AMD Ryzen 5800X3D: Let's Chat! 

Scooped one for $289 other day, wonder how low it’ll go before it runs out of stock

Just picked mine up for $289 also paired with the 3090fe . PBO tuner set all cores to -30 performs so good and stays cool . Undervolting is excellent on this cpu

I think you'll have to wait until Zen 5, but other than that, good luck! You'll surely enjoy this CPU over R5 3600, and will serve you very well for several years.

@@Winnetou17 You're right, I should have said "AM5 has matured". I'll let others pay for the uplift Zen 4 will bring by being beta testers of the new socket. Lol

@@benjaminoechsli1941 For longevity sake, I would think about a 5900X. Purely for more cores, clock speed, PBO and the ability to overclock. Do not get me wrong… for PURELY gaming the 5800X3D is a good chip. The 5900X simply games well enough and does every other task much better than the 5800X3D. Want you to have an amazing investment after these last crazy 2 years!

The most interesting thing is that none of the reviewers compared the 5800X OC, 5900X OC and 5950X OC with the 5800X3D processor. But it's interesting to look at

@@silenciummortum2193 I really don't understand why you guys keep comparing it to the 5800x. Literally costs more than the 5900x which is on sale right now, or retails for just 50 bucks more USD, and where most of us who'd be thinking of buying this thing already are on AM4 and so our end stage CPU choices are going to be the 5800X3D, 5950x, or 5900x. Those are the only three CPUs I really care about being lined up in different tasks, which obviously the 5950x crushes it (albeit more costly) and to where our real gaming potential is going to be what that can achieve on a usually mid to higher end board (because if you're thinking of this good chance you have at least a decent b450 if not high end b550 or x570) using an all core "overclock" with sustained turbos. Like the only thing I actually care about is 5900x, 5950x, or 5800x3d. I have no clue why every reviewer keeps doing this because it's just not helpful. A lot of us probably are also on 1440p or 4k. I doubt anyone at all is thinking about the 5800x at this point, or where more or less each of these CPUs is in range of each other except 5950x but even that's only a hundred bucks higher depending on sale (identical gaming performance but the PBO/Ryzen Master tweaking allegedly offers slightly higher turbo clock + your final AM4 CPU has 4 more cores for the next 4-8 years of using this PC). I'm not thinking about switching to Alder Lake and I'd never consider the 5800x or 5700x upgrading from my 3700x, which likewise the 3600/3600x, 3700x, and 2700x users are probably also looking at how it all compares. I'm crossposting this in hopes that at least someone does this lineup rather than comparing to the 5800x or 12900ks (lol) neither of which clearly I'd ever get from 3700x. Seriously I get that it's to compare the quality of the hardware as per what the advertisers are saying, but it's literally not at all related to what any of us actually wants, because I'd never upgrade to the measly 5800x to be my computer's final form on AM4, nor would I ever build on a completely new platform even if it did beat me because DDR5 is unicorn blood tier expensive and why would I waste the money for negligible if any results. The only people this satisfies is Intel and AMD, not the consumers, all of whom are thinking about the 5900x, 5800x3D, or 5950x, which is funnier when it costs $70 more than the on sale 5900x right now. You'd have either already bought a 5600x and not be upgrading for many years like a 3570k user buying a used 3770k on ebay before junking his LGA1155 on ebay or storage/relative, where this won't be useful to you for 5 years, or be on an older generation and thinking about what your last CPU is going to be which there really is no point getting a 5800x and would either be getting a 5700x to stay octacore, or just getting the best last chip (which is 5950x unless you're solely gaming and the added cost isn't worth it) which is vs the 5900x for gaming. Like I'd only be considering 5900x/X3D at that point because there's theoretical sustained boost on this chip so long as you have proper cooling, whereas 5900x has much better clocks (so it smashes at CSGO for example) and sustained all cores, but has 30mb less cache so cache hungry games will perform substantially better. That's where I'd bet like 90% of all people watching these videos are weighing their options. I only don't consider 5950x more strongly because it costs so much more while being nearly identical in gaming performance.

100%. I care more about the 1% lows than the avarage.

Precisely.

Technically it's not even a GOG upgrade, its just the same zen 3 architecture with added cache. Goes to show how good this architecture is, given more cache.

