FM22 Performance Benchmarking Thread

[Brother Ben]

 

FM21 setup for comparison in spoiler

Spoiler

 

------------------------------------------------------------------------------

 

Benchmark A - Download - 12/09/2022 - 19/09/2021

Benchmark B - Download - 12/09/2022 - 19/09/2021

Benchmark C - Download - 12/09/2022 - 19/09/2021

Benchmark D - Download - 12/09/2022 - 19/09/2021

 

Save files to

\Documents\Sports Interactive\Football Manager 2022\games

 

------------------------------------------------------------------------------------------------------

 

Hello all and welcome to the Football Manager 2022 Benchmarking thread.

This year i've decided to keep the tests the same or as close to the same as last year.  I'm not too sure sure how useful a comparison between the two will be as despite the similarity in tests the amount of games played in the same time period could well be different.  The date also had to be moved slightly as the same date was an international weekend.

With that in mind any conclusions will be purely anecdotal but it may prove interesting none the less.  It is also worth noting that there are 4 more leagues this year.

Despite asking in another thread about what everyones standard setup is for FM with a view to changing Benchmark A i've decided to go with exactly the same setup as last year as I think it represents a decent cross section of players and allows those with lower end systems to get involved too.

Benchmark B & C are again the standard "all leagues from all nations" tests with a little extra detail added to Bench C  to put those multiple core systems through their paces.

Back again this year by popular demand is Benchmark D.  This benchmark is maximum everything, all leagues, players and detail.  I would recommend this benchmark for only those with top end systems. A big thanks to @Daveincid for setting this up again.

 

--------------------------------------------------------------------------

Previous years.

FM20 Benchmarking Thread

FM21 Benchmarking Thread

---------------------------------------------------------------------------

 

Posting Results

In order to get the most accurate / consistent result possible, I advise you to restart your PC / Laptop and then close everything that is unnecessary, leaving only FM open.  If you run the game on a laptop, don't forget to plug it in the socket before you begin and also for laptop users make sure your system is in high power mode.

After you load the save, ensure that auto save interval is disabled, and also set processing to Fast (Less Responsive), both can be found by using the search bar in Preferences.

Below is the format in which I advise you to post your results, it's pretty much the same as previous years, don't worry if you aren't sure of the Base/Turbo frequencies of your processor as I can find this out for you as long as you know the CPU Model.

To find out the specs of your PC open the spoiler below

  

Spoiler

You can download this program HERE

once you've installed it the CPU tab will give you the type of CPU, it can be found under specification.  don't worry about CPU speed I can work that out from the model number

click on the "memory" tab and you will be presented with this

The type and size are top left

The speed is DRAM frequency.  You need to double this figure. (DDR = Double Data Rate)

So the above is 534 x 2 = 1068MHz 

 

-------------------------------------------------------

Quick reminder again this year i'll be asking for your Graphics level in 3D.

What this means is what graphics level you set in the game to get a smooth experience.  This should prove invaluable for the guys over at the New PC/Laptop recommendation thread so be sure to fill that in too

Below are my results for you to follow as a template:

Type: Laptop

Model: Macbook Air

CPU Model: M1

CPU Base Frequency: 3.20 GHz

CPU Turbo Frequency: 3.20 GHz

RAM: 8GB

RAM Clockspeed: 4266Mhz

GPU: M1 7 Core Apple

Graphics Level in 3D: High

Storage Type: SSD

Benchmark A: 01 min 07 Sec

Benchmark B: 05 min 40 Sec

Benchmark C: xx min xx Sec

Benchmark D: xx min xx Sec

-------------------------------------------------------------------

We had 97 benchmarks last year so if we can break the 100 barrier that would be awesome!

Any queries about the benchmarks feel free to let me know

Happy testing folks and massive thanks in advance!

