fixed Performance Loss 2.7.7 OB v 2.7.6 Stable

[Boosterdog]

Im putting this up so it may add (if needed) to Limaros previous excellent submission. Im also posting this as the comparisions have been done on a new PC, using a different brand of Chip and using a brand new Win 10 Pro install. Sadly I patched the install with the MS updates prior to thinking. For this comparison I have no outboard controllers plugged in. FRAPS is used to measure the average FPS. The Chip is a Ryzen 5600X running stock with 2 x 16GB 16-16-16-32 double ranked RAM @ 3600. Win 10 Pro 1904 updated, Latest Bios on MSI Tomahawk X570, Nvidia 496.49 drivers, Crosair RM850x PSU, System and game on separate 4th Gen M2 cards (Sammy 980 pro and WD Black) respectiviely..

The two miz files are used. The first is simply the default Caucaus map with a tracked unit placed at Kobuletti. I created the same mission in both the OB and in the stable and captured images from the same view point in both. I also used ech versions mission in the other and, save for the stasble throwing up a warning of being created on a different ME, saw no difference in ether version on either mission. Settings for both versions are identical in both Game and the NCP.  OB 100 fps, Stable 128 Fps.

 

OB

Stable 2.7.6.

Next I defaulted the NCP and repeated. OB 100fps, Stable 128 fps. The images are in the zip but are essentially identical to the above.

This is approximately the same ratio of performance difference on a low impact mission that I intially experience in my last build. On that occasion I had both the OB and Stable fully populated with all my modules. I have not added all the modules to the stable on this occasion as it would appear that these do not degrade the performace in this Stable version. I do not know if they degrade performance in the OB.  IF REQUESTED, I AM HAPPY TO COMMENCE REMOVING THE MODULES FROM THE OB TO MEASURE ANY DIFFERENCE. 

Of note in the last build was the performance increace on Stable that a generic optimization session had (YouTube ADAM Xs optimisations if anyone wishes to risk it). The result was a further increase of about 7-10 fps in stable but, oddly zero impact on the OB. I have not done this to the new PC as I have yet to pick apart what exactly it is doing and what it is not. 

Next I used the Mil24P instant mission Caucaus Wepons Range, keeping the NCP on default and taking the FPS from the intial cockpit view. OB 74 fps, Stable 86 fps. This is about 4 fps higher on both versions than on my old system both otherwise is consistent to earilier results in the experienced difference. The overall narrowing of the differences over the previous, low impact, mission, I assume is down to the limiting nature of the GPU currently installed bottlenecking the performance (GTX 1080).

Whilst in this mission I used the Riva element of the MSI Afterburner software to measure the CPU use for both OB and then Stable. The image below first shows the use in OB then Stable. I hope its obvious where one ends and other starts but , if not the swap occurs approximately halfway across where you see a period of little use.. The image does not show all cores but these were viewed and showed little to no activity of note. The cores captured however do show behavioural differences. I do not know it this is anything but Ive included it anyway. The result is basically Stable appears to be demanding less CPU use and peaks much lower than the OB in the same scenario.

Finally, Logs for both versions and a DXDiag are included along with the screen captures and the miz files. 

If there is anything more I can help with please ask. 

Supporting Data.zip

Edited November 10, 2021 by Boosterdog

[TAW_Impalor]

Stable was updated to 2.7.7.15038, now there is no difference lol. 

[Boosterdog]

38 minutes ago, impalor said:

Stable was updated to 2.7.7.15038, now there is no difference lol. 

Not sure I get your point

[Flappie]

@impalor Boosterdog made his test using a previous stable version (DCS/2.7.6.13436) as seen in the logs he attached.

Thanks for the data, @Boosterdog

[Boosterdog]

To add to the OP, I had previously, on my previous sytem, experienced a difference in performance on the current 2.7.7. Stable in comparision to the 2.7.7 OB. This equated to about 6 fps in the Mil24 Caucaus Weapons Range Mission at spawn in. At the same time the optimizations put in place using the Adam X guide resulted in a further FPS increase of 6 fps in the Stable 2.7.6 but had no discerable impact on the 2.7.7 OB or 2.7.7 Stable. This difference was not replicated by other people trying the same. I was careful to check my expeiencefor placebo effects but found the increase to be consisent accroos various stock missions and modules tried. 