@@arkeynserhayn8370 goes to show how good the 3D V-cache technology is, as well

@@dexter2392 Yeah, that's the main point... Relieves quite a lot of loading in stress by having a monster cache... Also it's right above the cpu cores... not to the sides which is much closer. That helps with speed and signal integrity.

Socket “dead ends” are overplayed anyways. Most users will run a CPU for 4+ years and then upgrade straight to a new socket, and those who do upgrade are upgrading into a dead end. Most 2000 series systems don’t have chipsets compatible with 5000 series CPUs anyways. 3000-5000 series is a 2 year upgrade schedule.

Can't wait to see what comes from this over time!

Which board? I run the msi B450-A pro and haven't seen documentation on this

Nice, awesome upgrade

It's likely going to be combination of the 2 that end up mattering. 12 thread CPUs are becoming the lowest common denominator of PC, and the consoles have 16. A 3700X is basically the baseline for what a maxed out EOL console port would demand. There is an argument that any CPU with equal or greater computational potential than that would be enough regardless of core/thread count. The 5600X will likely always be at least on par with the 3700X if we see games demanding 12-16 threads. It's unlikely we'll see even PC focused titles requiring more than that anytime soon.

I'd point you to Daniel Owen's coverage of the sample for Matrix Awakens, which is a UE5 title. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-0V5TCZzzgOo.html It's pushed a 5950X from around 29 - 32% most of the time. Now, he was doing a flyover so maybe the game isn't as intense otherwise, I don't know. But I calculate that out to about 60% for the 5800X3D (better IPC gaming) and 85 - 90% utilization for the 5600X (doesn't run as fast). So while it's true that is should still handle it, it's right there at the border and I certainly wouldn't buy a 5600X to put into a gaming rig at this point now that's I've seen what a UE5 title can do, and it makes me REALLY happy that UE5 can do this because game engines have needed to get better usage from a CPU, instead of being bottlenecked by the CPU but an 8 core is still only at 25% utilization, just because it takes so long to process threads.

​@@johndoh5182 I've actually watched that already. When I said "demand" I wasn't implying more would be of no use, just that it would make or break the experience. Keep in mind that is a unoptimized tech demo, and basically no game ever has come out being anywhere near any of the UE tech demos. It's intentionally not offloading GI calculations to the GPUs dedicated hardware, as a proof of concept. Most actually built games are more than likely going to partner with Nvidia or AMD (even Intel). Epic themselves have a long history of working with Nvidia to integrate their tech in their engines. Direct storage also intends on taking load off of the CPU.

The future looks bright and competitive at least with so many innovations. 🙂

@@blkspade23 nvidia just gives a lot of engineers for Unreal Development to work their magic. trying to integrate their special sauce directly for support. same from Sony and Microsoft for their consoles the RTX for Minecraft video made that very clear.

What temps have you been experiencing? I'm seeing an average of 58C with a Noctua NH-D15

Man I am thinking of getting this one to upgrade my 7 2700x too and I think your comment has convinced me

I'm on the same page as you with this one, been holding out with my 2nd hand 2700 for a while and now I'm considering making the jump because I won't have to do a whole new build

except $450 is not awesome for 8 cores. The 5900X is currently selling for less than this $449 MSRP.

@@bojinglebells agreed. But when it comes to gaming for people who only do that on their computers the 5800x3d is the better deal.

Probably significantly, as they will have more time to better integrate V cache timings, and games will better allocate important game functions, and resources to V cache. Can you belive it? the 5800X3D almost the same amount of SRAM as DRAM on my dads 2004 Athlon 1500+ system. Mind blowing 🤯

There was a video about the amd epyc cpus and how they benefit only certain workloads like some simulations. Not to mention that it seems that they dialed back boosting and disabled OC due to the fabrication process still being relatively new Cache is great for compiling stuff tho, I'd love to have a special server for building stuff at work :p

@@MarcABrown-tt1fp I believe you. My first pc had a Athlon 64 x2 and it was a little beast. Still miss it.

@@Mpdarkguy Milian x? Yeah, I saw that video. I'm going to guess that 3D V-cache will be optional, like a 7600X3D and 7800X3D versions for gaming. Wonder if it'll come to their workstation CPUs, ie TR? I'm sure there's some workstation workloads can benefit from more l3 cache like CFD.

