AMD will be releasing the Ryzen 9 9950X3D likely during CES 2025, & it looks good. Thanks to Vip-cdkdeals for sponsoring this video. 30% off code: GPC20 ▬ Windows 10 pro ($17):www.vip-cdkdeals.com/vck/GPC20w10 ▬ Windows 11 pro($23):www.vip-cdkdeals.com/vck/gpc20w11 ▬ Windows 10 home ($14):www.vip-cdkdeals.com/vck/gpc20wh ▬ Office 2016($28):www.vip-cdkdeals.com/vck/gpc20of16 ▬ Office 2019($47):www.vip-cdkdeals.com/vck/gpc20off19 Sources AMD Ryzen 9950X3D dual X3D Vcache Leak: www.techpowerup.com/327057/amd-ryzen-9-9950x3d-and-9900x3d-to-feature-3d-v-cache-on-both-ccd-chiplets AMD Ryzen 9800X3D Official Launch: ruclips.net/video/ZTMBqNbHUrs/видео.html
@low_etc In gaming. Not so much in everything else. Not to mention Arrow Lake needs CUDIMM ram to run at its full potential which allows it to actually beat the 7800x3D/7950x3D in some games. There also seems to be an issue with Windows hampering Arrow Lake performance, we need to wait a few more weeks to maybe even a couple months to really guage how well Arrow Lake actually is since it's a paradigm shift and a gen 1 product for Intel that still has quirks and bugs that need ironing out.
@03chrisv rofl intel fan boy grasping at straws! Yeah in two games, and in specific sections of those games, the Intel chip has faster 1 percent lows!
I want one monster CPU that doesnt differ between work and gaming. I hate that I have to choose between X3D and X, gaming/work. Just make one monster CPU. One CPU to rule them all.
How hard is it to base your choice on what your usage is? Do you game more often than run office suite software? AMD. Do you run productivity software more than game? Still AMD so you are not abusing the environment 😅
@@jasonvors1922 I mean for some reason people complained when 9000 series used less watts which is weird cos 1 those chips aren't their gaming chips and are getting similar performance at half the wattage its like hows that a bad thing?
Waiting on X3D feels like the smart move moving forward. I’m uncertain why anyone should continue caring about their base lineup if they will outdo themselves months later anyways.
that's why it was a non fireworks launch with accompanying fuck ups. they didn't care. those are the chips for Dells and the likes. X3D is the flagship but not their bread and butter if you follow.
@@Zlimee only if you are unimpressed with the performance that you were completely aware of when you ordered. 🙄 and of course it will cost much more. Is that in your budget? silly questions 🤗
I wonder if the end goal is a single large X3D cache that bridges over all CCDs so that 1 core could use all 192mb or all cores can access the same data in the cache no matter what CCD the core is on without the huge latency of going through the I/O die?
It's possible, though that involves a larger die and thus lower yields and thus higher prices. The upside though is that with a larger die you pack significantly more cache, 256mb wouldn't be farfetched. Though diminishing returns could be a thing here.
Zen 6 rumours are for stacked dies so the CCDs talk to the IOD/cache via low latency method like V-cache. This dual V-cache, will NOT run optimally without assigning application tasks to a CCD, the reason is having a lot of data in the other cache will result in a round trip via IOD to fetch it and it's possible that a thread on other CCD will then want to access the same data. On servers system managers know to keep jobs onto a single CCX, from Zen2 days even without the potential problems of swamping the IF by using both links.
@@the_retag They probably could, but that would require either support silicon or complex laser cutting on the ccd dies to get everything to fit together without overhangs. This mean an expensive increase in complexity of assymbly and the higher failure rate that goes with that and also increases costs. Now which is better than the other and what the optimal level of cache I couldn't say, This sort of thing goes beyond what they were teaching in most undergrad EE degrees when I was college 30+ years ago. Indeed on die cache was 8kb on the 486DX2 that came out around then. It was the only on die cache as you only had one core to have a cache for.
AMD came out months ago advising that it will be only on one CCD from memory for this cpu, i remember one AMD staffer quoting it in press. But we can wish.
