Assassin's Creed Shadows Performance Benchmark

[Vayra86]

Users, I guess

View attachment 390992

Look, TPU defines the benchmark suite not by (always varying) player counts, but by notions such as:

- what engine is used, covering all relevant game engines in the industry
- game genres
- diversity
- results that can be replicated and have no variance

Player count is just a tiny sliver of what matters. GPU reviews are not game performance reviews.

 

[njshah]

What a disappointment this game has been honestly , no wonder only 60k people on weekend on steam.

An optimization mess that channels like DF have been bribed to praise about , threat interactive was right about digital foundry.

 

[cinemaware]

Lol..."bribed"..."unpotimized"...etc...etc. Another day, another big game release hysteria/drama.

Now, I don't know how much more this game could or couldn't be really optimized, I guess the only people that do are the devs. What I do know, though, is that I can't really recall any instance when a very good looking, AAA open world game wouldn't push the hardware to the limit. KCD2, you say? Hmmm....not sure as I haven't played it myself yet, but tbh it doesn't look that much an upgrade from the 6 year old 1st one. It's also dangerous to do direct comparisons, since no two engiens/games are the same.

Well, I guess I will see when I finally get over that thing we can't mention here and get this game eventually. So far it seems maybe higher end config with FG could run it at 4K with RT, which is quite unsurprising.

In other news, big kudos to PC Gamer for doing what everyone else in the benchmarking biz should: using different CPU/GPU combos to test.

 

[Macro Device]

KCD2, you say?

Does kinda look worse than this Ubisoft product but when framerates on 6700 XT are higher in KCD2 than on 5080 in Assassin's Creed you do realize it's not how it's supposed to be optimized.

When you release a game, any game, no matter how big and ambitious it is, you should make the entry point reasonable. When the entry point is a 5 hundred dollar GPU (1080p Low, 1% lows at >60 only start with the likes of 4070) it desperately begs for a cancellation.

Why releasing a game that no one can play? We don't need no more Crysis games. Especially when GPUs faster than the middest of mids are made of pure unobtainium. Not only you can't play it on your current machine, you also can't upgrade it to be able to.

This is why this game despite maybe being awesome in everything that's not performance has said performance way too low to pass. It's expensive within no reason.

 

[JustBenching]

Look, TPU defines the benchmark suite not by (always varying) player counts, but by notions such as:

- what engine is used, covering all relevant game engines in the industry
- game genres
- diversity
- results that can be replicated and have no variance

Player count is just a tiny sliver of what matters. GPU reviews are not game performance reviews.

Player count is the most important metric apparently. The new TPU gamelist will be

1) League of legends
2) Dota 2
3)Counterstrike
4) Minecraft
5) PUBG battlegrounds
6) The banana game
7) Stardew valley


Oh well, those 5090s can't wait to stretch their legs in the newly added games.

 

[cinemaware]

Why releasing a game that no one can play?

Why use extreme hyperbole when the fact is that 2 million people are already playing this game?

The PC Gamer bench I linked shows the likes of 5600x + RX 6750XT pulling 54-65 fps in 1080p on Medium, 40-50 in 2K, and that's before upscaling. That should be enough for plenty of folks and, I'm sorry, but in 2025 if you want to play in bigger leagues you do need to spend more.

And yes, we absolutely do need more "Crysis games". I want the envelope being pushed, not being stuck at the same point ad infinitum. In fact, it's somewhat depressing that 5-year-old CP2077 is still the best looking game out there.

Btw, I do play basic -looking retro, indie, roguelikes etc, games too, perhaps more than AAA titles.

 

[Macro Device]

5600x + RX 6750XT pulling 54-65 fps in 1080p on Medium, 40-50 in 2K, and that's before upscaling.

With horrible 1% lows. Only 30 FPS adepts consider that smooth. It's no more than technically playable.

I'm sorry, but in 2025 if you want to play in bigger leagues you do need to spend more.

You're just being an enabler here. No one should ever need to spend more "because it's year X and things ain't how they used to be." Nonsense. It's never been easier for game devs to do their job than it is now but we have a flop on top of a flop on top of a flop over and over and over again. Where are the AAA games worth spending on?

