Total War: Warhammer III Benchmark Test & Performance Analysis

[W1zzard]

On previous games the framerate would absolutely tank when the AI was taking turns - before the massive performance patch in TWWH2 which made the turns a lot faster it was very severe into the 2-4ish FPS range, and even after the patch it was a big dip (down to like 16-20 FPS). I am actually very surprised at how fast the end turn is in this release (given poor performance elsewhere), and how little the frames drop during the AIs turn, although I'm used to the full mortal empires map so we'll see how it performs when they launch an ME map with the TWWH 3 factions and map as well.

Given this config option, I suspect previously the model was running on the render thread, which kills FPS, now it defaults to threaded

 

[Vayra86]

Ultra presets do make a lot of visual difference compare to High, though it's just too demanding for current gen GPU. These game companies are good at making people buy new GPU .

This release really does smell (stink) of that principle yes.

I mean, why isnt the performance equal to WH2 at least? I think they crammed way too many effects on those maps, with very high detail levels or at least, expensive/inefficient detal levels.
Warhammer 2 got many performance boosts down the road, I'm expecting similar here.

That said, WH2 is no performance king either, but at least it runs smooth now. CA is making a shitload of money on this trio of releases, but I think adopting a new engine for the last part would not have been beneficial to us either, I mean, that would kill a lot of cross-compatibility that the games actually do use. Mortal Empires for example in WH2. Also I think we'd have been waiting for installment 3 a LOT longer in that case.

And last... I'm not specifically happy with the way the engine works in 3 Kingdoms. But that was mostly down to the mechanics they chose to implement, man, what on earth was that anyway. And those concepts echoed on the battle maps, which was basically an open field with a bit of fog and two chunks of units on either side of it.

So maybe this 'step back' is actually a blessing in disguise. I don't recall Troy being a smash hit either.

 

[droopyRO]

I would kill am hoping for a Total War 40k in the next 5 years or so But Lord Of The Rings might be more likely. For any of them, CA might change the engine.
On topic, i installed my "old" 1070 Ti, and with my custom playable settings i get this results. A 6600 XT = 1070 Ti. This game dose not like AMD hardware in it's current state. In Cyberpunk 2077 the 6600 XT is 50% faster than the 1070Ti but equal in Warhammer III ? Something is not right.

Also check out the graps and the CPU load, Nvidia has different results for some reason:

 

[boidsonly]

The Grandson is playing WH III on my 5950x/RTX 3090 PC and it seems to run well. I've only just peered at it occasionally while speaking with him. The monitor is a 27" 4K 60hz.

 

[nguyen]

This release really does smell (stink) of that principle yes.

I mean, why isnt the performance equal to WH2 at least? I think they crammed way too many effects on those maps, with very high detail levels or at least, expensive/inefficient detal levels.
Warhammer 2 got many performance boosts down the road, I'm expecting similar here.

That said, WH2 is no performance king either, but at least it runs smooth now. CA is making a shitload of money on this trio of releases, but I think adopting a new engine for the last part would not have been beneficial to us either, I mean, that would kill a lot of cross-compatibility that the games actually do use. Mortal Empires for example in WH2. Also I think we'd have been waiting for installment 3 a LOT longer in that case.

And last... I'm not specifically happy with the way the engine works in 3 Kingdoms. But that was mostly down to the mechanics they chose to implement, man, what on earth was that anyway. And those concepts echoed on the battle maps, which was basically an open field with a bit of fog and two chunks of units on either side of it.

So maybe this 'step back' is actually a blessing in disguise. I don't recall Troy being a smash hit either.

Three Kingdoms is actually my most played single player game on steam (400h+). Since I grew up reading about Romance of the Three Kingdoms, CA really captured the romaticized side of Three Kingdoms (where Lu Bu wipes out an entire army by himself ).

Just checked back with WH2 @ 4K Ultra + 8xMSAA, I get 35-40FPS on the map while I get ~45FPS on WH3 + TAA (which looks better), frametimes consistency and input lag are very good with WH3 that 50FPS could be considered playable.

Spoiler

 

[Vayra86]

Three Kingdoms is actually my most played single player game on steam (400h+). Since I grew up reading about Romance of the Three Kingdoms, CA really captured the romaticized side of Three Kingdoms (where Lu Bu wipes out an entire army by himself ).

Ahhh yes. Flavor wise and the romanticized part of it, that definitely struck me as refreshing, it just didn't stick with me so well. I used to play Dynasty Warriors back in the day on playstations, but all I could really do was zerg/button mash and laugh at the whole thing for being so idiotic Fun, though, for sure.

I might go into it again with that pair of glasses on... maybe it'll work better that way. The whole ominous uber warlord thing is cool, for sure.