I didn't pull the trigger for the 7700x when I saw it for $211 waiting for zen5... Now I am waiting for the 9800x3d and you are advising me to wait for the 9950x3d?
the 9800X3D will be good enough for anything you throw at it I'm sure, unless you also do intensive workloads or rendering, but even then I'm sure the 9800X3D will be more than enough to get these done in a reasonable amount of time.
@@chillz739Amd stopped selling the 5800x3d cpus brand new because they know for a fact you are not required to upgrade for years so they discontinued them, Greedy bastards.
The 7950x3D hasn't had scheduling problems in a very long time... literally the only thing you have to do when a new game comes out is open Game Bar settings and select this is a game and that's it from now on it'll run exactly how it's supposed to
i manually schedule my processes using process lasso and it gives me massive gains in fps in cpu intensive games, i put every process on ccd1 (non cache) and only games on ccd0 (v-cache) so they get full 100% of the cores with cache
@zimxh hell yeah I have processe lasso too it works great But Here recently I've noticed that I haven't needed to use it near as much that game bar has been doing a pretty good job with the core parking but I like to Tinker as well and process lasso is an excellent application for that
@sengan2475 when was the last time you had an issue with scheduling on your 7950x3d? I haven't noticed anything in a long time and I usually keep processe lasso open on my second monitor just to keep an eye on my cores and see what they're doing even on brand new games once I check the box in Game Bar no issues.
@@danielabbott9312 I also have a 7950X3D but I am not so sure. Many recent reviews still show that the 7950X3D is slower in some games vs the 7800X3D while the clocks of the vcache chiplet is a little bit higher than the 7800X3D. So that does not seem right.
Overclocking isnt a significant improvement to ever be worth it, not ever. The only reason its advertised is ppl chasing benchmarks like its an addiction, while in real world performance it will be like 10-20fps improvement at a cost of a shorter lifespan. Which in essence is disgusting cause of e-waste and waste of money on good parts that didnt need to be rendered useless just for inflating someone's unimportant ego.
It will not change the situation too much since the inter-CCD latency will still be there. If any data has to be transferred between the ccds, it will still take way more time and way more latency.
thats the same with the non 3DVCache CPU's, and for heavily multithreaded game engines like Capcoms MT Framework and RE Engine it scales really well. I'm not saying that thats not a problem, rather that it will scale the same way the multiccd non 3dvcache cpus scale compared to the single CCD CPUs It makes it more predictable when it comes to performance
@@artorias89 well yes, there are onlx very few games than can use 8Cores / 16Threads and even fewer that also scale well with those threads. Only Games like City Skylines 2, Monster Hunter World or Monster Hunter Wilds (the latter hasnt been released yet just a beta which is no longer available). Those are some of those very few Titles that do scale well with high core counts and 3D V-Cache (there are more games that do prefer more L3 over slightly higher Clocks than there are games that scale with the number of threads). Altough eg MHWorld being an older Title now adays isnt hard to run and runs perfectly fine on a 6 Core CPU Zen3.
@@artorias89 depends uses said game a little more than 8 cores / 16 threads or is it able to spread across all the cores and threads. for example if it were to use only 18 or 20 Threads then the added latency between the ccd's might worsen the over all performance to the point where its slower than having only 8 core / 16 threads and overloading them. even the best scaling games do not double their performance on the cpu side when having 32 Threads compared to just 16. As one limiting factor is when running any multithreaded program you will always have one thread that is used just to schedule the workloads across all the other threads. meaning if that one thread isnt fast enough to coordinate such a large number of tasks across such a large number of threads the it wont go any faster potentially even slower than it would with just 12, 16 or 24 Threads (not to mention 1 or 2 threads are also being used by the OS, Linux switches automaticallly to the least utilized while windows is mor static and tends to run on thread 0)
@@funkydmunky1 Well, since I will much more than likely have both well before you, there's no "copium" here. You seem quite narcissistic and you might wish to pursue counseling, but then again since your I.Q. is likely about or less than 90, I wouldn't waste money on it.