I want the envelope being pushed, not being stuck at the same point ad infinitum

You can push it two ways:

A. Disregard everything, mix the most ridiculous wunschpunsch of whatever cutting edge you can ever come up with and pretend it's okay you need to wait for a decade till a GPU that handles max settings starts to exist.
B. Carefully innovate and bring more QoL rather than fidelity improvements. Provide visuals improvements by doing a better job in the design department in the first place. Only apply cutting edge tech in extremely conservative ways and only apply them for highest settings. Avoid TAA as a core mechanic. Make sure an average gamer can get 1080p60 on high settings without 1% lows dipping below 60.

Way A is what we see here with this release. Everything looks, feels and smells like Ubisoft never wanted us to have more than 30 FPS and it's actually a tech demo. And whilst the game actually looks good, it doesn't make any sense how GPU taxing it is.

 

[cinemaware]

You're just being an enabler here. No one should ever need to spend more "because it's year X and things ain't how they used to be." Nonsense. It's never been easier for game devs to do their job than it is now but we have a flop on top of a flop on top of a flop over and over and over again. Where are the AAA games worth spending on?

Your move into the ad hominem territory enables me to do the same and dismiss your learned opinion as that of just another armchair dev who thinks that a line such as "Where are the AAA games worth paying for?" constitutes an argument.

This kind of posting, which forums are rife with, is just lazy noisemaking. It might score some points with the gallery, but has zero substance overall. To wit:

-the fps numbers I quoted were actually min-max, so the "terrible lows" you bemoan are actually 54 in the 1080p example and the others aren't that bad either, even from this very website (which uses Ultra / Quality settings, so it's easy to figure that dropping presets ad bit lower will improve things). For anybody with a VRR capable display, this really shouldn't be a problem. In any case, saying this is just "technically" playable smacks of snobbery. Millions of people put up with much worse on the daily, both on consoles and PC, across a wide range of games.

-overall, the above, plus "it's never been easier for devs" is some kind of parallel-universe trip. The tired Crysis trope aside, taxing games have always been around, since Atari 2600. Perhaps you're too young to witness the glory of ~3fps Castle Master on ZX Spectrum, Elite II on Amiga, or R-Type on SNES, but surely Half Life 2, Deus Ex, or even The Witcher 3 ring a bell?

-continuing with this theme, I'm not sure in what year "things" (presumably CPU/GPUs) were so cheap that all gamers could afford blazing fast rigs and play everything maxed out? Certainly not in my living memory, since I was always forced to buy second hand and tweak the settings in AAA games if I wanted to have decent framerates.

-if you read the Digital Foundry breakdown you’d know that they actually tried pretty hard to make it both cutting edge and accessible too. Ah, no, wait, DF are paid for shills, so that doesn't count, I guess. Even so, if you actually think a bit harder about your claim that "Ubisoft never wanted us to have more than 30 FPS", do you really - I mean, really - think it makes any kind of sense?

 

[10tothemin9volts]

AMD RT performance really good in this game...

Good time for AMD Radeon

True in the sense that normally in ray-tracing at 1440p the 5070 Ti 19% faster (see further in the post for ray-tracing) and similarly more power efficient than the 9070 XT, but in this game the 9070 XT is 7% (67.1 FPS/62.5 FPS) faster and 6% ((300W [5070 Ti]/62.5 FPS)/(304W [9070 XT]/67.1 FPS)) more power efficient (if one assumes that both GPUs were full utilized, instead of actually measuring the energy comsumption).
However, in the much more visually spectacular path-tracing (which this game seems not to use), the 5070 Ti is over 50% faster and similarly more power efficient than the 9070 XT.

The game does indeed look dull (and is unoptimized(?)) when not using the ray-tracing setting.

 

[Macro Device]

-overall, the above, plus "it's never been easier for devs" is some kind of parallel-universe trip.

You're talking to an actual developer (sure, I only worked for 120 hours but I have actual experience and was stunned at how easy it is to do stuff in modern engines). I know how easy it is. Tried old school methodologies (not that old school tho; late 2000s/early 2010s), it's at least seven times more work for the same end result. I can't for the life of me stress enough how little devs put into actual optimization these days.

54 in the 1080p example

Yes, it's horrible. Anything below 60 is horrible.

in what year "things" (presumably CPU/GPUs) were so cheap that all gamers could afford blazing fast rigs and play everything maxed out?

