Just now, SteveW said:

I re-read the entire discussion and could not see where I recommend an AM - lol.

SteveW, are you on drugs?

1 minute ago, OzWhitey said:

To quote you:

"Turning off the first LP is mentioned in the P3D notes, that would require an AM, HT on or off.

[JOBSCHEDULER]

AffinityMask=14"

 

Yes, that's as it is in the P3D manual - you forgot about the rest of the post - lol.

...in the P3D manual, where it says four or more cores means the threading-out can take place onto other cores. The example isn't a good one but serves the purpose to show that AM=14=1110 - in other words avoid the first core and this makes sense whether HT is on or off.

...14 is not a good example because it restricts the sim to three cores (1110) and thus the primary sim process is loaded with other work it could thread-out.

5 minutes ago, SteveW said:

Yes, that's as it is in the P3D manual - you forgot about the rest of the post - lol.

Lol, I'm like literally rolling on the floor this is so funny. Actually, not. I'm trying to get some data, to avoid another cookie-cutter circular affinity mask discussion, as has been seen on Avsim plenty of times before.

Every time I ask you what affinity mask you're hypothesizing would work, you come back with some reply that doesn't even come close to answering the question.

I've tested affinity mask 14 because after asking you at least twice, that's the reply I (seemed) to get.

If it's not 14, what are you proposing does work???

OzWhitey - you asked me to recommend an AM - a number. That is not the same as recommending that Affinity Masks are useful which I have stated. Your problem is mis-understanding where you can help yourself

6 minutes ago, SteveW said:

...14 is not a good example because it restricts the sim to three cores (1110) and thus the primary sim process is loaded with other work it could thread-out.

That...doesn't actually answer the question of what a good example is.

Question: What is 2+3

SteveW answer: it's not 7, but actually that;'s just gravity.... 

Anyway, I've tested AffinityMask=14, and it appears to improves performance. I'll upload the data in a moment.

1 minute ago, SteveW said:

OzWhitey - you asked me to recommend an AM - a number. That is not the same as recommending that Affinity Masks are useful which I have stated. Your problem is mis-understanding where you can help yourself

My problem is talking to a guy for an hour trying to get a straight answer, when I should have just worked it out myself. I'm guessing you don't the answer, seeing as how you keep dodging the question.

1 minute ago, OzWhitey said:

That...doesn't actually answer the question of what a good example is.

Question: What is 2+3

SteveW answer: it's not 7, but actually that;'s just gravity.... 

Anyway, I've tested AffinityMask=14, and it appears to improves performance. I'll upload the data in a moment.

14 with an i7 HT on?

26 minutes ago, Cargostorm said:

It has nothing to do with stable OC or not. If you enable HT, your CPU core temps will rise. If you are at the edge, it might be too much.

Oh, come on guys. Stable means stable. If you CPU starts throttling because it gets to hot due to the overclocking and you either reduce the clock or you turn off HT, it is per definition not stable. Stable OC means, you can run whatever software you want, whatever stress test you want and you do NOT have to adjust anything further. Turning off HT is one of the biggest adjustments you can do, no? What is so damn difficult to understand here? I do not get it...

4 minutes ago, OzWhitey said:

My problem is talking to a guy for an hour trying to get a straight answer, when I should have just worked it out myself. I'm guessing you don't the answer, seeing as how you keep dodging the question.

You have all you need to get your answers, I gave all in my posts so you can work it out for yourself - but you missed it.

OK, test pilot report follows.

Question: Does an affinity mask improve performance in P3D v4.1

AffinityMask=14 on a 5820K (six core) processor.

Hyperthreading on.

Test vehicle: Mooney Bravo

Test location: Lovell (KCHA) in TN (generic mid-sized airport, default)

Addons: nothing running at time of test.

Clarification: why affinity mask 14? Answer: some possibly drunk guy on the internet, who claims to be an expert on affinity masks, seemed to suggest this, and then said it was a bad choice. But I test what I'm given, that's what test pilots do.

Frame rate assessment: point sampling technique, reported in frames per second.

Frame rate: no affinity mask - 234/285/289/221/211/259/239

Frame rate: affinity mask - 261/253/247/225/282/296/293

(Do the math yourself, it's after 1 am here, but the average rate is higher on the second set of readings).

Data before affinity mask change: (default)

Data after AffinityMask=14 applied:

Conclusion:

- Preliminary data suggests a moderate improvement in frame rates.

- It is noted that the test aircraft produced very high frame rates, not seen in typical simulator use. The raises questions regarding the external validity of the data.

- Further qualitative flight data is needed.

- Testing with alternative AffinityMask settings is recommended.

- Testing on multiple systems, including testing with typical add-on programs running simultaneously,  is strongly recommended before these settings be considered routine for use in everyday flight simulator practice.

Report written by:
Robert J Whitehead

P3D Test Pilot

 

 

 

 

 

8 minutes ago, SteveW said:

You have all you need to get your answers, I gave all in my posts so you can work it out for yourself - but you missed it.

Steve, you rambled on for 6 pages and said nothing of value to the world. But I've done the test at any rate, enjoy.

Note to self: do not listen to SteveW about affinity masks, ever again. Write this on your hand now - in ink that will not wash off. Possibly tattoo it. Ok, sorted.

Who's rambling?

 

AffinityMask=14 on a 5820K (six core) processor. Hyperthreading on.

This of course is plain daft as an example with 12LPs.

00,00,00,00,11,10=14

gives use of only two cores on three LPs why would anyone do that? Because they have no idea what they are doing probably.

 

 

Alright look at a simple four core non-HT:

We know that certain threading-out takes place with four (or more) cores available to the sim. So we leave the config with "no AM" - no AM being 15 in this case = binary 1111 - four cores enabled.

Place our addon exe apps on 1110 the 2nd, third and fourth cores and they don't sap the throughput of the first with the main P3D thread on it.

16 minutes ago, aushie said:

14 with an i7 HT on?

Well, yes. I wouldn't have gone with that. But I made the mistake of trying to comprehend SteveW when he was in a zen mood. "You must find the path yourself" etc ect babble something sort of technical etc

But hey, it seems to work. Hurrah! Half the great inventions in the world happened like this, viz penicillin etc.

Anyway, have a look at the data and see what you think.

And if anyone has an alternative suggestion for a 6-core processor, let me know - with an ACTUAL answer - and I'll test it out tomorrow (sleepy...)