What does “non-VCache” mean at the chip level?
Does it alter the amount of L3 cache?
What does “non-VCache” mean at the chip level?
Does it alter the amount of L3 cache?
For what it’s worth I benchmarked a 9800x3d vs i9 14900 in VR (both main thread limited, GPU 95 - 98%) and there was no performance difference in MSFS 2024, even the lows were on par
Yep.
The 7950X3D has 2 cpu chips (CCD). Both with 32MB L3 cache, but one has an extra 64MB cache on top to make up 96MB
Spoiler, the “3D V-Cache” memory is identical to the standard 32MB cache. But the 3D-V is mounted directly above one CCD - the shorter vertical connections are faster than horizontal connections to an adjacent cache. 3D-V is simply a marketing term, same memory type just assembled a better way by TSMC.
More details here:
Scientifically we need to compare X3D cpus to standard cpus with 64MB and 128MB cache.
I appreciate seeing this discussion here as I’ve had multiple CTDs in FS2024 with Hyper-Threading (HT) turned On on my Intel i9-14900K. I’ve had zero CTDs with HT turned Off. Ultimately, however, I was able to get FS2024 to run more smoothly with fewer CTDs with HT On and the Windows Power Plan set to “High Performance.”
Process Lasso sounds like an intriguing thread optimizer, but unlike AMD chips, Intel chips already have thread-scheduling built into Win11. I’ve never had an issue with the way Intel’s thread scheduler optimizes the cores and therefore see no need to add yet another background process to the mix, especially one that may conflict with Intel’s thread scheduler.
At the end of the day, both AMD and Intel make very solid chips for gaming. In my experience, however, both FS2020 and FS2024 are GPU bound, not CPU bound, especially when you dial up the resolution and the eye-candy. So your choice of CPU for this game is almost entirely moot.
I’ve spent a ridiculous amount of time tweaking the performance of my unlocked CPU. In CPU benchmarks, I’ve managed to gain about 15% in overall CPU performance. In graphics intensive games, however, the performance increase has been at most, 0-1 FPS.
While it’s certainly entertaining and fun to spend a lot of time tweaking system performance, any time spent tweaking CPU performance will have minimal, if any, benefit to graphics instensive games like FS2020 and FS2024.
I think that’s a fair point, when comparing AMD to AMD. A 9950X has 1.28MB L1, 16MB L2, and 64MB L3 (total 81.28MB.)
AFAIK they don’t make a ‘non-X3D’ with more than that.
I watched a video this morning that talked about Process Lasso being pointless (and sometimes detrimental) with Intel chips in most games for exactly that reason.
Also true (mostly…)
You can definitely stress the CPU with certain graphics settings.
With both HT and the E-cores enabled your Intel i9-14900K processor will have 32 threads.
This means there is a high probability that the CTD you’re experiencing is caused by a known bug in MSFS 2024.
edit
FYI, when I was using an i9 13900K, I disabled the E-cores and noticed better performance. I recommend that you try disabling the E-cores on your BIOS and test if it improves stability and performance.
@TenPatrol, you and I have already had an exchange on this subject on a separate thread that I posted here. As you know from that discussion, I have already speculated that my issue with HT On may well be related to the known bug posted by the FS2024 team. While having HT Off resolves the issue completely, I have also found relief with HT On while using the Windows “High Performance” Power Plan. Not zero CTDs, mind you, like I have with HT Off, but a significant reduction nonetheless.
That quite clearly says that 32 logical processors is expected to work, and that having more than 32 logical processors is a problem.
Here’s a corrected version
If you are using a CPU with 32 or more cores, the sim will crash. We are aware of this issue, and it is being investigated.
That’s not what the linked FAQ says. It quite clearly indicates that the recommended workaround is to set the maximum to 32.
So how did you set up Process Lasso with regards to this video posted by BegottenPoet228
Because I feel like he mixed up describing the vcache cores vs non vcache cores. Did you bind all other processes to 0 - 7 or 8 - 15 ?
The sim on 0-7, everything else on 8-15.
The biggest surprise takeaway for me was his suggestion to set most Windows OS processes to ‘Idle’ priority. It works. My computer is responsive and stable.
Note that I’ve also disabled Windows Game Mode and Xbox Game bar, and use Ultimate Power Plan. I also use Bitsum’s Core Parking app to disable core parking.
Regardless, my experience has been that by reducing the number of logical cores to fewer than 32 by turning HT Off in the BIOS, the problem is solved. So no matter what the precise language says, there does indeed seem to be a CTD issue when 32 logical processors are in play. I’m a lawyer too and can parse language like this with the best of them. I’m willing to bet the author of the FAQ answer is not.
In that case, I recommend contacting Asobo to get the FAQ fixed – because if what you say is true the recommended workaround of setting it to 32 CANNOT WORK. Can mods help?
Were you able to quantify an actual performance improvement for the sim when doing this though?
And how does one select the processes to set to idle? Is there a list?
I can only say that everything I’ve done has - in total - resulted in a very stable, responsive, stutter-free sim experience; one that is a substantial improvement over the disappointing performance increase when I first replaced the 5800X3D with the 7950X3D. I think I’ve posted a test I did where I compared a baseline with how things are set up now. There was a significant improvement seen. How does the ‘Idle’ setting fit into the delta? Don’t know…
There are too many variables (some of which remained constant through the CPU swap) to do true multivariate statistical analysis - nor do I have the desire to attempt such a project.
No list that I’m aware of. You need to watch the video if you want to do your own testing. For me it’s more about synergy, and results.
The testing Savitarax did in the video satisfied me. Did he test MSFS 2024, flying a particular aircraft in a particular location, with the exact same system I have, set up exactly the same otherwise? No.
One thing I’ve noted with my 9800x3d… is the game can perform with significant different fluidity from one startup to the next, all things been equal my end 500mbps up / down with 8 - 12 ms ping I can have a very consistent flight across London, re-start and it’s an absolute stutter fest! Re start again and back to stable frames. I still of the opinion that this game engine / cache or servers are not operating correctly
I would say that’s just FS for you, it’s always been that way to some extent. With the 9800x3D I’m actually getting the smoothest and most consistent performance (in high res VR) that I’ve ever had. It’s absolutely amazing!
Before I wasn’t quite sure whether it’s really worth the upgrade from a 7800x3D, but it definitely was! Especially since it only cost me about 100 euro after selling my old chip. Crazy how much those are still going for.