Hyperthreading off works great!

[bean_sprout 1,201]

I'm taking the standpoint that all the old FSX tweaks are unnecessary, inappropriate and could even have negative effects in P3D 4.4.  Rest assured that the LM team would have included them if they felt there was any positive affect in the sim.  As it stands they took the right approach and reworked the simulator engines(s).

 

Cheers
bs

[newtie 537]

2 hours ago, SunDevil56 said:

If you could give me the make and model number of the MOBO I might be able to help you get in to the BIOS screen...

Download and run Speccy.

It will tell you everything you want to know about your computer.

Cheers,

 

[newtie 537]

3 minutes ago, bean_sprout said:

I'm taking the standpoint that all the old FSX tweaks are unnecessary, inappropriate and could even have negative effects in P3D 4.4.  Rest assured that the LM team would have included them if they felt there was any positive affect in the sim.  As it stands they took the right approach and reworked the simulator engines(s).

 

Cheers
bs

For most people most of the time, you are correct.

When you factor in the unlimited variables that exist in our obsession, the argument weakens. LM can't include variables that they don't even know exist. A good example is their recommendation not to use HT. Imagine what their support forum would look like if they did? 

"I get stutters when I run 010011010100, but when I run 0100110100101, on Tuesdays when the humidity is more than 50%, there are less stutters. Can you help me with this?" 

Happy New Year,

Mark

[SunDevil56 183]

13 minutes ago, newtie said:

For most people most of the time, you are correct.

When you factor in the unlimited variables that exist in our obsession, the argument weakens. LM can't include variables that they don't even know exist. A good example is their recommendation not to use HT. Imagine what their support forum would look like if they did? 

"I get stutters when I run 010011010100, but when I run 0100110100101, on Tuesdays when the humidity is more than 50%, there are less stutters. Can you help me with this?" 

Happy New Year,

Mark

OMG, that is so funny, but also very true...🤣

[newtie 537]

E-yup...😊

[Cruachan 705]

Hi everyone, happy New Year!

This topic never seems to die, which speaks volumes about its perennial potential value to the serious Prepar3D simmer who wishes to spend some time exploring any performance benefits as might apply to this simulator.

Regarding the Hyperthreading ON/OFF debate and/or the use of the Affinity Mask to optimise Physical Core/Logical Processor assignments for Prepar3D, believe me when I say I have tried every which way to determine the ideal setup. There is no easy answer and, for that reason, I am reluctant to make any recommendations as I believe you have to discover the best configuration that suits your particular system.

My rig is not dedicated to Prepar3D alone. It is primarily a gaming rig although there are many other non-gaming applications installed. 

Hyperthreading = ON is the default setting in the motherboard BIOS for HT capable Intel Processors and will be assumed by most software developers. Some may exploit it and some may not, so why deny the application that does by switching hyperthreading off? Prepar3D appears to be a special case in that some users report better performance with HT=ON and some with HT=OFF. 

How many of you running high performance gaming software alongside Prepar3D or X-Plane 11 are aware whether these non-Prepar3D apps are exploiting Hyperthreading capabilities offered by their CPU? 

So, to me it makes perfect sense to leave Hyperthreading=ON and adjust the Core/Logical Processor assignments for Prepar3D by using the Affinity Mask.

Determining the best Affinity Mask can be a bit hit or miss unless you decide on adopting a more logical testing approach that will bring reproducible and reliable results. To that end I would suggest you look at this thread:

https://www.avsim.com/forums/topic/541073-using-loading-times-to-determine-affinity-mask/

There is lot of good stuff embedded in these posts and, as ever, Steve’s contributions proved to be invaluable. This may not be the only way but, as I have found, this method worked for me in Prepar3D v4.3 and now that same Affinity Mask also appears to be delivering good performance in v4.4.

My testing and results related to the 8-Cored/16 LP i7-5960X, but the assumption is that the principles behind the procedure will remain the same for other multi-cored CPUs. My CPU is water-cooled and temps are never a problem.

Those of you feeling reluctant to meddle with the System BIOS, and wish to employ an Affinity Mask for Prepar3D, might find that this approach can provide a more acceptable solution.

