Upgrading CPU and turn times

[RohirrimElf]

My question is quite simple. I want to upgrade my cpu and expect to get about 30% more performance gain. With a newer generation cpu. At the present i have a sandy bridge 2600k.

What will this do for turn times. I would like to play on a huge map with lots of civs. Which is just about unplayable for me at the present.

I hope someone could help me out. Thanx. (I dont really need advice on which cpu to buy)

 

[Dedrytus]

As you said you do not need advice about CPU... then what advice you want?
Personally I do not think you can gain 30% faster turn times with Skylake.

 

[RohirrimElf]

As you said you do not need advice about CPU... then what advice you want?
Personally I do not think you can gain 30% faster turn times with Skylake.

Please read what i`ve said. I dont need advise on what cpu to buy. I want to know how it affecs turn times. Just a good estimate on performance gain.
I also did not said i expected a 30% turn time gain. I have no clue on its affects for this game when i upgrade my cpu.

 

[Gorbles]

He saying he expects system performance to be improved by 30% I think (which is easily a fair estimation if the CPU is the main bottleneck, which at a 2nd generation Intel Core is likely to be the case).

I wouldn't know exactly how this will help your turn processing times, but mine are manageable (up to one minute lategame?) on an i5-3570K.

(sorry for the stupid advice but bear in mind if you want a current-gen CPU you're likely to need a new motherboard too)

 

[Dushku]

The thing is that Civ 5 is very poorly optimized. So while upgrading system resources WILL help, it's not as much as it would be if the game were better designed. I have a very strong machine and I have most settings on low so I don't burn my components out.

 

[danaphanous]

I have 2.8 GHz Intel Core i7 and play on a macbook pro.

I have heard the game is less optimized on the Aspyr mac port, however, I can play huge maps and the turns for everyone are usually over in 30 seconds or less even at the end of the game. It plays seamlessly for me So you definitely have room for improvement. I also play at mid or high graphics settings but if you haven't already you should try reducing your settings first.

It might not just be your processor though. Depending on how the AI code is being run increasing your RAM might have the same effect for cheaper. I have 16 GB of 1600 MHZ DDR3 RAM as well which is overkill generally for a single-player game like civ V. I haven't checked how much the game uses.

 

[ConjurerDragon]

My question is quite simple. I want to upgrade my cpu and expect to get about 30% more performance gain. With a newer generation cpu. At the present i have a sandy bridge 2600k.

What will this do for turn times. I would like to play on a huge map with lots of civs. Which is just about unplayable for me at the present.

I hope someone could help me out. Thanx. (I dont really need advice on which cpu to buy)

The CPU is not the only thing that has an effect on game performance. Depending on your system there might be other bottlenecks.
Do you meet the other conditions of the recommended sytem requirements (e.g. RAM, graphics card)?
http://www.civilization5.com/#/community/system_requirements

 

[RohirrimElf]

ConjurerDragon,

I easely meet the recommended system specs. Current system:

8 gb ram
3,4 ghz 2600k
gtx 780

I will upgrade after pascal/polaris to something like:
16 gb ram
skylake 6700k or broadwel-e
gtx 1080/polaris 10
500 gb SSD (for gaming)

So my pc is already not to shabby. Even do the turn times are getting a bit to rediculous on mid-end game. I`m using the Superpower mod which does kinda increase the number of units on the map i guess.

If i do get an improvement of about 25-30% i would continue playing civ 5 from time to time as my turn time issues got somewhat solved. Among more demanding modern games.
Civ 5 aged very well considering i was already playing 6 years ago i guess.

 

[rschissler]

Maybe before you upgrade you could stopwatch some of your turn times on a saved game. Then, with the new system, load the saved game and time it again. Then report back here. It would be curious to know how much of a performance increase you would get.

 

[RohirrimElf]

Maybe before you upgrade you could stopwatch some of your turn times on a saved game. Then, with the new system, load the saved game and time it again. Then report back here. It would be curious to know how much of a performance increase you would get.

Can write a review about it with a high-end system. Since i dont really upgrade purely for civ 5. This game still has hundreds of hours of gameplay left in it. Civ 5 might be interesting to try out at 4k. Since it is one of my most played games. (1200+ hours).

 

[kenrickandbros]

Before you upgrade you'll probably want to know what the bottleneck is. Load a game of civ and F4 out of it. Then open task manager and see which part of your system is at 100%. Upgrade that part first and then see if you really need to change your whole system.

