Starcraft 2 BottleNeck

[UaVaj]

sc2 @ 1920x1080p with everything maxed out including aa.

machine 1
i7-3770k
16GB
GTX 680
OS on 128GB SSD / SC2 on HD-7200RPM

machine 2
G1610
4GB
GTX 580
OS n SC2 on 128GB SSD

watching a 4v4c replay with machine 1 as host and machine 2 as viewer.

~20 min into the game when units are near maxed out. during a 1 minute battle scene.

machine 2 results are:
core 1 = 50%
core 2 = 75%
all core = 60%
ram usage = 2.85 GB (71%)
gpu core = 20%
gpu ram = 0.92GB (59%)

machine 2 fps is 5fps. given that machine 2 cpu/gpu are not even maxed out. what is causing this bottleneck on machine 2? to make this more complicated - machine 1 fps is 20fps.

 

[KentState]

Is machine 2 32 or 64 bit?

 

[Tweak155]

Something else could be using the cores too, thus it can't fully utilize them. If you mean for all processes, I'm not sure.

Even on my high end machine SC2 can slow down, but down to mid 20's.

 

[UaVaj]

both machine are running win7 x64. both machine are running z77 chipset.

only program running on machine 2 are SC2, Fraps and HWinfo.

1) gonna do a fresh win7 install on machine 2 and and see if the results changes.
2) gonna run the replay solo on machine 2 and see if the results changes.

all suggestions welcome.

-----

as for machine 1 cpu results.
core 1 = 40/5
core 2 = 70/0
core 3 = 0/0
core 4 = 30/0
all core = 18%

ram was not accurate because other programs were open in the background. ~6gb total was being used.

gpu was not accurate because SC2 was windows in the centered monitor, while doing other stuff on side surround monitors.

 

[Juddog]

both machine are running win7 x64. both machine are running z77 chipset.

only program running on machine 2 are SC2, Fraps and HWinfo.

How does it run when Fraps is turned off? It sounds like you're probably bumping into the memory limit.

 

[CuriousMike]

watching a 4v4c replay with machine 1 as host and machine 2 as viewer.

Does this mean machine1 and machine2 are tied over a network, with machine1 sending his packets to machine2?

If so, could it be a bandwidth issue ?