Currently you can better use D4 ram than D5 ram for MSFS. The current D5 ram is not that fast in many games . Perhaps the versions of D5 ram that will be released in 1.5-2 years will be faster. For D5 ram I want 6000 MHz cas 32 or lower.

Regarding D4 memory : 3200 MHz vs 4000 MHz. See the difference in fps.

 

 

Edited December 12, 20214 yr by GSalden

@Waldo Pepper

Many thanks for your extensive information. You are a great help.

I will post my results in this thread.

Btw : work in progress :

 

With only Windows from startup to desktop : 9 seconds…. I am sure that within a week it will take longer 🤓

Edited December 12, 20214 yr by GSalden

A really interesting thread for of useful information.  I have bookmarked the thread as am looking at going 12900k in the new year, probably with fast DDR4 Ram at the moment.

Just a note, do you thing using process lasso to set MSFS up on performance cores only is worth a try?  Windows and other background tasks could just use the economy cores then so they don't interfere.
Apparently, the economy cores are still reasonably fast anyway for that kind of thing.

9 minutes ago, bobcat999 said:

A really interesting thread for of useful information.  I have bookmarked the thread as am looking at going 12900k in the new year, probably with fast DDR4 Ram at the moment.

Just a note, do you thing using process lasso to set MSFS up on performance cores only is worth a try?  Windows and other background tasks could just use the economy cores then so they don't interfere.
Apparently, the economy cores are still reasonably fast anyway for that kind of thing.

Since June I stopped using PL as it caused MSFS to CTD when loading. But for that I found a workaround : set PL to not start with Windows. Start it up after MSFS has loaded into the scenery.

As I use Prosim I might start using PL again to put it on the slower E-cores. PSXT + RT , FSUIPC and WidevieW have to run on the E-cores as well. 

Probably the main thread and render thread will automatically use one of the P-cores but I will look into that to be sure everything from MSFS is running on the P-cores…

this thread belongs to Mobos, CPUs, RAM etc, where other interested parties can see it and reply.

Edited December 12, 20214 yr by turbomax

29 minutes ago, turbomax said:

this thread belongs to Mobos, CPUs, RAM etc, where other interested parties can see it and reply.

I thought that at first Turbo, but it can also be seen as performance related specifically to MSFS as well, as Gerard has posted the screen shots with the FPS etc.  An interesting thread for people looking to upgrade to gain performance in MSFS none the less.

8 hours ago, GSalden said:

Currently you can better use D4 ram than D5 ram for MSFS. The current D5 ram is not that fast in many games . Perhaps the versions of D5 ram that will be released in 1.5-2 years will be faster. For D5 ram I want 6000 MHz cas 32 or lower.

Regarding D4 memory : 3200 MHz vs 4000 MHz. See the difference in fps.

 

 

I'm confused.   Screenies say 5900x,  not 12900k.   Edit,  NVM.    That's a substantial difference.   Seems like the sim loves bandwidth.

Edited December 12, 20214 yr by Waldo Pepper

1 hour ago, Waldo Pepper said:

I'm confused.   Screenies say 5900x,  not 12900k.   Edit,  NVM.    That's a substantial difference.   Seems like the sim loves bandwidth.

This was about the difference in performance using 3200 mhz cas 16 vs 4000 mhz cas 16.

4 hours ago, GSalden said:

This was about the difference in performance using 3200 mhz cas 16 vs 4000 mhz cas 16.

I have just finished building my new system (see my signature). I've been skeptical about DDR5 RAM too because of high prizes and lower performance (ca 15 -16 ns vs 8-10 ns). My current DDR5 RAM is 5200 with 38 CL which is one of the best I could find. I'm not sure if we can really translate the significant different clock rates into fps. It is a complete new system PSU-CPU-Mainboard-RAM (I only kept my 3090 VC) so I will see during the next few days how good MSFS performs.

Edited December 12, 20214 yr by Nemo

4 hours ago, GSalden said:

This was about the difference in performance using 3200 mhz cas 16 vs 4000 mhz cas 16.

I just see, that you went for a 12900K with a DDR4 MB. Would be interesting to compare our performance differences if this is possible at all. Maybe with a standardized flight under defined conditions and settings. What do you think?

1 minute ago, Nemo said:

I have just finished building my new system (see my signature). I've been skeptical about DDR5 RAM too because of high prizes and lower performance (ca 15 -15 ns vs 8-10 ns). My current DDR5 RAM is 5200 with 38 CL which is one of the best I could find. I'm not sure if we can really translate the significant different clock rates into fps. It is a complete new system PSU-CPU-Mainboard-RAM (I only kept my 3090 VC) so I will see during the next few days how good MSFS performs.

I read somewhere that D5 memory is about 7% faster than D4 memory in calculation benchmarks (more bandwidth) and in games the difference is 2-3%. At this moment but that will change in the coming years

2000 * cas / mhz

2000 * 38 = 76000 / 5200 = 14,6 ns using D4 memory and  + - 13.6 ns using D5 memory..

 

 

 

6 minutes ago, Nemo said:

I just see, that you went for a 12900K with a DDR4 MB. Would be interesting to compare our performance differences if this is possible at all. Maybe with a standardized flight under defined conditions and settings. What do you think?

My server has a 2x 4K front view connected to the 3090, so I guess that would be comparing apples with oranges.

You have good hardware so performance won’t be an issue 👍

 

6 minutes ago, GSalden said:

My server has a 2x 4K front view connected to the 3090, so I guess that would be comparing apples with oranges.

You have good hardware so performance won’t be an issue 👍

 

Yes that's real difficult to compare ... I see you have already overclocked your 12900K, right? How are the temperatures in MSFS? I just made a CPU stress test with Realbench and my CPU became very hot (85-90 °C with some peaks around 95 °C!) and that w/o overclocking.

Edited December 12, 20214 yr by Nemo

1 hour ago, Nemo said:

Yes that's real difficult to compare ... I see you have already overclocked your 12900K, right? How are the temperatures in MSFS? I just made a CPU stress test with Realbench and my CPU became very hot (85-90 °C with some peaks around 95 °C!) and that w/o overclocking.

Ouch!  Isn't something very wrong with that, like a poor IHS to cooler interface?  I have a [email protected] on all 8 cores and temps rarely go over 52C in MSFS.

5 minutes ago, Noel said:

Ouch!  Isn't something very wrong with that, like a poor IHS to cooler interface?  I have a [email protected] on all 8 cores and temps rarely go over 52C in MSFS.

Temperatures like those what I have, have been reported by others too with 12900K CPU stress tests. That's why I was wondering how Gerard did overclock and what his temperatures were. My fans don't rise too much in MSFS, so I guess my temperatures in MSFS are still on a reasonable level. I have not yet installed my diagnostic tools so I cannot say exactly for now.