Suggestions for a video card upgrade.

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Sp12

Senior member
Jun 12, 2010
799
0
76
There was a huge thread with the TF2 dev's talking about performance on the Steam forums back in 08.

Basically, TF2 was originally coded for single-cores, but had multicore backported later. It's not very good, but it will utilize 4 cores. You may or may not have better framerate with multicore rendering on.

Tf2 sees the biggest benefit from cache. The best CPUs for TF2 are those with 12mb of cache, like the Q9 6/5 50 or a Quad Xeon with 12mb of cache (5620). Cache is the key element in Tf2.

Minicrits create ridiculous amounts of particles, which add additional stress on the CPU.

mat_reduceparticles 1 in console may help.
 

evolucion8

Platinum Member
Jun 17, 2005
2,867
3
81
My brother in law upgraded to the HD 4870, and can max Resident Evil 5 which is heavily multi threaded, and never dip that low, is a more CPU demanding game than TF2. It would seems that you have an issue with your setup, try to uninstall your drivers and use Driver Sweeper, also update your chipset drivers. And don't worry about the 8x/8x Crossfire setup, there's a review around that showed that can't even bottleneck an HD 5870 in Crossfire, I use 8x/8x setup and I see most of the time, near 100% scaling, and the HD 5770 performs about the same as an HD 4870.
 
Last edited:

mingsoup

Golden Member
May 17, 2006
1,295
2
81
At 280$ dollars however, would a Q9550 beat out a i5-750 for 195$? From looking at the prices and some of the benchmarks, older more expensive 775 quad's are a pretty damn hard sell, even without a platform change. My memory is worth just as much now as when I bought it. A new motherboard is a new motherboard. You'd save on the i5. I'd probably ditch 2gigs of ram in the process, I really just don't think I need that much. I certainly noticed very VERY little difference adding 4 gigs on top of 2 gigs. Windows 7 ate more ram, but I preen my systems anyway. Noticed little if no difference.

I still wonder about this 5770. My 24" 1920*1200, does dictate a >=5850. And I don't have an SLI board. I think I need to XFire 5770's now. Or sell off this 5770 and upgrade.

Good deal on the mobo, by the way. In Store pickup only though. :\
 
Last edited:

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
Spending $280 on socket 775 CPU is a bad idea. Perhaps you are better off selling off your parts and upgrading to a Core i5 ? Have you considered used Core 2 Quads?
 

edplayer

Platinum Member
Sep 13, 2002
2,186
0
0
I have a Q9550 and I luv it. But for $280, no way!

Maybe if you have a Microcenter close to you, you can find them for $170. But the used Q6600 suggestion was the best so far. Easy to find. You can look for a used Q9550 also but there are not many for sale and people don't usually price them cheap.
 

mingsoup

Golden Member
May 17, 2006
1,295
2
81
An interesting thing I found.

So I benched World in Conflict: Soviet Assault. Here were my results. Maxxed at 1600*900 on a 5770 with E2160 @3.3Ghz.
Min:8 (Nuke going off) Average 31.

Xbit Labs found that with a 5770 and an i5-750 the results were:
Min:19 Average 34.
http://www.xbitlabs.com/articles/cpu/display/cpus-and-games-2010_10.html#sect2


Hmmmm, I would be paying close to 400$ for an i5-750 platform change to add 11fps to my min and 3 to my average. If I hypothetically did sell my crap, I could get that down to maybe 80$ if I was really lucky, but still. :D...........Something interesting I'd thought I'd share.Also of note is those results are the most CPU dependent results because of the resolution, correct? I bet if I bumped the 3.3Ghz to 3.4 Ghz, I might lift that Min....I wonder.......
 
Last edited:

deimos3428

Senior member
Mar 6, 2009
697
0
0
In my experience fps can be significantly server-dependent in TF2. I don't know why that would be, so I won't venture a guess...just an observation.
 

mingsoup

Golden Member
May 17, 2006
1,295
2
81
In my experience fps can be significantly server-dependent in TF2. I don't know why that would be, so I won't venture a guess...just an observation.

Very odd. Wonder if there is an explanation for that anywhere. How to go about picking good servers. Is that why certain servers say 1000fps? I never knew what that meant.
Also, for the caring, I fixed the Natural Selection Problem as well as goldsrc. -gl and -autodetect on startup along with a gammaramp 0 in console seems to have alieved 95% of the problems. Without the -gl I as hovering at 25FPS. Again, I have no idea how this helps.

Also, in the process of doing a complete platform overhaul to i5 land. Stay tuned. (TF2 CPU BOttLes Beware!)