5800x3D still better than new 7000 series

Moved to #self-service:pc-hardware category to match the topic.

Tom’s Hardware MSFS results at 1080p. The 5800X3D still kills them all.

To quote Tom’s Hardware:

"The Ryzen 7 5800X3D is simply outstanding in this title, highlighting that the chip can provide huge speedups in some titles. However, that can skew overall rankings. This is why the chip is, on average, the fastest gaming chip in the world, but be aware that can change drastically based on the game you’re playing.

The stock 7950X and 7600X lead the 12900K and 12600K by ~14%. The Intel 12900K and 12600K profit heavily from overclocking, gaining 19% and 16%, respectively, while the Ryzen 7000 chips gain roughly 5% from tuning. Despite Alder’s big overclocking gains, the tuned Ryzen chips land within a few percent of their price-comparable competitors."


Is there a reason they only released a 5800X3D and not a 5900X3D or 5950X3D? It seems a weird decision, the 5800X3D is outstanding in some gaming titles but pretty terrible at productivity.

With the 7000 chips I’d much prefer to buy a high end X3D chip than a cheaper model.

1 Like

It was a test bed for the technology and as such aimed squarely at gaming performance, despite having a slightly lower clock speed than a regular 5800X and no overclock ability, due to concerns over heat dissipation. How that pans out with the much hotter and power hungry 7000 chips remains to be seen, but since the 7700X is their current 8 core/16 thread model, I’d guess that the first 7000 3D cache model will be a 7800X3D, again with 8 cores/16 threads, and possibly with a lower clock than the 7700X. Not sure if the higher core models will then (or can) follow suit.

What’s also interesting is that the next gen Radeon GPUs will, according to rumours, eventually include a 3D cache version.

1 Like

It will not be included for the first time in the next generation, it is already included.
The Infinity Cache in Radeon GPUs that are already shipping is almost identical inside to the VCache in EPYC Miran-X and 5800X3D, and will be loaded in the future as well.

1 Like

I am not disputing this benchmark. (And of course I have no objection to yours either.)

However, when we MSFS players look at the average FPS numbers, we may think that they are a little too high.
The data is probably being taken under light load conditions at high altitude, as the FPS is well above 120 FPS.
More and more MSFS, which is a heavy title, is also being put on the AMD 7000 series CPUs currently being benchmarked by the media.
However, the average is over 60 FPS in the majority of conditions. That’s a bit different than what we routinely suffer from on a
I think this is a little different from the fairly heavy conditions.

The 5800X3D in particular has a significant FPS increase when not hampered by other conditions at high altitude, which pushes the average FPS number up.
This is an area that is already too fast to be relevant to feel, and looking too much at the numbers alone may lead to misjudgment.
(I actually use a 5800X3D and know that in a typical scenario there is a +40% FPS improvement over the same generation CPU)

I mention it in the following thread. If I may.

FPS difference verification between AMD 5950X and 5800X3D at Haneda Airport (RJTT) 34L straight out flight - #30 by KanadeNyan

1 Like

I think the most realistic reason for the absence of the 59 series is that AMD could not imagine an enterprise scenario such as a multi-CCD+VCache.

The 5800X3D uses the same chip as the EPYC Milan-X series, which is a server CPU, in its VCache chip.
The 7000X3D is most likely the same VCache chip as the EPYC Genoa-X series, which is a server.

Depending on the time of year, of course, a 32MBx2 chip, the first generation VCache, is estimated to cost around $70 on its own. When integrated into the processor, it requires polishing and other processing on the processor side, which costs about another $30, for an estimated cost of about $100 per CCD.
For the 5900/5950, the cost will simply increase by $200 due to the 2CCD configuration.
(Genoa-X uses 8 sets of 32MBx2. (The processor is so expensive that it pays for itself.)

Test Production Products
CPU VCache technology is essentially the main target for virtual machine acceleration of server CPUs.
The 5800X3D was most likely a production test product before selling these expensive processors.
For this reason, it was not produced in very large quantities.
When the first Zen3 CPUs shipped, the interface for VCache was implemented from the beginning, but there was no cache chip yet.

It just happened to be suited for gaming.
VMs require a large amount of memory from multiple VMs in a wide range, and the timing is discrete. A large cache is well suited for this application. This scenario is very similar to gaming, especially MSFS.