 

 

 

Edited November 22, 2021 by Brother Ben

[Brother Ben]

Results

Spreadsheet 

(Contains full details, Cores, Ram speed, Graphics cards etc)

 

 

OVERVIEW - LAPTOP

 

 

 

 

 

 

 

 

 

 

 

OVERVIEW - DESKTOP

 

 

 

 

 

 

 

 

 

 

 

 

Edited November 8, 2022 by Brother Ben

[Phil930]

Great to see this thread back.  Interested in year-over-year change, albeit non-conclusive as you note due to changing test parameters.  Kept my own results for at the time a 8700K, then 5800X, and now 5900X for FM22.  Will dig them up.

Type: Desktop

Model: Self build

CPU Model: AMD Ryzen 9 5900X

CPU Base Frequency: 3.70 GHz

CPU Turbo Frequency: 4.80 GHz

RAM: 32GB

RAM Clockspeed: 3600Mhz

GPU: EVGA RTX 3080 FTW3

Graphics Level in 3D: Very High

Storage Type: SSD (game installed on C: Samsung 980 Pro 1TB NVMe)

Benchmark A: 53 Secs

Benchmark B: 04 min 29 Sec

Benchmark C: 07 min 32 Sec

Benchmark D: xx min xx Sec

[roykela]

Type: Desktop

Model: Self build

CPU Model: Intel Core i7 9700K

CPU Base Frequency: 3.60 GHz

CPU Turbo Frequency: 4.60 GHz

RAM: 16GB DDR4

RAM Clockspeed: 2600Mhz

GPU: Nvidia Geforce GTX 980Ti

Graphics Level in 3D: Very High

Storage Type: SSD (Kingston SNVS 2TB NVMe M.2)

Windows 11 21H2 (22000.318)

Benchmark A: 01 min 05 Sec

Benchmark B: 05 min 43 Sec

Benchmark C: 11 min 59 Sec

Interesting to see that Benchmark A & B is an improvement from FM21 but Benchmark C is not. C is about a minute slower on my system.
Also a bit disappointing to see that FM22 consistently used only 22% of my CPU capacity, with very few spikes up to 100% (Mainly match days). I would have preferred FM to utilize a few percentages more on days without matches. Maybe around 10%-20% more?
In benchmark C it did utilize 100% on match days. 

I expected better results on Benchmark C. 

Edited November 13, 2021 by roykela

[PineappleBlender]

Type: Desktop

Model: Custom

CPU Model: Ryzen 5950X

CPU Base Frequency: 3.40 GHz

CPU Turbo Frequency: Varies. PBO overclocking is enabled, typical all core load would be something like 4.6GHz on all 16 cores. Single core boost can do up to 5.1GHz.

RAM: 64GB

RAM Clockspeed: 3133Mhz

GPU: Nvidia 2080 Ti

Graphics Level in 3D: Very High

Storage Type: NVME PCIE 3.0 SSD

 

Benchmark A - 0:54
Benchmark B - 4:43
Benchmark C - 6:59

Again there is still almost no parallelisation with the QME and general game engine, if this took advantage of multiple cores like every single computer on the planet, there would be massive speedups. Benchmark A would complete in about 4-5 seconds for me at full load.

I have no idea why SI don't improve this. Even if the QME does take into account other results in the league (unlikely), you could run say PL games on one core, Championship games on another core, Liga 1 on another, and so on.

Checking CPU usage during benchmark C there were brief periods where it would use all 16 cores, but this would quickly drop off back to a thread or two. When the game actually bothers to load the CPU the work completes very quickly.

In 2021 despite multicore being dominant for a well over a decade, the game is almost completely single threaded, despite the fact that it should be embarassingly parallel.

This game could be so much faster for literally 100% of users. Not just people on the 0.1% high end machines. Everyone, from a basic laptop to near supercomputers. How can you get a better return on investment than improving performance for 100% of people?

Extremely disappointing that even with Benchmark C with an extra 33% more cores than the 5900X the speedup is about 7% when you compare to Phil's results.

We have a game based entirely around crunching numbers, and practically 90% of it is still to this day single threaded even under the best case scenarios for multithreading.

Edited November 13, 2021 by PineappleBlender

[yohanes1985]

Type: Desktop

Model: Custom

CPU Model: Intel Xeon E3-1240 V2 (Ivy Bridge 4C/8T) Q2'12

CPU Base Frequency: 3.40 GHz

CPU Turbo Frequency: 3.80 GHz

RAM: 8GB DDR3 

RAM Clockspeed: 1600MHz (CL 9-9-9-24 1T)

GPU: nVidia Geforce GT 1030 2GB GDDR5

Graphics Level in 3D: High

Storage Type: SSD

 

Benchmark A: 01 min 58 Sec

Benchmark B: 09 min 09 Sec

Benchmark C: 25 min 35 Sec (CPU Temp Reach Max 92° Celcius) 

Benchmark D: Computer says No!

" I'm optimistic that my Xeon will be the oldest CPU tested here... " LOL

Edited November 15, 2021 by yohanes1985
Updated Benchmark C

[DarJ]

Type:  Laptop

Model: Dell G5 5590

CPU Model:  intel i7 9750H

CPU Base Frequency: 2.60 GHz

CPU Turbo Frequency: 4.50 GHz

RAM: 16GB DDR4 

RAM Clockspeed: 2667MHz

GPU: NVidia Geforce GTX 1660 Ti

Graphics Level in 3D: Very High

Storage Type: HDD

 

Benchmark A: 01 min 33 Sec

Benchmark B: 07 min 57 Sec

Benchmark C: 16 min 45 Sec 

Benchmark D: xx min xx Sec

[Brother Ben]

1 hour ago, roykela said:

Interesting to see that Benchmark A & B is an improvement from FM21 but Benchmark C is not. C is about a minute slower on my system.
 

Having compared, I think the difference is that there are more midweek matches for other nations this year (slower for C) but no English transfer window deadline day (quicker for all)

Hope that makes sense?!

[Brother Ben]

42 minutes ago, PineappleBlender said:

This game could be so much faster for literally 100% of users. Not just people on the 0.1% high end machines. Everyone, from a basic laptop to near supercomputers. How can you get a better return on investment than improving performance for 100% of people?

Would that not require a complete rebuild of the entire thing though?

[DarJ]

9 minutes ago, Brother Ben said:

Would that not require a complete rebuild of the entire thing though?

Even if it does, I don't think it's like ME where tweaking one thing can cause multiple issues elsewhere (not saying it's an easy job either)

[Kruj]

Type: Desktop

Model: Self build

CPU Model: AMD Ryzen 9 5900X

CPU Base Frequency: 3.70 GHz

CPU Turbo Frequency: 4.80 GHz

RAM: 32GB

RAM Clockspeed: 3600Mhz

GPU: AMD Radeon R9 200 Series

Graphics Level in 3D: Very High

Storage Type: SSD

Benchmark A: 54 Secs

Benchmark B: 04 min 35 Sec

Benchmark C: 07 min 39 Sec

 

@PineappleBlender: Very well said! I did a huge hardware upgrade on my computer and I'm glad that the FM finally is playable for me now. But it's still disappointing to see that there would be so much more possible. This game is all about calculation, it's the core of the game, together with its large database, which unfortunately is unenjoyable even with the best CPU's as long as there is no proper parallelization.

[Brother Ben]

1 hour ago, PineappleBlender said:

Extremely disappointing that even with Benchmark C with an extra 33% more cores than the 5900X the speedup is about 7% when you compare to Phil's results.

This is true of a lot of software to be fair

Benchmark D when its ready will be a better test of multiple cores

Edited November 13, 2021 by Brother Ben

[jyny]

Type:  Laptop

Model: Dell G3 3500

CPU Model:  Intel i7 10750H 

CPU Base Frequency: 2.60 GHz

CPU Turbo Frequency: -

RAM: 16GB DDR4 

RAM Clockspeed: -

GPU: NVidia Geforce GTX 1660 Ti

Graphics Level in 3D: Very High

Storage Type: SSD

 

Benchmark A: 01 min 21 Sec

Benchmark B: 06 min 34 Sec

Benchmark C: xx min xx Sec 

Benchmark D: xx min xx Sec

[Jolyon Chen]

@Brother BenThx for this topic once again. To add to @PineappleBlender, when modded my I9 does use all cores for 95% of the whole simulation (even having to block FM accessing the 10th core cause of the full CPU load), so it also depends on the workload. This game (running it modded with over 200 leagues and 400k players) is the sole reason I upgraded from Kaby Lake to Comet Lake last year (yeah...Intel was way cheaper here). Anyone contributing with Alder Lake in benchmarking would be greatly appreciated. Wanna see if I should jump cpu brand.

