AMD vs. Intel GeForce RTX 3080 Benchmark

[Endymio]

I would expect that in many if not most games, a 3800xt would best the 3950x -- for much less money. Wouldn't that be a better comparison, especially since that chip is closer in price to the 10900K?

 

[Aryassen]

Wow, I did not expect those difference in certain titles. OK, granted, at 4K, where the card is positioned, there is no difference, but at lower res...damn! I wouldn't be affected as I am not a competitive gamer (e.g. 60FPS is all I care about), but those who seek high(est) FPS at 1080P, Intel is still the way to go.
Probably all that will change with Zen 3 cores, and I am eager to see, but that is just specualtion at this stage. For now, the crown clearly still goes to Intel (saying that as a satisfied owner of a Ryzen 5 3600).

@Steve - Many thanks for testing this!

 

[hahahanoobs]

With the hype Ampere received, and the fact I'm a 1080p/120Hz gamer, I wanted to see 1080p results even if it was just out of curiosity after buying an RTX 2060 back in May.

"Ryzen is more popular now so we used the 3950X in our review."
I thought the idea was to remove bottlenecks? Isn't the 3900X(T) faster?
32T for gaming?

I'm so confused.

 

[hahahanoobs]

"For example if we didn't list the CPU tested and instead just showed the results (the overall margins and cost per frame) you'd have no issue with the review."

WHAAAAT?!
LOL What's happened to this website???

 

[dragosmp]

Honestly, for 10900k it is the same. Not everyone is buying 10900k. Most people buy some i5, maybe 10600k or at best 10700k. There are also a lot of people who still cling onto their older CPUs like me e.g. with my 6850k.

What they do here is to simply compare the best available on Intel side with best available on AMD side. It has no favouritism this way. But you are right, I would also, like many others I assume, see more of mainstream CPUs being tested even with 3080 and 3070. Like 10600k vs 3600x or 10700k vs 3700x/3800x.

That's spot on. This review illustrates perfectly that the peformance increase between a 2080Ti and 3080 is the same at all resolutions, irrespective of the CPU being used. At this point wouldn't you just use the most popular CPUs for your reviews, which by all accounts are the Ryzens? I would have gone for a 3900X, as it's tons cheaper and about as fast. Can't wait for the midrange CPU scaling article

 

[Irata]

I would expect that in many if not most games, a 3800xt would best the 3950x -- for much less money. Wouldn't that be a better comparison, especially since that chip is closer in price to the 10900K?

You must have different prices where you live.

In my market, Ryzen 3900XT is closest to the 10900K in price, but still €73 cheaper. 3950X is roughly €170 more expensive.

The point is probably to use the top of the line model regardless of price.

 

[Endymio]

The point is probably to use the top of the line model regardless of price.

But in this context, doesn't "top of the line" mean highest performance? I seem to recall both the 3800xt and 3900xt beating the 3950x in gaming benches, albeit slightly.

 

[Evernessince]

Thank You TECHSPOT for the comparison.

based on your findings I will build my next desktop using an Intel CPU + my 3080.

every little bit of advantage is appreciated.

"Looking not so far beyond, we predict we’ll be moving to a Zen 3-based test system in a few months' time, so regardless of which way we went, having to do it all over again relatively soon is inevitable."

If every little bit of advantage is appreciated, you should be considering the soon to be released Zen 3 processors.

Of course I would have no issue, cause I would assume you tested with the fastest gaming CPU. For example, right now I have a 1080ti and I'm thinking about upgrading cause I can't really play metro exodus on my current resolution with max settings. But looking at your review, it seems like a waste of money to upgrade to a 3080 for that game, cause the gains are minimal. Little would I know that that is because of the CPU you are testing it with

Didn't you just state in your last comment that they should be testing with a mid-range CPU?

"Your reasons for testing on the 3950x don't make sense to be frank. PCIE 4 makes little to no difference, and most of your viewers are not buying a 3950x. It's either the 3600 or the 3300x. So yeah...."


"For example, right now I have a 1080ti and I'm thinking about upgrading cause I can't really play metro exodus on my current resolution with max settings."

I mean sure, you just so happen to be playing the game the 3950X happens to be weakest of all those tested and a resolution below 4K. Sus, very sus.

