FSXMark11

[Speedbird 217 126]

Wow thanks! It seems I really need to consider a new Nvidia card then. And also interesting to see that we basically have the same FPS, even though you have a Sandybridge and I have the older i7 920...

[kludger 7]

Wow thanks! It seems I really need to consider a new Nvidia card then. And also interesting to see that we basically have the same FPS, even though you have a Sandybridge and I have the older i7 920...

Yeah I think it is good evidence that in FSX for this FSXMark11 the CPU was not our limiting factor with that 5850/5870 card combo.

[Speedbird 217 126]

Today I got an EVGA GTX 570 with 2.6 GB Memory to replace my HD5850. I was very excited to run the FSXMark again, after benchmarking a little while ago with my old card.Here's the results from the first run with the HD5850 again:And here's today's results with the GTX 570:As you can see, there is an average improvement on the AVG frames of a little under 10 FPS. The Min frames seem to have stabilized at 21 now and didn't really change. The Max frames on the other hand went up by almost 17 FPS. It is fascinating to see that against all the talk that FSX is only dependent on the CPU, there actually are better results with NVIDIA cards. What is really interesting to see now is, how FSX will perform with addons and traffic. If the improvement melts away in this case, the new card is still not worth keeping. But this is a promising start.

[SpiritFlyer 987]

Hi Max,Thanks for the tests with both the 5850 and the 570. The resullts are clearly inline with what has become known of the superior FSX performance of Nvidia cards over ATI. Your numbers also show, when compared with others that have been accurately tracked, that the 920 at 3.8Ghz is now clearly limiting the 570's performance. If you overclocked your CPU further, say to 4.2 you would see a corresponding increase in FPS. On the other hand, if you jumped to a Sandy Bridge K model with a high overclock the increase would be a much greater impact still. Here is my point: The performance difference of the 570 GTX is even greater than is immediately apparent from your tests. Your top numbers would be in the high 40s average if your CPU was no longer bottlenecking the results. In other words, equally fast components rely upon each other to keep up.Not sure I am making sense, but hope so. :smile:Kind regards,

[Speedbird 217 126]

Yes, you're making a lot of sense Stephen! Actually the nice FPS improvement is just in a fresh FSX installation. I added some addons and when I load up the NGX in bad weather with traffic in Heathrow, well, the FPS go down to about 15 again. That's a little better than before but still not enough for me to switch from FS9 to FSX for IFR flying. VFR could be something different now. Funny thing is that it still doesn't matter if I set the sliders to medium or all maxed out, the FPS won't change at all.I noticed a huge performance increase in other games though. BF3 or Batman Arkham City run like a charm on the new card. Not really very surprising if you consider that these games use the GPU a lot more efficient than FSX does, and also the PhysX-stuff improves performance. They were very enjoyable before, but 60-70 FPS in BF3 on highest settings is not too bad. :biggrin:I guess I'm gonna keep the card and upgrade to an i7 2600K with a P8P67 and CPU Cooler later this year. Not sure if I'll need new RAM though, I already have 6GB of DDR3 1600 high performance memory from Corsair.

[Kosta 122]

Autogen or Scenery sliders impact FPS the least with i7 machines. The sad thing though is that the first gen i7 really perform bad when it comes to FSX. 2nd gen is killing FSX for that matter, compared to 1st. As I had the 920, I also couldn't bring myself to FSX, but since I have 2600K, no more FS9 finally.You gotta do some decent comparisons between those two to notice differences. Clouds for instance.Anyway, I put your results into the list, thanks for benching!

[Ben Cap 89]

Glad to see that you pulled the trigger on the GTX570. If you paired that card with a 3770K @ 5GHZ, you would see at least 40% boost in performance over what you have right now.

[Kosta 122]

Whoops, I just noticed that after all this time, me, a hardware forums MOD didn't have a benchie in this FSXMark11 thread, at least not the actual one.That just won't do.And yesss... I clocked back to 4.8. 5.0 was just pushing the hard limit (and was causing some rare crashes...).

[Rex251 0]

Hi everyone, I am posting here three results with different clock on cpu, hope will contribute to all results.I followed clean install procedure as described in pdf.rex251

[fghdgdfdfgfgf 111]

Hi Word Not Allowed-What switch did you use in the buffer pool section of your FSX config file?

[Kosta 122]

Hi Word Not Allowed-What switch did you use in the buffer pool section of your FSX config file?

I think it was a default one by bojote, as per manual. I believe it was rt 131072 or so.

[fghdgdfdfgfgf 111]

Thanx, man!

[Kosta 122]

hot&spicy,Your results are impossible.You must have done something wrong with the test.The results of the best system with the SB@5.2Ghz+GTX580 is 51fps average, and you are averaging around 63fps? No way...Please go through the pdf again, setting things exactly as it say, and by installing FSX completely clean. No other tweaks than the ones through venetubo.And if you post again (which we hope you do), attach the shots on the forum (not the link to the terribly slow server, otherwise I'm gonna delete it due to the slow loading times). But just the results. What you posted is unnecessary.Previous post deleted.

[hot&spicy 0]

ok i will all uninstall and try again

[JohnFromCO 15]

Here is the GTX680 on my i5 system.

 

The card was OC'd in Precision X to an offset of 200 on the GPU and 527 on the memory. Adaptive Vsync was turned off during these tests.

 

 

After running the 4 tests I decided to see how much of the GPU is being used. I performed 3 more flights (with Precision X frame rate goal of 120) and used OSD to watch the GPU clock, GPU load, and memory clock. During the run the GPU clock varied but maxed at 1013mhz (far less than the 1324mhz it runs at max stable OC). The interesting part was the GPU load never exceeded 65%. FPS were similar to the regular 4 test flights.

 

Here is a log from the last flight.

 

 

One nice thing you will notice the temps (bottom) stayed at 50 degrees throughout the flight.

 

I am going to load in REX, FTX and the J41 then bump the settings in inspector to see if I can max the GPU.