[Brother Ben]

29 minutes ago, Jolyon Chen said:

@Brother BenThx for this topic once again. To add to @PineappleBlender, when modded my I9 does use all cores for 95% of the whole simulation (even having to block FM accessing the 10th core cause of the full CPU load), so it also depends on the workload. This game (running it modded with over 200 leagues and 400k players) is the sole reason I upgraded from Kaby Lake to Comet Lake last year (yeah...Intel was way cheaper here). Anyone contributing with Alder Lake in benchmarking would be greatly appreciated. Wanna see if I should jump cpu brand.

There was a late entry from an Alder lake in the FM21 thread and the performance was very impressive.  Top 4 in all tests and top for benchmark b.

Edited November 13, 2021 by Brother Ben

[JamieTC13]

Type: Desktop

Model: Self build

CPU Model: Intel Core i7 9700K

CPU Base Frequency: 3.60 GHz

CPU Turbo Frequency: 4.8ghz all core

RAM: 16GB DDR4

RAM Clock speed: 3200Mhz

GPU: Nvidia GeForce RTX 2070

Graphics Level in 3D: Very High

Storage Type: SSD (Adata sx8200 1tb NVMe M.2)

Windows 11 21H2 

Benchmark A: 01 min 05 Sec @ 4.8ghz. and  A 59 secs @ 5ghz. So really no point having at 5ghz just for an extra 5 seconds 

Benchmark B: 05 min 25 Sec

Benchmark C: 11 min 25 Sec

I have got an i7 11700k to do just got to install FM22 on that system

Edited November 20, 2021 by jckc221013jamie

[Whatawaster95]

I should be getting an Alder Lake build hopefully in next few weeks once some DDR5 Ram is available...may also be interesting to see if DDR5 makes any difference in FM (Doubt it though going by other games benchmarks)

 

Will post my results once i've got it all set up.

Edited November 13, 2021 by Whatawaster95

[DarJ]

Does anyone know the difference in speed between installing the game on an SSD Vs HDD?

[grimbo88]

Type: Desktop

Model: Custom

CPU Model: i9 - 12900k

CPU Base Frequency:  3.2 ghz/2.4 ghz

CPU Turbo Frequency: 5.2 ghz/3.9 ghz

RAM: 32GB DDR5

RAM Clockspeed: 5200 Mhz

GPU: NVIDIA GTX 1080 - 8gb

Graphics Level in 3D: Very High

Storage Type: SSD (PCIe 4)

 

Benchmark A: 41 sec

Benchmark B: 03 min 43 sec

Benchmark C: 06 min 04 sec

Benchmark D: 16 min 37 sec

 

@Jolyon Chenhere are my results

Edited November 14, 2021 by grimbo88
Added Benchmark D results

[Pompeyboyz]

Type: Desktop

Model: Self Build

CPU Model: AMD Ryzen 5 3600

CPU Base Frequency: 3.60 GHz

CPU Turbo Frequency: 4.20 GHz

RAM: 16GB DDR4

RAM Clockspeed: 3200Mhz

GPU: GTX 1060 6GB

Graphics Level in 3D: Very High

Storage Type: HDD

Benchmark A: 01 min 32 Sec

Benchmark B: 06 min 11 Sec

Benchmark C: 13 min 26 Sec

[Brother Ben]

1 hour ago, DarJ said:

Does anyone know the difference in speed between installing the game on an SSD Vs HDD?

It would be a lot quicker on an SSD, save times would be better too but the game itself wouldn't be much different

[DarJ]

9 minutes ago, Brother Ben said:

It would be a lot quicker on an SSD, save times would be better too but the game itself wouldn't be much different

What about if I had the game itself on an SSD but the save files on my HHD?

