Game Speed Tests

GAME SPEED TESTS

As of late, there have been a lot of threads about the perfect game set-up and database size for a new savegame. Now that FM11 is only a couple of weeks away, I found it the perfect time to run some game speed tests to see how certain options impact the processing time. Primarily, I've done these tests for myself but then I thought I might as well share them with you. However, these tests are not highly scientific and results will vary from computer to computer but they may show how huge the impact of certain options is. By the way, the specs of my computer:

HP Pavilion p6503de

Processor: AMD Athlon II X4 635 (2.9 GHz, 512 kB L2-Cache, Quad Core)

RAM: 6 GB DDR3-1333 RAM

HDD: 1.5 TB, S-ATA, 5.400 rpm

Graphics: NVIDIA GeForce GT230, 1.5GB dedicated

OS: Windows 7 64-Bit

Moving on.. I did five tests, each with a running time of 125 minutes. At the end of the test I noted down the in-game date to calculate the passed days. The first test should be the reference savegame the others are compared to, so I always changed one option for each of the other tests (marked in the table). Please keep in mind that these tests don't show the long-term effects of options like database size and number of selected leagues. So if you go beyond these two hours of holidaying (which most of us will likely do), you might experience a different processing speed. Once again, these tests should only be regarded as a rough guide.

Now let me explain you the columns and rows you will find in the table. The first column shows the different options. The following columns show the 5 savegames and the options that were chosen for them.

Player Count: The amount of players loaded at game start (database size).

Leagues: Total number of selected leagues.

view-only: Number of leagues that were selected as view-only.

Playable: Number of leagues that were selected as playable.

Detail level: The detail level chosen for processing fixtures.

Savegame size: The size of the savegame at the beginning of the save. Only now I realise I should have noted the savegame size at the end of the test as well...

Holidayed for: Passed real-time while FM10 was holidaying. 125 minutes for each test.

Days holidayed: Number of passed days in-game during this timespan.

Days/Minute: Passed in-game days per real-time minute. This is the crucial row where you can compare the speed of the different game set-ups.

Comparison:

Savegame 1 / 2: The number of leagues has been reduced in savegame 2.

Savegame 1 / 3: The number of loaded players has been decreased in savegame 3.

Savegame 1 / 4: All view-only leagues from the first savegame have been set to playable in savegame 4.

Savegame 1 / 5: The detail level has been changed from none to maximum for all competitions in savegame 5.

That changing the options would result in a different processing speed was clear before. I was interested in how huge these differences would be. Obviously, the number of loaded players has the biggest impact on processing speed whereas the number of leagues loaded has some mediocre impact in the test (of course this is also affected by the amount of players and leagues you actually load). Also, the difference in processing speed between view-only and playable leagues isn't all too big. But keep in mind that setting leagues to playable usually also increases the number of players loaded. Setting the detail level from none to full slows the game significantly.

Please feel free to share your opinions or even contribute your own tests. And please don't ask for further tests, I won't do any more tests. By the way, I'm not a native speaker so please forgive me any mistakes.

As you said, a very difficult thing to quantify accurately but as a general guide, this is a good piece of work and should be a help to many posters. Thank you.

Mostly its up to the user to select the number of leagues depending on he way they play the game.

BTW Its a good test. Maybe I will do one myself.

I think you've just saved me £500 on a new PC.

I'm running a machine far inferior to yours: an X2 4400 2GB, XP SP3. and have run about 8 leagues on full view with 100,000+ players all year, and don't mind the delay, so I guess I can live with those settings

Was going to upgrade for FM11 - But I don't think I can justify the outlay. I'll just halve my player count for FM11 and add more leagues and watch it fly!

If I remember correctly from my testing a while ago - league matches are simulated in parallel, using all CPU cores, where day-to-day player processing (attribute changes for example) are using only one core, hence the results. Since I like to play with a lot (all) players loaded I've been bitching about this several times, but nobody seemed to understand what I'm saying

I think you've just saved me £500 on a new PC.