@@molly_mallard probably doesn't even have the 7800x3d. Enjoy you CPU, I got the same and it's a little beast and doesn't break a sweat on anything so I'm eating good for a good while. Granted I'm running a 7900xtx, everything so bury another so no worries there
Honestly i dont really know enough about processors to make a good descission based on various things. So my question being this. If i have the Intel i9 14900K currently and im thinking about getting the AMD 9950X3D how much more performance would i be getting? also how much more frame rate?
I'm waiting for the 9950X3D, though I'm building an entirely new system for 4k gaming. I remember watching Jays video a while back about sanding and lapping the IHS down for improved thermals. I might try something out like that for some additional overclocking. I'm open to some options, just depends on what it's like outta the box.
from having a 7950x3d ( which died ) and then moving to a 7800x3d as a replacement I miss the extra cores when starting games up and loading also when installing the updates and games
I personally had the 3950X & moved to the 7900X as I didn't need 16 cores. If the 9900X3D has VCache on both CCD's that might be my end game for a few years!
I've been very happy with my 5900X (sometimes I'll use more than 8 cores and my GPU isn't fast enough to make a CPU bottleneck much of a concern) so I can't really see myself upgrading unless AMD releases a 9900X3D with V-cache on both CCDs and/ or I upgrade my GPU... I do really think it would be a huge step in the right direction if AMD does double X3D so people who like to game can get the 12+ cores without as many scheduling headaches, and the normal X (non 3D) will still be available for people who don't care about it.
Even if both ccd have vcache, there is still the cross-ccd latency issue in multithreaded workloads. Assuming they haven't found a way to unify the cache or allow both ccd direct access to the entire cache, there will still be that latency caused by one ccd accessing cache located on the other ccd. It's still faster than going to system memory, but it might be a problem for performance in games that a single ccd processor wouldn't experience. There might still be a scheduler fix that forces all of a game's threads onto a single ccd to avoid this. But then you still have the question, why not just get a 9800x3d if games are only going to use those 8 cores anyway?
There are very rarely cases that multiple threads work on shared data. Normally each thread works on its own chunk of the data. And if you pin the threads to certain cores, they won't have to move from core to core either.
@AngelicStreak there are cases where one thread generates data to be used by the other threads. It's not the most widely used strategy, but when it comes up, it could cause latency. And yeah pinning a thread to a single core/ccd is a good idea. I wonder if they'll need another software solution for that, because Windows will shuffle threads around if left on its own.
thats the same with the non 3DVCache CPU's, and for heavily multithreaded game engines like Capcoms MT Framework and RE Engine it scales really well. You could check out CPU benchmarks for the various MonsterHunter Games (except stories/stories2 that one has a 60fps hardcoded cap) I'm not saying that thats not a problem, rather that it will scale the same way the multiccd non 3dvcache cpus scale compared to the single CCD CPUs It makes it more predictable when it comes to performance scaling.
@@limeisgaming yeah I know it does the same thing on other multi-ccd cpus, but now I'm wondering how much benefit you might actually get from forcing windows to avoid crossing ccds when shuffling threads around.
@@kevin9218 These cases where you do data pipelining with multiple threads would actually greatly benefit from shared L3 cachem which I believe would be the case with this X3D cache. It would be shared among all 16 cores. The inter-CCD communication would happen only for registries and L1/L2 cache moves.
The placement of the cache die wouldn't matter for single core performance, if you think heat is the only problem. Single core speed is lower because the voltage rail is shared with the CCD and cache, and the cache can't handle more than 1.3v. All core clocks operate with lower voltages, but the collective heat from them all is why X3D has lower all-core clocks. Lastly the scheduling issue never had anything to do with asymmetrical cache layout, because it's been happening on R9's before X3D existed. Windows normally attempts to balance light loads across the entire CPU bumping threads to the next fastest available core. It's a non-issue when all CPUs were monolithic, which is how Windows is treating R9s. The CCD opposite of where the threads spawn, will have no relevant data in it's cache whatsoever, but the scheduler can dynamically push a thread there trying to load balance. This is what happens without core parking to invalidate the opposite CCD, so adding the cache to the other CCD fixes nothing. The 7950X3D makes sense if need the cores but you're willing to split the difference of losing around 8% all-core performance with what you get from the cache in games. Without one of those standard dies being faster you'd be losing 16%+, which would do more to invalidate the 16 core than anything else. Lightly threaded productivity still gets up to 5.7Ghz clocks. Maybe 5.5Ghz max boost on a dual v-cache 9950X3D is still a reasonable compromise.