[Brother Ben]

3 minutes ago, DarJ said:

What about if I had the game itself on an SSD but the save files on my HHD?

It would be slower to save but its quicker these days anyway

[fc.cadoni]

No big difference between i9 12th and 5950X . Interesting

[obasa_G]

Type: Laptop

Model: DELL G5 15 5587

CPU Model: Intel Core i7 8750H

CPU Base Frequency:  2.20 GHz

CPU Turbo Frequency: 4.10 GHz

RAM: 24GB (Laptop originally came with 8GB)

RAM Clockspeed: 2400Mhz & 2667Mhz

GPU: NVIDIA GeForce GTX 1050 Ti

Graphics Level in 3D: High

Storage Type: SSD

Benchmark A: 01 min 34 Sec

Benchmark B: 07 min 14 Sec

Benchmark C: 20 min 48 Sec

[MataCro7]

Type: Desktop

Model: Custom

CPU Model: Intel Pentium G4560

CPU Base Frequency: 3.5GHz

CPU Turbo Frequency: N/A

RAM: 8 GB

RAM Clockspeed: @2400Mhz

GPU: NVIDIA GeForce GTX 1050 Ti - 4 GB GDDR5

Graphics Level in 3D: Very High

Storage Type: HDD

 

Benchmark A: 02 min 11Sec

Benchmark B: 08 min 40 Sec

Benchmark C: -

Benchmark D: -

It's definitely improvement over FM21. A and B tests are 2x faster than last years game (A - 3:59, B - 16:25)

[PineappleBlender]

I've run another set of benchmarks, this time disabling 12 out of 16 cores in the BIOS i.e throwing away basically 75% of the CPU.

Benchmark results, Ryzen 5950X 4/16 cores enabled, otherwise as above:

Benchmark A - 1:02
Benchmark B - 4:47
Benchmark C - 14:14

Notes

This was achieved by naively disabling the cores in the BIOS, with no attempt to select optimum cores. With Benchmarks A & B we're within margin of error. For most practical purposes under normal usage, effectively zero parts of the game use multiple cores. In theory this should allow PBO to boost to higher core speeds because there's extra thermal overhead. For scientific accuracy more detailed testing would be needed, but it shows the general idea.

As a reminder, results using the full 16 cores i.e. ~4x more processing power were:

Benchmark A - 0:54
Benchmark B - 4:43
Benchmark C - 6:59

Conclusions

If you look at Benchmarks A & B which is how almost everyone will play the game, the result is frankly embarrassing. I remember years back asking Miles to look into this on Twitter, and he was refusing to accept the game doesn't scale with multiple cores, insisting that telemetry from AMD and Intel showed how good his team's programming was. Send him these results. Or get someone to test it themselves, because this is blatant denial.

Some people might look at this and say "it doesn't matter about optimising for people with 16 cores". And you're right, it doesn't. There's probably only a handful of FM players with this many cores. But it's badly optimised for anything modern. That basic laptop you got from PC World? I'm sure you'd like FM to be 3-4x faster if it was optimised properly. There is zero reason to think the game scales great on 4 cores, but badly past that, because the same technique that would allow it to scale for 'normal' processors would work on high end stuff.

The point is not that the game isn't designed efficiently for supercomputers, it's that it's not designed to run well on anything past 2005.

Every single aspect of this game is able to be parallelised. QME could simulate different leagues per core. Transfer targets could be done per core. Training improvements could be done per core. Even for stuff like the full match engine which is complex, you don't need to parallelise that, because you're doing that already by simulating 20 games as part of the league.

FM should be scaling near linearly with core count, like video encoding. It is probably one of the best games in the world for scaling potential, but poor coding practices and a refusal to modernise or even accept this on the part of SI means the game is incredibly suboptimal.

It isn't easy to rewrite the game to use multiple cores. No one says it is. But at least Miles and the like should honestly admit their software does not work well on modern hardware, instead of dismissing criticism even with evidence.

  

9 hours ago, Brother Ben said:

Would that not require a complete rebuild of the entire thing though?

