It fair to say that almost everything that need to be done pre overclocking has been done.

Sometime in the next couple of day I will remove the Thermal Right Silver Arrow SB-E Extreme

And remount it with Coolaboratory Liquid Pro. I will also turn the fans around on the Thermal Right Silver Arrow SB-E Extreme to pull instead of push. That configuration will work better for air flow on this build. I will not at this point do heat spreader Mod or lap the CPU but hopefully I will do that at ta later date.

That plan is to bench mark the effect of Overclocking with an FSRecorder file of a flight over DRZEWIECKI DESIGN NYCX with the locker spot view. Using the Frame rate counter. See the recorded flight below:

This scenery is challenging probably for any build and the lowest FPS in the clip is 9.5

So with each change in the BIOS I will report the lowest FPS in the playback of the FSRecorder file.

Not a perfect benchmark but hopefully it will illustrate the effect of the Overclocking and other BIOS changes.

Instead of just picking on the lowest frame rate in the FSRecorder playback I'll try to be a little more scientific and try to make each playback of the FS recorder flight file as identical as possible. So my protocol will be;

 

1. Boot the system.

2. leave alone for 5 mins

3. Start FSX

4. Close the addon's that appear in the system tray: in my case Cumulus X, Winch X and EZCA

5. Start FRAPS

6. Load the saved bench mark flight

7. Start the FSRecorder file playback.

8. Pause the simulation.

9. Hold alt to get ride of the Menu bar

10. Un-pause the flight.

11. Start Fraps recording the instance the Bell Jet Ranger skids move.

12. End the Fraps recording when the Bell Jet Ranger touches down.

13. Exit FSX

14. Play back FRAPS recording and in 15sec intervals plot FPS on graph.

 

The First recording is with Optimal defaults loaded in the BIOS which has Hyper-threading on.

So the second recoding is with HT off. I know that benchmarking FSX is a bit of a misnomer but I like the correlation between the first 2 lines on the graph and I guess it is what one would expect to see: a performance improvement in FSX with Hyper-Threading OFF. :lol:

a performance improvement in FSX with Hyper-Threading OFF

 

 

Anyone explane this HT on/off frame rate difference?

I always thought it was more a case of HT on just not giving any advantage.

 

Dave was there an Affinity Mask entry in the cfg and if so was it set for each run?

 

gb.

 

 

Dave was there an Affinity Mask entry in the cfg and if so was it set for each run?

 

No affinity mask in FSX.cfg in either case.

Bios setting where exactly the same for each run

except for HT on, HT off.

 

More on the way.

 

I got kicked out of school for 2 weeks!

reminds me of my youth! :lol:

_{Cutting to the chase here we are at 4.8ghz.}

_{Not yet stability tested but ran 3DMARK06 for just under 40,000 3DMarks.}

_{Ambient temp in the garage is 4C}

_{The clock need to be tweaked and I anticipate a bit of improvement at 4.8Ghz}

_{but better still I suspect that 5Ghz is possible with this cooling technique.}

_{I am very happy to have been lucky enough to get a CPU capable of 4.8ghz.}

_{I must have a horse shoe you know were because my i7 920 was golden at 4.85ghz.}

_{Also I am working on a system to soft shut down the system at the first sign of condensation}

_{inside the case. It will cost about $180 but more about that latter!}

 

_{Have to say so far the performance difference between 4.3Ghz and 4.8ghz is not as much as I thought it would be.}

_{Its about 2 frames a second.}

 

Nice build. I'm also gonna be rebuilding my system next month with an i7 Haswell and an amd 290x

 

Sent from my EVO using Tapatalk

 

 

Nice build. I'm also gonna be rebuilding my system next month with an i7 Haswell and an amd 290x

 

Sent from my EVO using Tapatalk

Good luck with the Build! If you are building a dedicated FSX system you might want to Go with Nvidia over AMD!

 

Performance goal is achieved! I pushed it to 4.9Ghz and went above 40,000 3DMark06 points getting me 41st place in the 3DMARK06 HALL OF FAME :LMAO:

 

I can run the FSRecorder file without fraps running and it is nailed to 30FPS until I turn south toward downtown Manhattan when it drops but never lower than 22FPS. So it is possible to run DRZEWIECKI DESIGN NYCX smoothly on high setting :lol:

 

Here is a screen shot at 4.9ghz:

 

 

Here is the FPS graph for 4.9Ghz:

 

Here is a comparison graph between 4.5, 4.8 and 4.9Ghz:

 

That's it for pushing the limits but note that all 4 cores get to 100% here and the MAX CPU temp is 60C

That's 4.9ghz at 100% load at 60C with an Air Cooler! I think it is far to say that this is truly a superb approach to cooling a system and that is without the CPU heat spreader Mod

 

Next order of business is moving the HDD's to the warm side of the wall and ordering the condensation detection system before it gets really cold :crazy:

What was the ambient temp for those tests?