I'm very happy with my 7950x3d =D Zen 5% will not take it's glory away, i waited 7 months to buy it, yet i have it 14 months already, more then a year, and my old cpu (wich i considered 3 cpu generation upgrading already but didnt) was really needing an upgrade. The 9950x3d doesnt cut it. Other then clock speed increase and some saving if CCD 1 cache goes to CCD2 cache, this cpu will NOT be faster. Amd has 5950x3d dual chiplet, they have reason not to release it. By my experience with 7950x3d if the game moves to another CCD complex, the info in cache gets removed and written in cache/ram if to small of other CCD, that alone takes latency, that he cache is insanely fast could safe it some grace, but it's still a latency penalty. Also 'double the cache' is marketing, unless scheduling (the thing you are so mad on, so dont get your hopes up) gets improved, and each CCD can seperately control different things, the cache will have lot over overlaps and thus not double in size but be more of a copy register.
If that's the price, going for the 9800x3d makes more sense, question is of course, what will be the difference be between the 9800x3d and 9950x3d then?
If they bring dual 3D cache it will cause latency, because trip across 2 CCDs is still longer than single CCD This is why 9950X still games on first 8 cores and not on dual chiplets I know it, because I own both 7950x3D and 9950X and they game exclusively on first chiplet Adding 3D cache to both is a waste, as there is no scheduling for both chiplets on games
are you on windows or linux? Linux tends to scale better in gaming in my experience with multi ccd cpus compared to windows. -> my mothers PC has a 5600X -> my Sisters PC has a 7600X -> my old PC had a 5950X -> my current has a 7800X3D I was able to test on all of them on Linux with the same Kernel version except for the difference that the Zen4 CPUs used the Kernel build for x86-64_v4 while Zen3 CPU's used x86-64_v3 (because Zen4 has AVX512 and Zen3 doesnt, Its possible to use the v3 Kernel for Zen4 but it means you cant use AVX512, for Zen4 it doesnt make as much of a difference as it would with Zen5 since Zen4 emulates AVX512 by Concatinating the two AVX2 FMA's and Ports using a mask into 1 AVX512 FMA and Port, Zen5 on the other hand has true native AVX512 capabilities and 2 ports and FMA's per Core) had a 5950x before i got a 7800X3d and in some games liky Cities skyline 2 and MonsterHunter world the 5950X actually outperformed the 7800X3D (In MHWorld it didnt by absolute FPS but by powerefficiency, which is kind of weird)
Any game trying to use more than 8 cores was always going to perform horribly no matter if the second CCD has V-Cache or not. The 7950X3D is definitely better off using one non-3D CCD, it gets you a top binned 7700X in the same package as your 7800X3D, and background tasks were never going to use V-Cache anyway. The only reason to not carry this strategy forward to Zen5 is because the basic Zen5 cores have no advantage over Zen4 so the 9950X3D really would be pointless if they don't put 3D V-Cache on both CCDs.
The issue that exists with dual CCDs in games is more specifically a thing because games aren't needing more than 16 threads. It would be less problematic for games that do use more. Right now without core parking Windows is treating the CPU as if it's monolithic and will bump a thread to any available core to load balance across the entire CPU. None of the data it expects is in the L3 in the other CCD, but it checks for it, fails to find it, and then has to fetch it. Core parking stops the random excursion of threads to a core devoid of relevant data. If you have a game using more than 8c/16t, then the other CCD is constantly filled and there less idle space for threads to as arbitrarily get bumped around. StarCitizen for example is capable of using around 20 threads and is fine on Ryzen 9 CPUs. It needed a patch to specifically ignore Intel's E-cores because the huge difference in IPC and clocks caused really bad constant stuttering. The 7950X3D is about 8% slower in all-core productivity, but up to 20% faster in games than the 7950X. The all-core drop is basically linear to the clock drop of 1 CCD. The 16 core 3D wouldn't make sense to lose 16%+ by slowing down both CCDs. The scheduling issue still exists which is why standard Zen 5 R9s get the core parking driver. So a dual v-cache 9950x3D would still be no benefit for gamers, while making the entire package slower for nearly everything else. With higher overall clocks it would be less slow than if they did it with Zen 4, so maybe that makes it more acceptable. You don't go for the 16 core if you literally only care about gaming, but you would want the least hit to performance everywhere else you'd get the cores for.