It would require a lot of work, it's not easy. But we're not talking about only getting a 5% speedup for 10% of users. We're talking about massive speedups for every single player.

Edited November 13, 2021 by PineappleBlender
Edit

[EdL]

I don't think anyone at SI (who is technical at least) doesn't realise that outside certain scenarios, the game is laregly single threaded, as its self evident!

However it is probably a re-write of the game to multi-thread most the bits that aren't, the game is so interdependent, its just not paralleiseable atm without having locks everywhere which would destroy the performance and memory usage. The basic structure of the game was designed before multi-core CPU's were really a thing (very early 2000's).

Sorry you are disappointed but no one is in denial, but it is just not an easily fixable problem.

btw the QME is fully threaded, thing is they are so quick as that its virtually irrelevent.

[maaartio]

Hello everyone, My results:

Type: Desktop

Model: Custom

CPU Model: Ryzen 5900x

CPU Base Frequency: 3700

CPU Turbo Frequency: N/A

RAM: 8 GB

RAM Clockspeed: @3600Mhz 14,14,14,14,28

GPU: radeon 480

Graphics Level in 3D: Very High

Storage Type: SDD Samsung 980 pro

A: 51 s
B: 4:12
C: 7:08

 

Edited November 13, 2021 by maaartio

[Jolyon Chen]

7 hours ago, grimbo88 said:

Type: Desktop

Model: Custom

CPU Model: i9 - 12900k

CPU Base Frequency:  3.2 ghz/2.4 ghz

CPU Turbo Frequency: 5.2 ghz/3.9 ghz

RAM: 32GB DDR5

RAM Clockspeed: 5200 Mhz

GPU: NVIDIA GTX 1080 - 8gb

Graphics Level in 3D: Very High

Storage Type: SSD (PCIe 4)

 

Benchmark A: 41 sec

Benchmark B: 03 min 43 sec

Benchmark C: 06 min 04 sec

 

@Jolyon Chenhere are my results

Nice one. Thx a lot. You seem quite quicker in Benchmark C than last year without any update whatsoever. Interesting to see what Benchmark D will do.

[PineappleBlender]

4 hours ago, EdL said:

I don't think anyone at SI (who is technical at least) doesn't realise that outside certain scenarios, the game is laregly single threaded, as its self evident!

However it is probably a re-write of the game to multi-thread most the bits that aren't, the game is so interdependent, its just not paralleiseable atm without having locks everywhere which would destroy the performance and memory usage. The basic structure of the game was designed before multi-core CPU's were really a thing (very early 2000's).

Sorry you are disappointed but no one is in denial, but it is just not an easily fixable problem.

btw the QME is fully threaded, thing is they are so quick as that its virtually irrelevent.

I can't prove what Miles said on Twitter several years ago to me on an account I've long since deleted, so I'll understand not taking some random person on the Internet at their word making such an allegation, though it was completely dismissed.

It's not easy, I completely agree. I'm sure it would be a lot of work and you'd have to rewrite probably almost the entire code base to convert it into a format suitable for multi core. It would however be a lot of work that would pay massive dividends. Every single player having a faster game, twice as fast, four times as fast, ten times faster. Players being able to run large amounts of leagues with acceptable loading times.

It's also frankly crazy that this hasn't been a priority since c. 2010 when multi core started becoming a major thing. Especially these days as single core performance is much harder to improve, and there's a trend towards more and more cores.

One thing Miles has done recently is say women's football is being implemented, not for economic factors, but because it's the right thing to do. Yet apparently no one seems to have seriously argued about the massive benefits over the past decade of using 21st century programming design to achieve massive game speedups that other companies could only dream of! If Miles thinks "our game is 10x faster" isn't a good selling point, then I'm not sure what to say. If that isn't a good reason, how about "it's the right thing to do".

Edited November 14, 2021 by PineappleBlender

[Ivalice]

MBP 16 Inch, 32 GB, M1 MAX

high power mode on battery and plugged-in result pretty much the same.

A: 1.07

B: 5.54

3D: All high, AA off.