Would there be a more close race if the resolution is cranked up towards 4k? Or does the 5800X3D still have the advantage here?

At 4K, as with VR, the GPU plays more of a role, so the results become closer. Even so, the 5800X3D is on par with the fastest Intel CPUs. Here’s one benchmark I found that shows different resolutions flying from London City to Heathrow. Follow the link for more benchmarks and more details at different resolutions.

AMD Ryzen 7 5800X3D review.


Hmm, seems then that the 12900K is still better than 5800X3D when it comes to 4K. Even though it’s just be a few fps.

I don’t think many folks use MSFS in 1080.


I wonder if I should wait for an upgrade from i7-10700K to AMD Ryzen 9 7950X3D or should I get 5800x3D. Since I’ll have to change my motherboard as well, if I buy a mobo for 5800x3D then I can’t upgrade in future. I really can use some help here. Or should I get i9-13900K before even waiting for 7950X3D?

It’s up to 50% faster in CPU-dominant situations for MSFS. Search around the forums. It’s a solid 15-20 FPS. Those benchmarks don’t know how to push the CPU in MSFS. They simply load 200/200 Ultra and fly outside the cockpit. They don’t sit at O’Hare (KORD) or SFO with a little AI and a G1000 with TLOD at 250-350. That’s where the separation becomes most notable.

Yeah, that’s a tough spot if you really want an upgrade. They are saying Jan for 7000 series x3D but those are rumors. It could be spring 2023 or according to rumors, it could be Dec. Maybe. Certainly will need a new board and memory for the 7000 series.


Usually at 4K, most titles are GPU bottlenecked and performance doesn’t vary much. Just look at the Cyberpunk benchmarks at 4K. MSFS is unique in that it can still be CPU-limited at 4K, usually when using terrain LOD at 400. I think most reviewers don’t push terrain LOD too high, which doesn’t show off the benefits of the 5800X3D in CPU bottleneck titles.

Cyber benchmarks as a nice example of benchmarks showing a clear GPU bottleneck:

Yup I think folk just dont understand the timescales involved between planning and initial production and even then the extra delay scaling up for the big hitters. A flick of the switch it most certainly is not and then of course there’s always the next generation to worry about.

1 Like

I’m in the same boat. Looking to upgrade from I7 7700K and still can’t decide between the 5800x3d or the new zen 4 7000 series. I do video editing as well so maybe I would lean toward the 7000.

I think I’m gonna wait for 7950x3D in any case.

I’m waiting for the newer chips to (hopefully) drive the 5800x3d price down, then I’ll go for that - my PC is just for flight sim and I don’t have a spare COUPLE OF THOUSAND POUNDS lying about to do a worthwhile upgrade to latest gen tech in the next 12 months, so just sticking with 5000 series and my 3070, will make the hop from my 5800x to a 5800x3d when the price is right!

Just out of curiosity, if I buy a mobo that suits for 7950x3D, can one also run 5800x3D on the same mobo or they are incompatible ?

They are incompatible. 5800X3D runs on the older AM4 socket; 7950X3D, if it ever exists in the future, will run on the newer AM5 socket. You’ll need to replace both the motherboard and RAM when changing from one to the other.

1 Like

The OPs assumption is also nagging me no end…
So I’m still on a (heavily overclocked) i7-4770K at the moment, and did not even think about upgrading until I jumped onto the MSFS bandwagon earlier this year.

I want to get the best CPU to specifically tackle MSFS, and the PC will not be used for much else really.

I can see from several benchmarks, that 5800X3D outperforms even the 13900K in very high FPS scenarios
But then again, I’ve also come across benchmarks, that set up more demanding scenarios, with the FPS well below triple digits, and in such cases, the new 7000 was equally fast or even ahead, with higher 1% FPS as well.

With the actual scenario being LODs ramped up to 200-300 (wishing, lol), airports full of FSLTL AI traffic, and FPS probably more in the range of 30-50 in most extreme conditions, my assumption would rather be that the 5800x3D might not even be faster then current gen CPUS, maybe even suffer from lower 1% FPS, due to lower clocks and slower memory speed…but it’s a guessing game really :man_shrugging:

Then there is a possible Zen4 x3D looming, but nobody can tell for sure when/if will hit the market, and if it will offer the same performance boost as Zen3 3d cache did.

1 Like