I have to today installed the current 2.7.7. Stable having again applied the "optimizations" in the ADAMX package. Before dowloading I renamed all installed version's root and saved games folders to prevent crossover with the latest Stable and, additionlly re established the NCP settings for each specific exe, carefully checking the path to each. On this occasion the results are much more in line with those reported by others, specifically:

A very slight overall decrease of frametime low resulting in a very marginal fps/sec increase of 2 fps across all version (2.7.6, 2.7.7 OB and 2.7.7 Stable)

Zero difference in the performance of the 2.7.7. Stable to 2.7.7. OB as should be given there identical status other wise. 

No significant further increase in performance on 2.7.6 Stable over the overall 2 fps increase. 

I cannot account as to why this would be. Had it been down to patching and repatching the original beta, I would have expected to see the increase across all the versions with the 2.7.6. Stable continuing to show a slightly bigger increase. What Im seeing instead is the prevously better performing 2.7.6. Stable which acheived an average fps of 92 in the Mil24 mission now only acheives a stable 88 fps (the same 2 fps increase as seen in the other version and likely more the result of changing the power profile in the NCP) whilst the previously unimpacted versions not showing small improvments. 

I am not attaching logs as, without the orginal logs from the previous PC it would be meaningless. 

@BIGNEWY hate to be a PITA but I notice the orginal (now thankfully locked) thread was tagged as both "Reported" and "Investigating" orignally but is now simply tagged "reported", as is Limaro's thread. Are you able to comment on whether or not the issue is still under active investigation? Thank you.   

[TAW_Impalor]

1 hour ago, Boosterdog said:

Not sure I get your point

Oh, sorry, forgot to end with /s! I mean to say the devs never look back ((

[BIGNEWY]

16 minutes ago, Boosterdog said:

ate to be a PITA but I notice the orginal (now thankfully locked) thread was tagged as both "Reported" and "Investigating" orignally but is now simply tagged "reported", as is Limaro's thread. Are you able to comment on whether or not the issue is still under active investigation? Thank you. 

if it is marked reported the team will be looking for solutions, any fix will be in the change log for a future patch. 

thanks

[Boosterdog]

@BIGNEWY Thank you

[BIGNEWY]

Please note, I am still unable to reproduce, FPS has never been better for me, so its not going to be a simple issue to resolve, if the other team members and devs can not reproduce it makes it very difficult to find the reason. 

[Boosterdog]

1 hour ago, BIGNEWY said:

Please note, I am still unable to reproduce, FPS has never been better for me, so its not going to be a simple issue to resolve, if the other team members and devs can not reproduce it makes it very difficult to find the reason. 

I totally understand. At least Flappie experienced it. Whilst Im not one for threatening tantrummy unilateral boycots it does make me somewhat circumspect about future purchases at this time, especially newer modules. Its not a reactionary thing  but soley based upon the fact that my abilty to enjoy them is likely to be limited without the significant outlay required for a new GPU which, in turn, would pretty much wipe out my free cash for a good few months in any case. The Hind, as a case in point, can really only be used in 4 of the existing 12 instant action missions across the maps without issue on my current set up in OB. Even in the faster running 2.7.6 Stable, its marginal. Spending hard earned on supporting the brand whilst knowing it really needs to be committed elsewhere just to use what I already have is always a feature of simming I guess. 

I fully understand that if ED could fix it they would and also that tech marches on. Im not ranting. Im not even "Dad Dissappointed".  I regret the hit at this time but equally there is always an iceberg on the horizon. Sadly we appear to be somewhat in the middle of an unmoving perfect storm right now when it comes to remaining comfortably current on a casual hobbist budget.

I shall continue to hope for the best, be as helpful as I can to others and enjoy all the other stuff in the game that I can in the meantime.   

[BIGNEWY]

I would suggest continuing to be patient and wait for multicore improvements. 

thanks

[scampaboy]

2 hours ago, BIGNEWY said:

I would suggest continuing to be patient and wait for multicore improvements. 

thanks

Is ED going to be pushing any more patches which affect performance in the meantime? or will there be patches with any optimisation before multicore improvements come about ? 
Just wondering, reading between the lines from your post it seems that’s the only optimisation we can expect.