@@blkspade23 The one gaming application for 16 core CPUs is Shader Compilation, there's been at least a few games where I've heard about people waiting 10+ minutes for a Zen3 CPU to finish compilation where my 7950X3D crushes the same workload in under a minute. I would like to keep that capability when I upgrade to Zen5, but just getting 9800X3D comes with peace of mind that it cannot go wrong.
Hmm, get a CPU that will still be bottlenecked by a 4090 and likely a 5090 at 4K, or wait two and a half months to get a CPU that will still be bottlenecked by the 4090 and likely a 5090 that will cost more? Decisions, decisions... How about: Get a 9800X3D, which will be more than relevant for at least a few years, and then at the end of the AM5 lifecycle, slot in the last X3D that is released that will work with it?
I have been flipping back and forth on my current computer build. Originally I planned on AMD, but was put off by high thermals. Turned to Intel, just to find their chips actually degrading with physical damage to the cpu. Back to AMD and ran into instability, latency, and core parking. Took a hard look at new Intel Core ultra 9 285k for cooler thermals-but appears to be a dead-end for socket and cpu line. So I am waiting once more. I am praying that AMD comes through with two large caches. I plan to have this new computer for the long haul and being able to upgrade the cpu sometime down the road sounds rather enticing.
Honestly if they do put the 9950x3d dual ccd dual v cache and make it over clock capable??? You're talking a cpu they can produce and sell at whim until someone else makes something better. They can put more time into RnD before their next line because nothing would compare to that for workload or gaming at that point. They would have 5+years to develop. I'd replace all my chips to that. I run a team of esports gaming and streaming pcs. Please make my workload a little easier I'm beGGing you AMD
As someone who's had the 7950X3D since day one, the challenges are there with the CPU, but once you get it to sing, it's awesome. I have customized my entire pc to work around those cores. Things that are meant for nonX3D cores run in the background so the X3D cores run through the game flawlessly. I think it comes down to how big of a nerd you are for gaming. I love AMD tho, I will never go Intel for CPUs
I'd considering upgrading from my 7800X3D if all of these numbers ended up being close to the actual product. That's a big overall performance gain. I could probably recoup half the cost by selling my 7800X3D.
I've seen other sites that for gaming the difference is about 8%....also depending on you gpu you might be maxed already...about 18% difference in other benchmarks/encoding. I would stay where you are. just my 2 cents
Companies love ppl who want to upgrade every single new iteration, for sales. But in reality doing that is just trolling everyone else who doesnt and waits a proper long while then ends up with inflated prices cause ppl want to get the newest even when they dont need it.
I might have to skip the 9800X3D and go straight to the 9950X3D for Microsoft Flight Simulator 2024. These chips ought to do well on CUDIMM 9000MT/s RAM then the crappy 6400MT/s RAM .
i just want the 9950x3d to be as good in games or slightly better than 9800x3d and ill get it. Need to replace my 14900k and just use it to render and crap in the background secondary cpu.
If ur on top tier the only replacement necessary currently is how to think. Consuming products unnecessarily is not going to fill the void where thoughts are supposed to go
I don't think AMD would not have at least 1 cpu with X3D-Vcahe on all CCD's now. If AMD doesn't have 1 cpu with Vache on all CCD's now ; Good bye AMD hello Intel. This is AMD's BIG chance to recover from the earlier initial release of the 9k series and really put Intel further into the ground within a Pine-Box.
As much as I hope it’s true I don’t think they will add 3D cache to both CCDs because they will literally render their own 9950x useless, no one will buy it if they can a 9950x3D with the cache on both sides. I just don’t see AMD shooting themselves in the foot like that unless they make the 9950x3D VERY expensive as to not destroy the standard 9950x sales. But hey I guess when your only competition is yourself you get to ponder these questions lol.
