 3DMark scores to show the differences are only 0.5-17%... Running a normal standard fan cooled with power limit % increase in MSI Afterburner (like you should.. yet still runs under 50-70 degrees under load). Asus RTX 2080 Ti OC 11gb with a Intel i7-10700k @ 5ghz

Here are the the scores posted vs my scores with the 3DMark URL .

• 3DMark Fire Strike Performance: 31919 (+25% 2080Ti, +43% 2080S ) -> 30181 https://www.3dmark.com/fs/23188420 mere 0.5% difference, not 25%!
• 3DMark Fire Strike Extreme: 20101 (+24% 2080Ti, +45% 2080S ) 17576 https://www.3dmark.com/fs/23188361 (14%)
• 3DMark Fire Strike Ultra: 11049 (+36% 2080Ti, +64% 2080S ) 9400 https://www.3dmark.com/fs/23442247 (17%)
• 3DMark Fire Strike Time Spy: 17428 (+28% 2080Ti, +49% 2080S ) 15667 https://www.3dmark.com/spy/13196272 (11%)
• 3DMark Fire Strike Time Spy Extreme: 8548 (+38% 2080Ti, +59% 2080S ) 7259 https://www.3dmark.com/spy/13784438 (17%)
• 3DMark Fire Strike Port Royal: 11455 (+45% 2080Ti, +64% 2080S ) -> 10129 https://www.3dmark.com/pr/304202 (9%) **certainly not 45% ? da fk****

Don't go selling those 2080 Ti's cheap. Especially for a 3070 with the same type of vram but less. At higher resolutions it will stutter and the 1% will be crud. Learned this when had to sell my 1080 for a 1080 Ti within a week back in the day due to running at 4k and the stutter sucked in games that use graphics like BF1 @ 4k.. not 8k friendly cartoon games like Fortnite.

 

disappointing tbh

 

Nice job!

My timespy score with 10900k and 2080ti: http://www.3dmark.com/spy/13103810

 

I can't figure out how a GPU that is 30% faster than current fastest one and costs half the price could be a disappointment for anyone.... it's something like +160% price per performance. Hoping AMD can do even better but still.. these GPUs look very appealing to me.

 

Well there you go. Your 17,102 with a 2080 ti vs the 3080 at 17,428 is nice, since you are using the same i9-10900k CPU as the benchmarks. Compared to my i7-10700k

When comparing my result with yours (https://www.3dmark.com/compare/spy/13103810/spy/13803124) you can see that the actual graphic scores are about the same (3.6%) with your fancier CPU but gives a more noticeable bump in the scores for the portions not related to the GPU. Stupid 3DMark

 

https://twitter.com/_rogame

From the tests I have done with my 2080 Ti only seeing about a 1-15% difference. The poverty 2080 Ti being used in the nvidia comparisons seen so far are off by a fair chunk.

Even the Ashes of Singularity 4K test on 'Crazy' I got 84.8 fps compared to 88.3 for the 3080. While a tech dude on twitter beat it with a 89.7 avg. https://twitter.com/_rogame

Guess we will see when actual benchmarks we can all use come out and the settings used. Not interested in 1080p stuff, so hopefully GPU bound tests vs decent 2080 Ti's @ 1440p-4k+ .. Like to see how the vram holds up at 8k in a game with HDR, RTX and all that jazz

 

Well this was hugely disappointing. Still, I wait for official benchmarks before deciding what to do.

The interesting thing here is the power of marketing. Just think for a moment the huge impact this has had on people and on the used card market.

 

I'm not surprised one bit. They hype around the nvidia's own slides was insane, all the youtubers/media etc. did blow it up to astronomic proportions. I had guys that aren't even into hardware ask me if I heard about the new nvidia cards... so yeah

I'd wait for a 3080Ti, but I don't know if I have the willpower. Probably not.

 

Don't know if I believe those scores but we shall see.

By the way, does anyone know what the CPU/GPU monitoring software in use in those screenshots is (that shows in top left corner green/blue/red colour)? I see it everywhere, but don't have it, and want to get it.

 

One of my result with an unlocked (shunt mod) 2080Ti:

https://www.3dmark.com/spy/7007126

Guess the 3080 can do it better when pushed, but right now the 3070 came to my mind and want to see if it really outperforms an unlocked 2080Ti, specially if the new 3070's come power locked.

 

MSI Afterburner + RTSS

 

2080Ti SLI ! , get a second 2080Ti for $350
NVIDIA driver456.16

 

Am still convinced AMD has a beast in the wings and Nvidia knows it just looking at the specs of both the consoles coming out they are looking mighty impressive remember this is all in a 250w envelope as well
Xbox has RT confirmed PS5 is 4k and can upscale to 8k the ps5 36 CUs @ 2.23ghz gives 10.28 TF the Xbox does more again all within a 250w envelope incl CPU HDD etc.

Am waiting lets see what AMD has I think its going to be very surprising and hopefully I can finally move back to Team Red.

 

I dont think these would even fit in my case.

 

very underwhelming tbh if these are true. has going samsung 8N hurt their performance gains? Each shader SM in these new GPU's seems a lot weaker than previous generations. How can you almost, and in some cases, double SM count but only achieve 30-40% when you are saying they are upwards of 80-100% improvement?

I have literally everything crossed......EVERYTHING... for AMD to bring something "BIG" as they say.... but I have my doubts.

 

Those percentages are way off.

E.g. Timespy: 17828 vs 14854 is +20%, not +28%
Port Royal: 11455 vs 9106 is +25.8%, not +45%

I agree with the people who are disappointed. I wanted to upgrade, but if these numbers are correct I will probably give this generation a pass.

 

80% performance improvement is with raytracing and dlss... straight up performance gains, not impressive.

But it honestly doesn't surprise me. 3090 will probably only be 35% faster than the 2080 ti, without using any "tricks"...

 

This generation isn't really intended to coax people who have RTX 20xx to upgrade, it's squarely aimed at GTX 10 owners. a 50-60% uplift over my 1080Ti (which is about right, considering the 2080 Super is on average 10% faster), plus the addition of RT and Tensor cores is more than enough to make me get one. I'm all about the eye candy, and can't wait to play Cyberpunk and replay Witcher 3 with RT on, and hit the new expansion for Control without needing to play on GeForce Now.

I really don't get the responses here. Since when is a single gen jump EVER worth upgrading? Why are we even bothering to compare 20 series cards to 30 series when we KNOW that the results will be between 25 and 50%, like they always are?

 

The 3080 has the same number of SM's as the 2080Ti.

Perhaps you're talking about FP32 (CUDA) cores, which were doubled but, if you read further they aren't always operating in the doubled mode. If integer is required then it's back to normal 2080Ti levels per SM.