Never? But we've had it better than today and there's no valid excuse for that.

Millions of people put up with much worse on the daily, both on consoles and PC, across a wide range of games.

Millions of people are also dying of cancer. Is cancer okay now?

if you read the Digital Foundry breakdown you’d know that they actually tried pretty hard to make it both cutting edge and accessible too

The only way I see it being legit is by Ubisoft running into engine limitations and not being able to say, "whoopsie, we need a better engine" before too late.
But I'll read it tomorrow.

do you really - I mean, really - think it makes any kind of sense?

I do. 300-dollar GPUs are marketed as legit 1080p devices. Sure, both AMD and NVIDIA keep forgetting to deliver since 2022 but that's not the point. The point is this game is only good at 1080p if you get into the lowest preset on these GPUs.

 

[Luminescent]

I know it sounds crazy, but, what if just like in politics, there are people paid to bitch about some game that is Nvidia or Amd sponsored/optimized, the competition trashes that game even if it looks and works great at reasonable settings, everyone blames the "unoptimized game" and doesn't blame Nvidia or AMD for that 1000$ gpu that he/she clearly overpaid.
If they sabotage themselves with the usual don't implement FSR or DLSS why wouldn't they go a step further ? Just a thought.

 

[R-T-B]

View attachment 390903

lol. Always 60+ eh?

You did not read his whole post eh?

So interpolated frames with artefacts count as performance for you? And all games support this, yes?
Well, enjoy then.

Where on earth did you get "all games" from "I play this one game my way and it gets 60FPS+?"

Strange. I'm on max settings but everything looks grey and washed out like the photos on low settings. Maybe a patch will fixed this.

Are you possibly enabling HDR on a display that supports it poorly (or not at all)?

You're talking to an actual developer (sure, I only worked for 120 hours but I have actual experience and was stunned at how easy it is to do stuff in modern engines). I know how easy it is. Tried old school methodologies (not that old school tho; late 2000s/early 2010s), it's at least seven times more work for the same end result. I can't for the life of me stress enough how little devs put into actual optimization these days.

As a dev who worked years on a few massive (and one incomplete homebrew) projects this sounds very much like someone who doesn't have a very big project yet, no offense intended (as we all thought like this at some point).

Making an actually large game quickly and also optimizing it is basically incompatible. Every big publisher wants stuff quickly now because time is money. So one thing gets a pass. Guess what it is?

 

[Macro Device]

Making an actually large game quickly and also optimizing it is basically incompatible.

Somehow devs from yesteryears had it even worse and still managed to do roughly the same job optimizing stuff, if not better. Today it's "some 50odd seconds on Google and you found a dumb, bad, but working solution/workaround" on top of most everything working despite you writing zero code lines. Then, it was like "maybe you can find a Greek guy who barely speaks any English who solved this problem but he spiralled into heavy drinking so good luck making him remember what he did." And thousands upon thousands manhours of implementing baked lighting.

doesn't have a very big project yet

True but 'twas a smooth sailing. Surprisingly smooth since I am an idiot but this thing works on my machine that doesn't even have a UE5-friendly GPU.

 

[R-T-B]

Somehow devs from yesteryears had it even worse and still managed to do roughly the same job optimizing stuff

It's called not having a publisher that wants it by yesterday. The problem is higher up then the devs in 90% of cases.

I'm sure there are some lazy devs out there, but most I worked with were more borderline OCD types than lazy. And it drove our higher up nuts when they asked to take time to optimize something that was "already hitting the needed performance targets."

maybe you can find a Greek guy who barely speaks any English who solved this problem but he spiralled into heavy drinking so good luck making him remember what he did.

Ironically, of every line you've written,I relate to this one the most. God, we've all seen this and why?

Oh you meant "then." Oh it still happens, you just haven't had an obscure enough project issue yet. The odds go up exponentially as the codebase grows, and also if the engine build you chose becomes dated.

 

[Macro Device]

It's called not having a publisher that wants it by yesterday. The problem is higher up then the devs in 90% of cases.

I'm sure there are some lazy devs out there, but most I worked with were more borderline OCD types than lazy. And it drove our higher up nuts when they asked to take time to optimize something that was "already hitting the needed performance targets."

