Can a CPU at 80% utilization be a bottleneck?

R

Raxxath

Limp Gawd
Joined
May 29, 2007
Messages
491
So I just built a second rig out of the following spare parts:

Core 2 Duo E6700
P35 Asus Blitz Formula
4GB Corsair RAM
EVGA GTX470
Ooooold WD1200JB HDD

I installed WoW and maxed out the settings (other than shadows) and turned on Vsync just like I do on my main rig. I get a very solid 60 fps on my main rig with brief dips into the 40s or 50s, but I was getting mostly 30-50 fps on this one. My CPU utilization was around 70-80% and my GPU was only about 40%.

So I did a simple overclock from stock 2.66ghz to 3.2ghz and it seems to have made a difference. I'm getting closer to 40-60 frames now, with GPU utilization slightly higher and the CPU about the same.

So I guess my question is... If my processor is holding me back, why isn't it showing closer to 90 or 100% utilization?
 
wow can use up to 2 cores. so the other 20% may be coming from your other core.
 
World of Warcraft still does the majority of it's processing in a single thread. Aside from that, there are several additional threads that handle model loading. So while the game has the ability to run processes on as many as 3 or 4 different cores, in practice the extra model loading processes won't even be enough to fully load a 2nd core. This is why you see almost no benefit in WoW going beyond 2 cores when all else is equal.
 
From my WoW experiences, the CPU architecture is the biggest thing holding that rig back. WoW is very, very cache intensive. The bigger and lower latency the L2 cache, the better the performance. That 4MB cache is what's holding you back.

(I played with various things trying to narrow down where WoW runs into bottlenecks, and turning off the L3 cache on my Core i7 caused the biggest slowdown. Running with just the 256k l2 cache on those cores had it running slower than a dual core AMD machine. CPU speed helps, but not that much. I think that's because most chips link the speed of the L2 and L3 cache with the speed of the processor.)
 
Raxxath said:
Really? You think that would have a noticeable affect on fps?
Click to expand...

WoW loads most of it's textures on-the-fly. A slow hard drive won't necessarily affect your FPS all the time but you will get frequent hitching and stuttering any time it is loading something and has to wait for the hard drive. Sometimes an area will have textures still popping up around you 15+ seconds after you entered the zone.
 
GotNoRice said:
Well the 2500k with 2MB less cache than the 2600k and a 100Mhz clockspeed disadvantage still manages to outperform it in WoW, so perhaps that is only true to an extent?
Click to expand...

Why is the 2500k outperforming 2600k? Does HT slow down single threaded performance?

Wow I can't seem to find any good benches comparing 2500k and 2600k at the same clock speed. This is getting on my nerves. Why do dragon age and wow consistently run better on a 2500k that's clocked 100mhz under a 2600k???

I'm about ready to turn off HT.
 
Last edited:
The processor usage only monitors a particular factor of the processor execution. It does not monitor other factors of processing such as an interrupt request to handle and the time it takes to switch from this to another context, etc. The bottlenecks can come from these factors and/or its ability to execute the instructions.

Why is the 2500k outperforming 2600k at the same clock speed? Does HT slow down single threaded performance?
Click to expand...

The latency to access the LLC (last level cache) can be less due to the fact that it is less in size. Therefore, possibly giving the minor increase in particular circumstances related to cache accesses. This is one of the reasons to the 256K second level cache in the Core i7 architectures. Its small size reduces the penalties; but also reduces which it can hold.
 
flegg said:
Why is the 2500k outperforming 2600k? Does HT slow down single threaded performance?

Wow I can't seem to find any good benches comparing 2500k and 2600k at the same clock speed. This is getting on my nerves. Why do dragon age and wow consistently run better on a 2500k that's clocked 100mhz under a 2600k???

I'm about ready to turn off HT.
Click to expand...

I think me and others have posted this in about every thread comparing the 2, http://www.anandtech.com/bench/Prod...337.338.339.340.341.342.342.63.64.50.49.48.47 - but it usually gets glassed over (I guess because the 2600k costs more?). It was the same way with the first gen chips in games (i5 750 vs i7 920).

Raxxath have you tried running HDTune on that hard drive? It could still run fine but the seek time could be shit due to age, and that could be your problem. Good place to start.
 
Raxxath said:
Really? You think that would have a noticeable affect on fps?
Click to expand...

Yes i do, Wow is very large, and eventually your going to fill up your ram, and constant harddrive file swapping is indeed going to drag down your fps, also it will degrade your OS performance as well.
 
QuiteSufficient said:
Raxxath have you tried running HDTune on that hard drive? It could still run fine but the seek time could be shit due to age, and that could be your problem. Good place to start.
Click to expand...

Thanks for the insight, everyone.

I just ran HD Tach and got:

Random Access: 13.9ms
CPU Utilization: 1%
Average Read: 28.4 MB/s :p

I wasn't planning on using a drive this old, but the 150gb Raptor I was going to use turned out to be dead when I tested it the other day
 
28.4 MB/s, is that a floppy drive you're running off of??? =P I do believe a faster HDD will help you
 
thegtproject said:
28.4 MB/s, is that a floppy drive you're running off of??? =P I do believe a faster HDD will help you
Click to expand...

Like I said, wasn't expecting my raptor to be dead! It was working fine when I took it out of my system a couple years ago and it's been untouched in a plastic box since then. This was the only other drive I had laying around.
 
You must log in or register to reply here.
Back
Top