cpu choices for faster turns in the later game

intel ? amd? more threads? more cores? M4 apple silicon?

So I currently run an 8840 laptop AMD with a 4070 video card as ROG laptop. In all normal tasks this laptop feels fast and non laggy. To be clear i have it pretty software barebones as my game machine with only a chrome browser and steam and epic installed along with some games. When i play GC4 after about 200 to 300 games on a gigantic galaxy the turns get longer. I have a second monitor pulled up with system monitor and my cpu is never cresting over 25-30 percent and my video card is doing nothing hovering around 5% or less. My 48 GB of ram is not even breaking a sweat at over 32 available (but used as cache). I am playing the biggest maps with one or two "streams" only. If the PC is computing between turns why is the game not maximizing cpu or using the video card for next turn computations? Will a new M4 apple silicon play better? Again not sure if its an integration problem but it seems that my system is not being fully utilized by the game but if some other CPU will be better ( intel or apple silicon) then im willing to spend the money on my hobby. I have actually played a game with more sectors ( alot more streams) where the turns were approaching 30 seconds between yet the cpu was not even breaking a sweat and ram use was under 20GB. So weirdness of the underlying OS (win 11) or the game is unable to utilize all the power available ? For the record this is not me complaining or bitching. Love the game just want to know if other hardware would make the game go faster? Only basing on CPU SSD and memory usage my pc does not seem to be the limitation. Playing on a desktop with faster memory (7500?) There was a program that i used years ago where i would load the games up in a ram disk to speed them up and that was heaven but since SSDs are actually fast now it seems unnecessary but if that will help i would be more than willing to do that! THere was even a steam program call dim disk ( or something like that) it was a Ram disk program that would autosyn back to your drives but last i checked it was abandonware. I have read through some posts here but most make reference to having more ram or latest CPUs and im only 1 AMD generation from newest right now and my RAM is complete overkill according to usage. I had posted on another subject before and they said the game was developed on apple silicon. So if that is true it would run faster on apple silicon. Before i make that investment i would like to know! Sorry for the long post but was trying to give as much information as possible.

TY!!

Roy

galactic civilizations iv

Advanced Search

Watch this post

Do not email me updates for this post Email me updates for this post

Successfully updated karma reason!

scifi1950

Join Date 02/2005

+119

I'm bumping this in the hopes that someone with the right knowledge can give you an answer.

Successfully updated karma reason!

ty!

Successfully updated karma reason!

I used to work on AMD video cards, and I have 2 theories:

1). Let's say you have a quad core machine, and Galciv4 is coded to have two threads: the thread doing the actual AI crunching between turns, and the MEL--which basically lets you click your mouse around but not much more than that. The AI thread would be at 100% and your MEL thread would be very low, resulting in a 25-30% CPU utilization on a quad core.

2). Cache misses. Cache misses even on the L1 can easily cost as many as 16 CPU cycles. Cache misses on the RAM and...let's just say your CPU utilization would be very low while it waits to retrieve a page from the RAM. That's one of the reasons why we multithread: because the CPU is just sitting there. Might as well be doing something else. But you are barebones. You are not doing much else.

First I would look at per-core CPU utilization, and if that is also at 30%, then I would look at how much of the time it is in an I/O wait state. I am more familiar with Linux than Windows at that kind of analysis.

Successfully updated karma reason!

Quoting tetleytea,

reply 3

I used to work on AMD video cards, and I have 2 theories:

1). Let's say you have a quad core machine, and Galciv4 is coded to have two threads: the thread doing the actual AI crunching between turns, and the MEL--which basically lets you click your mouse around but not much more than that. The AI thread would be at 100% and your MEL thread would be very low, resulting in a 25-30% CPU utilization on a quad core.

2). Cache misses. Cache misses even on the L1 can easily cost as many as 16 CPU cycles. Cache misses on the RAM and...let's just say your CPU utilization would be very low while it waits to retrieve a page from the RAM. That's one of the reasons why we multithread: because the CPU is just sitting there. Might as well be doing something else. But you are barebones. You are not doing much else.

First I would look at per-core CPU utilization, and if that is also at 30%, then I would look at how much of the time it is in an I/O wait state. I am more familiar with Linux than Windows at that kind of analysis.

So The game coding might be the answer but i thought that since GALCIV 3 the machine will use all the threads and cores it can find. Am i wrong? My video card seems not in use at all by this game going between 1 and 2 percent ( seems idle normal?). Not even sure if the game uses the video card for anything other than putting the simple graphics on screen. But my 16 threads seem very underutilized even at later stages in the game.

On your number two point you have truly exceeded my hardware knowledge That said what you are describing in your next paragraph ( cpu per core under 30) is definitely happening as all 16 threads dont seem to be utilized. at any kind of demand. Its not like one is full tilt while the others are napping. They are all at about the same 10-25 percent when i hit the next turn button ( less than 5 between turns).

Based on your knowledge here its more of me needing a CPU that is not out there yet. I just bought an m4 pro max with a pro m4 chip so i will try a bit of parallels on it and see if it does any better!

TY!

Successfully updated karma reason!

That is odd that all 16 threads are 10-25%, adding up to a total 25-30% utilization. Not all threads are equal (or at least, not supposed to be). Cores are. Cores are like CPU's. But threads are like programs--they are software, not hardware. It's like 16 different programs running in parallel to form one program. They can be identical programs, just running on different data in parallel in an attempt to divvy it up and run the whole thing more quickly (those are called worker threads), but not all 16 should be exactly identical. Different sub-programs running different code should have dramatically different utilizations.

That leads me leaning more toward a matter of cache misses. Or at least a serialization problem in the shared memory between worker threads, which is a similar problem. And if that's the case, I don't think a new CPU helps you, since you are already well below 100% physical RAM utilization and an M4 already has 16 cores. There are obvious balance and QA problems in Galciv4 still, as it is. Do you think Stardock really extensively verified the multithreaded benchmarks running AI's between turns in a Gigantic galaxy? I doubt it. This looks to me like a software issue that you can only help with good old-fashioned clock speeds. And raw clock speeds just aren't going up much nowadays.

Successfully updated karma reason!

I think it’s more about the game than your hardware. Even on desktop setups, some guys have this issue.

Even with all that power, if the game isn’t coded to fully utilize multicore CPUs or GPUs, the bottleneck stays. I don’t think switching to Apple Silicon will magically fix this unless the game is specifically optimized for it. So Devs need to address this kind of stuff for big maps.. scratch geometry dash

Successfully updated karma reason!