That big cache is good for compiling software too.

production?

5800x3d is super crap for rendering, video etc... so I really hope you've got two computers to make that switch...5950x is the champ by miles I've got 29k cinebenchr23 at 69°C amd is so ahead for productivity..in the same loop the 12900k can't really hit 28k and it's at 90°+ same cooling ! whatever you decide definitely don't go intel 12900k ddr5 Z690 lanes/bandwith management is really bad and ddr5 is a giant lie, you cannot have more than 2 sticks and there's a mafia like code of silence when it comes to that, no one talks about it...when amd release their version if it's the same garbage be ready for a dumpster fire because there arre more productivity people using AMD and they won't like that at all

I hope there is R9 5950X3D a last hurah for AM4 users.

@@jayvee8502 the die space is pretty much maxed out on the 5950. It's got 64mb of L3 cache anyways so it's not too far off the x3d for gaming, but it still smokes it in productivity apps.

@@amirpourghoureiyan1637 2x32mb cache (not 64mb) the L3 cache between ccx has to be kept in sync if the process runs across ccx groups

That will be huge upgrade. Which GPU are you using?

Did the same thing with from a similar build's running off a 1060. best upgrade you will ever make!

Hi, did you see and feel any difference?

Zen4 without 3d cache is faster than one with 3d cache in productivity... so like now it depends on what you do with your computer. We can be quite sure that we see some special edition of Zen4 with 3d cache, but it won´t be the fastest in everything! And it will be more expensive than non 3d cache version. But if you look for gaming CPU... these can be really sweet!

@@haukionkannel wrong. Its not hte 3d cache that makes it slower.

@@santiagoferrari1973 It is the clock speeds that has to be low because of 3d cache... So it is because of 3d cache, or the limitations with it. If the Zen4 can go even higher with high voltages. The difference is even bigger.

I was gonna do the same upgrade... but picked an MSI mother board and the bios isn't gonna support 5000series (as of mid-april anyways)... much pain. If I went Asus, or one of the other ones, I probably could had this upgrade path. Really disappointed MSI basically lied.

@@QuantumConundrum Thankfully I'm running a 370x Prime Pro. It supports 5000 series. When in doubt, I always go asus, and it usually doesn't fail me.

go for a 5800x or 5900x trust me

@@chadbrick67 if he cares more about gaming performance and less about productivity then 5800X3D is the logical choice. However, something else to keep in mind is what resolution he is going to be gaming on. At 1440p and above the gains are 11% or less (average) according to hardware unboxed.

@@lakibadhikari7930 5900x is still cheaper and on par with gaming

yes me too. I recently switched to linux and I have this channel bookmarked specifically for linux info. I know the lvl1techLinux exists but that doesn't get a lot of content.

@@odizzido Hey, I'm looking to switch, got some channel recommendations for starters (proton etc)?

@@adriankoch964 Chris Titus Tech, Linux Gaming PC, DistroTube ...

@@adriankoch964 check out the channel "Awesome Open Source" there is many interesting Linux applications being shown there, also some alternatives to Proprietary and Windows Software which might help you out.

@@adriankoch964 Chris Titus Tech has a few good streams and videos and Blog write ups to follow along . I must warn you the rabbit hole is deep. start with arch it's not as scary as it seem . if you break something it will happen then just have timeshift installed to revert back to the known good version . All the best

Silicon lotto my guy. Sometimes it plays into our favor, my 3800x holds a solid 4.6 all core at 70c with a 360 aio. Other people.. have done jack shit with PBO, and have called it bogus.

@@crisnmaryfam7344 in comparison to you I would say I got the Bronze award in terms of maximizing your CPU 😅

silly go for a 5800x or 5900x

Gamer's Nexus does. They have a Wendell-porter.

Intel have more cores doesn't exactly beats a 5950x at the time

@@chriswright8074 Not to mention 5950X is still a general purpose CPU.. their productivity-focused lineup is the Threadripper family and Intel literally has nothing to compete there.

he must be doing different tests to mine I have a I9 1200ks it hammers the amd cpu

@@chadbrick67 Yeah, there has not yet been any reviews where this CPU is paired against other CPUs in optimized systems. It is just "here is this running in stock" (because it cannot do anything else) and "here is this competitor running in stock" (even though you can push ~10% more from it if you want to.

