Wolfenstein II: The New Colossus Performance Review

[Yannick_G_THFR]

Wolfenstein II: The New Colossus marks the franchise's return, this time with MachineGames and the id Tech 6 game engine. The engine is packed with all kinds of technology, it utilizes the Vulkan API, and it promises a lot in terms of rendering quality.

Wolfenstein II: The New Colossus Performance Review : Read more

 

[HEXiT]

something i noticed about this game. i have an ancient 920@3.5 and a gtx 970
i can run the game at near max settings (just turn the textures down from ultra) and when vsync is on it shows a solid 60fps... i turn vsync off and get between 40 and 55 fps. so im guessing the fps counter isnt that accurate in game.

 

[toddybody]

Not sure why the RX 64 and GTX 1080 Benchmarks are separate as a "bonus"...and why other top tier (i.e. 1070, 1080ti) cards arent included.

Don't have one myself, but 4K/2160p benchmarks would be nice too. Kinda underwhelmed, considering how long the title has been out.

 

[Sakkura]

Would have liked to see a 4GB card in the mix, and ideally even the 1060 3GB. So we could see how much VRAM is actually needed for solid performance. VRAM utilization is imprecise because games can allocate a lot more VRAM than they really need.

I would also like to have seen other CPUs tested. It's great that the game uses 6 cores and 12 threads, but will it still run well on 4c/4t? Lots of people are still on processors like that.

 

[spdragoo]

Would have also been nice to see some actual CPU benchmarking, especially since the spread of minimum/recommended CPUs represents a very wide range (i.e. Ivy Bridge Core i5/i7 CPUs & 4C/8T FX CPUs all the way up to current Ryzen CPUs), as well as some idea as to whether newer Intel CPUs have much of a boost over the 3rd/4th-gen versions.

Something strange as well: the minimum/recommended Intel CPUs were 4C/4T or 4C/8T CPUs, implying that you need at least 4 physical cores to run this (i.e. just having 4 threads won't work, so no 2C/4T Core i3/Pentium CPUs). And that's kind of supported by the listed FX CPUs. But why would Machine Games say that you can't use a Ryzen 3 (4C/4T) CPU to run this game? The R3 1200 is almost identical to the minimum R5 1400 listed (same Boost/XFR speeds, only 100MHz slower on base, & is a 4C/4T CPU vs. the 4C/8T 1400), & the R3 1300X runs almost as fast as the 6C/12T R5 1600X. Also, would this perhaps be a game that a Coffee Lake Core i3 (4C/8T) could handle, or would you still need to use a Core i5 or i7?

 

[spdragoo]

Would have liked to see a 4GB card in the mix, and ideally even the 1060 3GB. So we could see how much VRAM is actually needed for solid performance. VRAM utilization is imprecise because games can allocate a lot more VRAM than they really need.

I would also like to have seen other CPUs tested. It's great that the game uses 6 cores and 12 threads, but will it still run well on 4c/4t? Lots of people are still on processors like that.

Just what I was wondering, especially since they listed Ivy Bridge/Haswell Core i5 (4C/4T) CPUs in the minimum/recommended CPU sections.

 

[Kahless01]

did you read different articles than i did? there is a damn 1060 3g and several 4gb cards included in the test. the 3g 1060 takes a huge hit compared to the 6g.

 

[quilciri]

Looks like the game requires a large amount of GPU memory, and doesn't necessarily need the highest end GPU. I don't know how they came to that conclusion that they did when the 8GB 390 did so well in the test.

 

[quilciri]

Would have liked to see a 4GB card in the mix, and ideally even the 1060 3GB.

There were four 4GB cards tested, and so was the 1060 3GB.

What it looks like they should have tested, given their results, is something along the lines of an 8GB 560 - a mid or lower range GPU with a large amount of memory.

 

[phobicsq]

I thought it runs really well all maxed out. I was a bit sad there wasn't a lot of openess though. It was rather linear.

 

[spdragoo]

Would have liked to see a 4GB card in the mix, and ideally even the 1060 3GB.

There were four 4GB cards tested, and so was the 1060 3GB.

What it looks like they should have tested, given their results, is something along the lines of an 8GB 560 - a mid or lower range GPU with a large amount of memory.

Don't know about that, since the minimum was supposed to be the GTX 770. But I do think they should have used a different GPU list:

• 
  ■ They probably should have skipped the GTX 1050 or RX 460, as both are 2GB GPUs (well below the supposed minimum 4GB VRAM threshold) & well below the minimum GTX 770/R9 290 minimums. Although it did confirm that low-end cards aren't going to cut it. Maybe they would have been better in a follow-up article, i.e. "Can low-end GPUs handle Wolfenstein II?".
  ■ They should have tested the GTX 770 & R9 290, since both are listed as the minimum GPU needed for the game. Yes, I know that the 6GB GTX 1060 is roughly comparable (1 tier up from the 770, same tier as the 290), but there have been a number of games where similarly-tiered GPUs don't always have similar performance.
  ■ Not only was it strange that the GTX 1080/RX Vega 64 testing was "bonus" testing, but they didn't even bother testing with the GTX 1070/1070TI (or even anything like the Fury X or Vega 56). Considering that those GPUs are the current recommendation for 1440p gameplay (which was a resolution they tested), it would have been nice to see that testing.

 

[RCPG]

Well, I have a FX-8350 with an ASUS Radeon R9 390 8G STRIX and 16GB DDR3 2400MHz CL11 and play the game at 1080p, max settings (mein leben!), with V-SYNC on, almost always at 60 fps. So, you can play well with a "moderately weak" system. My hardware is not "properly recent".

 

[Sakkura]

Would have liked to see a 4GB card in the mix, and ideally even the 1060 3GB.

There were four 4GB cards tested, and so was the 1060 3GB.

What it looks like they should have tested, given their results, is something along the lines of an 8GB 560 - a mid or lower range GPU with a large amount of memory.

Only 6GB and 8GB cards in the VRAM utilization test.

 

[Blytz]

Secondary conclusion, apparently no-one has a Vega 56 for this game

 

[jdwii]

Very happy to see a new engine be made properly now we have several engines that support multithreading correctly one thing that surprised me was it looked like no single core in the 1600X was pegged to 100% meaning no bottleneck. In a lot of older DirectX 11 games it would support 4-6 cores but often one core was pegged to 80%+ all the time.

Also i think with Doom and now this game Vulkan has earned its name and i believe it has shown more success then Directx 12

 

[cinergy]

Good to see Vega shine bright here.

 

[scoffmeister]

I had pre-ordered the game months ago, and when I finally received it and then looked at the system requirements I was thinking I wasn't going to be able to run it. I have a venerable AMD Phenom II X4 965 BE clocked at 3.9 GHz, and the minimum specs for Low Settings @720P list a Core i7-3770 or FX-8350 and so I was thinking I was going to have issues due to the number of cores/threads of my CPU vs. the newer ones (the requirements listed on the box do not include the Core i5-3570 or Ryzen 5 1400.) I have the minimum recommended 8 GB of system RAM, but I have a pair of nVidia GTX 960's with 4GB of RAM each running in SLI and have had no issues playing the game at all. I'm running it on Ultra Settings @ 1440P (2650x1440) and have been averaging 45 FPS, max (capped) 70, and min 30 with no screen tearing or artifacting, etc. Some might think avg. 45 FPS to be too slow, but I haven't seen any issues with smoothness. I'm actually running a second monitor as well without issues. So I agree with many of the assessments given here; it's far more GPU and GPU RAM dependent and the CPU requirements are exaggerated.

 

[quilciri]

Would have liked to see a 4GB card in the mix, and ideally even the 1060 3GB.

There were four 4GB cards tested, and so was the 1060 3GB.

What it looks like they should have tested, given their results, is something along the lines of an 8GB 560 - a mid or lower range GPU with a large amount of memory.

Only 6GB and 8GB cards in the VRAM utilization test.

Lol, that's because all the 4GB cards were capped out in memory utilization. If you looked at the charts you see that the *least* memory utilized is 5.2GB. Including 4GB cards there would have been worthless information. They are in every test where the results are actually in question.

 

[Sakkura]

Lol, that's because all the 4GB cards were capped out in memory utilization. If you looked at the charts you see that the *least* memory utilized is 5.2GB. Including 4GB cards there would have been worthless information. They are in every test where the results are actually in question.

But that's the point, the game isn't really using 5.2GB actively, that's just what it preallocates when there's more available.

 

[quilciri]

But that's the point, the game isn't really using 5.2GB actively, that's just what it preallocates when there's more available.

What part of the article gave you that idea?

 

[Sakkura]

But that's the point, the game isn't really using 5.2GB actively, that's just what it preallocates when there's more available.

What part of the article gave you that idea?

3GB card outperforming 4GB cards.

 

[quilciri]

But that's the point, the game isn't really using 5.2GB actively, that's just what it preallocates when there's more available.

What part of the article gave you that idea?

3GB card outperforming 4GB cards.

It is a 1060 (the lesser GPU) and the cards it is outperforming, it outperforms on pretty much any game.

Someone correct me if I'm wrong, but GPU memory doesn't get reserved like system memory. If there is GPU memory usage, it is because it is actually being used by game assets copied from system memory.

 

[Sakkura]

It is a 1060 (the lesser GPU) and the cards it is outperforming, it outperforms on pretty much any game.

Someone correct me if I'm wrong, but GPU memory doesn't get reserved like system memory. If there is GPU memory usage, it is because it is actually being used by game assets copied from system memory.

Yes, but it shows that it isn't being hurt by having 3GB of VRAM. Because if it was, weaker cards with more VRAM would be catching up to it.

And I already made that correction before. VRAM does get pre-allocated.

 

[quilciri]

It is a 1060 (the lesser GPU) and the cards it is outperforming, it outperforms on pretty much any game.

Someone correct me if I'm wrong, but GPU memory doesn't get reserved like system memory. If there is GPU memory usage, it is because it is actually being used by game assets copied from system memory.

Yes, but it shows that it isn't being hurt by having 3GB of VRAM. Because if it was, weaker cards with more VRAM would be catching up to it.

And I already made that correction before. VRAM does get pre-allocated.

I actually looked it up after I posted. It appears memory on a video card is *not* pre-allocated in any form meaningful to these benchmarks. What I mean is that the only pre-allocation happens immediately before something is trasnferred to the space. For all intents and purposes here, If memory is shown as used on a video card, it is because there is an asset stored, not because it's reserving the space.

Also, the average fps difference between the two different 1060 models isn't as big for most games as it is here, which points to memory being an issue.

https://www.gamersnexus.net/hwreviews/2604-gtx-1060-3gb-vs-6gb-benchmark-review/page-4

 

[Sakkura]

I actually looked it up after I posted. It appears memory on a video card is *not* pre-allocated in any form meaningful to these benchmarks. What I mean is that the only pre-allocation happens immediately before something is trasnferred to the space. For all intents and purposes here, If memory is shown as used on a video card, it is because there is an asset stored, not because it's reserving the space.

It's already well established that many game engines do in fact allocate VRAM they don't really need to run smoothly. Here's a hairy guy explaining it, at 15:24:

[video="https://www.youtube.com/watch?v=yTMnP59we-w"][/video]

It doesn't really matter if there's something actually stored in that chunk of memory - what matters is how much is actually in use, in such a way that having less VRAM would hurt performance.