 

Not trying to be a hater or nothing and I respect the enthusiasm, but you could have saved a lot of time, effort, and money going water-cooled. If you look at my spec, I'm running 48x on 2700K and achieving the same temps and I live in Florida. Rarely goes over 60C on full load with a simple Corsair H100 push/pull.

 

I will say that I really like your home built cockpit setup. :good: I may try to replicate that down the road.

OK! So this is an i7 4770k at 4.9ghz at 60C under 100% load! i7 4770k at 4.9ghz is about the same as i7 3770k at 5.5ghz

60C is 32C lower than the Maximum allowed for i7 4770k! There is not a snowballs chance in hell any normal air cooled or even high end liquid cooled system would achieve these results. Even a chilled liquid cooled system could not do this!

BTW! HERE IS MY DISCLAIMER:

 I AM NOT ENCOURAGING ANYONE ELSE TO TRY THIS. YOU WOULD HAVE TO BE CRAZY OR LOADED OR STUPID OR ALL THREE TO TRY THIS! I AM A 75%, 50% and 25% RESPECTIVELY. :LMAO:

One hour of AIDA64 Extreme complete at 4.8Ghz  4.1Ghz NB frequency and 1.39 CPU CORE VOLTAGE OVERRIDE. Max Temp was 78c. I forgot to take a screen shoot :(

Any it looks like I have a horse shoe you know where. This i7 4770K is golden. And all this without the Heat sink Mod.

 

Still have to pass OCCT4.4.0 before I can say its stable.

_{I finally have a system that is a match for FSX and P3D V2 is out next week :Cry:} 

_{Just kidding I am sure this machine is exactly what P3D V2 will call for.}

_{Anyway made 4.8ghz with 4.5Ghz NB frequency and tested stable with}

_{OCCT 4.4.0 with temp and voltage to spare. That's it though I go this far and no further.}

_{Just another couple of tweaks and I am done and I will post the final Benchmark graph.}

_{Hope I am surprised by a 2 FPS increase but I doubt it.}

 

 

Just kidding I am sure this machine is exactly what P3D V2 will call for.

 

Great to see someone trying something different.

 

I think most of us are hopeing that P3D V@ will not need 4.9Ghz though .

 

gb.

 

 

 

 

I think most of us are hopeing that P3D V@ will not need 4.9Ghz though .

 

I don't think it will but I am sure we are stuck with Overclocking for top performance.

I am also sure that 95% of the problems we have all had with FSX over the years will disappear in a puff of smoke.(Hopefully not my Haswell exploding) :lol:

 

Once P3D V2 is released I will install DRZEWIECKI DESIGN NYCX and try to install FSRecorder and play back the FSRecorder file I have been using for benchmarking the Haswell. It won't be perfect but hopefully I can get it close enough to make a reasonably objective comparison.

 

And I was right to doubt it. No 2 fps increase! I changed the command Rate of the memory to T1 without any extra voltage required so I am to all intent and purpose done with Overclocking. So I can tell you at least with my approach to bench marking the clock increases moving from a NB Frequency of 4.1ghz to 4.5ghz and changing the memory command rate from 2T to 1T made virtually no difference. Maybe 1 fps a one or two of the 15sec interval in the benchmark graphs above.

 

Having said that it made a more noticeable difference is 3DMark06 and put me over 40,000 3DMarks. Not as good as 4.9ghz but a definite increase in performance but not enough to make a discernible difference in FSX.

 

Based on NickN's Guide I do have a little room to increase the voltage and I am inclined to use that to try for 4.9Ghz which does make a tiny discernible difference.

 

Proof of God's existence: 1. Unique cooling plan for overclocking that would in fact require a cherry picked Golden CPU to be worth the effort. 2. By random chance the CPU purchased from newegg is as good as if it had been a cherry picked golden CPU or perhaps better. 3. It could have happened to a less deserving person. 4. God loves bad people more than good people. QED. :LMAO: