What about more multithreading optimization and GPU using ?

When i play FM (day processing), my 8-core CPU load not more 20%. I think it is not good, because FM data is totally parallelable and CPU can't modelling day matches for leagues with this load (20%).

May be you use random for day modelling ? FM manipulate digits and what about GPU using to calculate it?

Please, tell us about game engine (not match) ? it is so old for modification ? Why so many bugs with any release ? You try to rewrite it?

Because sadly it also needs to work on these minimum specs.

OS: Windows Vista, 7, 8, 10

Processor: Intel Pentium 4 Intel Core AMD Athlon 2.2GHz+

Memory: 2 GB RAM

Graphics: NVidia GeForce FX 5900 Ultra ATI Radeon 9800 Pro Intel GMA X3100 256MB VRAM

DirectX: Version 9.0c

Hard Drive: 3 GB available space

Stats from Steam to put everything into perspective, please note that 2.76% XP users don't count as it seems that XP is not supported in this version :

http://store.steampowered.com/hwsurvey

^^^ WHS

Although I'd take out the sadly part. They'd be making a massive mistake - business-wise and just generally - if they were to pander to the high specs, no matter the results they could manage. If they could do both, and scale so that higher specs get big gains while still leaving it playable to lower specs, then great, but I'd be absolutely amazed if this hasn't come up in conversation before. And if it's come up before and not been brought in, then you can bet there's a good enough reason behind it. Probably related to the above.

^^^ WHS

Although I'd take out the sadly part. They'd be making a massive mistake - business-wise and just generally - if they were to pander to the high specs, no matter the results they could manage. If they could do both, and scale so that higher specs get big gains while still leaving it playable to lower specs, then great, but I'd be absolutely amazed if this hasn't come up in conversation before. And if it's come up before and not been brought in, then you can bet there's a good enough reason behind it. Probably related to the above.

I agree that FM2016 would be too soon.

Sadly for me and for ZOXEXIVO and many people who have bought a laptop/desktop in the last 5 years. Or people who are still running 5-8 years old high-end desktop CPU's.

I'm expecting a fully 64Bit game for FM 2017, 32Bit users are dropping like flies (0,5% per month). Singlecore CPU's are getting really rare if you link this to the windows XP story. So FM 2017 could be a gamechanger and for me that would already be enough to buy the new version.

Not expecting anything on the DX12 front for years to come...

PS: multithreading is really a difficult discipline in programming

Because sadly it also needs to work on these minimum specs.

DirectX: Version 9.0c

Hard Drive: 3 GB available space

Stats from Steam to put everything into perspective, please note that 2.76% XP users don't count as it seems that XP is not supported in this version :

http://store.steampowered.com/hwsurvey

I know about this fact, but possible to create scalable game model. it is not so difficult, but need to reorganize current game engine version and it possible to run game with 2 cores, but it will be slow.

Situation like PS4 vs Xbox One where PS take only 30FPS, because XBox One is .....

Multicore computer not so expensive. 2015 year!

I want more scalable game processing structure and i no need new UI features e.t.c. WE NEED NEW ENGINE !!! I want to feed that game physically calculate games and not use random goals and players stats, because some Peter in some town have one core computer with Windows XP.

I agree that FM2016 would be too soon.

Sadly for me and for ZOXEXIVO and many people who have bought a laptop/desktop in the last 5 years. Or people who are still running 5-8 years old high-end desktop CPU's.

I'm expecting a fully 64Bit game for FM 2017, 32Bit users are dropping like flies (0,5% per month). Singlecore CPU's are getting really rare if you link this to the windows XP story. So FM 2017 could be a gamechanger and for me that would already be enough to buy the new version.

Not expecting anything on the DX12 front for years to come...

PS: multithreading is really a difficult discipline in programming

I trust that when the time is right to do it, they will. They'll probably be having to answer different questions than are visible to the average user, so important to bear that in mind.

As long as they don't begin to alienate their customer base by putting the specs too high, then good. I think it's a massive credit that if my fairly decent laptop packs in, I can load up Steam and play on my wife's lesser laptop, or even go and pick up a dirt cheap unit that will play if it came to that.

Like I said, if they can manage both, wonderful.

I know about this fact, but possible to create scalable game model. it is not so difficult, but need to reorganize current game engine version and it possible to run game with 2 cores, but it will be slow.

Situation like PS4 vs Xbox One where PS take only 30FPS, because XBox One is .....

Multicore computer not so expensive. 2015 year!

I want more scalable game processing structure and i no need new UI features e.t.c. WE NEED NEW ENGINE !!! I want to feed that game physically calculate games and not use random goals and players stats, because some Peter in some town have one core computer with Windows XP.

All I can say from that is that I'm glad you're not in charge of decisions at SI. It's great that you can afford multicore computers, and only care about your own experience. Luckily, SI actually have to cater to a community and userbase than just you.

I agree that FM2016 would be too soon.

PS: multithreading is really a difficult discipline in programming

Depends on the situation. FM have many independent from each other data and it possible to calculate it in separate threads

I want to feed that game physically calculate games and not use random goals and players stats, because some Peter in some town have one core computer with Windows XP.

If you have the processing power and the patience, switch the leagues to full detail.

I know about this fact, but possible to create scalable game model. it is not so difficult, but need to reorganize current game engine version and it possible to run game with 2 cores, but it will be slow.

Situation like PS4 vs Xbox One where PS take only 30FPS, because XBox One is .....

Multicore computer not so expensive. 2015 year!

