Whoa, I think you misunderstand me.

I thought he might not have known that using Windows could not only give him much better performance, but sidestep troubleshooting his issue, too. Installing a Windows partition on a Mac is super easy and copies of Windows aren't hard to come by. I was trying to help. Really trying to be respectful here, I'm a little baffled by the hostility.

It's still on my docket. I need to clean it up and organize it so that it'd be more palletable and readily customizable for the public.

Truth is though, there's not much to it or room for differences in a performance config. We know which variables help. Mostly you'll want to decide on your max shadows value or having rag dolls on/off and such. A couple of the threading cvars only work on Intel cpu's. I want to organize it so the stuff we know helps is tucked away and the values a user might want to or have to customize to their build/preferences are neatly up top.

Glad to be reminded to do this :)

Well, wouldn't switching to Windows also solve the problem? I thought it might be useful for him to know, sorry if I'm violating forum etiquette.

It happens to me (gtx 760) when using the lod tweak + mat_picmip -1. Setting it to "2" fixes.

Would it be possible to get Clean TF2 working again with a method similar to the preload_map fix used with ytrium's viewmodels? http://www.teamfortress.tv/34834/yttriums-competitive-viewmodels

Hate to be that guy, but anyone that can and cares about performance should run TF2 in Windows instead of the Apple OS; the frame gains are huge.

Try changing your mat_picmip value to -1 or 2.

Server owners: just say NO to bison impact. Because, you know, doors n shit.

There's a file in "tf" folder that I believe stores them--I think "stats" is in the file name. I'm not sure how to resolve the problem, but it's a start. Good luck.

Couldn't hurt to validate your cache files (right click tf2 in steam>properties>local files).

It seems like having a main caller is something that would be an objective advantage so long as the caller were proficient at the role. So the more pertinent question is whether or not that's something "we" want as part of the format. From the perspective of a captain trying to build and manage a team at any level of competition it seems like a nightmare to me. It's hard to gather and coordinate six people, and adding a seventh non-player but super important person to the equation is ultimately a big complication.

I'd say keep it simple and ban it. Part of the greater feat of excelling in competitive play is the formulation and communication of higher level calls while also jumping around killing dudes.

Interesting. Best to leave the smoothtime value at default, I guess, and picmip to visual preference. It is weird that a networking value has any impact, especially when benching offline, isn't it? I figured that the engine having to perform "smoothing" less often was responsible for my small performance boost, but it was odd to find any difference to begin with. In all likelihood it simply has 0 impact and the differing benchmarks are a result of something random and unrelated.

Specs, out of curiosity?

I'm using an i7 4790k @4.0ghz and an nvidia gtx 760

I made a couple minor finds:

-mat_picmip -1 (highest quality textures) counter-intuitively gives slightly better frames than 2 (lowest) again. For a while using 2 actually caused a very major performance drop, but it was fixed and they yielded the same frames...not sure why there's any difference again.

-cl_smoothtime affects FPS. Increasing it to the max value of 2 (default 0.1, minimum 0.01) yields a small performance boost. If you use cl_smooth 0, the cl_smoothtime value shouldn't have any effect other than on FPS, so those users might as well max the value. If you use cl_smooth 1 though leave it alone, 2 would be bad.

I'm thinking about releasing an updated config, as I fear comanglia has become inactive (you there buddy?).

2639 frames 13.608 seconds 193.92 fps ( 5.16 ms/f) 14.320 fps variability
2639 frames 13.599 seconds 194.06 fps ( 5.15 ms/f) 13.640 fps variability
2639 frames 13.610 seconds 193.90 fps ( 5.16 ms/f) 14.000 fps variability

2639 frames 13.112 seconds 204.38 fps ( 4.89 ms/f) 14.957 fps variability
2639 frames 13.013 seconds 203.80 fps ( 4.93 ms/f) 14.336 fps variability
2639 frames 13.158 seconds 204.56 fps ( 4.99 ms/f) 14.106 fps variability

2639 frames 12.887 seconds 304.79 fps ( 4.88 ms/f) 22.259 fps variability
2639 frames 12.581 seconds 305.23 fps ( 4.91 ms/f) 22.301 fps variability
2639 frames 12.208 seconds 304.90 fps ( 4.73 ms/f) 22.756 fps variability

That's quite an improvement! The .3ghz increase in CPU speed and any other changes in your setup definitely clouds how much the optimizations are to thank, though.

Using my own stress-test demo (taken from spectating "action" in a 32 player pl server for ~2 minutes), I went from about 70 to 80 average FPS with an i7 4790k at 4.5ghz OC.

The performance optimizations increased average FPS on my benchmark demo from about 80 average to 90.

Most of the replies here are ad hominem...boo.

He's asserting that the active player-base is too small to make Match-Making viable unless it displays an overwhelmingly enthusiastic response, which he considers unlikely.

While I sincerely want MM to succeed, I think he's being entirely reasonable.

Instead of attacking him or bemoaning his not simply playing cheerleader, perhaps we could discuss the numbers and our own perceptions on whether or not the greater TF2 player-base does indeed have sufficient interest to make MM viable? Maybe if we confront the obstacles and valid concerns that MM faces we can make it more likely to succeed, or at least have a stimulating conversation that doesn't involve being nasty.