I'm running a machine far inferior to yours: an X2 4400 2GB, XP SP3. and have run about 8 leagues on full view with 100,000+ players all year, and don't mind the delay, so I guess I can live with those settings

IMO if you don't mind the delay, why load less. I have a similar PC and I run 11 leagues with 40,000 players. Can easily go to 15 to 20 if you load 30 to 40k players

If I remember correctly from my testing a while ago - league matches are simulated in parallel, using all CPU cores, where day-to-day player processing (attribute changes for example) are using only one core, hence the results. Since I like to play with a lot (all) players loaded I've been bitching about this several times, but nobody seemed to understand what I'm saying

Got a link to that.

How did you conclude that.

It all depends on how long you plan on playing.

There is no point maxing out your system if you plan playing a long term game because it will still cripple your PC(or MAC) if you have too many leagues after several years.

Also, the game won't use more than 2GB Ram no matter what you do, so 6GB is no advantage for this game.

Also, the game won't use more than 2GB Ram no matter what you do, so 6GB is no advantage for this game.

4GB on 64bit.

No, 2Gb on 64bit as well.

No, 2Gb on 64bit as well.

NO.

A 32bit app gets 2GB user memory and 2GB kernel on a 32bit OS. But a 32 bit app gets 4GB (as it is still 32bit) on 64bit app.

NO.

A 32bit app gets 2GB user memory and 2GB kernel on a 32bit OS. But a 32 bit app gets 4GB (as it is still 32bit) on 64bit app.

For fm games it uses 2GB regardless.

For fm games it uses 2GB regardless.

You won't listen then?

You won't listen then?

I'm testing now

You won't listen then?

I'm testing now

Nope sorry 64bit Windows 7 with 8GB

... the game is not using over 2GB

Nope sorry 64bit Windows 7 with 8GB

... the game is not using over 2GB

You are not using it right. FM isn't something different, its still a software.

Is there a limit of what FM will use? RAM wise.

. .

2GB on a 32-bit OS, 4GB on a 64-bit OS.

You are not using it right. FM isn't something different, its still a software.

I think I'm using it right.

I enabled all leagues, loaded all players and guess what .. 2GB!

Am interested to see who is right, i thought it was 4 gig.

Am interested to see who is right, i thought it was 4 gig.

He's right

I did the test again and manged to use over 3GB

He's right

I did the test again and manged to use over 3GB

Fair play, least you admitted he was right, good man!

It took me all leagues, all players and about 2 seasons (on holiday) to get over 2GB so it is not obvious when you start a game.

but yeah - apologies, to all concerned.

It took me all leagues, all players and about 2 seasons (on holiday) to get over 2GB so it is not obvious when you start a game.

but yeah - apologies, to all concerned.

How many cores have you got, was it using all of them?

He's right

I did the test again and manged to use over 3GB

Just when I dug up sources........

Fair play, least you admitted he was right, good man!

Yeah

How many cores have you got, was it using all of them?

I told you that i've seen 8. I even have a screenie.

4 cores took about 30 minutes and it nudged over 2GB

Tbh I was watching TV so it might have been much sooner than that - when I checked it it was over

I told you that i've seen 8. I even have a screenie.

No worries, just wanted to see how fast it was running.

I really wouldn't recommend playing a game with tons of leagues though. The first few seasons are fine but it grinds to a stuttering halt around the 5-10 season mark.

The save games are absurdly huge.

I really wouldn't recommend playing a game with tons of leagues though. The first few seasons are fine but it grinds to a stuttering halt around the 5-10 season mark.

The save games are absurdly huge.

It's going to be trial and error. I'm going to be running it on an amd phenom 1055t 6 core at 2.8 ghz overclocked and 4 gig of ddr3 RAM. So I may go with a third to half of all the leagues.

Great experiment and test. This will certainly help a lot of people.

What do you guys think slows down a game more, loads of leagues or loads of players?

The experiment indicates that it's player.

Yeah, I guess it does. I wonder what would slow down your game more in the long run, though.

The experiment indicates that it's player.

