Ryzen 7 5800X3D vs. Ryzen 7 7800X3D, Ryzen 9 7900X3D and 7950X3D

[STbob]

All this "You will have a upgrade path"
You do not need one at all. I've never used and I have wasted money on that before. The truth is in 20 years I don't think I've done that. Something better comes out and I get entire new MB/RAM/CPU. I mean it sounds nice for sales but in reality don't worry about.
Its all about price.

I built two AMD systems last july and chose not to go with DDR5 for many reasons. Stability, prices, bugs etc. I would have spent 500.00 more because I bought two of them.
And would not be much faster than what I have now since I'm not using a 4090 I have a 4070. ZERO regrets and have a 5600X3D so I do have a REAL upgrade path to boot + 500.00 in my pocket.
I wont upgrade either I will skip the next and probably the generation after it unless I get at least 2X performance jump with less electricity.

Take the "I need a upgrade path" with a gigantic grain of salt.

 

[Lew Zealand]

All this "You will have a upgrade path"
You do not need one at all. I've never used and I have wasted money on that before. The truth is in 20 years I don't think I've done that. Something better comes out and I get entire new MB/RAM/CPU. I mean it sounds nice for sales but in reality don't worry about.
Its all about price.

Do you also stick with the same GPU the whole time? Same idea.

You may not need an upgrade path but some people take full advantage. My kid's B350 went from R3 1200 to R5 1600AF to R5 5500. Why spend another $100+ on a new MoBo if you don't need any new features and all you need is more CPU? I built a second PC with a B450 for myself from that R3 1200 and now have an R7 5800X3D in there.

Upgrade options are good.

 

[Q Wales]

Shame we can't get this kit in the UK for anywhere near the prices you guys get it for in the US. I can buy stuff from China really cheaply but stuff from the US costs a fortune.

 

[ScottSoapbox]

The One main thing that has kept me from not upgrading from my 3950x to any of the newer 5950x, or 7950x is that when using a 4090 to 4k game (the main use of this pc), all the benchmarks I read show very little difference between the 3 choices as everything is so gpu bound at 4k. Sure I'd get ddr5 and pcie5 and usb4, but all the benchmarks I've read don't lie........ I'd be spending hundreds on new mb, ram, cpu, and potentially new NVME's to get a marginal upgrade. Maybe the 9950x will give me a big enough boost to pull the trigger. we shall see! If only I used more of the creator side of this cpu's abilities, maybe I could see the reason to justify....... Anyone else out there have a 3950 and 4090 combo, that are also on the fence? (or not even on it, but looking at it going "What for?")

I had a 3900x and 3080 combo and upgraded the CPU for Spider-Man. I did the drop in 5800x3d because (as you point out) it's a lot to do the full upgrade. It was a noticeable difference.

At this point there are a handful of games where the CPU makes a significant difference. If you play those games it is worth it, otherwise you are generally fine. That said, the 3000 series has amazing multicore capabilities and adequate single core. So there is a good amount of improvement in newer chips.

I personally am waiting for the 9000 series to upgrade again. I held off on the 7000 series because I knew the RAM was going to get significantly better and cheaper once it wasn't brand new tech. Also, I've realized I don't do as much video editing as I used to, so the 12 cores are kind of wasted these days (my primary PC is 5900x), so I'm eyeing the 9800x3D and 5080-ish (depending on pricing) next.

 

[ScottSoapbox]

All this "You will have a upgrade path"
You do not need one at all. I've never used and I have wasted money on that before. The truth is in 20 years I don't think I've done that. Something better comes out and I get entire new MB/RAM/CPU. I mean it sounds nice for sales but in reality don't worry about.
Its all about price.

I built two AMD systems last july and chose not to go with DDR5 for many reasons. Stability, prices, bugs etc. I would have spent 500.00 more because I bought two of them.
And would not be much faster than what I have now since I'm not using a 4090 I have a 4070. ZERO regrets and have a 5600X3D so I do have a REAL upgrade path to boot + 500.00 in my pocket.
I wont upgrade either I will skip the next and probably the generation after it unless I get at least 2X performance jump with less electricity.

Take the "I need a upgrade path" with a gigantic grain of salt.

Upgrading just the CPU is a nice option. Depends on the user/use-case on if you take that option vs a full rebuild, but having the option is always nice.

The alternative was Intel forcing MOBO sales for no reason. That's not nice. The downside for Intel, though, was it lowered switching costs. I.e., a customer looking at a new CPU had to buy a new MOBO staying with Intel or switching to AMD.

 

[Jyrkz]

Irrelevant for a CPU benchmark. All running 4k would do is compress the results and make worse CPUs look better then they really are.

That is very true, but I didnt say only 2k, 4k. For me, I would have really liked to see both 2k and 4k benchmarks cause I am playing at those resolutions and I also want to see whats best cpu at those res. I would have liked to see the whole picture.

 

[Lew Zealand]

That is very true, but I didnt say only 2k, 4k. For me, I would have really liked to see both 2k and 4k benchmarks cause I am playing at those resolutions and I also want to see whats best cpu at those res. I would have liked to see the whole picture.

The easy conclusion is the CPUs are still in the same order, but their performance differences are compressed to be close to random variance and well below what you could discern without an FPS counter active.

TechPowerUp tests at 720p to 4K in their day 1 CPU reviews and what's best at 720p is the same at 4K, the margins are merely reduced from 24% between the best and worst current-gen 6+ core CPUs to 6.6%. There's a small amount of reordering but they're within 0.5% variance, so effectively equal.

 

[gamerk2]

That is very true, but I didnt say only 2k, 4k. For me, I would have really liked to see both 2k and 4k benchmarks cause I am playing at those resolutions and I also want to see whats best cpu at those res. I would have liked to see the whole picture.

The results would be basically the same, due to them being compressed. Which is why CPU benchmarks at high resolutions is not only useless, but gives outright false information. One could incorrectly assume CPU x and CPU y are the same, and will remain the same years in the future. And that wouldn't be true.

 

[Gastec]

I've been meaning to ask this for a while.
In "The Best AMD X670E Motherboards" it is written:
"Therefore, you wouldn't say purchase the Aorus Master over the Taichi because the Gigabyte board's VRM ran 4c cooler. [...] As for the other three options, we think the least impressive is the X670E Aorus Master. [...] Perhaps the bigger problem for the Aorus Master is that for the same price the Asus ROG Strix X670E-E Gaming and Asrock X670E Taichi are better equipped. Both boards wire in their primary and secondary PCIe x16 slots to the CPU for PCIe 5.0 bandwidth, the E Gaming is the most affordable X670 motherboard to provide three PCIe 5.0 enabled M.2 slots while the Taichi supports USB4 and eight SATA ports.
Frankly, picking between the Asus ROG Strix X670E-E Gaming and Asrock X670E Taichi is difficult, they're both excellent boards, albeit mighty expensive at $500, but they're equally good in our opinion."

So, why has the Gigabyte X670E Aorus Master motherboard been used in most of these tests and benchmarks?

 

[Gregtombstone]

Just upgraded from 3700x 16gb 3600 cl 16 to 5800x 3d 32gb ram. 6800xt for gpu.
Bf2042 ultra at 1440p before 109 avg now 139 avg . Very happy. 5800x 3d gift from my Kids