12 hours ago, tooting said:

dan when you run 4k what Texture filtering setting do you use ? 

I have always used the maximum of Anisotopic x16, believing that this variable has very little impact on performance.

Hyperthreading OFF often makes a significant improvement in reducing the CPU bottleneck. I have an i7-8700k and I went from 25FPS  with the MilViz Beaver to >60FPS just by turning off hyperthreading.

7 hours ago, downscc said:

I have always used the maximum of Anisotopic x16, believing that this variable has very little impact on performance.

thanks alot. 

11 hours ago, giantkillerrobot said:

Hyperthreading OFF often makes a significant improvement in reducing the CPU bottleneck. I have an i7-8700k and I went from 25FPS  with the MilViz Beaver to >60FPS just by turning off hyperthreading.

Sorry, but that is simply not possible. Why people continue to post such nonsense? I don't get it...

IF your rig performance improves that much (over 240%!!) simply by turning off hyperthreading, either your whole system is set up somehow strange or you have a totally misconfigured sim.

simple: giantkillerrobot has a giantkillerpc with HT OFF.😁

High temps can throttle the CPU with HT ON, especially when overclocked.

The only thing that I could watch with my CPU (4790K) with HT ON in fact was increased temperatures with no real benefit when it comes to fps or smoothness. I also leave it off so my overclock can go higher. But each CPU / setup is different.

Also, 2x16GB DIMMs is a suboptimal memory config--X299 is a quad-channel architecture, so 4x8GB would use all four memory channels, where your 2x16GB configuration uses only two.  And fast memory would be worth having--2666MHz at CAS16 ain't zackly tearing it up.

The hybrid HDD looks a little anemic for a performance system build.  SSDs are pretty inexpensive these days...you don't need a high-end one to really speed up your load times from that hybrid drive.

Good luck

 

14 hours ago, AnkH said:

Sorry, but that is simply not possible. Why people continue to post such nonsense? I don't get it...

IF your rig performance improves that much (over 240%!!) simply by turning off hyperthreading, either your whole system is set up somehow strange or you have a totally misconfigured sim.

Well, it gets posted from an understanding of how HT works...

Prepar3d is often bottlenecked by CPU - we need single-process, single-thread, single-core performance to be maxed out. HT does the opposite of this, attempting to run several threads/processes on a single core, allowing it to multi-task - for example, if one process is waiting on memory access, HT allows another "thread" to ustilize the core to do some work. This is exactly what we don't want with Prepar3d.

My rig is water cooled, GTX1080, i7-8700, with 16gb dual channel DDR4 - HT OFF made a quite significant improvement in performance. 

The 8700 CPU has six cores which is plenty for P3D.

Hence the recommendation to turn off HT for an 8700.

On a quad core CPU, HT actually helps, but in that case, it makes sense to block off the main P3D core, using an affinity mask for P3D, like

244=11 11 01 00

That allows P3D to use 5 (virtual) cores.

Edited July 19, 20187 yr by Bert Pieke

37 minutes ago, giantkillerrobot said:

Well, it gets posted from an understanding of how HT works...

Prepar3d is often bottlenecked by CPU - we need single-process, single-thread, single-core performance to be maxed out. HT does the opposite of this, attempting to run several threads/processes on a single core, allowing it to multi-task - for example, if one process is waiting on memory access, HT allows another "thread" to ustilize the core to do some work. This is exactly what we don't want with Prepar3d.

My rig is water cooled, GTX1080, i7-8700, with 16gb dual channel DDR4 - HT OFF made a quite significant improvement in performance. 

Proof please.  Scientific proof.

11 minutes ago, lownslo said:

Proof please.  Scientific proof.

You have multiple folks who run an 8700K (myself included) giving you "boots on the ground data." What else do you need to see? A benchmark in an article.

17 minutes ago, ZLA Steve said:

"boots on the ground data."

That's too funny!  "Boots on the ground data"... for a simulation!  And from flight simmers! 😀

Actually, he posted some specific claims about HT... just would like to see some proof of those claims.

Each system is different and there will be a different opinion for each system AND - for THAT system - it is correct. There are very few statements about P3D/systems that can be cast in stone.

I run a 7700k at 5hgz with HT off and no AM - and for ME on MY system - it is much better, cooler, more efficient and flat out runs better than with HT on and any one of the myriad of possible AM's. I have recently started using Project lasso to shift a few processes off core 0 and that all seems to work as I like it. I have also tried running at 30hz with vsync on and tribuffer on and switched back to my 60hz setup. It runs better and I never had any issues with tearing etc without Vsync, so, for me, 30hz is not the way to go.

Again, just pointing out that there is usually more than one way to accomplish your tasks - nothing is set in stone so, please have respect for someone else's opinion, especially if it differs from yours.

Vic

@paulh1

 

Your system has potential to be a good flight sim PC, though it could use some tuning, and a different choice of RAM.

First thing's first, the 7900x does have a base clock of 3.3GHz but this is the guaranteed minimum clock.  I have a 7900x that I run far more stressful workloads on than P3D and I've never seen the all-core clock speed dip below 4GHz, so things aren't nearly as bad as they may seem at first glance.  Additionally, the 7900x has a Turbo 3.0 maximum single core clockspeed of 4.5GHz for lightly-threaded workloads.  Typical workloads will produce clockspeeds between 4.2-4.3GHz, depending on the number of cores being utilized beyond a certain threshold and the presence of AVX instructions.  This is not bad, but not great for a program like P3D which is still mostly affected by the performance of a single core.  You can and should push for more clockspeed, if you find that your FPS is lacking.  Typical 7900x overclocks range from 4.4-4.6GHz on all cores, and can often see single core clocks pushed to 5GHz or so.  

As for the RAM, the amount is fine, though the number of DIMMs and clock speed is less than optimal for your platform.  X299 is a quad-channel platform so you really want 4 sticks to feed those cores the bandwidth they want.  Furthermore, RAM clock speed can have an effect on performance and is worth pursuing at least to a certain point.  You should target RAM speeds above 3000MHz to get the best performance out of your system.

I hope this helps.

Edited July 20, 20187 yr by TechguyMaxC

7 hours ago, vgbaron said:

Again, just pointing out that there is usually more than one way to accomplish your tasks - nothing is set in stone so, please have respect for someone else's opinion, especially if it differs from yours.

Sorry to disagree, but "opinion" is not something that is valid for such discussions. As you might know better than me, being a moderator, there are hundreds of threads (including this one) from people complaining about the performance and desperately looking for a "solution". I am totally aware that the ESP-based engine is somehow strange regarding same results on different hardware, but statements such as this we discuss about are simply of no use for anyone. Especially not if it is not underscored by hard facts (a video for example). Exactly due to the fact you mentioned: each system reacts differently. Of course the topic starter could now try HT off, but as he has completely different hardware and the sim reacts different even on identical hardware, what is the benefit of posting things like "turn HT off and you get 200% more FPS"? Besides driving those guys nuts that do not achieve such a performance increase by turning HT off...

Edited July 20, 20187 yr by AnkH

Sorry but an opinion is just that. Doesn't have to be "correct" just that one person's opinion. To some, the mere fact that someone else perceives a 200% increase by doing x is important.  These are OPEN discussions not just for the technically elite. It is up to the user to take what's offered and apply it to their own system - as the saying goes - "if the shoe fits, wear it".  The idea is for the statement to cause the reader to look at his system possibly differently.

Sorry Chris but these are open discussions and, as such, are going to get some very valid information and some wild speculations. Telling someone their opinion does not matter is a tad snobbish IMHO.

Vic