Yes it does. But one thing is that this guy used a quad core. Surley he will have to run a lot more league to make a difference to that CPU.

I might run a test on my Low-med end laptop if people want......

Sure, would be interesting.

Got a link to that.

How did you conclude that.

CBA. I never did so detailed post, but remember asking in the tech support forum with no answer. Just watch the cpu load graph on the holiday screen. During the processing only 1 core at a time is active, while during matchdays all lines jump.

Sure, would be interesting.

OK thats one. I thought it out, will do a test to find out what effects CPU and what effects RAM and by how much.

CBA. I never did so detailed post, but remember asking in the tech support forum with no answer. Just watch the cpu load graph on the holiday screen. During the processing only 1 core at a time is active, while during matchdays all lines jump.

Oh OK.

OK thats one. I thought it out, will do a test to find out what effects CPU and what effects RAM and by how much.

Oh OK.

Will this mean a dual or quad proccy with high speed 3.5 or above would be better than my 6 core at 2.8? Little concerned now.

Will this mean a dual or quad proccy with high speed 3.5 or above would be better than my 6 core at 2.8? Little concerned now.

Err no. Where did I say that??

Err no. Where did I say that??

You didn't but it was said above the only time lots of cores are recognised is when a match is on. When processing only 1 core is used.

You didn't but it was said above the only time lots of cores are recognised is when a match is on. When processing only 1 core is used.

But when you play FM(not holiday it), Day to day processing doesn't matter much. You won't even notice much difference. Its matchdays that take the most time.

IMO you worry too much. Wanna get old sooner.

But when you play FM(not holiday it), Day to day processing doesn't matter much. You won't even notice much difference. Its matchdays that take the most time.

IMO you worry too much. Wanna get old sooner.

Only worry when I spend alot of money mate, but ok i'll take your advice!

Yes it does. But one thing is that this guy used a quad core. Surley he will have to run a lot more league to make a difference to that CPU.

I might run a test on my Low-med end laptop if people want......

I was running EVERY league at full with all players loaded. It is not the CPU - it is the RAM (whatever you have - it won't be enough!)

It becomes unplayable somewhere between 5-10 seasons.

Don't ever try to play with that set up unless you want to wait 30 minutes for the game to save.

Only worry when I spend alot of money mate, but ok i'll take your advice!

I'll tell you that its a awesome build especially for FM. FM also has an award for great utilization of multicore CPUs.

But when you play FM(not holiday it), Day to day processing doesn't matter much. You won't even notice much difference. Its matchdays that take the most time.

When you load all the players and sane amount of leagues (lets say 5), then daily processing is all that matters. Match days are not every day and take significantly less time than crunching through all the players.

I'll tell you that its a awesome build especially for FM. FM also has an award for great utilization of multicore CPUs.

Where is that, I find it really hard to believe?

You can also run FM in a window and keep an eye on task manager, you'll see the normal processing doesn't go above 50% cpu (for dual core).

When you load all the players and sane amount of leagues (lets say 5), then daily processing is all that matters. Match days are not every day and take significantly less time than crunching through all the players.

We have matchdays 3 times a week in the EPL. Add more leagues and it multiplies. BTW 5 isn't sane, its v.v. less.

Where is that, I find it really hard to believe?

IDK intel gave it for their core i series.

You can also run FM in a window and keep an eye on task manager, you'll see the normal processing doesn't go above 50% cpu (for dual core).

Does for me.

Great contribution!!!

I'm surprised to see how little impact it has to have leagues playable as opposed to have them on view only.

Now I'm wondering if I should sacrifice the detail level in future saves. Does anybody know what exactly we lose when being on low detail?

Whats low detail? Small DB?

We have matchdays 3 times a week in the EPL. Add more leagues and it multiplies. BTW 5 isn't sane, its v.v. less.

Oh, well... I never had problems with the leagues, the amount of players is what cripples my games.

The number of cores always shows the correct number, the processor usage graph on the right is what matters. Holiday a while and you will notice how it is at full utilization only on matchdays.