Jump to content
fabio_chavez

Skylake vs Zen for A3 Thread

Recommended Posts

There was a small improvement from the Sandbridge 2600k to the Haswell processors but hardly enough to make it worthwhile. Skylake is another big increase in GPU performance and no movement in CPU performance from Intel so I would hazard a guess its not going to bring much improvement again unless something amazing happens that we haven't yet heard about.

As to Zen based on the fact its a 12 cores (probably stretching that definition like the 8350 did with 8 cores) it may or may not help Arma 3. If they get single thread performance to be roughly the same as Intel's (unlikely considering the vast difference in resources of the two companies) then the extra cores will help the rendering process go a little faster and give the Zen an advantage. But just a 10% drop in single threaded performance will hand the advantage to the Intel CPU again.

Fact is the game just isn't written with todays or future processors in mind, thus it doesn't care much about the architecture changes coming and the general trend towards more cores. Its extremely unlikely we'll see a big change to that basic fact in Arma 3's cycle, DX12 is only going to reduce the rendering time in the API which right now is pretty minimal but it wont fix the serial thread issue over the rest of the games processing. I know people want hope but its more honest to just accept the reality that it'll always perform this way and no magic CPU is coming to save the day. The only time to show any hope is once BI themselves freely admit that the client has severe performance problems due to its design and commit to fixing that, until they do that (they haven't done yet) we aren't going to see any improvement.

 

^^^^^this.

Share this post


Link to post
Share on other sites

Cutting 10 AI out of your scenario takes 5 minutes and have the same performance effect as a new $300 CPU. :)

  • Like 3

Share this post


Link to post
Share on other sites

Cutting 10 AI out of your scenario takes 5 minutes and have the same performance effect as a new $300 CPU. :)

1+. The time buying a cpu only for the arma series is long ago....performanceimprovements via patches are unprobable so we have to reduce the range of arma3-scenarios to "simulate" improvement.

Share this post


Link to post
Share on other sites

Proof positive! I'm sold! :P

Yeah, buy the Skylake cause you see some Bars on a graphic from a Polish site.

 

After you bought it dont forget to spam the Steam Discussion Threads with this sentence "have Skylake but no Fps gain"... I bet there will come by some real experts and they will start to "Bottleneck" your system ^^

Share this post


Link to post
Share on other sites

It's from a Polish site, wich is greek to me, so take it for what it is.

 

kkpgKcE.png

 

Wow, so much more performance...

Share this post


Link to post
Share on other sites

Here is another benchmark with skylake and arma3.

 

http://www.hardware.fr/articles/940-15/cpu-jeux-3d-crysis-3-arma-iii.html

 

I would gain like 10 fps against my 3770k @ 4.5Ghz

 

Better wait on official DX12 support (comming with Tanoa DLC), imo. These tests only show that A3's multi threading capabilities are very poor. A i7 5960X has three times the multi threading ability of a i5 2500K yet they both get the same performance. The engine is too heavily dependant on single threaded performance which DX12 should atleast improve.

Share this post


Link to post
Share on other sites

DX12 not gonna help much since the real bottleneck is the simulation engine..

How do you explain the low fps in DayZ then. And only in cities just like in Arma 3. You point your camera at a wall and your CPU gets wrecked by every building in the city, even if you dont see them. From the DirectX12 presentation its obvious that it should help in games like DayZ. Mantle raised my FPS by a lot in Battlefield 4 where i was CPU bottlenecked, why wouldnt dx12 do the same in Arma 3?

Share this post


Link to post
Share on other sites

DX12 may improve the performance in the editor on an empty map, but how should it solve the #1 problem in multiplayer, which is the simulation thread?

Share this post


Link to post
Share on other sites

The age old concern regards performance for this series. Many players have pondered on this very subject, but its un-answerable isn't it. Because its down to the players expectations, the server they use, the amount of players and AI there are etc.

 

BI make the game as a sandbox, then let the players make the 'game-play'. I know its old tech, but it hasn't really given many players that much of a problem over the years, yes it could be better, but name a game that you could say is perfect..

Will this new idea or that new idea make a difference, no not really, until they change the engine, some players will cry-out. Well yes, they go ahead and change the engine, then we'll get the reaction of 'well it won't do this, or it won't do that', why!

Because its a sandbox and the players set out the table.

Share this post


Link to post
Share on other sites

DX12 may improve the performance in the editor on an empty map, but how should it solve the #1 problem in multiplayer, which is the simulation thread?

Didn't they solve that with DayZ standalone? Your FPS doesn't depend on server population there. Also if simulation thread is the only problem in Arma 3, how do you explain the FPS difference between objects detail on low and high. I believe that if they manage to increase the performance in editor it should increase in multiplayer as well, not the same as i singleplayer but difference between 25 and 35 fps is huge.

Share this post


Link to post
Share on other sites

 

 

BI make the game as a sandbox, then let the players make the 'game-play'. I know its old tech, but it hasn't really given many players that much of a problem over the years, yes it could be better, but name a game that you could say is perfect..

