Can anyone provide advise on 5800x vs 5900x for wow? From what I understand is that higher clockspeed per core is more important than utilizing every core in the CPU. 5800x has a higher clockspeed but lesser cores. I’m not sure if wow can use the 5900x optimally.
For majority of games 12-cores won’t be useful. If you want only gaming and no video encoding, streaming, code compiling, kubernetes, docker… then 8-core is the best value. As for clock speeds - those can be changed by various AMD overclock systems. And RAM is also very important as best/basic RAM config is 10-15% performance delta.
1 Like
5900x has higher single core clock boost being 4.8ghz
I think the 5800x has 4.7ghz range.
As long as you have good cooling you can stay on the boosted speed longer. I’ve heard the 5800x runs a bit hotter than the 5600x and 5900x however.
Got 3600mhz working on my 5900x with single core pretty much in the plus 4ghz when playing WoW. Not seen much slowdown in busy areas as of yet.
But here’s the thing, even a 5600x will do Shadowlands no problem. It’s the cheapest option, runs cool. So there are options.
1 Like
Thanks for the replies, guys!
I’m coming from a 8700k that was never overlocked due to the cooler. Planning to go all AMD now with the following:
ASRock X570 Taichi and either a 5800x or the 5900x. The goal is to primarily play games with it with no streaming or super heavy workloads involved. Playing on an ultrawide 21:9 with 160 refreshrate screen.
As for cooling, it’s going to be the Corsair H115i RGB platinum (First time going for watercooling) Read good things about this one.
For memory I read that high speeds are mandatory with an AMD CPU. Currently using 16GB of Corsairs 3200, 25600. Am looking at the G,Skill F4-4266C19D-16GTRG 32GB memory (4x8GB). Question: The CL is a bit high with 19, but the speed is also high. Decent enough? And Im seeing mixed views on choosing 2 or 4 slots to maximize the potential/squeeze out a little more performance. This motherboard is better for utilizing 4 slots.
Lastly, will I bottleneck the M.2 or the GPU if I use the PCIe 4 slot for the M.2? The GPU also uses PCIe 4 iirc. Can it be shared equally between the GPU and M.2 without downgrading the bandwidth?
The higher the frequency the same absolute latency will be expressed with a higher CL number. 3600 CL16 is quite strong, 3200 CL14 etc. If you buy like 4000-4400 CL19 RAM then it’s pretty much a high bin designed for RAM overclocking (if your CPU will be able to handle 1:1:1 clocks at 4000 then you would lower a 4400 ram to 4000 and set tighter timings, if not then you lower both to 3800 at even lower timings). If you don’t want manual RAM OC then stick to 3600-3800 CL16 RAM or alike.
F4-4266C19D-16GTRG is 4266MHz, CL19 so purely OC RAM. You won’t be running it at 4266CL19 but 4000 or 3800 at like CL16-18 depending what will be stable.
This is quite complex topic. Zen 3 likes 4 ranks of memory (where as 1 stick can be single or dual rank so 4 single rank dimms or two dual rank dimms). Plus higher clocks are mostly limited to only 2 dimm configs (as AM4 boards use only daisy chain RAM banks configuration, no T-topology boards from what I know). 2x8GB could be a good way to start (or 2x16 if you want the extra)
You use the PCIe x16 slot for the GPU while the M.2 drive goes to the M.2 slot, they don’t interfere with each other… and you don’t really need PCIe 4.0 M.2 SSD as that’s beneficial only to specific pro-workloads.
No, there’s lanes for the CPU that ensure that even with PCIE4.0 graphics card, it won’t affect the 4.0 M2 drive as that has it’s own four lanes from the cpu.
The second M2 4.0 would take 4 lanes from the X570 chipset so there won’t be a bottleneck there.
Ryzen’s have 4 extra lanes than Intel processors right now. 16 lanes for Intel compared to 20 lanes for Ryzen.
I am running a Gigabyte Aorus Elite x570 with 4x8GB 3600 CL15 RAM, but cannot get it stable past 3200Mhz with 4 DIMMs.
So make sure you research on the web if people can get 3600 on that board at all.
1 Like
not unique to zen 3, or even zen from what i understand. seems intel cpus also prefer 4 ranks.
and here’s an interesting thing as well. at 16GB, you can either do 4x4 or 2x8, and from what i understand most of those dimms are SR, so with 2x8 you’ll generally be using 1 rank per channel. 4x4 should then give you 2, but 4 dimms generally clocks worse than 2. don’t know what’s best at that point.
at 32, most 2x16gb kits will be fine cause most 16gb dimms are DR from what i understand.
also shows that there’s a lot more to it than what the memory is rated for cause i’m running my 2x16 3200 c16 kit at 3600 c16 with stock voltages on my 9900k/z390-f strix, but with looser sub timings.
memory is incredibly complex in general, and i don’t really know enough about it i feel like.
On Intel board that uses T-topology 4 dimms would not have such disadvantage. All AMD AM4 boards from what I know use only daisy-chain and that’s why they clock higher only on two dimms instead of four. They didn’t really make it “easy” there 
1 Like
True but some work okay on four.
I could never get the rated speed on four dimms with 2000 cpu, I tried but think my 2700x wasn’t a good binned silicon.
This 5900x though has managed to take 3600mhz no problem. Actually first time I can use the full speed on these dimms. So happy there, but think me overclocking the cpu memory controller 400mhz over the upper supported speed is adding to the temp a bit.
Same motherboard as mine those results are interesting to read if or when I upgrade to a 5000 in the future. Thanks for the info.
I need a better cooler if I’m going to use PBO though, the AIO I have right now isn’t good enough. Still good enough cpu, just don’t have PBO on right now. Don’t see why I should see any overclocked over 4.8ghz right now.
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.