 

[cliffordcooley]

"What We Learned" and was not mentioned in the conclusion.
PCIe 3.0 is still not a limiting factor in single GPU gaming.

- - - -
I don't understand why people are screaming for PCIe 4.0.

 

[theetails]

I'm sorry I couldn't get past the part where you used 3200 ram for both when AMD uses 3600 and Intel does not.

Is that really not going to make a difference? If so, why hasn't that been brought up? With low latency 3600 is pricier, but performs faster than 3200.

 

[Irata]

But in this context, doesn't "top of the line" mean highest performance? I seem to recall both the 3800xt and 3900xt beating the 3950x in gaming benches, albeit slightly.

Good question. Overall and price wise the 3950x is top of the game but am not sure about gaming in particular.

 

[Squid Surprise]

Let's just hope that the Zen 3 CPUs match or defeat Intel in gaming, then this argument will be moot.

 

[Jyrkz]

I would still go amd, its only few fps lower then intel, but amd owns almost in all other non gaming places. + also if you start with a good MB you can upgrade without problems. @zen3

 

[jbc029]

Let's just hope that the Zen 3 CPUs match or defeat Intel in gaming, then this argument will be moot.

I'd expect the 8c/16t models to be on par. Not the 12/16 core, though. Those will still have ccd crosstalk.

 

[theruck]

If those 3 fps would be in favor of ryzen you would go mad how well amd was performing

 

[nodfor]

Since almost every other site tests with 10900k, it is nice to have a reviewer running tests with 3950x. The 3950x seems kinda of irrelevant right now because of its high price, but if its price gets reduced when next gen Zen launches it could become a reasonable upgrade option for those who want to upgrade their Ryzen cpus.

 

[Adhmuz]

Thank you for this revisit, glad you allowed the 3080 to run at it's full potential.

 

[mAdmAnDingo]

Thanks Steve, I really enjoyed the article. I would love to see a tweaked timings RAM/IF scaling side piece for Zen2 CPUs. I am running an 3700X, and I would love to see what benefits we would see with tweaked timings and higher frequency RAM/IF with the 3080. I would like to see how much ground can be made up with an optimal RAM/IF setup for Ryzen, and what gains we could see at the different resolutions.

If pushing the RAM/IF and tweaking the timings to much is an issue for system stability when benchmarking, maybe just push as far as you can go while remaining stable. Even if that is only 3600mhz/3733mhz with a bit looser (but still tweaked) timings, whatever is actually doable. I am sure it will be a ton of work, but I would love to see it and would very much appreciate it.

I would even love to see it with lower end CPUs, 3600, 3700x, 3900x etc. I am not averse to similar articles for Intel either, but I am more interested in Zen2 as they need the performance the most because they really depend on an optimal RAM/IF/Timings setup (and it is what I own and am running at home). And it is something that can be done on all Ryzen systems to increase performance, unlike overclocking which does not really help on Zen2.

And I do appreciate the AMD test system, because I actually want to see which games Zen2 falls behind compared to Intel. I wouldn't want to purchase an 3080 and find out my GPU is performing 20% less than an Intel CPU in certain games at certain resolutions. I could just purchase an 3070 in those cases and save some money instead, so thank you Steve, I honestly appreciate it.

Most other sites are testing on Intel CPUs, and those that do include Ryzen testing, only do non-tweaked Ryzen testing, and that makes it very difficult to find out where Ryzen owners such as myself stand in comparison to Intel. And I don't expect any other sites to do RAM/IF/Tweaked timings testing with Zen2 and 3080 GPUs, so you are our only hope to see such testing.

So that is also why I would love to see the tweaked timings RAM/IF scaling article with Zen2 and an 3080, to see if I should grab an 3080 or just an 3070. I am running 3733Mhz RAM/IF with tightly tweaked timings myself, and I would like to see what differences I could expect before making any purchase.

In my country an 3080 costs me $1 262,32 (American dollars), so you can understand where I am coming from. And those kinds of prices are normal for my country, we are always price gouged. So I really do want to know if it is actually worth the added cost over the upcoming 3070 with my current setup (highly tweaked timings and 3733 RAM). And I am planning an monitor upgrade as well, but I would like to know where I stand before I make that choice. It won't be a 4k monitor, so likely a 1440p monitor. But I want to know where I stand before I commit to any specific model and refresh rate.