It's pretty impressive that AMD has been able to neutralize Alder lake, Intel's latest and greatest with a 17 month old architecture on a CPU that costs significantly less and uses significantly less power, as well as DDR4 memory vs DDR5. Factor in that Intel has an R&D budget 650% larger and an annual revenue over 800% larger than AMD's and its pretty embarrassing for Intel. If AMD had Intel's budget and resources, Intel would be embarrassed on a whole other level.

@PHD DOOM I too have an X470 but its an Asrock wifi SLI and with the same CPU as you do, so if i were to upgrade to 5800X3D. Will I have to do do Bios update or just drop it in?

Yes. Before you drop a newer CPU in your Asrock motherboard make sure you have the most up-to-date BIOS to insure you unlock all of it's features and for stability. Go to the Asrock website site and on the downloads or support page you just need to look up your motherboard and look for the most recent BIOS update.

@@MrAddis aaah i don't have to download fron the start and go up to the most recent release? I'd just have to go to the newest one?

same dilema, not helped by that 5900x is cheaper now

its nowhere near the KS I've got one don't believe that rubbish alderlake hammers it

@@chadbrick67 Ok fanboy.

@@jotunheim5302 so ok I'm a fanboy because I told the truth..hope this helps

@@chadbrick67 Told the truth yet there a plethora of reviews out that prove you're a liar.

@@jotunheim5302 slight gaming gains but productivity not as good. And I'm a liar you say. The triggering is complete

Yess. I was also waiting for an upgrade from my trusty 3700x while increasing core count.

Some have said it was to hard to add extra 3D cache to a cpu that has more then one CCX. Like it would added to much latency on the infinity cache bridge too the two or more CCX chips. So in short it would have hindered a 5900x3D hence why they did not make one.

@@digitaltactics9234 Don't get me wrong, I'm all read up on it and I understand why there's no higher core variant, it is just something I would have liked to see.

There is one big problem... 5900x and 5950x are mainly CPU for productivity. 5900x3d would be worse in productivity than normal 5900x, so it make very little sense to produce CPU like that! You make CPU that is slower than then old one and would cost 2*$110=$220 more than the normal one... It would be hard market for CPU like that! Even 5800x3d has hard time. You can get 5900x at the same price or cheaper and 5900x beats 5800x3d hands down at everything else except gaming. So for pure gamers 5800x3d makes sense, but 5900x3d for content creator does not make any sense at all!

@@haukionkannel No problem as it IS meant for gamers

Workloads not optimized to take advantage would be small workloads with a small memory footprint usually. Not much you can do about that

@@Jaker788 it feels like there must be some way to step through the code and quantify the cost of all the hoops some programs jump through to maintain a small memory footprint. Far beyond my skill level but it feels like it should be possible

@@christopherjackson2157 Most of the time when you have a small footprint memory-wise, you either just didn't give up anything(cause you where small to start), or you gave up having large features you may not really have wanted anyway. There will be cases where more cache means you can have bigger data for the same performance, but anything that didn't care now will probably not later. x86 had big caches to start with, this is just application specific overkill.

The real question is - who would do the optimization that will only help

@@CZmiho well I think they developed the 3d cache knowing that large physics simulations like those discussed in Wendells earlier video were a large enough market to justify the existence of the product and hoping that others could hopefully run faster algorithms if they had the cache to work with. When a single customer is explaining how something scales across 30000 epyc cores to optimise the design of their product... if the 1 percent of customers who would benefit all buy 1000 cpus at a time... I really don't know how the economies of that would work it's entirely outside of my wheelhouse. But certainly both and Intel are willing to develop new technologies to the spec of single clients if it's a big enough contract

Honestly... I'm glad I'm not the only one willing to shell out $$$ only to play stellaris a little faster lol

@@tubawest9768 I play games like skylines, football manager, and stellaris. Pretty much nobody does cpu/gpu testing on those sort of games. Makes it really difficult to figure out what the game wants for cpu. About the best I could find was some of the community doing performance testing of different CPUs in football manager. But it’s still difficult to tell if that is dependent merely on clocks or if cache would make a significant difference. Blah

@@tubawest9768 I mean it slows down hard in the end game.