Edited November 14, 2021 by Ivalice
clarify

[XaW]

Type: Laptop

Model: MSI GS66 Stealth

CPU Model: i7-11800H

CPU Base Frequency: 2.30 GHz

CPU Turbo Frequency: 4.60 GHz

RAM: 16GB

RAM Clockspeed: ?

GPU: NVIDIA GeForce RTX 3060

Graphics Level in 3D: High

Storage Type: SSD

Benchmark A: 01 min 08 Sec

Benchmark B: 06 min 02 Sec

Benchmark C: 11 min 39 Sec

Benchmark D: xx min xx Sec (waiting for the file)

[TopToffee]

5 hours ago, Ivalice said:

MBP 16 Inch, 32 GB, M1 MAX

high power mode on battery and plugged-in result pretty much the same.

A: 1.07

B: 5.54

3D: All high, AA off.

Just as a general FYI, you can 100% max out the 3D settings with that spec. I've got it maxed out - including high AA - on the M1 Air (the 16GB version with the extra graphics core) and it runs buttery smooth.

Shows how single core FM is, that your A and B times are almost identical to the base model Air posted above. God I love the M1 Air - incredible machine for the price.

[gothenburg83]

Type: Desktop

CPU Model: Intel i7 8086k

CPU Base Frequency: 4.0GHz

CPU Turbo Frequency: 5.0GHz

RAM: 32GB Corsair Vengeance LPX DDR4

RAM Clockspeed: 3600MHz

GPU: Asus NVIDIA GeForce RTX 2080 Super 8gb

Graphics Level in 3D: Very high

Storage Type: Samsung 860 EVO SSD 500GB

OS: Windows 11 Pro 64-bit  

Benchmark A: 1 minute 09 seconds

Benchmark B: 5 minutes 51 seconds

Benchmark C: 12 minutes 22 seconds

Benchmark D: 39 minutes 15 seconds

Edited March 25, 2022 by gothenburg83
updated for new RAM

[beltza]

Type: Mac desktop

Model: Mac mini (M1, 2020)

CPU Model: 8-core CPU with 4 performance cores and 4 efficiency cores

RAM: 8GB

GPU: 8-core GPU

Graphics Level in 3D: High

Storage Type: SSD

OS: macOS Monterey 12.0.1

Benchmark A: 1 minute 11 seconds

Benchmark B: 6 minutes 4 seconds

Benchmark C: 14 minutes 27 seconds

Was wondering if there were any updates on the native (ARM) Apple M1 version of Football Manager 2022?

[AlWakRa]

Finally early to the benchmark 

 

Type: Desktop

Model: Custom

CPU Model: AMD Ryzen 9 5900X

CPU Base Frequency: 3.70 GHz

CPU Turbo Frequency: 4.80 GHz

RAM: 32GB

RAM Clockspeed: 3600Mhz (16-18-18-38)

GPU: NVIDIA GeForce RTX 3080 TI - 12GB

Graphics Level in 3D: Very High

Storage Type: 2 x Samsung 970 evo 2TB in Raid0

Benchmark A: 51 Sec

Benchmark B: 4 min 14 Sec

Benchmark C: 7 min 7 Sec

Benchmark D: xx min xx Sec (To be updated)

 

Notes: PBO enabled, tried test A twice to see if there is any difference, both runs were 51 seconds. 

[Brother Ben]

4 hours ago, TopToffee said:

God I love the M1 Air - incredible machine for the price.

Yeah me too, been a PC guy all my life but the Air at its price point is a game changer

[Brother Ben]

2 hours ago, beltza said:

Was wondering if there were any updates on the native (ARM) Apple M1 version of Football Manager 2022?

I can't see it being a priority sadly, probably not economical given the size of the player base

[Ivalice]

17 minutes ago, Brother Ben said:

I can't see it being a priority sadly, probably not economical given the size of the player base

No, actually if the code can run in ARM aka Nintendo Switch natively, there is a real chance to do this with minimal cost.

[Brother Ben]

Benchmark D is here, big thanks to @Daveincid for sorting it out.  

