Early look at Alder Lake performance in MSFS

[Waldo Pepper 1,953]

 

[bobcat999 8,484]

Is that a Cat 6 cable?

Edited November 8, 2021 by bobcat999

[psolk 3,040]

30 minutes ago, Seth2021 said:

I have great temps with 11900k (adaptive boost + multicore enhancement enabled) on MSFS. Average temp while playing 40-55c. Sometimes on loading when all cores are working you could reach 75c( max) for just 1sec.  So i guess its going to be the same with 12900k. 

 

Even at 55C for extended periods of time with my office/flightsim room in the back of the house and a lot of sun exposure it made this room 10-15 degrees hotter than anywhere else in the house.  Well that plus 8 monitors, 2 tv's, 2 desktops, 2 laptops and two video conferencing units LOL .

I actually invested in a really easy homemade venting system.  I got an inline duct fan connected to a portable A/C vent hose.  You can turn it either way so it can pull fresh air in or vent heat out.  When it gets really hot I use a $25 portable A/C window vent kit and point the duct hose to the front or back of the case if I want the airflow in/out.  

Cost about $150 all in, completely portable with humidity control as well.  I think it was made by AC infinity.  

Edited November 8, 2021 by psolk

[peloto 213]

Humm, in that case, I will leave the fridge doors open then day and night, and if I have too much cool in the room I turn on the boiler.

[MrFuzzy 2,118]

On 11/7/2021 at 2:59 AM, Republic3D said:

Here's a little look at how the 12900K performs vs the 10900K in MSFS at 1080p and over ocean (non-complex scenery). I'm not sure how useful the test is, because both CPUs are chilling at 25% and the GPUs around 85%. Still the 12900K pulls ahead by 10-20 fps.

I would love for someone to test over complex scenery.

https://youtu.be/83ipj6WXBRU?t=182

CPU isn't chilling, it's the main thread bottleneck. However 15 fps / 165 fps = less than 10% which reflects the difference in IPC in single thread operations. Nothing new or revolutionary.

[MySound 403]

First and foremost: Hello everyone.

I've been a reader for quite some time here, so now I want to join in :-)

@rob You should use the Ryzen DRAM Calculator if you need help with tuning your RAM. And I guaranbtee you your score will considerably go up in Timespy.

I run a similar setup: 5900x + 3080ti, and my Timespy score is 20100 overall, 21600 GPU, 14700 GPU. It took me quite some optimization to reach this.

[Maxis 1,865]

7 hours ago, Rob_Ainscough said:

I worked thru the PBO Curve Optimizer (All Cores) and PPT, TDC, EDC and made some improvements (SMT off for TimeSpy).

PPT = 220w

TDC = 147

EDC = 145  (tricky one with temps)

All core curve optimizer Magnitude = 30  (sign negative).

On the CPU side about a 10% increase.

Next I'll start in on memory frequency/timing and see if I can find anything there.  I updated the EVGA 3090 with latest firmware but performance went down a little (perhaps within margin of error).

CCD0 - Core 3 was flagged as best, then Core 1

CCD1 - Core 15 was flagged as best, then Core 13

Temps increased but well below thresholds at only 57😄

AMD is very different approach to OC than Intel.

Cheers, Rob.

Congrats and very nice. 

However .. all core -30 Curve optimized equals golden sample (Granted you will be able to keep most of them at -30). Here's how it normally works .. 

Under load (all benches and games) it will work perfectly. Then randomly when the pc is doing absolutely nothing at all the system will crash and reboot. This is usually because your best cores "usually" cannot run a -30 negative curve and your going to have to back those numbers down till those random crashes (WHEA) errors go away. 

Please watch out for this and scour the error log for the offending core and adjust down accordingly.

I did mine the manual way scouring the error logs but i believe there are a few new programs on the scene that will take most of the guesswork out because it is a rather difficult way of testing for instability (system crashing under very light load instead of very heavy load) OCCT core cycler i believe is one of them.

 

Take care and have fun.

Edited November 9, 2021 by Maxis

[MySound 403]

Another question to Rob:

why are you limiting EDC if you have so much temp headroom?

Ryzen is designed to hit these high temps (80-90c). There’s is a dedicated video by AMD on this. You waste CPU performance.

[Republic3D 1,232]

6 hours ago, MrFuzzy said:

CPU isn't chilling, it's the main thread bottleneck. However 15 fps / 165 fps = less than 10% which reflects the difference in IPC in single thread operations. Nothing new or revolutionary.

Are you sure? The CPU usage is around 20-30%. 


 

[Republic3D 1,232]

5 hours ago, MySound said:

First and foremost: Hello everyone.

I've been a reader for quite some time here, so now I want to join in 🙂

@rob You should use the Ryzen DRAM Calculator if you need help with tuning your RAM. And I guaranbtee you your score will considerably go up in Timespy.

I run a similar setup: 5900x + 3080ti, and my Timespy score is 20100 overall, 21600 GPU, 14700 GPU. It took me quite some optimization to reach this.

Welcome to the party! 🙂

[ckyliu 1,749]

On 11/4/2021 at 2:27 PM, ryanbatcund said:

1080?  Is this 2008 lol? Let's see 2k and 4k?

2K is basically 1080p, to be exact it's 2048x1080 px, the slightly wider cinematic version of 1920x1080. Do you perhaps mean 1440p (which would be 2.5K at 16:9)?

Sorry but it's a real bugbear of mine, the term 2K is misused so often it's meaningless.

Edited November 9, 2021 by ckyliu

[ryanbatc 12,857]

1 minute ago, ckyliu said:

2K is 1080p. It's 2048×1080, the cinematic version of 1920x1080. Do you perhaps mean 1440p? Sorry but it's a real bugbear of mine, the term 2K is misused so often it's meaningless.

Hehe yes technically - I just call it 2k though.... 1440P.  We all have our pet peves lol.

Edited November 9, 2021 by ryanbatcund

[MrFuzzy 2,118]

31 minutes ago, Republic3D said:

Are you sure? The CPU usage is around 20-30%. 


 

20-30% across all threads, but it's not balanced. If you open the task manager you can see that one thread is maxed out while the others are working at 10-15% capacity. The average is 20-30% but it's not like you have 70-80% headroom. You have zero headroom because one thread bottlenecks the CPU rendering, in fact the GPU is not maxed out in that "benchmark".

[Republic3D 1,232]

5 minutes ago, MrFuzzy said:

20-30% across all threads, but it's not balanced. If you open the task manager you can see that one thread is maxed out while the others are working at 10-15% capacity. The average is 20-30% but it's not like you have 70-80% headroom. You have zero headroom because one thread bottlenecks the CPU rendering, in fact the GPU is not maxed out in that "benchmark".

That's true. But personally I don't think I've been main thread limited outside of complex scenery. I'd have to experiment a bit with that. 

Edit: But I've never tried in a lower resolution than 3440x1440 so that might be why.

Either way, the 12th Gen chip seems to increase the FPS by around 10% or so. 
 

Edited November 9, 2021 by Republic3D

[MySound 403]

@Republic3D you see what limits you when you display the stats in dev mode.