Promised performance increases observed - with Patch 16.1 - updated 05.02.2015

[Mr.Freud]

it seems they lowered LOD again... "optimization: CA style" LOL

[alQamar]

Hello lads,

what I can tell meanwhile there is no performance influcence by the driver 335. I ran all tests meanwhile and will upload the results here and the OP soon.

[alQamar]

So there are performance increases as you can see in the OP on maxed out settings and extreme settings we are now close to the release performance we had in September. This is indeed not a big deal, however they at least managed to recover all negative performance drive we had before so up to 12 % performance more performance compared to patch 9 final.

Interesting enough it seems patch 10 indeed seems to reset the whole game to the state it was back in September, as also the performance increases for lower settings are now quite equal to those we had on release.
Now people on low end systems have enough reasons to cry, that's why I also renamed the thread.

This all seems be a really freaky show by CA :o

[Mr.Freud]

I decided to uninstall the game and download it again. Sometimes I think that new patches mess up each other somehow. Next step: clean windows installation...
I just don't understand how they promise better performance with every patch - and we get worse performance.
I also bumped 2 threads over NVIDIA forums - still hope they release SLI profile so I would try downsampling.

[alQamar]

Hi william, thanks for joining. I know the benchmark is not the ultimate tool but it is the only one that meets objectivity in terms of reproduceability of a set on different computers. Unfortunately we did not manage yet to create an own scripted scene that would provide better testing.

Still the results here are very important. The issues you describe having a fluent FPS but stutter, certainly speaks about pathfinding issues again. If the CPU would be overloaded you would not have decent FPS as the CPU has to provide the GPU data.

I haven't monitored the CPU / GPU usage in patch 10. And I doubt I will step into it again, as the changes are too neglible.
If you are interested on further information please consult this thread http://www.twcenter.net/forums/showt...ormance-issues

[Totalheadache]

Hi william, thanks for joining. I know the benchmark it is not the ultimate tool but it is the only one that meets objectivity in terms of reproduceability of a set on different computers. Unfortunately we did not manage yet to create an own scripted scene that would provide better testing.

Still the results here are very important. The issues you describe having a fluent FPS but stutter, certainly speaks about pathfinding issues again. If the CPU would be overloaded you would not have decent FPS as the CPU has to provide the GPU data.

I haven't monitored the CPU / GPU usage in patch 10. And I doubt I will step into it again, as the changes are too neglible.
If you are interested on further information please consult this thread http://www.twcenter.net/forums/showt...ormance-issues

instead of using the benchmark just use a custom siege with all the same variables/settings ets and then record over a certain unit at a certain time...something like that perhaps..

[alQamar]

I thought about that with Samuele to make an own setting. The problem I have is: With some efforts I could assure the quality of a custom scene, but then it gets difficult to be sure others can do it exactly the same. Afterall it cost a lot of time. The simpliest thing would be to make a own battle, save the replay and replay it and measureing this particular performance but I am not even sure if CA ever fixed the replays till today. I think this topic is off the agenda since a while now.

[Denco]

From official forums

My ingame settings:
http://i1251.photobucket.com/albums/...5-28-02-13.jpg

My replay ( It requires DeI mod )
https://www.dropbox.com/s/8v31slai8e...or%20CA.replay

DeI mod ( just in case )
http://www.twcenter.net/forums/showt...ed-11-03-14%29

This replay mimics the battles that I have in my Spartan campaign. I've fought many, and I mean many, battles that were the way this one is set right now. As soon as all of the enemy units are fighting my front line pikes the fps drops to 20. Using patch 9.0-9.1 my fps always stayed above 30 in situations like this ( tested again just to make sure ).

As you can see the performance increase that patch 9 brought has been diminished and I am now again back to were I was pre patch 9. Now I've heard many people saying that patch 10 improved performance but none of them threw any conclusive evidence. I really don't understand what could be causing this "performance differences" between patches but this is getting a bit silly.

Here is my proof how patch 10 ruined performance for me. I do not use benchmark and the only way to properly test the fps is doing it the way I did it. Can anyone here that had performance increase with patch 9 try my replay with patch 10 and 9 and tell me if they have the same fps loss.

[Humble Warrior]

Let`s be thankful it`s a beta for us beta testers to test.

Hopefully the actual official patch will fix all that.

[Achilleslastand]

Let`s be thankful it`s a beta for us beta testers to test.

Hopefully the actual official patch will fix all that.

What developers put their patches thru a beta test? Are they that incompetent that they cannot test the patches themselves. Also odd they beta test the patches now but did nothing of the sort before the games release.

[Humble Warrior]

What developers put their patches thru a beta test? Are they that incompetent that they cannot test the patches themselves. Also odd they beta test the patches now but did nothing of the sort before the games release.

You`re preaching to the choir, m8. It really does look like CA does not (or just barely) test their games at all before release.. If it don`t fall apart after 2 runs they just release, even if obvious cracks are seen. I know, it`s crazy.


No other industry would do this. No one with common sense would do this with anything, but CA does. I think they may even have admitted it. It`s an insult.

[Denco]

I've made a small error in my previous post. The smallest fps in patch 9 was 24 not above 30.

[Condottiere SOG]

CA, listen to your fans...stop listening to that devil on your left shoulder...

[alQamar]

ok I will try what I can do. Any other mods running?

[billydilly]

Hi AlQamar,

I did a benchmark test with Patch 10 on my not-so-new computer:

CPU: AMD Phenom II Quadcore @ 3.2Ghz.
Motherboard: Gigabyte 870A-UD3.
RAM: Kingston, 8Gb DDR3@1333.
HD: Western Digital 500Gb (System)
HD: Kingston SSD 200Gb (For Steam)
GPU: ATI Radeon HD 5770 (1GB) (Driver 13.12)

---

Resolution 1920x1200

I chose quality in the graphics menu:
Extreme: 20.4 FPS
Ultra: 22.3 FPS
Very High: 27.2 FPS
High: 46.0 FPS
Medium: 55.3 FPS
Custom settings: 49.0 FPS (A combination of Low and Ultra)

It should be mentioned that I use RadeonPro with Dynamic VSync wich limits the framrete to max 60.

The FPS is about the same as with Patch 9.1

HIGH


VERY HIGH

[alQamar]

Hi Billy! Thank your for your time and participation. Very interesting, as for my computer I noticed a drop on (Very high 85,4 to 82,8) / (High 155,2 to 141,3) with the same drivers 335.23. Do you have screenshots for me so I can have a look on the performance curves?

[billydilly]

I'm at work at the moment But I'll add it to my post later tonight.

[alQamar]

cool, I am looking forward! Thanks a lot and have a nice day.

[billydilly]

Hi alQamar,

I'v added two screenshots of the benchmarks (High and Very High).

I also tried and added Medium, Ultra and Extreme (no screenshots).

[alQamar]

thank you but those screenshots are done with p10 or patch 9.1? I could need all four runs for comparison.