you are 100% incorrect. the 14900 doesn’t even keep up with the 5800X3D.
Most gaming workloads really benefit from the extra vcache on the X3D chips, MSFS especially loves it.
The days of intel’s superiority in gaming workloads are long behind us.
yes, it shows the 13900K there, but the 14900K is basically the same.
Does TH show a chart for Flight Simulator (DX12)?
don’t think so, also, doubt it matters for CPU performance.
FullHD and DX11. Nobody cares.
DX 12 and 4k please.
And oops: Intel
ehm, you know there’s a reason they’re benchmarking CPU’s on 1080p right? it’s to make sure there’s no GPU bottleneck whatsoever.
I had only ever bought Intel CPUs but the 7800X3D was too good to pass up as an upgrade. No regrets.
Pretty sure I’ve only ever owned Nvidia GPUs other than my 3DFX GPUs back in the day (anyone remember the Voodoo cards 
)
Indeed I do. I had a 1st gen Celeron 300 (overclocked to 450MHz) with a 12MB Voodoo II. That thing was the shiznit back in its day. The leap ahead in graphics tech was incredible. Particularly in OpenGL and GLide games. That was next-level stuff back then.
Apart from all the benchmarks and all, I think, this video says it all:
From what I have read there still apparently appears to be at least a couple of unresolved issues with AM 5 and DDR 5:
I am still reading of overly long boot up times due to memory learning which increase as ram sizes increase. A bit of a concern for me since I am looking for at least 64 gb ram (6000) for my next PC. I gather there is an option (fast boot) to reduce boot times but that can occasionally lead to system instability. My current (oldish) system boots up pretty rapidly and a long boot up time every time I boot the PC will I think for me become a bit tiresome after a bit.
I am also still hearing that AM 5 is still less tolerant of achieving stated ram speeds with 4 sticks due to memory controller issues. Hopefully, though, this might not be an issue if I get 2 sticks of 32 gb. Intel can also apparently have issues with achieving stated speeds with 4 sticks but that platform doesn’t seem to be as bad from what I’ve read.
Tbh I personally think the whole launch of AM 5 has been pretty much a mess with stories of burnt MBs, dodgy BIOSs plus various other issues.
So … not really sure whether to bite the bullet in the next couple of months and just go the current AM 5/7800x3d route, go the intel route or wait for the next gen AM 5 stuff to come out in the second half of the year (July/Aug ??) or wait for Intel Arrow Lake to come out (late 2024??).
Yeah it’s definitely a bit flaky still and down to the luck of exactly what came off the production line I guess… I think I lost in the luck lottery but I am in the minority.
I am running 100% stable now with 4 sticks of 16gb, but had to reduce speed to 5600 (5800 was mostly fine, but would BSOD once in a while so I prefer to just run it a bit lower to avoid that frustration). I could not get 6000 to post with all 4 sticks, but I am on an older Bios now. Keep meaning to try the latest and see if things have changed but can’t face the work at the moment just to end up back where I am now.
The “Fast Boot” you wrote is a different thing. That (if I read you right) is for WINDOWS to resume faster. The faster boot up time to skip the memory training is called Memory Context Restore, whereby it ‘saves’ the previously passed training timings and uses that, which for sure speeds up the initial boot (before Windows) but frankly it doesn’t take that long since the Bios revision I am on now. I would say 20 seconds with my 4x 16gb at that 5600 speed. Actually pushing the speed higher is what seems to take longer in training so this time is variable by a large degree. Not sure if 2x 32 would make any difference over 4 x 16 in this regard. Probably a BIT but nothing that noticeable. The overall boot time compared to how long MSFS takes to start makes this a moot point in my mind anyway! I just turn it on then go to the loo or make a brew (whilst MSFS loads - the post boot is nothing compared!).
At the end of the day, once you run stable for a month or so with the speed you want, you can comfortably turn on the Memory Context Restore and only activate it again if you have a problem or change anything (I think it trains again automatically if the PC crashes but not sure).
As you probably saw in my crisis thread, it took me a long time to get things working but now it’s dialled in I have had no problems. I am sure it should or could be running better but it’s still awesome so one day I will find the time (and be brave enough) to try a Bios update and run 6000 again. Or maybe some better RAM will come out (?) and I’ll just buy a kit of 2 x 32gb and give that a shot again. Happy for now though!
Hi Baracus. Thanks for your reply 
Not too sure which way to jump at the moment.
My current system dates back to 2017 (with a few enhancments along the way) so I am now getting really impatient for a high end upgrade. I’m still getting really good performance though in FS20. Not sure how 
I followed the thread you were posting in at the time you got a new PC and you definitely had a very tough time. You stuck at it though and I’m really pleased you got the PC working ok in the end. Not sure I would have had your patience/resilience though.
I myself was originally going to do a self build (even bought a H7 flow PC case) but I’m beginning to think that I might just have an easy time and just get a customised prebuilt. Currently I’ve been playing about looking at builds at Cyberpower PC UK (where you bought your PC I think). Would you recommend them? I have read good things about them but I’ve not bought from them before.
Regarding boot up times, yes, I was really referring to the ‘Memory Context Restore’ you mentioned. Good to know that your boot up time is only 20 secs. I can live with that so maybe that might not be an issue for me after all.
