Yeah turn times are just not improving. ugh.

Well I'm glad you do but I wish the turn button was more responsive and there was more to do on each turn but there isn't.

Quoting Echillion,

reply 4

Turn the speeds up then KD7BCH

How?

I am already overclocked as far as it will won Air lol but it isn't the cpu speed its the programming. It is wasteful and not efficient. It also doesn't take advantage of all of the cores.

Quoting KD7BCH,

reply 5

Quoting Echillion,

reply 4

Turn the speeds up then KD7BCH

How?

I am already overclocked as far as it will won Air lol but it isn't the cpu speed its the programming. It is wasteful and not efficient. It also doesn't take advantage of all of the cores.

If you have a concrete example that it is not using all the cores, please send that in as a bug ticket and let us know the response.  They have been loud and proud about that feature, and I expect they would want to know about it very much.  I would.  That would be a significant bug indeed.

Can you clarify what you mean by the programming is wasteful and not efficient?  How are you measuring that?  I mean, we all have subjective of responsiveness, etc. Is there a more objective factor I should be looking at?   I know the AI is doing its "thinking" while I am doing things in my turn.  I know that when I end my turn, the delay seems to be in moving ships around as the AI takes their turn.  Is that the inefficiency you are talking about?  Would quicker moves change that perception?  I'm not trying to say you are wrong, but with my middle of the road machine, I am not seeing anything I would put down to programming inefficiency.  Can you give some clues?

Quoting erischild,

reply 7

Quoting KD7BCH,

reply 5

Quoting Echillion,

reply 4

Turn the speeds up then KD7BCH

How?

I am already overclocked as far as it will won Air lol but it isn't the cpu speed its the programming. It is wasteful and not efficient. It also doesn't take advantage of all of the cores.

 

If you have a concrete example that it is not using all the cores, please send that in as a bug ticket and let us know the response.  They have been loud and proud about that feature, and I expect they would want to know about it very much.  I would.  That would be a significant bug indeed.

Can you clarify what you mean by the programming is wasteful and not efficient?  How are you measuring that?  I mean, we all have subjective of responsiveness, etc. Is there a more objective factor I should be looking at?   I know the AI is doing its "thinking" while I am doing things in my turn.  I know that when I end my turn, the delay seems to be in moving ships around as the AI takes their turn.  Is that the inefficiency you are talking about?  Would quicker moves change that perception?  I'm not trying to say you are wrong, but with my middle of the road machine, I am not seeing anything I would put down to programming inefficiency.  Can you give some clues?

I have 8 logical cores on my AMD FX-8350. When I use a program such as Handbrake to encode video it will utilize all 8 cores if I permit that with core affinity. This is the default. It will not use 100% but it will capture between 50-60% of the overall cpu resources in this task, depending on what I do, sometimes it is closer to 75-80% and it is highly parallel meaning that each core gets a good balanced chunk of the workload.

Handbrake will scale to however many cores you have and how fast/much ram you have to feed it. If you have 16 cores it will do the same thing, if you have 4 cores it will do the same thing with 4 cores. The FX-8350 doesn't have the most outstanding cache or memory controls so that is why it doesn't fill the pipe up to 100%. This is both good and bad. Bad because a single project will take longer on the FX than on an i7 albeit it for $150 less silicon for silicon. Sometimes I run two or even three instances of the program which will fill the pipe to 100% and overall it will complete the two projects faster than doing it sequentially.

GC3 will usually load 4 cores at 100% and it will not do this with the other 4 cores, even when set to high priority and all 8 cores are given affinity to GC3. Pure and simple it is not processing the turns as fast as it could if it was more parallelized. I also don't feel turn lengths, minimum time from click to click, are good. I also don't see the game crunching much AI or anything while the player is doing their thing. The interface will freeze up and become unresponsive to varying levels while the AI is executing their turn. So you can't check a report or move about the UI you just need to sit have a cookie, scratch your ass, give it the 15-30 seconds it needs and then you can do what you want.

The efficiency question I have is why doesn't the AI do it's thinking and load the execution while you are taking your turn and then when you hit the turn button all it should have to do is simultaneously move all the units, resolve all the battles, and prepare the battles you can watch, make all of its actual thinking decisions while you are doing your move. I just don't get it. The reason I say it is not doing this is that it only hits 100% on those 4 cores, and only after I hit turn. It doesn't use the other 4 cores, and it doesn't peak CPU usage at all while I'm doing my turn.

To me this is inefficient. Am I whining, call it that if you want. but it is a contributing factor to why the game plays sooo slow.