 

 

That's strawman. The game can't keep up at a decent fps (and choppy 30fps or under is not ok) even on their campaign. Forget about 60fps v synced, it's just not possible. A lot of players also play at 30fps or under on consoles and don't complain, that doesn't mean it's an "ok" state of affairs.

 

Other games do much better what they intend to do. Take for instance GTA, it can render vast areas with different assets, basic AI going on, all in a smooth and performance friendly package. BF can have 64 players shooting at each other with fairly complex physics going on  and the performance doesn't go down like it would in ArmA.

 

Didn't they solve that with DayZ standalone? Your FPS doesn't depend on server population there. Also if simulation thread is the only problem in Arma 3, how do you explain the FPS difference between objects detail on low and high. I believe that if they manage to increase the performance in editor it should increase in multiplayer as well, not the same as i singleplayer but difference between 25 and 35 fps is huge.

 

This isn't about server population, is about occlusion which the engine can't do properly. You can go on an empty server, in a big town, enter a house. Look at a wall towards the city and see the fps drop massively. Then look at another wall facing opposite to the city to see the performance go up. Basically whenever you have those types of assets close together in one place, the game doesn't like it although a colleague posted a draw call test in which they were quite low. Not only that, but the game also renders far beyond what you see (wasted resoursces). It works ok for woods when you're gonna be GPU bound, but in cities is a mess. Most likely the reason why the houses are empty (in ArmA).

Share this post


Link to post
Share on other sites

That's strawman. The game can't keep up at a decent fps (and choppy 30fps or under is not ok) even on their campaign. Forget about 60fps v synced, it's just not possible. A lot of players also play at 30fps or under on consoles and don't complain, that doesn't mean it's an "ok" state of affairs.

 

Other games do much better what they intend to do. Take for instance GTA, it can render vast areas with different assets, basic AI going on, all in a smooth and performance friendly package. BF can have 64 players shooting at each other with fairly complex physics going on  and the performance doesn't go down like it would in ArmA.

(in ArmA).

 

Lets say there are 20,000 members here to the forum/board (possibly, was much smaller years ago but what with dayz etc). Lets say 50% of those post regular (probably much lower). Of those 50% lets say 50% are unhappy with performance. That's 5000 players unhappy, I think that's far too many 'here', but lets be conservative.

Now these players go across all titles in the series, but giving the benefit of the doubt to you and those that struggle to run the game, lets stick with those figures '5000'. Lets say BI have sold 3 million across the titles (more I would think), so what is the percentage that complain regards performance. If my maths do me correct its one sixth of one percent (don't hold me to that, I've not had my tablets yet :wacko: )

 

So your probably right, its a known fact Arma can't run above 30fps... No, not really. :rolleyes:

Yes, to those that can't run the game, it will and does feel that way, but in reality that's not the case. Well lets say its highly unlikely. ;) I would add, that's probably the case for the vast majority of players that play the series, they don't have too many issues based on what they want from the game.

Share this post


Link to post
Share on other sites

It has been a lot of talk regarding poor performance, some accept it and play further, others had stopped. After all there are tens of millions who play at 30 +/-.

Play the campaign, the FPS is all over the place and quite abysmal under draw distance and AI load that will make ArmA ... an ArmA game. Try also MP, you won't get constant 50-60 fps in larger scenarios/map.

Share this post


Link to post
Share on other sites

It has been a lot of talk regarding poor performance, some accept it and play further, others had stopped. After all there are tens of millions who play at 30 +/-.

Play the campaign, the FPS is all over the place and quite abysmal under draw distance and AI load that will make ArmA ... and ArmA game. Try so MP, you won't get constant 50-60 fps in larger scenarios/map.

Exactly, the best way to see how ARMA 3 performs is in campaign, which has already a quite large number of AI by itself.

In MP there are many other factors to have in consideration, server hardware and performance, stances running in same machine, external scripts for mods and some other gimmicks, also would be needed a miracle and a insanely powerful machine for to have an acceptable performance in servers with 64 players or above.

Share this post


Link to post
Share on other sites

 

This isn't about server population, is about occlusion which the engine can't do properly

Yeah but in Arma 2 dayz mod your FPS dropped with more players on the server, which doesnt make any sense. I believe thats the case for Arma 3 as well.

Share this post


Link to post
Share on other sites

Those are two different issues that affect the FPS. View distance/occlusion is still there, not sure about the other. It's logical that the performance drops if the server can't keep up with the amount of data it needs to process. I really don't understand why people keep making servers into 50-100 players, when choosing half of that will give better and maybe stabile results. Perhaps some just enjoy or are ok with relatively low fps.

Share this post


Link to post
Share on other sites

It's logical that the performance drops if the server can't keep up with the amount of data it needs to process. I really don't understand why people keep making servers into 50-100 players, when choosing half of that will give better and maybe stabile results. Perhaps some just enjoy or are ok with relatively low fps.

Why  does the FPS drop on the client if the server cant keep up, that just doesn't make any sense. Also on many mods more population means more action and fun.

Share this post


Link to post
Share on other sites

Someone else could probably tell you more about the server side. I prefer 20-30 people with 50-60 fps than 100 with 20-30.

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now

×