P3D 5.3 Hotfix 2 released

[IanHarrison 522]

2 hours ago, Mejjo said:

Biggest issue for me is the ugly horizon with no haze. I use EA off.

This was fixed previously and seems to be present again with the latest hotfix.

Never noticed that. AS seems to fix it. Anyway I use EA on which with AS gives a superb rendering.

[Mejjo 33]

31 minutes ago, IanHarrison said:

Never noticed that. AS seems to fix it. Anyway I use EA on which with AS gives a superb rendering.

I am in between trying my sim with EA on. The only thing that puts me off is at night if there are any foggy conditions around it shows quite a bright white layer. Do you get this too?

[IanHarrison 522]

1 hour ago, Mejjo said:

I am in between trying my sim with EA on. The only thing that puts me off is at night if there are any foggy conditions around it shows quite a bright white layer. Do you get this too?

If I fly at night it is usually unattended and a long Inter-continental flight. That is, I will take off in the evening and the let the A/P control the plane until I get up the next day. So I can't give you an answer, unfortunately.

[mpo910 2,112]

5 hours ago, Chapstick said:

So did LM really not address their broken BGLs that are causing massive stutters at airports like EFHK? Still seeing them with HF2.

In addition to what Rob A answered, the BGL are not broke. They contain to many data and due this, loading takes time which can result in stutter.

[mpo910 2,112]

5 minutes ago, IanHarrison said:

If I fly at night it is usually unattended and a long Inter-continental flight. That is, I will take off in the evening and the let the A/P control the plane until I get up the next day. So I can't give you an answer, unfortunately.

This sounds very familiar to me! 😂

Marcus

[Wise87 1,281]

Looks like LM posted a small edit for those of us with the lack of Haze with EA off. 

https://www.prepar3d.com/forum/viewtopic.php?p=253081#p253081

[chapstick 2,758]

1 hour ago, Rob_Ainscough said:

Correct.  LM didn't feel they could address that issue in this HF2.

Cheers, Rob.

Ah, so they punted on fixing a bug that they introduced. Cool.

48 minutes ago, mpo910 said:

In addition to what Rob A answered, the BGL are not broke. They contain to many data and due this, loading takes time which can result in stutter.

I know that. Broken.

[Nemo 1,305]

1 hour ago, Wise87 said:

Looks like LM posted a small edit for those of us with the lack of Haze with EA off. 

https://www.prepar3d.com/forum/viewtopic.php?p=253081#p253081

Should be posted in a separate thread here on Avsim.

[SteveW 1,792]

On 1/21/2022 at 7:26 PM, sonny147 said:

Hey Steve what would you use for a 9700k eight straight cores no HT ?

With no HT there's little to think about because there's no core sharing so the simulator choice would be OK.

9700K 8 cores 8 threads - use 255, 255, 0,2,4 (or 0,1,2 works the same). 

[SteveW 1,792]

On 1/21/2022 at 10:41 PM, joepoway said:

Steve,

Why 253 instead of 255 for eight logical cores?

Also doesn’t this assignment put the RenderThreadScheduler on HT0 meaning the hyper thread of Core 0?
 
See link:  http://izn-flightsim.s3-website-us-east-1.amazonaws.com/affinity.html#assign

I would  assume :

AffinityMask = 255
P3DCoreAffinityMask = 255
MainThreadScheduler = 0
RenderThreadScheduler = 2
FrameWorkerThreadScheduler = 4

If you want all cores used for P3D and Main on C0, Render on C1, and Frame on C3 for instance.

What am I missing?

Joe

Hi Joe,

The problem with HT is that we want to avoid the main thread to share with background threads which sap throughput from the main task.

With only four cores we can enable HT and we have 8 logical processors. However they are still only four cores,

Let's look at the Logical Processor map:

[JobScheduler]
AffinityMask=253
P3DCoreAffinityMask=253
MainThreadScheduler=0
RenderThreadScheduler=1
FrameWorkerThreadScheduler=2

03,02,01,00=core number
76,54,32,10=LP number
11,11,11,01=AffinityMask = 253
11,11,11,01=P3DCoreAffinityMask = 253
00,00,00,01=MainThreadScheduler = 0
00,00,01,00=RenderThreadScheduler = 1
00,00,10,00=FrameWorkerThreadScheduler = 2
 

We can see the main task is on LP0, the first LP of core zero which we know often hits 100% or get's close to 100% when we have more fps. If this can't get the throughput consistently then the fps suffers.

There is nothing allocated to the second LP of core zero and so there is no contention, LP0 runs without sharing that core.

The AffinityMask is the same because we don't want the simulator to use it for add-on processes which could otherwise share the core.

The RenderThreadScheduler and FrameWorkerThreadScheduler usually get to around 50% so that can sit on core 1 on one LP each (LP2 and LP3) which is still quite efficient.

The other two cores run 4 LPs with background tasks, scenery gathering often hits 100% so we don't want those shared with the more sensitive task on LP0.

With no HT there's little to worry about.

[IanHarrison 522]

On 1/22/2022 at 10:45 PM, B777ER said:

This is why I have never seen it. I was too cheap to install Office365 on my sim computer. 

Open Office all the way.