 

[Gorbles]

The thing is that Civ 5 is very poorly optimized. So while upgrading system resources WILL help, it's not as much as it would be if the game were better designed. I have a very strong machine and I have most settings on low so I don't burn my components out.

1. "poorly optimised" could mean one of about a dozen things.

2. "better designed" could mean one of about a dozen things, ten of which don't relate to performance of a specific game mode (or even all game modes).

3. CiV is not going to fry any of your hardware components.

Misinformation doesn't help anyone, please don't attempt it.

 

[phillipwyllie]

i7-2600k v's Xeon e5-2698

Would produce pretty much the turn times in CiV as their single threaded scores are the same. Which is the same for all CPU's since about 2009. Atm it's overclock not upgrade.

 

[Kakuhen]

I will give you my insight after buying a new Skylake "budget" computer.

I played before on two computers :
an old core2quad + HD 4890 + 4go ram DDR2 with a samsung EVO SSD
a Mac Book pro Retina mid 2014 with a i5-4278U: 2.6GZ dual core, with 1500Mo Iris IGPU, may turboboost to 3.1Ghz, Sata E SSD

And my new computer :
Z170i
Skylake I3 6100 - 3.7Ghz Dual Core with hyperthreading (simulate 4 cores)
2*8GO DDR4 - 3200Mhz
Asus STrix GTX 960 - 2GO
SSD from the old computer
...

It's kind of a budget computer (it depends on your definition of budget ...) with a 230€ GPU and 130€ CPU. I spent little more money to get good DDR4 Ram a good Mobo to be able to upgrade later if I want. I was pretty worried about that CPU performances as it's the cheapest i3 CPU, it is the cheapest skylake cpu with hyperthreading atm.

Well I can play Civ5 and Endless legend with all graphics + AA at highest settings, no lags, turns are really faster.
I didn't benchmark the turn times but on my mac I had to play 6 Civ map on single player to get doable turn time, it was getting a little too slow on 8 players.
Now I get at least same speed with 12 Civ map on single player as I used to on my Mac Book Pro Retina on 6 players, but it feels faster !
I think I will play a 12 Civ map and post turn times on both computers to benchmark it if some people are interested. I wonder how long will be turn times in late game.

 

[TheMeInTeam]

I wouldn't know exactly how this will help your turn processing times, but mine are manageable (up to one minute lategame?) on an i5-3570K.

I don't think having turn times longer than player turns in many cases is manageable . Unless one's really micromanaging the end game the only serious barrier to sub-90 second turn times is that the game itself hampers you from physically giving orders by refusing to let you select and move the next unit as quickly as you'd like.

My specs are pretty far above "recommended" at this point, though I'm on AMD. I don't think my late-game turns are quite that long, but haven't done them in a while admittedly.

2. "better designed" could mean one of about a dozen things, ten of which don't relate to performance of a specific game mode

Lol I was just envisioning something ridiculous on seeing the "better designed", like all wonders cost 1 hammer. I don't think this would noticeably affect performance at all, and it's doubtful it would hold up as "better", but it's still a good illustration of design change that wouldn't do crap for performance.

I do wonder what the heck the AI is doing with its algorithms though. Generally speaking, I can't issue inputs or make calculations as fast as a machine can do those things, so it's vexing that in some cases I'm markedly faster than the average AI turn on recommended specs late game, if we're both at peace.

What IS causing most of the slowdown? Is it the AI's pathing against other units/ZoC? Are algorithms slowing the game down drastically while getting (extremely) marginal tactical utility out of the AI? Why is it a game like EU or Sc2 can handle more units on the board moving simultaneously in real time faster than AIs can take their turns in Civ V?

What's *really* happening that the CPU is taking longer to move its units than I would, absent input lag? Why do I have input lag on my own turns on that note? Is unit cycling really so hard that issuing a move command then clicking on the next unit is something where I should be having to wait in a game made in the last 15 years? What's being computed that takes so long?

 

[joncnunn]

What's *really* happening that the CPU is taking longer to move its units than I would, absent input lag? Why do I have input lag on my own turns on that note? Is unit cycling really so hard that issuing a move command then clicking on the next unit is something where I should be having to wait in a game made in the last 15 years? What's being computed that takes so long?

