Matrix Games Forums

Will game performance cap out on high end gpu/cpu combo. As a for instance will layering benefit from gpu memory? Will the game work better with more cpu cores.

Let’s say and i7/GF 2080 vs an i9/Titan. Thoughts and any insight into key info about what elements improve performance.

ORIGINAL: afelix1

Will game performance cap out on high end gpu/cpu combo. As a for instance will layering benefit from gpu memory? Will the game work better with more cpu cores.

Let’s say and i7/GF 2080 vs an i9/Titan. Thoughts and any insight into key info about what elements improve performance.

I've got an i7-8700 (6-core), 32 GB RAM, and Windows running on an SSD.

Works fine, can run in fastest if I wish on some pretty hefty scenarios.

Seems to work fairly well with multi-core.

Laptop Lenovo Legion: i7-9750, nVidia 1660 ti, 32 gb RAM, SSD. Op Brass Drum runs about 50:1 on fastest
Surface Book 2: i7-8750, nVidia 1050 mobile, 16 gb RAM, SSD. Op Brass Drum runs about 35:1 on fastest

I use Op Brass Drum as my personal benchmark. btw, CMNAO ran Op Brass Drum at 30:1 on the Lenovo.

To me, raw single core speed is the key. Multi-core helps, but that single core speed is the critical part. If you run TacView as the 3D window, the GPU is very important also. But without the 3D window, I have found only minor differences between nVidia and Intel GPUs.

I have nearly identical specs as your Lenovo and Operation Brass Drum performance gradually degrades as the game progresses and becomes nearly unplayable after about 10 game time hours.

Then there is something wrong with your setup. I have run it multiple to times to the full 24 hours. It will slow a bit when a lot of missiles are in the air, but comes right back quickly. What scan rates are you getting at the beginning and then during missile attacks?

I've not noticed any degradation but I've not played for as long as that at a time.

If it's an issue, I'd first want to check memory usage using resmon. That could reveal if any memory leaks are present - the bane of 24/7 software. That can, for example, slow down sorts and lookups of internal data structures, and even cause excessive paging (swapping to disc instead of pure RAM usage). I'm speculating here, but at least look to rule it out.

i have an I9-9900k with 32 GB RAM. I have run CMANO and CMO on a very large scenario (over 5k units.) CMO does run faster than CMANO, but CMO's best speed on this very large scenario was about 3:1.

Define "very large" scenario if you can.

I will investigate and report back.

Thanks to all for the feedback. I am still wondering if the GPU helps render the graphic layers as you zoom in and out. Any thoughts on that?

Not really. As I already said above, I have seen only minimal improvement switching from Intel graphics to nVidia graphics.

"I have found only minor differences between nVidia and Intel GPUs."

The developers wrote in one of the pre-release posts the following:

"Turbo-charged for your pleasure

One of the most immediately noticeable features, especially when running large or elaborate scenarios, is the jump in performance. Command now offers both greater absolute speed in simulation execution (ie. reduced “pulse time” as displayed when the “Show Diagnostics” option is enabled) as well as significantly increased CPU utilization, as a result of improved parallelism. (As an example of the latter, on a recent trial run of “Northern Fury #9: Hold the line”, CPU load was observed fluctuating between 65%-75% on a Core i7-6700 system. This may vary across different systems, of course). Maximum practical unit count has also increased thanks to better memory management, with the result that even more massive scenarios are now feasible."

As you see they reached 65-70 % CPU utilization on a i7 6700K. If I understand things correctly games are almost always limited by some main process that runs on one core and one core only and uses it to the max and then, demanding on the amount of multicore optimization, the game also runs several other less demanding processes that gets spread out on the other cores by the operating system and that constitutes the rest of the CPU utilization (That is not taken up by other programs).

I guess you don't need much of an graphics card to run this game. CMANO used to run fast in 4K with a cheap Intel Pentium laptop processor GPU, like the 605 series. I might imagine that CMO is slightly more demanding but any post-2015 graphics card from Nvidia like a 960 GTX or an AMD R9 270X that I have in my HTPC seem to be more than enough. CMO is just not a very graphics intensive game so the game speed is practically always CPU-limited.

I can't guarantee I have understood things exactly right but this is my impression.

I have fixed my problem. I cleaned out some programs that had snuck into my start up. That has fixed the memory leak or whatever problem I was having. Now scenario Bass Drum runs very well. Thanks for the tips.

ORIGINAL: mavfin

ORIGINAL: afelix1

Will game performance cap out on high end gpu/cpu combo. As a for instance will layering benefit from gpu memory? Will the game work better with more cpu cores.

Let’s say and i7/GF 2080 vs an i9/Titan. Thoughts and any insight into key info about what elements improve performance.

I've got an i7-8700 (6-core), 32 GB RAM, and Windows running on an SSD.

Works fine, can run in fastest if I wish on some pretty hefty scenarios.

Seems to work fairly well with multi-core.

I'm having issues with CMO and a similar config, albeit only with The King of The Border scenario.

I'm having issues with CMO and a similar config, albeit only with The King of The Border scenario.

There's a known issue with KotB that relates to area complexity. It's scenario specific (i.e. won't affect any other scenarios) and it's being worked on right now [:)]

ORIGINAL: apache85

I'm having issues with CMO and a similar config, albeit only with The King of The Border scenario.

There's a known issue with KotB that relates to area complexity. It's scenario specific (i.e. won't affect any other scenarios) and it's being worked on right now [:)]

Yeah, I read something on this regard so I'm not really complaining [:)]

ORIGINAL: LargeDiameterBomb

...
I guess you don't need much of an graphics card to run this game. ...

I had asked in another thread if it had been considered during development to include support for nVidia's Cuda for mathematical functions in CMO (similar to the way that graphics cards are used for Bitcoin mining), and if there was a particular reason it wasn't done, but I don't think there was a reply to that. I definitely don't think it's needed for graphics processing of the game (though it may help a bit w/ TacView).

ORIGINAL: thewood1

Laptop Lenovo Legion: i7-9750, nVidia 1660 ti, 32 gb RAM, SSD. Op Brass Drum runs about 50:1 on fastest
Surface Book 2: i7-8750, nVidia 1050 mobile, 16 gb RAM, SSD. Op Brass Drum runs about 35:1 on fastest

I use Op Brass Drum as my personal benchmark. btw, CMNAO ran Op Brass Drum at 30:1 on the Lenovo.

To me, raw single core speed is the key. Multi-core helps, but that single core speed is the critical part. If you run TacView as the 3D window, the GPU is very important also. But without the 3D window, I have found only minor differences between nVidia and Intel GPUs.

Update with two newer latops:

Gigabyte Aorus - i9-10980HK, RTX 2080 Super MaxQ, 32 Gb RAM, M2 SSDs - Op Brass Drum ran about 53:1 over 10 run throughs

Surface Book 3 - i7-1065G, GTX 1650, 32Gb RAM, SSD - Op Brass Drum ran at 37:1 over 5 runs.

Just trying to keep the benchmarks going. Keep in mind these are all using the latest public beta. So game optimizations might have as much impact as upgraded hardware.

ORIGINAL: stoltz

I have nearly identical specs as your Lenovo and Operation Brass Drum performance gradually degrades as the game progresses and becomes nearly unplayable after about 10 game time hours.

Try clearing the message log. I've seen slowdowns when the message backlog gets into the hundreds.

That's a good piece of advice and the effort is appreciated, most likely. But that comment is over a year old. I was just updating the benchmark.