High cpu usage+low gpu usage/low fps?

[chuy409]

Why do I get such low fps in tf2? Gpu usage is around 20-30 percent MAX while CPU is around 50-60 on ALL 4 cores I dont know about you but thats too much. I get around 30-40 fps during heavy fights  I would get more fps with my old gtx 460 768mb. There is definately something wrong.

My system specs:
AMD Phenom x4 II 955 @4.1ghz
GTX 770 2gb stock
12GB ddr3 1600mhz

I mean, my phenom should easily be able to handle tf2. I run max settings including AA at 1080p.
When I play crysis 3, cpu usage goes higher like around 80's but gpu usage is almost always 95+.

[ApatheticMango]

The best answer I've read for this is it's just that TF2 is dated. The Source engine is pretty damn old, and it's just not as efficient as the other shiny new engines. I get around 60-70 on BF3 but sometimes TF2 drops to low 30s for me too. No clue why switching to a newer card made it worse though.

[Christian]

Hmm weird I'll have to test my system what the % is. I get 200+ fps and hovers around 150 when I play. I'm using GTX 480 and i7 920. I don't play on 1920x1080 tho. I have mine on 1366x768.

[chuy409]

This is a proper bottleneck im getting in Crysis 3. The percentage left most in the first line is GPU usage. and the 4 going down vertically is each cpu core usage.



In TF2, i get somewhat high cpu usage while gpu usage is just a joke. and it gets even worse when the battle intensifies. It even goes as low as 20 percent on the GPU.

[Vic-Dog]

What program are you using to get those on screen specs? 

[chuy409]

What program are you using to get those on screen specs?

I am using afterburner and hwinfo 64 I believe what the program is called for the cpu

[Dave]

I'm not sure TF2 utilizes multiple cores. It does show around 50% cpu on all four cores, but that may just be one or two threads bouncing between cores.

Also, TF2 uses only d3d9 whereas crysis 3 uses d3d11. d3d9 uses a much older shader model, which means that it may not be able to fully utilize the gpu. Take a look at this table and look at the 2.0 and VS2.0 columns (d3d9) and compare to the 5.0 and VS5.0 columns (d3d11).

[rvns]

tf2 is very dependent on single threaded performance, doesn't really matter how many cores you have as long as it's over 2. your phenom proc is pretty old so that's definitely what is bottlenecking your performance. you could put a titan in your machine and it probably wouldn't make that much of a difference

the drop in performance from the new card is probably related to how well it does d9 stuff as dave has stated, newer cards may possibly be worse in this regard as they optimize for d11.

[Christian]

edit

[chuy409]

tf2 is very dependent on single threaded performance, doesn't really matter how many cores you have as long as it's over 2. your phenom proc is pretty old so that's definitely what is bottlenecking your performance. you could put a titan in your machine and it probably wouldn't make that much of a difference

the drop in performance from the new card is probably related to how well it does d9 stuff as dave has stated, newer cards may possibly be worse in this regard as they optimize for d11.

 Yea thats true of the single threaded thing of tf2. The problem is phenoms are almost at the top of the chart when it comes to performance despite their age. Vishera is marginally-but noticiably better core for core than phenoms. The 8 or 4 cores of the cpu doesnt matter. Phenoms straight out beat bulldozers and is a bit-but not much behind of visheras. Does that mean to find something exaggerated differences I must go Intel? I find that fishy. And we must note that Tf2 came out in 2007 while my phenom in 2009 and extremely OC at 1ghz+ than stock.

[cowking]

Turn off v-sync

[chuy409]

Turn off v-sync

it has been off ever since i started playing TF2. I never play with vsync on. There is just so much lag like in movements to actual show in my monitor.

[bluezhaan]

the difference in video cards is probably negligible for performance.

tf2 is a cpu bottlenecked game.  Added content keeps reducing performance.  I run a i7 2600k oc'd to 4.6 with chris maxframes config at 1080p and framerate will still will drop to 100 hz on a 32 person server in a heavy firefight.

If you want better frames, get a better cpu, lower the graphic settings.

[chuy409]

oh man. I need more than a modern CPU to run TF2 at decent frames. A 6 year old engine needs a CPU from 2013. I doubt it. So does that mean TF2 ran like sh*t with high end CPUs of 2007? Even without all the fancy enhancements we have now?

And yes, I know TF2 only uses 1 or 2 cores instead of all of my 4. Well, shouldn't TF2 at least try to use ALL of core 1 and 2??!?!?!! If CPU usage is around 50 percent each, then i believe 100 percent utilization would fix that.

My god, Crysis 3/BF4/ Red Orchestra 2 are less of a CPU hog than TF2..... Does that sound right? Yes i know it some of them are GPU dependent but you still need decent horsepower to use all of your GPU. They should remove Battlefield 4, Far Cry 3 and other "demanding" CPU games in benchmarks only to be replaced by TF2-if you ask me.... I mean, Red Orchestra 2 is a 64 player game that has HUGE maps that it takes like half an hour to run side to side and needs a BUNCH of CPU power to calculate all the physics of the bullet like bullet drop, body hit detection, tank armor penetration from other tank's shells and Anti tank rifle fire, penetrable walls etc-and yet I need more CPU power for a 6 year-old cartoony game with 32 players. I seriously doubt the little effects they have added since 2007 are severely going to start bottle-necking CPUs. Source engine wouldn't max my and other better CPUs even if it TRIED. And in those games i mentioned, I never fell below 30 fps at max settings @1080p. Yet in TF2, i go down as far as like 20 fps at max settings (even with chris' config file, the lowest i get is 70 fps but in exchange i get a game that looks like my n64 could run).