I really am still debating, so the more information I have the better of course. With everything PC related being so expensive in my country, I don't make any decision quickly or without lots of thought on the matter.

If you guys are to busy with all the current content you are working on I do understand though, so I will have no complaints if you are unable to do the extensive testing required, I am not unreasonable and I understand if it is unrealistic with your workloads you find yourselves under. I can always make an approximation to what I think my system would achieve with an 3080, and I think I wouldn't be to far from the actual outcome.

 

[Badvok]

"the vast majority of our audience either already owns or are looking at buying a Ryzen processor over anything Intel."
Interesting statement, I assume that means that your audience behaves exactly the opposite to the people who use Steam. I wonder who they are?

 

[Endymio]

"the vast majority of our audience either already owns or are looking at buying a Ryzen processor over anything Intel."
Interesting statement, I assume that means that your audience behaves exactly the opposite to the people who use Steam.

Odd, I didn't know Steam surveyed what their users were considering buying.

 

[Tommy Lee]

If I'm a serious competitive gamer, aren't high frame rates more important than higher resolution? However, I'm not a serious competitive gamer myself, but I play RuneScape, and I notice that on my 1080p screen, the interface is kind of cramped. So maybe 4K does have enough benefits that it's very important to get decent performance at that resolution.

Bro a few frames behind means jack anything over a 100fs is overkill to begin with you can't even tell at a point how fast the refresh is many will lie and say they can but their full of it

The Results for Ryzen chips is good enough to be happy with I own many Intel Rigs as well AMD Rigs there comes a point that people want the most frames but you only need 30-60fps to really enjoy a game anything after is just scratch

Real Players know 60FPS and a good Internet connection is just as good as someone on a 144hrz refresh it's all the same on the internet your connection matters more than refresh and frames if you have a mid to top tier card just saying

You have people playing on less and last time I checked most people are still using 1080p and 1440p slowly becoming the standard

I have a 3950x Overclocked and used my 1080ti and 2080super with zero issues or lags it's people who think that a few stupid frames matter when you can't tell anything after 60fps

This is a FANBOY argument because at the end of the day they all do the same crap and 8 to 10 frames less can be made up in other ways

We don't know if they overclocked what ram etc so I mean most of these test benches are on stock configs as well

 

[leeghoofd]

So were you comparing boxed clocks? Knowing that most motherboards overclock out of the box all the Intel 10900K cores to eg 4800MHz. If so up to Techspot to find out what happens if you OC the Ryzen CPU to 4.0 or even higher... I bet the performance difference will become marginal at the lower resolutions...

Happily upgraded from 9900K@5Ghz to 3700X@4.2, not even looking back

 

[Tommy Lee]

"the vast majority of our audience either already owns or are looking at buying a Ryzen processor over anything Intel."
Interesting statement, I assume that means that your audience behaves exactly the opposite to the people who use Steam. I wonder who they are?

Most of my Rigs on Steam have been AMD and Intel from Phenoms 965BE - FX 8320 -8350-3770k-6700K - 6850K 6950X -8700K-8086K- Ryzen 1600x-Ryzen 3700x and Ryzen 3950x.

Bottom line if you're playing on 1080p you don't need this card period 1440p and up should only be using this card the higher the resolution the less restriction that's why 1080p Gaming can be done with lower end GPU's and still bang out massive frames anyone playing on 1080p with this card is pretty much wasted their money .

The 4k results tell the story pretty much 1440p the frames go way beyond a 100fps most people's goal is to achieve a steady 60 lol still I don't get why people are complaining lol it's not a slide show that's for sure !

 

[emmzo]

It was kind of a debacle and I see you complain about limited time for testing, but there was another site that very day where they had tested side by side both top AMD Intel CPUs in all three resolutions for the 3080, plus OC, plus 20 other cards in 13 games! So, I wonder: do you have a staffing problem?

 

[Lionvibez]

This patch hope you guys use it for the next review.

https://www.techpowerup.com/272486/...t-simulator-patch-improving-performance-by-10