Regards,

Mike

 

 

[SunDevil56 183]

9 minutes ago, Cruachan said:

 

......How many of you running high performance gaming software alongside Prepar3D or X-Plane 11 are aware whether these non-Prepar3D apps are exploiting Hyperthreading capabilities offered by their CPU? 

So, to me it makes perfect sense to leave Hyperthreading=ON and adjust the Core/Logical Processor assignments for Prepar3D by using the Affinity Mask.

Determining the best Affinity Mask can be a bit hit or miss unless you decide on adopting a more logical testing approach that will bring reproducible and reliable results. To that end I would suggest you look at this thread:

https://www.avsim.com/forums/topic/541073-using-loading-times-to-determine-affinity-mask/

There is lot of good stuff embedded in these posts and, as ever, Steve’s contributions proved to be invaluable. This may not be the only way but, as I have found, this method worked for me in Prepar3D v4.3 and now that same Affinity Mask also appears to be delivering good performance in v4.4..........

Those of you feeling reluctant to meddle with the System BIOS, and wish to employ an Affinity Mask for Prepar3D, might find that this approach can provide a more acceptable solution.

Regards,

Mike

 

 

Wow...  Great post Mike, I'm definitely going to be checking out your findings in the thread you posted...

It never fails to amaze me, the time, effort, and expertise that people like you, Steve, and many others put in to finding solutions to very complex questions like this HT "on or off" and "what AM to use" subject. 

In my personal experience, this is one area of of fine tuning P3D that has been more elusive than the proverbial Unicorn, and I just want to extend a thank you to all of you guys that put countless hours in to this... 😎

[bean_sprout 1,201]

2 hours ago, newtie said:

For most people most of the time, you are correct.

When you factor in the unlimited variables that exist in our obsession, the argument weakens. LM can't include variables that they don't even know exist. A good example is their recommendation not to use HT. Imagine what their support forum would look like if they did? 

"I get stutters when I run 010011010100, but when I run 0100110100101, on Tuesdays when the humidity is more than 50%, there are less stutters. Can you help me with this?" 

Happy New Year,

Mark

lol - I hear you clearly

bs

[fghdgdfdfgfgf 111]

I5 here,

Dedicated P3D machine powered by an 8600K.  I'll put $100 on it that it out performs 99% of the machines that were used to fill this thread and others like it. Not kidding!

 

 

[Cruachan 705]

 

9 hours ago, FunknNasty said:

I5 here,

Dedicated P3D machine powered by an 8600K.  I'll put $100 on it that it out performs 99% of the machines that were used to fill this thread and others like it. Not kidding!

Hi Ken,

Nice system spec and, as you say, in terms of raw performance it is likely that you will outgun the majority of rigs out there. Congratulations! I thought the same 3 years ago. However, statements like that without any Application-specific qualifications are fraught with difficulties and open to being challenged.

It all boils down to how we define ‘performance’ and comparing apples with apples. If our Prepar3D sim setups and flight testing scenarios are identical, and performance is perceived in each case as being smooth at 30fps with some headroom to spare, then it could be argued that we are seeing the same performance. Under these circumstances hardware specs become largely irrelevant.

On the other hand, running complex gaming software and seeing raw performance in terms of achievable frame rates at 4K then, yes, there will be some entitlement to some bragging rights. BUT, does it really matter as much these days as it did in the past since most recent hardware can produce what is needed in terms of fluidity and frame rates to satisfy the majority of end users?

When I built my latest rig I did so in the knowledge that component selection should serve me well for the foreseeable future. This is proving to be true. For many years CPU single threaded performance is the one factor that has dogged sims like FSX and Prepar3D. My prediction is that where Prepar3D is concerned this is all about to change and less reliance will be placed on Core 0 of the CPU to deliver the goods. This constraint continues to be largely irrelevant (or certainly seems that way) where modern gaming software is concerned as I seem to be able to run smoothly at high frames rates using ultra settings with my GTX 1080Ti outputting to my sub 4K 27” monitor.

I have installed Windows 10 Pro on its own 2TB SSD under a dual Boot arrangement and anticipate that Prepar3D V5 will install and run well without any pressing need for further hardware upgrades. As ever, time will tell.

Regards,

Mike

[Cruachan 705]

Hi Ken,

You have piqued my interest and curiosity...LOL!!