[SteveW 1,792]

On 1/22/2022 at 1:03 PM, fluffyflops said:

steve, 

can you run me some numbers for an  8700k with HT off.

this new HF is giving me blurries. 

thanks 

8700K is 6 a core CPU and with HT enabled it emulates 12 Logical Processors. So with HT Disabled there is nothing to do with six straight cores you will have three main tasks and three scenery gathering: '111111'=63.

6 core HT disabled:

[JobScheduler]
AffinityMask=63
P3DCoreAffinityMask=63
MainThreadScheduler=0
RenderThreadScheduler=2
FrameWorkerThreadScheduler=4

03020100=core number
111111=AffinityMask = 63
111111=P3DCoreAffinityMask = 63
000001=MainThreadScheduler = 0
000100=RenderThreadScheduler = 2
010000=FrameWorkerThreadScheduler = 4

With HT Enabled there will be more throughput available to the background tasks (scenery gathering) through hyperthreading at the expense of not allowing ultra high overclocking:

6 core HT Enabled 12 Logical Processors

4053 gives one core for each main task.

05,04,03,02,01,00=core number
...98,76,54,32,10=LP numbers 0-11
11,11,11,01,01,01=AffinityMask = 4053
11,11,11,01,01,01=P3DCoreAffinityMask = 4053
00,00,00,00,00,01=MainThreadScheduler = 0
00,00,00,00,01,00=RenderThreadScheduler = 1
00,00,00,01,00,00=FrameWorkerThreadScheduler = 2

4093 gives extra to the background tasks but shares core 1 to the Render and FrameWorker: 
11,11,11,11,11,01=AffinityMask = 4093
11,11,11,11,11,01=P3DCoreAffinityMask = 4093
00,00,00,00,00,01=MainThreadScheduler = 0
00,00,00,00,01,00=RenderThreadScheduler = 1
00,00,00,00,10,00=FrameWorkerThreadScheduler = 2
 

 

 

[IanHarrison 522]

46 minutes ago, SteveW said:

Hi Joe,[JobScheduler]
AffinityMask=253
P3DCoreAffinityMask=253
MainThreadScheduler=0
RenderThreadScheduler=1
FrameWorkerThreadScheduler=2

03,02,01,00=core number
76,54,32,10=LP number
11,11,11,01=AffinityMask = 253
11,11,11,01=P3DCoreAffinityMask = 253
00,00,00,01=MainThreadScheduler = 0
00,00,01,00=RenderThreadScheduler = 1
00,00,10,00=FrameWorkerThreadScheduler = 2
 

 

Hi Steve,

I tried this on a non- intensive scenery route from Kuala Lumpur to Dubai. FPS hovered around 28-30 (locked at 30). Vram 2.7/7.1. All good. Mini drops to 2-3 fps twice when descending, literally only a half second or so, and one 3-4 second drop to 2 fps when approaching airport to land.

LP % :0 = 100%: 1 = 5-10%: 2 = 40%: 3 = 30%: 4 = 30-80%: 5 = 30-80%: 6 = 90-100%: 7 = 90-100%. Taken on final approach.

Looks like this AM is the best so far.

Will try Dubai to Schiphol tomorrow, with TE Netherlands and Eham v2, a real test.

Thanks.

 

[SteveW 1,792]

12 minutes ago, IanHarrison said:

Hi Steve,

I tried this on a non- intensive scenery route from Kuala Lumpur to Dubai. FPS hovered around 28-30 (locked at 30). Vram 2.7/7.1. All good. Mini drops to 2-3 fps twice when descending, literally only a half second or so, and one 3-4 second drop to 2 fps when approaching airport to land.

LP % :0 = 100%: 1 = 5-10%: 2 = 40%: 3 = 30%: 4 = 30-80%: 5 = 30-80%: 6 = 90-100%: 7 = 90-100%. Taken on final approach.

Looks like this AM is the best so far.

Will try Dubai to Schiphol tomorrow, with TE Netherlands and Eham v2, a real test.

Thanks.

 

That is about all can be done with four cores which is 8 HT LPs.

All we can do with AMs is mask out cores or Logical Processors. In effect this is really only useful to have P3D avoid using some cores or LPs. So long as the main core is not shared (MainThreadScheduler) we will avoid losing fps. With HT Disabled we are by definition not sharing cores. With HT enabled we use the AMs to have P3D avoid using shared LPs for its own tasks and Add-ons as well so the second LP in from the right (LP 1) is defined as zero in the binary so that will not be used by P3D or any addons it runs which is the main thing.

[SteveW 1,792]

23 minutes ago, IanHarrison said:

LP % :0 = 100%: 1 = 5-10%: 2 = 40%: 3 = 30%: 4 = 30-80%: 5 = 30-80%: 6 = 90-100%: 7 = 90-100%.

...also we can see that the RenderThreadScheduler and the FrameWorkerThreadScheduler share the second core LP 2 and LP 3, which are running at around 30-50% so they can share an HT core quite happily. If any of the background tasks (LP6 and LP 7 are near 100%) were running on a core with any of the main three tasks the sim will be having a hard time getting the throughput those need to maintain consistent fps. And finally with the masked LP 1 it is only seeing 5-10% which is at least not robbing the main core too much and allows LP 0 all the throughput it can get which equates to about 90% of possible.