KD7BCH,

You must have a conflict within your own system. I say this because literally 1000's of players play on Insane (me being one) who do not have the slowing issue you are stating. The game flies for me. Around turn 300 with all AI revealed and most of the map I average about 14-17 seconds on each 'turn' button press before it comes back to my moves. 

Hmm, I read that it is indeed supposed to be working on itself "during" your turn.  I dont play insane, but I'll keep an eye on it, as I too have an AMD FX-8350.

I am running a 2 year old chip, 3570K I5.

I do know that AMD chips have had issues being slower than intel ones. There are a number of posts related to it you will have to dig around a bit. That may be your issue. It sucks because I have always been an AMD guy but after years of tests, the proof is that Intel while more expensive are hands down just better at gaming. 

I cant get it to use more than 6 of my 8 processes.   But that's really the OS's decision, not GC3s.  I remember readinging a year or 2 ago, when I decided to get this AMD processor, that there will be issues with multi threading for gaming (think this is the first true multi-core supported game I've played).  Now, if I happened to be doing a lot of other computing, like watching a movie, running huge excel files, all of this simultaneously, then the AMD extra cores truely kick in and begin to shine.  But, this is really the OS's decision, as it determines scheduling and cpu assignment.  An application doesnt have this ability.

Hence, Intel will be better for gaming (single process using multi-threads), AMD will be better for true multi-process (not multi-thread) use.  

Interestingly enough, there do seem to be some GC3 issues as well.  I was running an insane galaxy, with 26 races, and I turned on soak.  If I kept FOW Off, it would run turns every 8-9 sec. If I disabled FOW, it took 14+ secs per turn.  However, when I zoomed into my own homeworld, where only 1 planet was displayed on the screen, it still took 14sec.  GC3 should realize that I not currently looking at the entire galaxy and be just as fast with FOW enabled or disabled...but its not.

I enabled quick move as well, but it didnt seem to do anything, while watching all races.  So, I think that's broken too, and that would help speed up games as well.

So, I really think the issues belong to Windows and Stardock, not AMD.

FX-8350 is a kick ass CPU isn't it? For the money I think it rocks. Where I find it really shines and earns it's keep is video encoding. It rivals some of the i7s for half the price.

On my system it is pretty concrete using only 4 cores, and I don't see it hitting the other 4 at all except for in the very beginning when it appears to be building something for the AI but I have no idea what.

I have looked at the Corsair h105, which I guess is pretty good for the FX-9590, which is a 225w CPU while the FX-8350 is a 125w cpu. Anyway the h105 is rated as good for the FX-9XXX so my understanding is it is great for the FX-8350. For about $100 I don't think you can go wrong and it is my next upgrade too. I am presently conservatively clocked at 4.6 on air cooling with an adjustable fan module. When I encode video or load it to max I bump the fan rpm to 1800 and I don't see temps above 66C. I would really love to get it clocked at 5.1 or 5.2 just to squeeze the last 10% of her but it flys right now anyway so I don't worry about it.

What do you have your OC to? Also what other coolers have you looked at?

Nothing from Intel that isn't hundreds $ more is coming down the pipe that interests me because their many core cpus are still over a grand and their fast i7s are too highly prices for the number of cores. I am excited by the prospect that HBM and a cpu with 16 cores from AMD might come for around $200 next year which will be on par with i7 single core IPC but yeah

Quoting dansiegel30,

reply 22

On the 8320, I'm at 3.92GHz.  The base rate is 3.5, and the specs say 3.75 is the turbo limit, so my air based cooling system is performing decently (haha, I smoke way too much in my computer room, and my fans and MB are always filthy - further, even though I have 3 fans in the case, none seem to be adjustable in speed).  Havent looked at coolers before, after writing that post I only started looking into it for the first time.  I saw the corsair.  The data I'm looking for are specific examples of people with my CPU, stating the perf improvement with water cooling.  So far, I have seen several examples of different CPUs...the ballpark average is about 25% over turbo speeds. that's convinced me its worth it for sure.  Maybe more if I would keep my PC clean

Yeah I have turbo disabled, and my cores set to always max speed in the bios. I think you can get the 8320 up around 4.6-4.7 on water. Not sure. The fan I have is an oldie but a goodie its a Zalman 9700 they don't make it anymore but they got something similar anyway it comes with this little knob dial that controls the voltage so you can slow your fan down to like 1250 rpm or drive it at max like 2250rpm. I keep it just below my audible range unless I need to crank the cpu then I turn it up a bit