dm_forlorn [Black Mesa]

[will2k]

dm_forlorn is a standard deathmatch map for Black Mesa.

http://steamcommunity.com/sharedfiles/filedetails/?id=813925575

This is a remake of my map de_forlorn for CSGO.

Recommended player load 12-16 players.

Enjoy the map.

 

 

[will2k]

Black Mesa vs CSGO fps

I had an interesting observation while working on the Black Mesa remake of Forlorn.

The fps in the BM version is significantly lower than the CSGO version even though the BM map version has an enhanced optimization system compared to the CSGO one. I was able to make a crude comparison since I now have both versions up and running.

I chose a random location for this quick study and recorded the fps

The fps is 133. Please note that all settings are on high/very high, with anisotropic filtering and anti-aliasing, running in full HD 1920x1080.

Now the same spot in CSGO with basically the same settings.

The fps is 223. That’s a full 90 fps difference, a very significant one.

BM ran originally on an updated and customized orange box version of Source engine, recently switching to Source 2013 version, while CSGO has its own “proprietary” version of Source that Valve does not license to others as far as I know.

I started a quick investigation into the BM version to try and narrow down the reasons for this fps variation.

I had a hunch that I should start my inquiry with the CSM (cascade shadow maps) system.

When BM was released, it did not have CSM since this is not a feature in Source (Only CSGO has CSM and Valve does not license this engine branch nor does it share its code)

The guys at BM decided recently to code their own in-house CSM system; they did a tremendous job and this turned out to be a beautiful system where you can even have named/switchable lights (something that is not feasible in CSGO’s CSM system). However, the coder himself admits that the system still needs tweaking and work.

Based on this knowledge, I decided to turn off the CSM system in BM and test again.

The fps is now 200; that’s a huge increase of 67 fps just by reverting back to the old, “ugly” lightmapped shadows.

In CSGO, there is no menu option to turn off CSM but one can do so using a console command that kills off the CSM entity.

The fps is now 241; a mere 18 fps increase when CSM is disabled.

The CSGO CSM system appears to be more efficient and less taxing on the engine resources and subsequent frame rate.

A second area of investigation was static props.

I hid all static props from the map in-game.

The fps is now 181, a 48 fps increase from the initial version with static props rendered on screen. This is still a noteworthy variation.

I tried the same approach in CSGO where all static props are hidden.

The fps is 252, a 29 fps increase from the full map with static props present. This number is still notable but it is lower than the 48 fps variation recorded earlier in the BM version.

This observation also leads me to think that CSGO handles static props slightly better than BM in terms of frame rate and engine overhead.

Finally, I tried hiding func_detail and displacement in both versions, however, results were very similar on this front where the increase was limited to 10-15 fps in both versions.

In conclusion, we can safely say that more or less the same map in BM has considerably fewer fps than its CSGO counterpart; 2 main culprits in this fps decrease could be attributed to the CSM system and prop statics rendering.

In another conclusion that probably ascertains a 100 similar conclusions before it, Source is really OLD and OUTDATED.

[blackdog]

I didn't know about the lighting improvements, so I was going to say that for sure CSGO has a better vis and handles props more efficiently

Not surprising that their own lighting is so taxing, they're engine base version is older than what CSGO started with I guess, and is stuff that in Valve they have decades of experience doing.