Prepar3D V2.4 on an i7 4770k/GTX 780 system at 5ghz

[Guest]

I decided to cut to the the chase since I already had a tested and stable 4.6ghz profile for my system. I duplicated the the exact flight and just as above took an FPS sample at the exact same intervals. Here is the graph:

I think this is pretty conclusive that Overclocking the CPU certainly does have an impact on performance in P3DV2.4. At least with my system it does. I am not sure what the poster that indicated that OCing had no benefit meant? Maybe that the the sim was just as smooth at 20FPS as it is at 30FPS? That's not exactly my experience.

I will upload the recording of the flight at 4.6ghz just in case anyone wants to verify the results for themselves.

I will try 4.8ghz next but its not cold enough yet for 5ghz. I have to wait for at least 0c for that:-)

[Guest]

Here is the link to the 4.6ghz video:

https://drive.google.com/file/d/0B4aX5Q3-OT6kQ0tsTktaTmZtZlk/view?usp=sharing

[JLuis 409]

Avidean,

 

Nice performance test!!!

 

Congratulations,

 

B)

[Guest]

 

 

Nice performance test!!!

 

Thanks for watching :smile: 

I draw the 2 following conclusion:

On the setting I am using there is a 1 frame per second increase for every 200MHz the CPU is overclocked. And an overclocked CPU stave's off OOM’s

The second conclusion is drawn from when the FSUIPC OOM warning first chimes in. In the 3.7GHz test it started at 6m 48s and in the 4.6ghz test it started at 7m 37s

Hopefully I will do 4.8ghz today and see it the results from that support the conclusion above

[JLuis 409]

Thanks for watching :smile: 

I draw the 2 following conclusion:

On the setting I am using there is a 1 frame per second increase for every 200MHz the CPU is overclocked. And an overclocked CPU stave's off OOM’s

The second conclusion is drawn from when the FSUIPC OOM warning first chimes in. In the 3.7GHz test it started at 6m 48s and in the 4.6ghz test it started at 7m 37s

Hopefully I will do 4.8ghz today and see it the results from that support the conclusion above

 

Good! I am anxious to see your results at 4.8ghz!!! B)

[sauviat 137]

And the purpose of this? Will it eradicate Ebola over the winter, or bring IS to its knees in a winter of discontent? Utterly futile in my opinion :huh:!

 

Sauviat

[Guest]

And the purpose of this? Will it eradicate Ebola over the winter, or bring IS to its knees in a winter of discontent? Utterly futile in my opinion :huh:!

 

Sauviat

There have been some posts that suggested that Overclocking did not benefit P3D V2.X

 

There are thousands of threads on AVSIM. I cannot image anything more futile than that making a point of posting in the ones that I have no interest in, that I have no interest!

 

Anyway back to business. I happen to write down my bios setting for 4.8ghz before my last bios update so getting to 4.8ghz was handy enough and here is the graph:

 

So that's it. Fair to say that P3D V2 does benefit from Overclocking

 

Also the FSUIPC OOM warning came just slightly later at 7m 41s

 

I will update when it gets cold enough for 5Ghz. I need an ambient temp below 0c for that. That's likely a month away.

[FlyIce 354]

P3D is almost always CPU-bound, so there is no doubt it always benefits from higher CPU clocks. Thank you for proving this by such a scientific approach!

 

Maybe in some scenery the GPU is overloaded so that higher CPU speed doesn't help, but in majority of cases of low fps/stuttering I believe increasing CPU speed is the only cure (or you reduce the settings). If we have 10GHz CPU now P3D will be running silk smooth in every scenery we can throw in...

[newtie 539]

I'm not so sure that's the case. If you search the FSX forum on this subject, you're going to find some cases where super high overclocks resulted in MORE stuttering. Backing off on the OC resulted in a smoother flight.

 

Best not to assume a high overclock = smoother flight. Many other variables involved here including the vsync problem.

 

Mark

[Guest]

Let's see what the effect is when the GTX 780 is overclock. I don't have time to do it now and there are a couple of warmer days coming so no colder outside than in. There are a few more lines to be added to the graph but they are not going to appear quick. They area:

 

GTX 780 max over volt and overclock. I need below about 5c for that.