I will compare my 12900k on it

Sheeesh you tryna build a thermonuclear reactor LMAO

yeah but no...the problem that they are already facing with this version is that the memory gets between the cpu die and IHS that's probably the real reason for the non overclockability more than the voltage

@@fredEVOIX The main reason for the lower Boost clocks are definitly the limitation to 1,35v instead of 1,5-1,6v . But yes , under full load you have one layer that blocks heat Transfer . The reason that there is no second layer is in my eyes that the one layer adds 4 ns in latency , if a second layer would add another 4 ns , that may be not worth it . And you dont have to forget that the 5800X3d is in the same 105W TDP enevelope , additional cache needs additional energy , not that much but enough to lower the Base clock by 100 - 200 Mhz . There is a reason why AM5 will go up to 170 W TDP

@@MK-xc9to This. It seems likely the core and cache are fed from the same source and cannot be set independently. Ryzen CPU single thread boost clocks demand up to 1.5v, and never produce as much heat in that when lightly threaded. Then refer back to Intel's 10th gen die shaving. Its purpose was to allow heat to escape the die faster. AMD had to do the same thing to keep the height even, but also provides the same thermal benefit. The boost clocks definitely fell from lower available voltage.

Those workloads will see a big benefit. That server market is actually the primary motivation for this technology, the 5800X3D is just a quick port to make a gaming CPU competitive with alder lake until Zen 4 comes out.

If you play heavily CPU bound games like MMOs the x3d is a huge upgrade, if you play GPU bound games the 5700xt is the limiting factor. World of Warcraft, for example, would get double the fps in 25 and 40 player content from a cpu upgrade. One of the most CPU bound games out there. A lot of first person shooters would get almost nothing.

@@p51mustang24 Thanks for replying :) I appreciate it. I did eventually end up buying the 5800X3D and changed my 5700XT for a 6700XT. I'm hoping the 5800X3D will keep my rig in service for another 2 or 3 GPU generations.

Imo just wait, I have a 3950x paired with a 3090 and I'm waiting for AM5.

Even if you don’t end up getting Zen 4, it never hurts to wait. At the very least, current Zen 3 CPU’s will likely get discounted around the time of Zen 4 launch. Already, I’ve seen the 5900X dip below $400 USD at some retailers, which is an insane deal. If you don’t need the 12 cores, then wait for the 5700X to go below $300. It’s likely to get down to or below $250 in the coming months.

While the CPU multiplier may be locked, I dont see how that would effect the voltage supplied to the chip by the motherboards VRM. Not overly familiar with AMD systems but Id assume they have something equivalent to SVID tables that Intel has. While the CPU may ask for whatever voltage AMD thinks it should have at a certain said frequency, one can override SVID and provide whatever voltage your motherboard VRM will allow as a choice in BIOS. Funny thing is this is usually thought of in regards to RAISING voltage for OC purposes, not lowering voltages as Intel and AMD i know have always erred on the side of too much rather than too little voltage to account for silicon quality in the lowest end and the highest end. They must know how many people get a kick out of undervolting everything they own nowadays, why not allow lowering vcore but not raising above that preprogrammed SVID tables max allowance? Dont know why Im asking you this question, more just wondering aloud.. :) Cheers

Rumors from reputable sources have consistently stated that Zen4 will have at least a 29% IPC increase over Zen3!! Add I to that a substantial frequency bump as AMD itself has claimed a minimum of 5ghz but probably more in Zen4 and the benefits of moving to 5nm from 7nm on the chiplets and 12nm to 6nm on the IOD and Zen4, and the move from DDR4 to DDR5 (with recent AMD statements saying that DDR5 overclockong on Zen4 will be groundbreaking) could easily be seeing a 35%-40%+ core for core performance increase over Zen3 which would literally be an unprecedented performance increase generation over generation in the last decade or more with respect to CPUs. Multiple leakers have even said that Zen4 will have no problem beating Raptorlake, so Zen4 should really be something to behold.

@@post-leftluddite Exciting times.