It’s a bit of a concern that you say your PC is still a bit flaky and also that you still can’t achieve the 6000 speed. That would hack me off a bit tbh especially given how much these PCs cost. I wonder if this will rectified when the new AMD 9000 cpus come out with the new 800 MBs? Thing is the second AMD gen might have its own problems especially at launch which is when I would probably pull the trigger. Maybe better the devil you know?
Thank you for your comments anyway. Very interesting to hear and glad you are now finally in a happy PC place
Correct. And no, I wouldn’t recommend them. I don’t know if any of the others are better at all but wish I had gone with Overclockers now. Just they didn’t have stock of the CPU when I was desperate to get something. In hindsight I should have been more patient too but hey!
Not what I was wanting to hear
Thanks though
I wanted to provide some addition color with regards to your concerns.
Long boot up times - I’ve got 32Gig of DDR5-7200 memory in my recently built PC and I really can’t tell that there is any extra time added due to memory retraining. I’m sure there is but it’s a very small amount of time that I don’t notice. While its not 4 sticks of 64Gig memory, I think its still relevant. Keep in mind AMD really improved memory retraining time with their AGESA releases late last year (motherboard manufacturers incorporate the AMD AGESA into their bios releases). @MrTonySM can give you more feedback with his 64Gig.
Not getting stated RAM speeds -
I know AMD has had their struggles with memory but I believe the majority of the memory issues have been resolved with AMD’s versions of AGESA released late last fall (1.0.0.7+). Perhaps @Baracus250 might want to update and try to hit 6000mt/s?
Also, think many people do not purchase memory that is on the motherboards qualified vendors list (QVL) of tested memory for compatibility. On my recent upgrade to AM5 with a 7900x3d I purchased 32gig of DDR5-7200 memory on my motherboards QVL list. First time booting up the PC with the most recent bios and memory speed set to XMP DDR5-7200 it just worked and with no stability issues. I’m currently running the memory slightly overclocked and am very happy.
I did that and it would not post even at stock speeds (safe defaults). There has been a few newer ones since so not sure if they would have changed anything but this is why I’m hesitating to waste all that time.
My current kit (Kingston Fury) and another kit I bought to try (GSkill Neo) are all on the QVL List so it’s strange.
Thanks for the feedback. Appreciated 
I do know @Baracus250 has tried one further BIOS revision without luck with 4 sticks. I am very hopeful that he can hit 6000mt/s, as there are other users reporting 4 sticks working at stated speeds on further updated bios. I am currently using the bios version containing Agesa comboAM5 1.1.0.0. It is best bios to date for AM5. Memory training takes seconds in this version, and the hic-ups have been worked out. The addition of DDR5 Nitro really helps with Stability, that came out a few bios’s back from this one. I would stay away from the couple bios versions past this one from now. Unless you have one of the 8000 series cpu’s. Too many problems being reported as they mitigate UEFI LOGO Fail vulnerabilities.
Myself, I run 64GB in a two stick package. They were a bit tricky to OC but I ended up with some decent results. They are hynix memory and I find they really can take the voltage. If you we’re ever not sure what kind of voltage… this kind This is a dual rank kit. 64GB
To sum up in a sentence… INTEL or AMD? AMD will give you a higher overall framerate, INTEL will give you better handling of 1% lows. If it’s a new build, always stay with 2 sticks of ram either intel or amd. 4 sticks is tough on the memory controller. You certainly won’t be OC’ing 4 sticks on either platform. INTEL is being put in the hotseat by game makers as of late. Specifically the 13900K, and the 14900K crashing Unreal 5 games (And I’m sure others)
EDIT: It’s a very tough choice right now. AM5 though does reportedly have a better memory controller (2 Sticks, 4 sticks are hard on boards)(Buildzoid has made many comments in his past videos about the poor quality of the Intel memory controller on the 13900k and 14900k and the ease of use with current AMD builds… he flat out states that memory over 7000mt/s qvl or not has no guarantee of hitting those speeds on Intel boards)
Understand, comments from RAD in the article above will come under intense scrutiny from Intel. They wouldn’t say this without proof, this would cause a major law suite otherwise (It may still)
EDIT2: This vid 
I don’t really understand the issue. Personally I don’t care about the brand. Whenever I decide to refresh my PC every few years I look at benchmarks and pick what’s fastest within my budget or slightly above. If I end up with multiple choices I look at other factors like power efficiency or compatibility. It’s that simple.
It’s really easy to choose a CPU. What’s hard is choosing tiles for your bathroom.
I’ve switched between Intel and AMD several times. Both work fine. If there is a problem it’s between the ears. Sometimes you need to do a little bit of research to get everything working optimally. But that’s half the fun of building your own PC.
I don’t like being ripped off, so lately I just pick the best from the previous generation. Right now it’s a 5800X3D (300$) and RTX 3090 (600$).
Another view on the 13000k and 14000k crashing in some games was that certain board manufacturers set the PL1/PL2 power limits to “unlimited” instead of the Intel recommended limits of 125/253w. So long as everything works out, you get better performance but if the circumstance are right, you get a crash.
That’s one of the reasons Alienware infamously limits the CPU power on their boards. They don’t want to have to deal with customer complaints when things go wrong in this way.
This link provides more details around what @wilco3563 is referencing above.
Intel 13th-14th gen cpu’s issues with power limits set too high