Download D

[Rivanov]

Type: Laptop

Model: 14" MacBook Pro M1 Pro

CPU Model: 10-Core

RAM: 16GB

GPU: 16-Core

Graphics Level in 3D: Very High

Storage Type: SSD

Benchmark A: 1:08.07

Benchmark B: 5:50.49

Benchmark C: 10:46.03

Benchmark D: 31:10.46

[fc.cadoni]

21 hours ago, EdL said:

I don't think anyone at SI (who is technical at least) doesn't realise that outside certain scenarios, the game is laregly single threaded, as its self evident!

However it is probably a re-write of the game to multi-thread most the bits that aren't, the game is so interdependent, its just not paralleiseable atm without having locks everywhere which would destroy the performance and memory usage. The basic structure of the game was designed before multi-core CPU's were really a thing (very early 2000's).

Sorry you are disappointed but no one is in denial, but it is just not an easily fixable problem.

btw the QME is fully threaded, thing is they are so quick as that its virtually irrelevent.

It’s time for game to be into next level; I am pretty sure all the community can accept any delay from the point when the deadline being announced. :-)

[Kruj]

Am 13.11.2021 um 12:26 schrieb Kruj:

Type: Desktop

Model: Self build

CPU Model: AMD Ryzen 9 5900X

CPU Base Frequency: 3.70 GHz

CPU Turbo Frequency: 4.80 GHz

RAM: 32GB

RAM Clockspeed: 3600Mhz

GPU: AMD Radeon R9 200 Series

Graphics Level in 3D: Very High

Storage Type: SSD

Benchmark A: 54 Secs

Benchmark B: 04 min 35 Sec

Benchmark C: 07 min 39 Sec

Benchmark D: 19 min 49 Sec

 

I'm a bit disappointed with the result. Benchmark D was a lot quicker last year, around the 13 minutes mark (I don't have FM21 so I did no Benchmark last year, but the results of others with the 5900X were a lot better).

I'm curious about further results in this thread.

[PineappleBlender]

Benchmark D (all cores): 16:33.

As a rough approximation if you compare this to Kruj's result, this means my result with 33% more cores is about 20% faster for Benchmark D. This is obviously better than ~7% faster in Benchmark C, but for most practical purposes as discussed earlier it's pretty much all single threaded.

 

1 hour ago, Kruj said:

Benchmark D: 19 min 49 Sec

 

I'm a bit disappointed with the result. Benchmark D was a lot quicker last year, around the 13 minutes mark (I don't have FM21 so I did no Benchmark last year, but the results of others with the 5900X were a lot better).

I'm curious about further results in this thread.

I don't think you can directly compare, because it might be that FM22 has more fixtures. The benchmarking thread is useful, but for a direct comparison between versions it would probably need SI to create a special save file with dummy data (i.e. no licensing to worry about) where they could use the same data each year.

[Brother Ben]

1 hour ago, Kruj said:

Benchmark D: 19 min 49 Sec

 

I'm a bit disappointed with the result. Benchmark D was a lot quicker last year, around the 13 minutes mark (I don't have FM21 so I did no Benchmark last year, but the results of others with the 5900X were a lot better).

I'm curious about further results in this thread.

It's a totally different test, I wouldn't compare to last year, best to compare to computers that have similar spec to your own this year

[Kruj]

Alright, I already supposed something like that

[Brother Ben]

The intention was to make them the same but there are so many variables.  With benchmark D all you need is a dozen or so nations to have mid week games that weren't present in the week tested on FM21 and you get an entirely different result

[grimbo88]

I have added my Benchmark D results for Alder Lake. I may run this again when the LGA1700 screws come for my cooler, my Mobo has both LGA1200 (currently using) and LGA1700 connections so there may be a difference in temps, I hit 100° in this test so there will have been throttling.

[Brother Ben]

25 minutes ago, grimbo88 said:

I have added my Benchmark D results for Alder Lake. I may run this again when the LGA1700 screws come for my cooler, my Mobo has both LGA1200 (currently using) and LGA1700 connections so there may be a difference in temps, I hit 100° in this test so there will have been throttling.

Did it hit 100% on all cores?