I want more scalable game processing structure and i no need new UI features e.t.c. WE NEED NEW ENGINE !!! I want to feed that game physically calculate games and not use random goals and players stats, because some Peter in some town have one core computer with Windows XP.

I have a feeling that this is linked with the singlecore support or it also could be due to other intrinsics characteristics in this engine.

The fact that Singlecore IPC has only improved marginally (Not forgetting the Ghz barrier due to heat and power draw) in the last years are also something that SI couldn't have expected in their future engine planning.

Very interesting thread where it was proven that running all leagues in full detail showed a significant increase in speed when running a 5960x. So I guess there is already a sort of scaling for the poweruser.

http://community.sigames.com/member.php/6912-yugo23

Football Manager strikes me as a game that for the most part is embarrassingly parallel. I don't know if the AI actually responds to other results in the league, but even if they did (which would make the game more realistic), it should still be possible to separate out different leagues. Quite easily. For example, running through Prem matches on Core #1, L1 on Core #3, and so on. EDIT: Just to clarify, I'm talking about this from a theoretical viewpoint; I obviously don't have access to the source code. But you have a lot of independent calculations, all that needs to be done is to implement a way of splitting them up across multiple threads /EDIT

The game certainly does this when running on full detail, but the quick match engine doesn't support it. Which given that this doesn't even simulate matches properly, it wouldn't make sense for it to care about results in the same league.

As basically almost every computer in the market is multi core these days, having the QME run across multiple cores would result in a large speed boost for pretty much every player. From someone playing on an integrated card on a 10 year old laptop, to someone running a supercomputer desktop. It makes no sense to say "think of older hardware" because older hardware uses multiple cores. It doesn't matter how old the hardware is, because the game still doesn't take advantage of it.

Are you telling me that players running an 8 year old dual core wouldn't mind the game being 1.5-2.0x faster? Quad core users wouldn't mind it being 4x faster?

According to Steam's stats, having FM fully support multiple cores (instead of just via the full match engine) would result in a faster game for over 97% of players. Dual core and quad core users make up over 90% of gamers on Steam.

I'm still waiting for SI to join the rest of the software industry and jump to 64bit

I'm still waiting for SI to join the rest of the software industry and jump to 64bit

This is really one of the most important things performance wise that the game needs, 16GB RAM computers are pretty common these days and it would be nice if FM could use more of it.

Football Manager strikes me as a game that for the most part is embarrassingly parallel. I don't know if the AI actually responds to other results in the league, but even if they did (which would make the game more realistic), it should still be possible to separate out different leagues. Quite easily. For example, running through Prem matches on Core #1, L1 on Core #3, and so on. EDIT: Just to clarify, I'm talking about this from a theoretical viewpoint; I obviously don't have access to the source code. But you have a lot of independent calculations, all that needs to be done is to implement a way of splitting them up across multiple threads /EDIT

The game certainly does this when running on full detail, but the quick match engine doesn't support it. Which given that this doesn't even simulate matches properly, it wouldn't make sense for it to care about results in the same league.

As basically almost every computer in the market is multi core these days, having the QME run across multiple cores would result in a large speed boost for pretty much every player. From someone playing on an integrated card on a 10 year old laptop, to someone running a supercomputer desktop. It makes no sense to say "think of older hardware" because older hardware uses multiple cores. It doesn't matter how old the hardware is, because the game still doesn't take advantage of it.

Are you telling me that players running an 8 year old dual core wouldn't mind the game being 1.5-2.0x faster? Quad core users wouldn't mind it being 4x faster?

According to Steam's stats, having FM fully support multiple cores (instead of just via the full match engine) would result in a faster game for over 97% of players. Dual core and quad core users make up over 90% of gamers on Steam.

The real truth is that Windows XP 32 bit (2.76) is probably a huge chunk in the singlecore stats.

And guess what according to steam, this OS is not even supported anymore for football manager 2016:

http://store.steampowered.com/app/378120

BTW, I have tried running the minimum specs on Football manager 2014 as a test. The game was almost unplayable.

This is really one of the most important things performance wise that the game needs, 16GB RAM computers are pretty common these days and it would be nice if FM could use more of it.

And you wouldn't run the WOW 64 layer anymore. That alone would already have a 3-5% speed increase.

Because sadly it also needs to work on these minimum specs.

OS: Windows Vista, 7, 8, 10

Processor: Intel Pentium 4 Intel Core AMD Athlon 2.2GHz+

Memory: 2 GB RAM

Graphics: NVidia GeForce FX 5900 Ultra ATI Radeon 9800 Pro Intel GMA X3100 256MB VRAM

DirectX: Version 9.0c

Hard Drive: 3 GB available space

Stats from Steam to put everything into perspective, please note that 2.76% XP users don't count as it seems that XP is not supported in this version :

http://store.steampowered.com/hwsurvey

That is not a problem, it will still run. Parallel algorithms can run on a single core. Match processing uses multiple cores when available. The massive issue is player processing (attribute changes, injuries, etc) that takes the bulk of the time spent waiting.

I registered on this board in 2008 to post it as a bug with the "threading" checkbox before I realized what's going on, but they don't ever comment on this topic. I can't think of a reason why SI hasn't touched that part in the last decade. It is extremely important, because it will allow people to play with up to 8 times the player count, which will vastly improve scouting and transfers.

FM is always in the top games by number of players in the Steam chart (above Civ 5 for example) and it is being sold every year. The resources to improve the game are certainly there, if SEGA is being especially greedy is another story.