It's definitely related to the graphics because in late game when I'm in strategic mode, it cycles thru units much faster than if I'm in normal mode. This problem got worse in BNW compared to G&K, so perhaps the trade units slow things down even though they can't be selected.

 

[Gorbles]

I don't think having turn times longer than player turns in many cases is manageable . Unless one's really micromanaging the end game the only serious barrier to sub-90 second turn times is that the game itself hampers you from physically giving orders by refusing to let you select and move the next unit as quickly as you'd like.

My specs are pretty far above "recommended" at this point, though I'm on AMD. I don't think my late-game turns are quite that long, but haven't done them in a while admittedly.

Lol I was just envisioning something ridiculous on seeing the "better designed", like all wonders cost 1 hammer. I don't think this would noticeably affect performance at all, and it's doubtful it would hold up as "better", but it's still a good illustration of design change that wouldn't do crap for performance.

I do wonder what the heck the AI is doing with its algorithms though. Generally speaking, I can't issue inputs or make calculations as fast as a machine can do those things, so it's vexing that in some cases I'm markedly faster than the average AI turn on recommended specs late game, if we're both at peace.

What IS causing most of the slowdown? Is it the AI's pathing against other units/ZoC? Are algorithms slowing the game down drastically while getting (extremely) marginal tactical utility out of the AI? Why is it a game like EU or Sc2 can handle more units on the board moving simultaneously in real time faster than AIs can take their turns in Civ V?

What's *really* happening that the CPU is taking longer to move its units than I would, absent input lag? Why do I have input lag on my own turns on that note? Is unit cycling really so hard that issuing a move command then clicking on the next unit is something where I should be having to wait in a game made in the last 15 years? What's being computed that takes so long?

r.e. my turn times, I dunno, I just expect it for games at that scale at lategame. I had it with SotS (the original, I never bought SotS II); the more planets I have to auto-process, the more factions, the more fleets . . . it all adds up. Sometimes exponentially! I will say that semi-recent-now-old patch for CiV where they (apparently) backported some Beyond Earth improvements into the codebase actually made a noticeable dent in end-of-turn times for me across the board, for the whole game.

(and yeah I like to micromanage a lot. Gives me that empire-building feel which I don't require, I just like doing)

From what I remember of Clever People talking about CiV (or it might have been earlier Civ. games) both ZoC impact on movement calculations and trade-route calculations in general were cost-heavy. Can't underestimate (automated) Workers either. Europa Universalis is a fun one because there aren't as many movement restrictions; combine this with in general a lower unit "count" (you don't tend to have anywhere near infinite army stacks, and each stack counts effectively as one unit) and you have a smoother calculation in general. This is mostly guesswork - I don't know how good EU4 is either; I've mostly played EU3.

In terms of the delay of moving between units, I really don't know. Mostly likely to be graphics-related as others have noted Strategic Display helps. Probably around how the camera pans / how it tries to render as the camera pans. Fast movement combined with deferred draw calls will add up in any theoretical situation to a bit of a delay. Given how people can encounter this kind of hitch panning the camera themselves - especially on larger maps (remember CiV is 32-bit, large maps encounter Issues) this is my best guess here.

In BE as well as CiV (though I play far more BE these days than I do CiV) the map size has a huge multiplier effect on these kinds of things. Duel maps are rarely badly-affected, even in lategame. Well, depending on Worker spam

 

[TheMeInTeam]

From what I remember of Clever People talking about CiV (or it might have been earlier Civ. games) both ZoC impact on movement calculations and trade-route calculations in general were cost-heavy. Can't underestimate (automated) Workers either. Europa Universalis is a fun one because there aren't as many movement restrictions; combine this with in general a lower unit "count" (you don't tend to have anywhere near infinite army stacks, and each stack counts effectively as one unit) and you have a smoother calculation in general. This is mostly guesswork - I don't know how good EU4 is either; I've mostly played EU3.

EU IV added zone of control to a new fort system in a patch last year, and there are hundreds of tags on the board, so even if each one only has one stack (severe simplification, in many wars AI will move anywhere from 5 to 20 separate stacks, some of them size 1 or so) you're still talking hundreds of potential units running afoul of ZoC. Granted, from a design standpoint fort functionality is trash right now, but from a movement/game speed perspective you're certainly not churning on recommended specs. Even if you count each stack as one "unit" the number of units in that game far outstrips most Civ V games, and Civ V doesn't have to do it real-time.