Performance and hardware comparison of Final Fantasy XIV and World of Warcraft
There is a lot of comparison between WoW and FF14 lately. You can compare Y'shtola to Y'shaarj and so on but today I've tried to compare the performance and hardware/software characteristics of both game. Let's do some benchmarks.
Final Fantasy XIV benchmark app
FF14 has a free and easy to run benchmark application that can be used to test your PC performance in the context of this game. It also has a character creation section allowing you to create a character and the see how it looks and behave in various scenarios.
You don't need any account or registration to run the benchmark, but it's only available for MS Windows (while the game also runs on macOS and PlayStation).
WoW versus FF14 - the technical stuff
World of Warcraft runs on Windows and macOS including Apple M1 native version. Recently support for AARCH64 Windows 10 (ARM) was also added. FF14 runs on same systems without Windows ARM support and handling Apple M1 via Rosetta emulation (no official support). However FF14 is also available on PS4 and PS5. Both games are not present on game streaming services like GeForce Now even though FF14 is also on Steam.
FF14 lists Intel Core i5 2.4GHz or higher, GTX750/R7 260X and Windows 8/10 64-bit as minimal PC requirements with GTX970/RX 480 as recommended. World of Warcraft lists similar i5-3450/FX-8300 and GTX 760/RX560 as minimal. Recommended specs are somewhat higher - i7-6700K/Ryzen 2700X and GTX 1080/RX Vega 64. Both games list 4GB of RAM as minimal and 8GB as recommended, but WoW already has problems with some areas on 8GB of RAM while Shadowbringers benchmark uses 4-5GB on average (but it's not the live game).
Overall both games seems to be in same ballpark of minimal requirements. WoW tends to also put some spotlight on integrated graphics and older hardware. I’m not sure what are Square Enix policies on deprecating/removing support for old PC hardware though.
Benchmarks
For the benchmarks I used FF14 Shadowbringers benchmark app - that got replaced by Endwalker version on the day I wrote this articles ;). This isn't the live game but gives a quick and repeatable way of benchmarking the game engine.
The benchmark app won't report in detail about FPS but will instead give a score. Score below 1000 means your hardware does not met the minimal specification, while scores below 4000 should be able to run the game on low settings after some tweaking. 4000-8000 is for default settings, while anything above is high - option to use a higher resolution or extra quality settings.
WoW does not have any benchmark mode and for my benchmarks I've picked specific in-game locations and activities to have as repeatable results as possible. You can read on that in my previous benchmarks.
FF14 benchmark app runs few different scenes/scenarios. For Shadowbringers it looks like so:
▸FF14 Shadowbringers benchmark
The benchmarks were run on Windows 10 with all updates installed and with latest drivers (AMD 21.6.2, Nvidia 471.11). If not stated otherwise 2x8GB 3200CL15 RAM was used. Resolution was 1080p and the apps were installed on a 512GB SATA SSD. Ryzen 4800U benchmarks were done on a Lenovo IdeaPad 5 14ARE05 laptop.
For FPS results each scene of FF14 was measures separately: duel, city, open world raid boss, open world and raid boss scenes respectively.
Integrated graphics
iGPU benchmarks of FF14 were done with the Standard laptop
preset and 1080p resolution:
The opening scene has a lot of particle effects in few attacks which tanks 1%/0.1% FPS while having high FPS in other parts of the scene. If you would constantly run into such particle effect density then the game would not be playable on those iGPU systems. Let us look at more realistic scenarios:
In the city scene we have lots of NPCs yet the CPU part of the APU handles that really well. When we get to open world big raid encounter or even the later open world travel the amount of effects/assets hits the FPS scores. To avoid stutter you would have to limit the density and quality of particle effects and assets. This is also true for WoW - and this game can scale really well on the GPU side as you lower the quality:
In a smaller encounter the performance is more consistent:
The benchmark app doesn't have a wide range of quality presets but standard laptop
and desktop high
do differ substantially:
If we take a look at Lenovo IdeaPad 5 laptop results (Ryzen 4800U with Vega 8 iGPU) the FF14 performance is quite low:
Results from Stonard to Necrotic Wake are WoW, while other are FF14 Shadowbringers benchmark app scenarios. WoW results are from 1080p mode 2 (low) setting which is likely lower than what FF14 benchmark app can be set to so take those results with a grain of salt.
Dedicated graphics
With dGPU it looks like a budget one (aside those hard bottom of the generation chart) can provide a playable 50-60FPS while to get more or handle absurd particle effects you need something mid-range or better.
The intro duel is the most extreme due to those effects during some attacks:
In more realistic scenarios even GTX 1050 Ti handles it:
FF14 has the Nvidia Gameworks logo on it but the benchmark results between AMD and Nvidia cards do seems to line up according to their relative performance levels. Sadly I don't have any recent mid/top range AMD card to compare it to better Nvidia cards.
Nowadays there is a lot of GTX 1650 ultra budget gaming laptops and pre-builds and that should be fine, although GTX 1650 Ti is way better than 1650... and newer devices will be equipped with RTX 3050 and 3050 Ti with respective performance uplift over mentioned predecessors.
