some1

Solid investigation about the recent issues.

The choice of the chipset does not affect cpu performance, it only increases the number of peripherials you can connect and bumps pcie for some slots to gen. 5, which isn't useful nowadays anyway. It's actually the same chipset used on b650 and x670e, the later simply has two of them.

Not really. His reviews may be good, if you can understand the broken English. But when it comes to things like this, he's just another guy wildly speculating on the internet, without much to back it up. All his theories about why nvidia gpu connectors melted were proven wrong. The previous generation he stirred a drama about capacitors at the back causing crashes, that also turned out to be a nothing burger. Even now, some damage photos on the cpus does not quite match his theories.

Now it's broken in CE too. Can we get a simple Throttle OFF and IDLE shortcuts like is the standard for other DCS modules? The current implementation is not usable with any throttle I'm familiar with (Warthog, Virpil, Winwing).

People there cluttered the thread with a lot of talk about system inacurracy and errors, but that's not the root of the problem. Heatblur's gunsight is consistently inaccurate in a turn fight, 100% of the time it points to the wrong place in the sky, always behind. Yeah, if the target is big enough or doesn't pull g's, then you may still hit it despite aiming short. The diamond (BATR cue) is more accurate than the gun cross, but it only shows once you start shooting.

Yep, and the issue is still there.

If you run out of VRAM, it will increase CPU frametimes significantly because it needs to juggle the textures between RAM and VRAM almost constantly. And I can't believe you're not running out of VRAM in DCS with 6GB GPU. The game can easily exceed 16GB usage.

Using this benchmark in 4k is pointless, it will only show GPU limits. In VR CPU frametimes went down by about 15%.

All tests were done in 1080p. If you see such a huge difference from lowering the resolution, then most likely you are GPU limited in this test. I'm on 4090.

Gigabyte X670E Aorus Elite, on AM4 I had X570 Aorus Elite.

My 7800X3D arrived earlier than expected, so I've been able to install all the components yesterday and ran some DCS benchmarks myself. Compared to my previous 5800X3D with 4x16GB 3600 CL16 RAM, the new setup has 7800X3D with 2x32GB 6000MHz CL36 DDR5 RAM. Which I set to CL32 optimized secondary timings using Buildzoid's settings from the video posted above. Overall, I see around 20% more FPS in DCS on 7800X3D vs 5800X3D, tweaking the memory timings and -20 Curve Optimizer adds another 2% on the Plazma test track. Most of that 2% is thanks to the timings, CO-20 is within the margin of error, even though the CPU boosts noticeably higher. Likewise, CPU frametimes in VR are around 15% better, but that's harder to measure reliably. Here's my old 5800X3D on Plasma test track, 1080p, no mirrors, clean DCS on High preset. Blue line is GPU usage: Here's 7800X3D, same setup, all settings stock, with memory on EXPO profile (6000MHz CL36): And here's after I added -20CO and tweaked memory timings. 20% increase in CPU performance is a substantial difference between CPU generations, but not really worth the upgrade from 5800X3D unless you have the money to burn. Also for those using older AM4 CPUs, 5800X3D is still a very good value upgrade compared to the cost of the whole new AM5/Intel platform.

Now it would be interesting to see how much that translates into actual DCS performance.

Last time I checked, I got 5800X3D to 192 avg. FPS in Plazma test on ST build and 228 avg. FPS on MT build. Clean DCS with mirrors off. I have 7800X3D on order, but not yet delivered, it will be a week or two.

People do these tests with different DCS versions, different settings and fps limit. So the numbers are not easily comparable. A 7kX3D is roughly 20% faster than 5800X3D, there's no way it could produce 55% more FPS in the same conditions.

Just check your GPU usage or, if you play in VR, CPU frametimes. If GPU is at 100% all the time, or your CPU frametimes are lower than GPU frametimes, then no point in upgrading the CPU. Overall, if you don't have a lot of disposable income, then 5800X3D is much more sensible option with your current hardware.

The model is still not fixed, 2 years and counting. Maybe ask ED for help, they've made the correct model of this radio for MiG-15.

It is still not fixed.

Such arrangements would be under strict NDA, not for the guy to just blow it up on his 10k subs Youtube channel before the actual press outlets get their embargo lifted.

Anything that pops on YT before the actual embargo lift is most likely faked. Especially if they don't even show the hardware on screen, just the results.

Workloads normally don't switch cores by themselves, scheduler does that. That's because it is in fact beneficial for the modern CPU to spread the load (heat) between cores rather than lock them to a single one and suffer from thermal throttling. Plus even the threads that seemingly need 100% CPU, usually are in a wait state multiple times a second and can yield the core for other tasks. Locking the game to one CCD is fairly trivial, as the scheduler knows which process owns each thread and can simply decide that it won't move these threads to the second CCD. At least as long as the game doesn't explicitly set the affinity mask, or create a dozen of threads with heavy computations inside that all want to run at the same time. From the reviews I've seen online, there are a few odd cases but 7700X is behind 7900X most of the time, if only marginally. Overall, it's not worth the extra $$$ just for gaming, but if the prices of 7800x3d will be high, and availability low, then 7900x3d starts to look tempting. Anyway, 7800X3D reviews should be up today, in about 2-3 hours.

The cores on the second CCD do not have to be parked, as long as DCS is not running on them. Other background tasks should be free to use them if needed. That's the job of the scheduler, not to shut down CCD completely. The only real test of the scheduler with 7950X3D and 7900X3D is to run the same DCS benchmark with and without second CCD enabled in bios, and compare the results. Like the guys at TPU did when simulating 7800X3D with 7950X3D. https://www.techpowerup.com/review/ryzen-7800x3d-performance-preview/

Personally I will try to get 7800x3d, but if it won't be available or it will be sold at a price close to 7900x3d, then I'd just buy 7900x3d. In most cases 7900x3d is still going to be the faster cpu, at least as long as Windows scheduler is doing its job.

449$, so 150$ cheaper than 7900X3D. Of course no guarantee that you'll be able to buy it at that price, demand may be higher than stock.

I play relatively simple missions that I created for myself at 90 FPS. However, if I wanted to play a campaign or multiplayer, 90 FPS is hard to achieve and CPU is sometimes the limiting factor, even with MT.

Ok, thanks for the heads up