This makes no sense whatsoever. The 9950X would still be the best choice for workstation tasks which don't benefit from extra cache, regardless of the 9950X3D has 1 CCD with Vcache or 2. The 9950X will be cheaper and will be able to clock faster across all cores. The 7950X has a similar advantage over the 7950X3D in workstation tasks.
@ this new x3d version will clock higher with the re design of how the 3d cache is implemented. The gap in clock speed between the 9950x and 9950x3d will not be as far apart as they were on last gen Ryzen 7000 thus making them much closer in workstation tasks then they were before. Did you even watch the video ?
Maybe in 4 or 5 years. Not liking the direction computers are heading they are already drawing power like a microwave oven at this point. I'm sitting at 850 watts on AM4 platform and not interested in going above that.
Zen 5 X3D is shaping up to be a beast in gaming and productivity. Especially in gaming. Zen 5 X3D is said to have significantly higher clock speeds than Zen 4 X3D, so that will substantially boost gaming performance. Gaming performance tends to scale well with higher clocks. A reason why Zen 5 X3D is expected to have considerably higher clocks is because the 3D V cache will be stacked below the CCDs, unlike the prior generations, which had 3D V cache stacked above the CCDs. This will significantly improve thermal temperatures and should address heat dissipation issues. All of this should allow the 3D V cache cores to clock just as high as the non-3D V cache cores. 3D V cache and high clock speeds massively boost gaming, and getting that in one package in Zen 5 X3D has me hyped. Zen 5 X3D also has an unlocked multipler for overclocking. This all looks really promising. Hopefully, they won't cost more than the Zen 4 X3D CPUs did.
I want dual socket 128 core EPYC 9785X with 3D V Cache and overclockable with 24 TB *(48 DIMMS on the motherboard ) 24 channel (12 channel per socket ) DDR5 6000 MHz ECC CL30 support with CXL support up to 2 petabytes of additional DDR5 CXL 6000 MHz ECC CL30 RAM and 8 RDNA5 Workstation flagship GPUs each having 384 GB of 48 Gbps GDDR7X VRAM (3 GCDs per GPU, 8192 SPs per GCD) and 64X Sabrent Rocket 5 8TB PCIE5 M2 NVME SSDs on 2 double sided M2 adapter cards each side having 16 M2 slots. 24 Kioxia Enterprise 128 TB PCIE5.0 SSDs all in E-ATX form factor without watercooling and with Dynatron CPU fans and 6X 8K 240 Hz monitors which can convert into 4K 480Hz and 1080p 960Hz on demand via software. all powered by 16X 2500 Watt FSP PSU. my case should support 8X 3 slot GPUs, 24 2.5 inch SSD bays and 64 140mm fans which run at 18450 RPM each and have zero decibels.
Even if this fake news about 9950x3d is true, switching from 9800x3d to that is by definition NOT an upgrade. I wanna see a blind test of you telling the difference between a 50fps boost on a 120+ avg fps range b4 i believe the claim that it should be considered an upgrade. And vsync has to be enabled cause counting disgusting screen tearing in non vsync is easy to tell apart
Yeah me too i really want to see these 2 Configs on Desktop, I mean they already did 4+8 on Laptop so where are them hybrid CPUs for Desktop ? Dual CCD Zen5 + Zen5c 6 + 12 as well as a CPU that has a single CCD but with 16 zen5c Cores just like the Biggest Epyc Server CPU has
@@limeisgaming Oooh I didn't know they were already making Zen5c chiplets... I'll take 8 core Zen5X3D with 16 core Zen5c. The only thing extra cores gets me is reduced shader compilation time, but I hate shader compilation so this configuration would definitely win as the best all around gaming CPU.
Theres no logic to putting vcache on both CCDs when it will go to waste due to cross CCD latency penalties. Even the current 9950X parks one of the CCDs to avoid this issue. Your reasoning doesn't make any sense for this very reason, the current non vcache 9950X already requires AMD software parks one of the CCD during gaming.