And of course older GPUs can use lower settings to get better performance in the game. The RTX 3070 was tested on a Ryzen 5900X system while all other GPUs on a Intel i5-9400F - all with Desktop High preset. Benchmark final scores line up like so:
CPU
WoW does like strong CPUs. Any higher amount of actors will increase load on the main
core and thus the stronger single core performance the better. The game scales up to 4 cores and then hits diminishing returns when moving to 6 cores. If you add addons or other apps you will need more than the game + OS uses.
In the FF14 benchmark it seems that in high actor count scenarios there is also the main
core effect but this is to early to tell - will have to do some in-game benchmarking to see how the game scales and which type of in-game activity uses multi-core or single-core more.
One core is always loaded higher than others
The intro duel is bottlenecked by the GPU (tested with GTX 1070) so all CPUs achieve same numbers. Small raid boss encounter in the end is also GPU bound but achieving high FPS at that:
In mass-actor scenes the CPUs start to differentiate:
Ryzen 3500X and i5-9400F are both 6C/6T CPUs that have a very similar performance in games. They have very similar results so the game looks Intel/AMD agnostic. Intel CPU will reach bit better result when the game can benefit from lower latencies within the CPU. If you have a Zen 3 part this difference should be even smaller.
Ryzen 4650G is an APU with integrated graphics and is very similar to 3500X in terms of CPU performance. For gaming 4650G has lower latencies but also much less L3 cache (game cache
) and lower score means that FF14 can benefit from larger cache found on the 3500X and newer CPUs. This is also one of reasons why AMD (and Intel) is experimenting with 3D silicon stacking to add even more L3 cache to the CPU design.
Overall CPU side doesn't seems to be stressed that much in FF14 while in WoW for raids or multiple addons it can be.
Even the open world scenario is CPU bound when tested with GTX 1070
If we look at core count (configuring i5-9400F to run with 6, 4 and 2 cores) we can see a similar scaling to WoW - big jump between 2-4 cores and small one when moving to 6:
RAM
WoW nowadays realistically needs 16GB while FF14, at least in the benchmark app could fit under 8GB - but realistically any modern DDR4 system should just have 2x8GB configuration.
System performance will be affected by RAM working in single/dual channel and then RAM specs - timings and frequency (megatransfer per second). For PC desktop you should try to always pick good RAM kit for you CPU (and motherboard) as it can be 10-20% difference in performance for CPU bound scenarios:
Comparisons
Aside of raw component benchmarks we can try do some performance comparison. Like let say scenarios with large amount of actors, cities:
As you can see performance is quite similar, although the scenarios are not 1:1 comparable. WoW Orgrimmar auction house result is higher here as it's a static scene (I did not move the character, just looking at a large amount of players there). Dynamic scenarios like doing loops around Dalaran is very similar to FF14 City scenario.
WoW Oribos loops do got quite low FPS due to occasional stutter - at the time of writing this article WoW 9.1 has some bugs/regressions affecting performance in newer zones.
It's also hard to compare large raids as FF14 has a lot of particle effects while WoW quickly becomes CPU bound even with bleeding edge hardware:
The WoW scenario uses TBC Karazan with large amount of trash mobs in combat which gives very consistent results but they are the worst case scenario
for this game - if you pull a raid boss the 1% low FPS from that raid boss fight will be around the average FPS from Karazan scenario.
FF14 FPS is higher in this comparison but once again - the FF14 benchmark app isn't putting that much actors into the raid scene as the WoW counterpart. So in the end I would say they are quite comparable. Also note that hardest FF14 raids are only 8-man and only easy
Alliance raids being big.
Conclusions
Having just first encounter with FF14 I can't go deep into this game performance scaling and nuances, but from what I've already benchmarked it looks quite similar to WoW. Due to it more flashy nature it's bit more GPU heavier than WoW - but then you can play with in game setting. I'll have to check later on what are the actual capabilities of those settings - will they allow smooth gameplay on an APU laptop for example.
World of Warcraft development team seems way more active than FF14 - they got Apple M1 native version quite quickly, they even released Windows ARM version after that (and Windows ARM is a super small niche right now). On top of that they do use DX12 with optional ray traced shadows. On the downside quite recently (9.0.5, 9.1) WoW is plagued with performance regressions/bugs and as usual the communication is pretty much none. FF14 using DX11 and not rushing with like Apple M1/Metal version should at least in theory gave more stable/predictable game performance.
As for the FF14 benchmark app I would say it's a really really good idea to have one - you can quickly test your PC, create and test characters before starting your playthrough. WoW doesn't have this and technical support forums are filled with lots of simple
cases of performance problems or sudden regressions that such app could help to solve quicker or without Blizzard support staff. What I would want to see in this app is more quality presets - we are in a post-Intel-UHD era and all modern iGPUs are capable of running most games. iGPU preset is a must have. Seeing 20-30 FPS is less appealing to a laptop owner than 40-60 FPS on fine-tuned lower settings.
And in the topic of benchmarking apps and both games - it would be really good if such apps also try to look for potential problems - when someone has an issue. SSD getting close to full and performance degrades to a point of stutter while loading and alike, overheating of CPU/GPU, Nvidia drivers conflict - this is handled by hand on WoW technical forums.
I'm planning to play FF14 and try to then have in-game and benchmark app results published to get a better and more accurate picture of this game performance.
Time to give this game a spin
Comment article