I would kill am hoping for a Total War 40k in the next 5 years or so But Lord Of The Rings might be more likely. For any of them, CA might change the engine.
On topic, i installed my "old" 1070 Ti, and with my custom playable settings i get this results. A 6600 XT = 1070 Ti. This game dose not like AMD hardware in it's current state. In Cyberpunk 2077 the 6600 XT is 50% faster than the 1070Ti but equal in Warhammer III ? Something is not right.

Also check out the graps and the CPU load, Nvidia has different results for some reason:
View attachment 237365

Sounds like a classic case of AMD driver regime versus Nvidia game ready drivers.

 

[droopyRO]

Or like someone said, it might like Pascal architechture. It would be interesting if someone here would have a moment or two to test out a GTX1080 Ti at 1440p.
I found an old video i made. The first TW Warhammer, benchmark at 1440p Ultra settings on a 980 Ti and 3770K. I got an average of 60 fps vs an average of just 34 in the newest title with a Ryzen 2700X and 6600XT. Bleah.

 

[Denver]

Gamers: "Wow, that's cool now that consoles are much more powerful, we'll have a big graphical advance in the next releases"

Game Studios: "Now we're going to release games with the same graphics as a decade ago, and skip the bug fixing and optimization step. Oh, and also raise the prices."

Just buy it

 

[droopyRO]

It was 48 Euros. That is not more expensive than Warhammer II or Warhammer I were at launch.

 

[kapone32]

I was intrigued by the Crossfire support question. I am borrowing a card from a mining rig to see if Crossfire is still supported by changing the script. It did not work with 3Kingdoms but was fully supported in TWWH2.

 

[pathstrider]

I was intrigued by the Crossfire support question. I am borrowing a card from a mining rig to see if Crossfire is still supported by changing the script. It did not work with 3Kingdoms but was fully supported in TWWH2.

I suspect not - the decision to stop supporting SLI/Crossfire was linked to the move to TAA in 3k. TAA has made it over to wh3, so same might apply there?

 

[kapone32]

I suspect not - the decision to stop supporting SLI/Crossfire was linked to the move to TAA in 3k. TAA has made it over to wh3, so same might apply there?

I know that they have said that but it seems to run worse than the 3K. When I get the card I will see what it can do. The script does look a little different than 3k's.

 

[pathstrider]

I know that they have said that but it seems to run worse than the 3K. When I get the card I will see what it can do. The script does look a little different than 3k's.

Fingers crossed then!

 

[kapone32]

Or like someone said, it might like Pascal architechture. It would be interesting if someone here would have a moment or two to test out a GTX1080 Ti at 1440p.
I found an old video i made. The first TW Warhammer, benchmark at 1440p Ultra settings on a 980 Ti and 3770K. I got an average of 60 fps vs an average of just 34 in the newest title with a Ryzen 2700X and 6600XT. Bleah.

You can't compare the first Game to this one. This Game is so much more than the first. The benchmarks are set to run at whatever your Maximum refresh rate is anyway (if your PC is powerful enough). If you have a panel that has variable refresh rates you will notice that the benchmark and initial screen will run at 120 or 165 FPS based on your setting and so will the benchmark (mostly). The actual Game though is a monster that really needs to leverage cores but has not been optimized to do that. The biggest influence on FPS though is Unit size. If you want Ultra you need to have the performance.

Fingers crossed then!

Yep I hope so. If there ever was a case for Multi GPU it is definitely Total War. Except 3 Kingdoms.

 

[pathstrider]

So as an interesting note, I tested out my 3060 with an old 4690k 4c/4t cpu and it worked surprisingly well on ultra. In the battle bench it wasn’t significantly cpu bottlenecked, which was a surprise.

The game does seem to use more cores than 2 did if it needs too but they seem to have optimised it so that it doesn’t need to most of the time. Unit size doesn’t seem to be a big of a hit as before.

This is unlike the last game, Troy, which spread itself out up to 16 threads.

 

[droopyRO]

I didn't notice a framerate difference between Large and Ultra unit size on my 3770K, 8600K, 2700X or 12600K during the 5 years since Warhammer 1 launched. Plus unit size matters only in battles, not in campaign map.
And i don't know what happens in WH3, or how it is ballanced, but in 1 and 2, spells worked different depending on the unit size. Some spells were more powerful the less units you had, so on Large unit scale for example. So i always opted for Ultra. Legend always plays on Ultra so that is what i stuck with since 2017.

I see no CPU bottleneck in WH3 compared to WH2.
Warhammer III is very good at using the CPU. Check out this comparison:

vs Warhammer II, don't have a 1024x768 test yet. but i have a 1280x720 one. The 2700X barley makes it to 60 fps in WH2. Yet the same 2700X exceeds 150 fps in WH3 no problem !
Yeah, so i did a quick test. In 1024x768 Low settings + Ultra unit size, The 2700X performs way better than in 720p with Ultra preset.
But the Blood DLC makes the framerate drop from about 130-150 fps(with it off) in the Skaven Benchmark to about 70 with it on. Yeah.

 

[pathstrider]

It was confirmed that wh2 was balanced around large.

As far as we can tell, it’s not supposed to matter for 3. Certainly single entities scale better this time. Unsure about magic.

 

[droopyRO]

Total War: WARHAMMER II NVIDIA's Recommended GeForce GTX GPUs For 60 FPS Gameplay

Play Creative Assembly’s Total War: WARHAMMER II at 60 FPS at 1920x1080 on a GeForce GTX 1070, and at 4K with max details on GTX 10-Series SLI setups.

www.nvidia.com

If you wish to increase graphical fidelity to Ultra, or beyond, look to the GTX 1080 at 1920x1080, and SLI configurations for 2560x1440. Or if you're happy with a lower framerate, pair a more affordable CPU and GPU config with a G-SYNC monitor to mask frame rate variances.

A 1080Ti was recommended for a 1440p 60 fps with High(not Ultra) preset. For 4K you needed a GTX 1080 TI SLI according to Nvidia
I remember this guide. I had a 980 Ti and a 1440p monitor. I really wished i had a 1080 Ti, but they were "expensive" just before the mining boom of 2017-2018. Biggest mistake i ever made with PC components was to not buy a second hand 1080 Ti in 2018 after the crypt crash. I waited for the new generations and the perfromance/price that we never really got.

PS: i have about 8 hours in WH3, all of them tinkering with benchmarks and hardware and about half an hour in to the Prologue. Good thing i have Spellforce 3 Reforged to entertain me while i wait a few weeks for a couple of patches
As i suspect that Mortal Empires will be delayed for at least 6 months even in to 2023, based on the actual state of the game. Both in hardware performance and in general gameplay balance and stuff.

 

[Devaraja]

Why does the comparison is 10 Nvidia cards vs 8 AMD cards? would it make more sense to remove the 2080 as there is already a 2080 ti on the chart and add the vanilla 6800. it would make a more fair comparison because for nvidia there is the 3070 and 3080 on the chart.

 

[W1zzard]

Why does the comparison is 10 Nvidia cards vs 8 AMD cards? would it make more sense to remove the 2080 as there is already a 2080 ti on the chart and add the vanilla 6800. it would make a more fair comparison because for nvidia there is the 3070 and 3080 on the chart.

Denuvo limits the number of GPU changes to five per 24 hour period, so I had to make due with a smaller set of cards

 

[Devaraja]

Denuvo limits the number of GPU changes to five per 24 hour period, so I had to make due with a smaller set of cards

but if you remove one Nvida card it would still be 18 cards total, 9 AMD and 9 Nvidia instead of 8 and 10.

 

[W1zzard]

but if you remove one Nvida card it would still be 18 cards total, 9 AMD and 9 Nvidia instead of 8 and 10.

If I remove one card, I have one less in the charts and still have to wait 24 hours for more activations. Next time I'll plan this a bit better. 6800 is roughly in the middle between 6700 XT and 6800 XT

 

[Devaraja]

If I remove one card, I have one less in the charts and still have to wait 24 hours for more activations. Next time I'll plan this a bit better. 6800 is roughly in the middle between 6700 XT and 6800 XT

Sorry i dont want to sound disrespectful of your work. i appreciate all the work you do, it just that i still don't understand why it is an issue you still have to wait for the next activation and is just swapping card for another, it means the same time just a different card. i understand that the 6800 is in the middle of the 6700 xt and the 6800 xt but we could say the same for the 3070 is in the middle of the 3060 and 3080 and that si still added.

I just wanted to understand the reason behind the card selection.

 

[W1zzard]

I just wanted to understand the reason behind the card selection.

I started benching, with the usual cards in random order. Then I was like "WTF" when I ran out of activations and had to prioritize cards
3070 Ti isn't included either

Steam Hardware & Software Survey

What if "fair" meant according to market share?

 

[Devaraja]

Steam Hardware & Software Survey

What if "fair" meant according to market share?

That would be totally ok with me if the idea of the article is to see how the game runs on the most used / common cards on the market. if they idea is to show how does it rum against different cards vendors, I think it should be or try to be like 50/50 until intel bring the new GPUS and in that case it should try to be 33.33% each brand. But is just my opinion and I know is hard to please everyone.