So I did a little research:

http://www.cpu-world.com/Compare/108/Intel_Core_i5_i5-8600K_vs_Intel_Core_i7_Extreme_Edition_i7-5960X.html

I think I can say that my then ridiculously overpriced 5960X (not sure if they are still available) is still holding its own in today’s competitive environment, even moreso when overclocked. When under load all 8 cores/16 LPs are currently dynamically overclocked stable to 4.6GHz as compared to your impressive 5.3GHz. So, your single thread benchmarks will be higher:

https://m.cpubenchmark.net/

Benchmarks (PassMark - CPU Mark)

8600K (12,775/3.6 x 5.3 = 18,807

5960X (15,948/3.0 x 4.6 = 24,453

However, the story is somewhat different for Single Thread Performance, which, as we know, is important for Prepar3D as applied to Core 0:

8600K (2,517/3.6) x 5.3 = 3705

5960X (1,996/3.0) x 4.6 = 3060

I have gone to 4.7Ghz stable for a couple of years but ‘only’ on 4 cores/8 LPs, remaining cores at 4.6Ghz. I note that the 8600K does not support hyperthreading, but it has 6 Cores which, as I’ve demonstrated, seems to be the optimal count for Prepar3D.

Regards,

Mike

[Bert Pieke 7,585]

Which one more time supports the notion that for those of us who are still happily flying with quad Haswell processors, like my 4770K@4.3, an affinity mask that provides 6 cores for P3D to run on, is indeed a good choice 

[fghdgdfdfgfgf 111]

1 hour ago, Cruachan said:

Hi Ken,

You have piqued my interest and curiosity...LOL!!

So I did a little research:

http://www.cpu-world.com/Compare/108/Intel_Core_i5_i5-8600K_vs_Intel_Core_i7_Extreme_Edition_i7-5960X.html

I think I can say that my then ridiculously overpriced 5960X (not sure if they are still available) is still holding its own in today’s competitive environment, even moreso when overclocked. When under load all 8 cores/16 LPs are currently dynamically overclocked stable to 4.6GHz as compared to your impressive 5.3GHz. So, your single thread benchmarks will be higher:

https://m.cpubenchmark.net/

Benchmarks (PassMark - CPU Mark)

8600K (12,775/3.6 x 5.3 = 18,807

5960X (15,948/3.0 x 4.6 = 24,453

However, the story is somewhat different for Single Thread Performance, which, as we know, is important for Prepar3D as applied to Core 0:

8600K (2,517/3.6) x 5.3 = 3705

5960X (1,996/3.0) x 4.6 = 3060

I have gone to 4.7Ghz stable for a couple of years but ‘only’ on 4 cores/8 LPs, remaining cores at 4.6Ghz. I note that the 8600K does not support hyperthreading, but it has 6 Cores which, as I’ve demonstrated, seems to be the optimal count for Prepar3D.

Regards,

Mike

Let me take you inside my world .... orbix so cal - with my hardware:

cpu 52/48 with 4266 memory at cl17.   

[Mace 1,935]

I'd rather have 8 physical cores than six though.  :)

9 minutes ago, FunknNasty said:

Let me take you inside my world .... orbix so cal - with my hardware:

cpu 52/48 with 4266 memory at cl17.   

Ah, the ROG Maximus X Apex.   There's the difference.   My ROG Maximus X Hero (a notch down from the Apex) and GSkill mem isn't liking 5.3.   But then, I haven't spent much time tweaking voltages, which might do it.

Good to see you doing that with an H100i v2.

Do you have the SoCal pauses thing people talk about here?  I don't know Orbx SoCal, but I'm curious since you're clocking pretty well.

[Rocky_53 1,583]

56 minutes ago, Mace said:

I'd rather have 8 physical cores than six though.  :)

Ah, the ROG Maximus X Apex.   There's the difference.   My ROG Maximus X Hero (a notch down from the Apex) and GSkill mem isn't liking 5.3.   But then, I haven't spent much time tweaking voltages, which might do it.

Good to see you doing that with an H100i v2.

Do you have the SoCal pauses thing people talk about here?  I don't know Orbx SoCal, but I'm curious since you're clocking pretty well.

I've just uninstalled Orbx Socal to leave me with NA LC and Global Vector... surprise, surprise, the frame rates are exactly the same.... abysmal about 22 at default KLAX...go figure!

Edited January 2, 2019 by Rockliffe