5ghz CPU. I need about -5c for that to be safe

and Memory timing of CL8. That needs big voltage and temps below -5c for me to risk it.

and then Hyper threading vs no Hyper threading.

 

Anyway it will be interesting to see what happens when the GTX 780 is overclocked.

I think the result will be surprising;-)

[Guest]

Well unfortunatley if I stick my PC outside ... even in the shade ... it's 91F ... inside the house I keep it at 71-72 F (thank you solar panels).

 

But this might be a safer option than sticking it outside: http://koolance.com/exc-800-portable-800W-recirculating-chiller

 

 

This is part of my new Xmas build PC I'm working on.  It'll be cooling CPU, motherboard, RAM, and 3 Titan Blacks.  Goal is 5Ghz on a 5960X with 16GB DDR4 3200 quad (CL 15).

 

GPU OC will work well for P3D but you really need a good water cooling unit like this: http://koolance.com/video-card-vga-nvidia-geforce-gtx-titan-water-block-vid-nxttn2

 

RAM cooler: http://koolance.com/ram-33-memory-water-block

 

 

You should be able to hit a reliable 5Ghz and considerable OC with the 780 using this setup.  You may need to max voltages but you shouldn't need to go beyond "Safe" voltages.  So over stock clocks you should see on average (not peak) about a 60% gain in performance.

 

Although OC done wrong can be the source of all kinds of issues with applications, OS, and flight sims ... if done right OC can be a big win ... doing it right is the tricky part.

 

Best of luck and encourage you to forge ahead ... just make sure you leave thermal protection enabled ... this should save components.

 

Cheers, Rob.

[gizmosellsbunnys 106]

Box Resembles a welder like mig hahaha. Get -30 below in Minnesota may need extra material protect get frost bite and furry critters lolol

[Guest]

That's quite the setup Rob! Sounds like that is going to be an amazing build. I did look at something similar a couple of years ago but it was not for me. My business is seasonal so I have a lot more time of in the winter than I do in the summer. And of course I don't fly in the winter either. FS is a winter sport for me so this solution works very well for me and my Sim Room is also silent despite the fans going at full tilt.

 

Anyway I have changed tack here. I discovered I actually had Hyper threading enabled for the above test when I thought it was disabled. Became aware of this when I tried to up the CPU Cache from 41 to 45 and could not get a minute into CPU OCCT without getting BSOD with Setting I knew worked before. Discovered that HT was enabled. After disabling it the system was stable.

 

I ran the benchmark again a got these result. I am going to drop lines from the graph as I go to keep it clear:

 

I think this a clear improvement over 4.8Ghz with a 4.1Ghz CPU Cache.

Changes two things here! Not a good approach but I think its a safe beat to say that the different results have more to do with the CPU Cache increase than turning HT off.

 

The other thing is that I did not get the FSUIPC warning of an impending OOM at all during the play back.

 

At this point the FPS in at 5m 30 has gone from 12 at 3.7ghz to 20 at 4.8ghz. Almost Double

[whitav8 98]

I've read that you could overclock just one or a few of the cores and leave the others at lower Ghz in order to keep the temps lower and therefore OC the one core (the main P3D core)  a lot more and maintain stability. Have you noticed the ability to try that? I really appreciate your careful benchmarking - please keep track of memory speeds and other settings (hardware and prepar3d.CFG) so the rest of us can consider following in your footsteps. Nick N. always used to talk about the "sweet spot" for memory timings - hope you find it. Not sure if you can try to measure smoothness - maybe make a spreadsheet with the FRAPS interframe timings (e.g. 35 34 22 39 msec where a stutter is a delta that is bigger than 5 msec?) or whatever Rob A. might suggest. We all want to find high FPS with smoothness - if only we could framelock well.

    Seems strange that OOM is affected by better performance.

 

Thanks

 

Dave

[gboz 157]

At this point the FPS in at 5m 30 has gone from 12 at 3.7ghz to 20 at 4.8ghz. Almost Double

 

 

Excellent work Dave.

 

Any chance of running a straight HT on/off comparison?

 

gb.