Sure, not spilling any blame onto programmers themselves as I'm a CEO myself so I'm strictly criticizing the management alone. Unrealistic goals + greed = grim results.

 

[R-T-B]

Well frankly yeah, that's it in a nutshell. See also why I fled that field for a IT security related job.

That said, don't let me scare you. It's still a massive market and if you can afford to let people do things right, ripe with potential. Especially if you take the time to know the tools yourself! Yes the new engines are pretty awesome compared to the complexity level of yesteryear so if you know whats needed you can do real well. That's the way a company should be managed frankly, so wishing you the best of luck!

 

[Macro Device]

Especially if you take the time to know the tools yourself!

I decided to go full YOLO and do ALL the code myself. With 0 previous experience. It's either I make a fool of myself (likely) or of AAA companies (desired result).

if you can afford to let people do things right

I can. I'm young (30 y.o.) and healthy enough to live for another thirty and I want to make a game I would've loved to play myself (because no one else will do that for me) so quality is a higher priority than speed. Or money. Fuck money. I'm doing it for the sake of it.

Thank you!

 

[wolf]

I keep telling people...This is what to start expecting. This isn't going to be abnormal, and is in-fact the new normal. And 9070 XT is doing exactly as it should.

Care to elaborate on why you believe this is the case? AMD's own slides and expectations for the product don't match it, and the averages from many games so far in independent testing don't either. There are examples like this, and examples where it also significantly underperforms its average across all games. What is it about this game that has you thinking this is the new normal?

Still genuinely curious if you have good reasons for this assertion, you've been active since I replied so I wonder if you missed it or don't want to reply.

 

[Arkz]

You did not read his whole post eh?


Where on earth did you get "all games" from "I play this one game my way and it gets 60FPS+?"


Are you possibly enabling HDR on a display that supports it poorly (or not at all)?


As a dev who worked years on a few massive (and one incomplete homebrew) projects this sounds very much like someone who doesn't have a very big project yet, no offense intended (as we all thought like this at some point).

Making an actually large game quickly and also optimizing it is basically incompatible. Every big publisher wants stuff quickly now because time is money. So one thing gets a pass. Guess what it is?

Yes, did you? His OP said "Always 60+"
I was implying "all games" facetiously as there's loads of games maxed in 4k on a 5090 with framerates massively below 60, so you can't in fact run everything in 4k max on a 5090 at always 60+. You Have to drop settings, use DLSS to upscale from a lower render res, and use interpolated crap to get there.

 

[R-T-B]

Yes, did you?

The one you quoted outlined it well. Pretty sure a 5090 can do damn near anything at 60 if we concede framegen anyways (which we are forced to do because it was his argument, not yours).

Now there is certainly an argument against those fake frames being "real" but thats not what I witnessed, at all.

 

[njshah]

Lol..."bribed"..."unpotimized"...etc...etc. Another day, another big game release hysteria/drama.

Now, I don't know how much more this game could or couldn't be really optimized, I guess the only people that do are the devs. What I do know, though, is that I can't really recall any instance when a very good looking, AAA open world game wouldn't push the hardware to the limit. KCD2, you say? Hmmm....not sure as I haven't played it myself yet, but tbh it doesn't look that much an upgrade from the 6 year old 1st one. It's also dangerous to do direct comparisons, since no two engiens/games are the same.

Well, I guess I will see when I finally get over that thing we can't mention here and get this game eventually. So far it seems maybe higher end config with FG could run it at 4K with RT, which is quite unsurprising.

In other news, big kudos to PC Gamer for doing what everyone else in the benchmarking biz should: using different CPU/GPU combos to test.

Except ubisoft were caught with their pants down with proof during starwars release , cant believe you missed that.

The game doesn't even look that great , getting 80 fps on a 5090 at 1080p is bizzare.

If you want technical breakdowns search threat interactive on YouTube , as a developer i can assure you we are aware of the issues the industry faces , problem is the MBAs incharge only care about the cash and time.

 

[vacsati]

Once again the fantasy vram usage not correlate with actual fps.

4k ray tracing vram usage almost 12gb, meanwhile even in minimum fps literally no difference between the 4060ti 8/16gb and 3070 8gb.