NVIDIA GeForce RTX 3080 Rips and Tears GeForce RTX 2080 Ti In New DOOM Eternal 4K Comparison Video

[Tsing]

Image: NVIDIA



NVIDIA has shared a new video that compares the performance between the GeForce RTX 3080 and GeForce RTX 2080 Ti using id Software’s latest DOOM game, DOOM Eternal. As you might expect, green team’s new “flagship” (we’re still having a hard time processing that due to the existence of the GeForce RTX 3090) runs the game like a walk in the park and exceeds the previous halo product by leaps and bounds.



Under the Ultra Nightmare preset at 4K resolution, the GeForce RTX 3080 appears to average around 110 to 130 FPS, hitting brief peaks of 170 to 180 FPS. The GeForce RTX 2080 Ti, on the other hand, struggles to reach or maintain 100 FPS in the majority of comparisons.



This is pretty cool, since you’re getting significantly better performance than last year’s flagship ($1199/$999) for $500 to $300...

Continue reading...

 

[MadMummy76]

Better than I expected. Of course it's cherry picked, but if even 60% of it translates to other titles it makes it a worthy upgrade. For gaming.

 

[Dan_D]

I doubt it translates into that much of a gain in other titles. After running various tests on Doom Eternal, it's a bit weird. Even at 4K, it's oddly CPU limited. I'm sure its a fairly solid upgrade, but I wouldn't be at all surprised if this is the largest gap we'll see between the two out of any game.

 

[Armenius]

I doubt it translates into that much of a gain in other titles. After running various tests on Doom Eternal, it's a bit weird. Even at 4K, it's oddly CPU limited. I'm sure its a fairly solid upgrade, but I wouldn't be at all surprised if this is the largest gap we'll see between the two out of any game.

Eternal is, surprisingly, very CPU-limited which is why I think this is actually a worst-case scenario for the GPU. I was seeing the gap from the 2080 Ti be around +40-50% in that video, so I imagine what I have been saying about it being around 60% faster will end up being accurate on average.

 

[MadMummy76]

I doubt it translates into that much of a gain in other titles. After running various tests on Doom Eternal, it's a bit weird. Even at 4K, it's oddly CPU limited. I'm sure its a fairly solid upgrade, but I wouldn't be at all surprised if this is the largest gap we'll see between the two out of any game.

What are you suggesting? That they paired the 2080ti with a slower CPU? Wouldn't that be false marketing?

 

[Armenius]

What are you suggesting? That they paired the 2080ti with a slower CPU? Wouldn't that be false marketing?

My take is that he believes Doom Eternal was an odd choice to showcase because of CPU performance with the game, in general. But In my opinion, I think he got his conclusion backward, which is what I was explaining above.

 

[Dan_D]

What are you suggesting? That they paired the 2080ti with a slower CPU? Wouldn't that be false marketing?

No, I am not suggesting anything of the kind.

My take is that he believes Doom Eternal was an odd choice to showcase because of CPU performance with the game, in general. But In my opinion, I think he got his conclusion backward, which is what I was explaining above.

You are sort of right. CPU performance is all over the map with this game in my experience. However, I only tested it with Vulkan. As you can see below, the performance isn't what you'd expect. This is at 4K, using a 2080 Super. Not a 2080 Ti. You would think that we would be more GPU limited here. In Ghost Recon, all these CPU's perform almost identically on the same GPU at 4K. FPS is only 1-3FPS across the entire test spectrum. In Doom Eternal, we have drastically different results. Based on what I've seen, the game runs very good on current hardware. In a case where you use a GPU that's not remotely challenged, you will see evidence of CPU limitations. This is why I'm saying that Doom Eternal may be an outlier and that 60% advantage probably doesn't translate to other games.



I don't have solid conclusions about Doom Eternal. What I am really saying is that based on what I've seen, it could certainly be an outlier as its performance isn't what one might expect for a AAA game at 3840x2160. I would also like to point out that these tests were manual run throughs, which do not provide the most consistent data. Although, some games do show more consistency than others.

I really need to find a way to script these tests rather than doing them manually. Not only would that save me some time, but it would also provide clearer results. Because the results I've seen were so weird for this game, I stopped using it in my reviews for motherboards or CPU's.

 

[Peter_Brosdahl]

To be fair to all, even in the other video from Digital Foundry(on another thread) that guy also said DE was strange title for NV to use for some of their comparisons, not to mention how they were essentially throwing the 1080 Ti under the bus(I know we're talking about the 2080 Ti here). He stated how frame rates can be significantly different at different parts, and then again, how various cards were also having different behavior with them. I admit I haven't watched this video yet but I think Digital Foundry was using a 10980xe, or something similar. So in essence, they used the fastest CPU available and witnessed erratic numbers with the 3080 in 4K at max settings.

 

[Dan_D]

To be fair to all, even in the other video from Digital Foundry(on another thread) that guy also said DE was strange title for NV to use for some of their comparisons, not to mention how they were essentially throwing the 1080 Ti under the bus(I know we're talking about the 2080 Ti here). He stated how frame rates can be significantly different at different parts, and then again, how various cards were also having different behavior with them. I admit I haven't watched this video yet but I think Digital Foundry was using a 10980xe, or something similar. So in essence, they used the fastest CPU available and witnessed erratic numbers with the 3080 in 4K at max settings.

This is in line with my experiences. Doom Eternal is very erratic in my tests. Since the game lacks a built-in benchmark, I used Frameview to capture the performance data. That means running through a specific area in the game manually. It was so erratic that its hard to draw conclusions with it. Again, this supports my opinion that the 60% performance gap between a 2080 Ti and a 3080 probably doesn't translate to any other titles.

 

[MadMummy76]

To be fair to all, even in the other video from Digital Foundry(on another thread) that guy also said DE was strange title for NV to use for some of their comparisons, not to mention how they were essentially throwing the 1080 Ti under the bus(I know we're talking about the 2080 Ti here). He stated how frame rates can be significantly different at different parts, and then again, how various cards were also having different behavior with them. I admit I haven't watched this video yet but I think Digital Foundry was using a 10980xe, or something similar. So in essence, they used the fastest CPU available and witnessed erratic numbers with the 3080 in 4K at max settings.

I don't really get their goal here. The caginess makes me suspicious. I'd rather have 40% confirmed, than than this 80% which might be an outlier performance step. The digital foundry video is clearly an outlier because they compared frametimes with ray tracing only.

Either way it makes me suspicious, not confident.

 

[Peter_Brosdahl]

I don't really get their goal here. The caginess makes me suspicious. I'd rather have 40% confirmed, than than this 80% which might be an outlier performance step. The digital foundry video is clearly an outlier because they compared frametimes with ray tracing only.

Either way it makes me suspicious, not confident.

It's rare a manfacturer's claims are on 100% spot on but the solid take away is that there's significant improvent. I think also, you and I were talking in another thread and I said I had extreme doubts they'd release a 2080 Ti class, or better card, for under $1K. Glad to be wrong there.

 

[MadMummy76]

It's rare a manfacturer's claims are on 100% spot on but the solid take away is that there's significant improvent. I think also, you and I were talking in another thread and I said I had extreme doubts they'd release a 2080 Ti class, or better card, for under $1K. Glad to be wrong there.

Yeah I didn't expect them to do that. Especially with supply being low, they could easily get away with selling it at $1000 or possibly more. But I'm pretty sure we'll see them listed for $1000 on ebay.