4 hours ago, Nyxx said:

I just wish Steve would do a stripped down to the very basic AM setting for.

4 core no HT = AM=xxx

4 core with HT ON =AM=xxx

6 core.....you get the picture

That would be nice without all the tech, graphics etc. 

So people can just think " I trust this guy and simple look what AM they need or not need.

Hi David,

Steve has offered a few helpful suggestions here:

http://www.codelegend.com/forums/viewtopic.php?f=2&t=532

Regards,

Mike

Best advice around, you guy's are great! Thanks for the mentions.

I'm basically only trying to drop the best and most relevant knowledge on here that I've found that makes a difference. I know it's a bit of a wind up for some, but that can't be helped. I think it was Budda that said you can't make a difference unless you upset at least one. 

On 1/14/2019 at 1:51 PM, SteveW said:

...it's also a complete professional test harness package ...

I'm afraid most of the comments about IF here miss this important part of it.

Dirk.

Edited January 15, 20197 yr by Dirk98

On 1/14/2019 at 12:28 PM, Ray Proudfoot said:

I'm still shaking with anger at your reply. To call be .... because I don't want to buy some software is totally out of order

Wow, what a scene! lol, I wish you took a selfie of yourself that moment and posted it here ))

Yes, I shouldn't have used the unallowable word, I'm sorry, Ray. 

I just don't understand why would you want to twist my words and their meaning? I called to be 'i.l.l.i.t.e.r.a.t.e' not because you didn't want to buy the software for Christ's sake, but because you threw your unenlightened judgment about the value of the software that was quite beyond your depth of understanding at that moment. More to that, your "I’m more than capable of deciding my next flight" sounded pretty angry or at least contemptuous in context, and that what was really out of order. And yes, we all are very sensitive, intelligent and understanding people on these boards lol.

Water under the bridge though, of course. We are all learning as we go.

Cheers. 

Well said. I offer no argument for things that make sense and make calls based on hard won research into the wee hours. I've not at any moment forgotten Ray's contribution to the FS scene, Concorde or in general. I hoped he might take the hint that if I mentioned I got into that and had all that software and so on, that I just might have learned a lot from his commentary in the past and never argued with any of it. So I'll personally thank Ray for that any time of day.

Eureka!

I've been very happy with v4.4 and was glad to leave all this AM stuff in the pass along with moving addons off & on core etc.

But!

After reading SteveW finding I thought I would open up task manager and watch my 8 "processors" do there thing 4 with HT ON.

Like I said I was very happy with 4.4 the only time I could not get perfect smooth flight was just after takeoff and sometimes on approaches. Looking out the VC at 90 degrees at the moment in flight was not perfect.

So I added in a AM=85 and loading up 4.4 watching very carefully. I saw, Logical core 0-1-2-3 working as it should be, Thread 0-1-2-3 doing nothing. Took off and did a lap of my airport.

Logical core 0-2-3 were doing all the work Logical core 1 was only running at around 30%. So I upped my setting till I saw Logical core 0-2-3 peaking out, this is with FPS set to unlimited as I wanted P3D to work as hard as it could. Then I Locked FPS to my normal 26, Logical core 0-2-3 then topped out around 90%.

So I went and did a full flight with AS/ASCA/ProATC/RyanairV/UTL/Chaseplane.

Since I saw Logical core 1 was not being used much, I used task manager to place all the above addons on all the Threads =0-1-2-3 and Logical core 1.

Result was amazing. So smooth at the most demanding times. I thought 4.4 was running well, really well, but this has placed it on another level.

O well back to AM and management cores lol, it's worth it.

Edited January 16, 20197 yr by Nyxx

Hi David,

4790K (4 Cores, 8 Logical Processors HT=ON)

Affinity Mask 85 (Decimal) = 01,01,01,01 (Binary)

In sequence from right to left, Logical Processors (LPs): 01,23,45,67

Hope you don’t mind me jumping in here but, in the interests of greater clarity, I wonder if I could stick my neck out by asking you to go over your experience again for us while referring to the above. I think I followed your drift, but your references to cores and threads made it all a little confusing. Maybe you could simply modify your post rather than having to create another....assuming you aren’t timed out by the forum software.

Good to see you are getting worthwhile results from these endeavours.

Thanks.

Regards,

Mike

If you get it right it certainly works since just simple mathematics can show it:

If two LPs are being time sliced on the core - even though the HT mode is saving the work of switching them over each time-slice, they push each other out. With the main task of P3D we give it a core to itself and the sister LP want to show zero activity. We say the task is monolithic in nature.

Where we have several tasks that are parallel in nature, for example loading up new scenery, these can co-exist on a core, so we can allocate two LPs to those (in some circumstances) and that uses more cycles of that core.

By the way: It would be better to use the MS/Intel nomenclature:

1 HT core  = 2 Logical Processors (LPs).

Not using an AM, means that every LP is used for a P3D task. The second task would share the first core with the sacred main task which we hope to give maximum throughput. If the second task shows 50% when the first task shows 100%, the first task has 75% and the second task has 25% of the total throughput. If we move the second task onto the next core we free up 25% of that throughput to the main task.

If we allocate more and more LPs then all that happens is they start to wait for each others time at drawing on the same resource, in turn the main task then is waiting for those. The system starts to show a reverse in the performance - while still increasing in the loading speed the sim performs worse overall.

And also all the shared cache in the CPU is used up needlessly. So when we see small increments in the loading speed adding another LP, we probably don't need that other LP.

 

...as always, if we don't see changes we expect, there's a bottleneck or a limit reached in there somewhere and we won't get the results we desire no matter how much AMs and HT is turned on and off.

Keep addons off of that main task core, even a tiny app or panel.exe showing 4% will upset that main task.

When corralling addons make sure to give them two LPs at least, even if they must reside on the same  core.

You win, Steve!

Just tried AM 85 with my 8600K .....WOW! Long story short, the AM (85) totally out performs no AM and does it 7c to 10c cooler than with no AM, which is saying something because I was completely satisfied with my system before the AM test.

Thanks for sticking to your guns on the AM topic(s) .....I would have muzzled ya long ago if AVSIM had the feature. <g>

 

 

side note:

The only reason I even considered trying  AM is because I have an 8086 on the table waiting to be installed and wanted to test  ...and I didn't the 8086 buy it for Hyperthreading, i bought it for the IMC (integrated memory controller) as a plan B in case the new mems I bought don't work with my current I5.

For those tinkering with AM for P3Dv4.4: please dont miss to try this Affinity below, that was said produced a very good result on a few machines:

xx.xx.xx.xx.11.11.10.10 - don't miss it in your tests!!

Edited January 16, 20197 yr by Dirk98

@SteveW lots of people getting it now, your advice proved invaluable to me also, forever respect O guru and master of AM's :p

Or more simply, thanks mate!

Thanks again! The research that divulged a lot of that, was essentially done with a test system that is in the most part included with IFPro. I can do a few more things here that may have been a help back then but not completely necessary in the long run, not stuff I can easily set up in an app anyway. Cheers Mates!

Thank Buddha that all ended well.