Will most likely be the last cpu generation where AMD are leading in gaming. Intel are going to catch up very quickly, which isnt a surprise. AMD are the massive underdogs to intel at the end of the day. They cant even make their own cpus 😂
I don't believe that, in the short term, AMD have been better since Zen3 and Intel with the same architecture for 10 years, making compromises on processors even adding fake "efficiency cores" that are useless on desktops. It does not compete with TR or Epycs.
@mojojojo6292 just look up backside power delivery. Intel has been developing 2nm 20A backside power delivery cpus which have superior power delivery and efficiency. TSMC who make AMD cpus are 2 to 3 years behind in the technology. Intel are going to have a huge lead in the cpu world. Yes intel cpus had a flaw. But this was to do something with the current 10nm older architecture node (lg 1700 socket V) which was naturally inefficient compared to TSMC's current 5nm node, and they pushed it to the limit. Intel are currently using TSMC chips with a 3nm node compared to AMD ryzen 9's 4nm, which is why intel ultra cpus are now more efficient, but performance is on par.
AMD will be releasing the Ryzen 9 9950X3D likely during CES 2025, & it looks good.
Thanks to Vip-cdkdeals for sponsoring this video. 30% off code: GPC20
▬ Windows 10 pro ($17):www.vip-cdkdeals.com/vck/GPC20w10
▬ Windows 11 pro($23):www.vip-cdkdeals.com/vck/gpc20w11
▬ Windows 10 home ($14):www.vip-cdkdeals.com/vck/gpc20wh
▬ Office 2016($28):www.vip-cdkdeals.com/vck/gpc20of16
▬ Office 2019($47):www.vip-cdkdeals.com/vck/gpc20off19
Sources
AMD Ryzen 9950X3D dual X3D Vcache Leak: www.techpowerup.com/327057/amd-ryzen-9-9950x3d-and-9900x3d-to-feature-3d-v-cache-on-both-ccd-chiplets
AMD Ryzen 9800X3D Official Launch: ruclips.net/video/ZTMBqNbHUrs/видео.html
No one has bench-tested a 9950X3D, You have no idea what it can or cannot do, AMD lied about AM5 and you shouldn't just blurt out performance numbers.
Even without benches we know it will completely destroy Arrow Lake.
@low_etc In gaming. Not so much in everything else. Not to mention Arrow Lake needs CUDIMM ram to run at its full potential which allows it to actually beat the 7800x3D/7950x3D in some games. There also seems to be an issue with Windows hampering Arrow Lake performance, we need to wait a few more weeks to maybe even a couple months to really guage how well Arrow Lake actually is since it's a paradigm shift and a gen 1 product for Intel that still has quirks and bugs that need ironing out.
@@FastGPU intel never shows bias in their performance numbers vs amd?
@@03chrisv lol bunch of deluded intel fanboy
@03chrisv rofl intel fan boy grasping at straws! Yeah in two games, and in specific sections of those games, the Intel chip has faster 1 percent lows!
I want one monster CPU that doesnt differ between work and gaming. I hate that I have to choose between X3D and X, gaming/work. Just make one monster CPU. One CPU to rule them all.
Is it that hard to make a cpu that just consumes 60watts of power no matter the workload and saves you tons of money on your electric bill
@@jasonvors1922 I think most cpus have an eco mode where max power draw will be like 65 watts while holding on to most its performance
How hard is it to base your choice on what your usage is? Do you game more often than run office suite software? AMD. Do you run productivity software more than game? Still AMD so you are not abusing the environment 😅
@@jasonvors1922 I mean for some reason people complained when 9000 series used less watts which is weird cos 1 those chips aren't their gaming chips and are getting similar performance at half the wattage its like hows that a bad thing?
5950X, but it runs hot. I might downgrade to the 5900XT.
Waiting on X3D feels like the smart move moving forward. I’m uncertain why anyone should continue caring about their base lineup if they will outdo themselves months later anyways.
I know I don't.
should i cancel my 9950X? i do 3D and gaming so will 9950X3D will really be that much better?
that's why it was a non fireworks launch with accompanying fuck ups. they didn't care. those are the chips for Dells and the likes. X3D is the flagship but not their bread and butter if you follow.
@@Zlimee only if you are unimpressed with the performance that you were completely aware of when you ordered. 🙄 and of course it will cost much more. Is that in your budget? silly questions 🤗
non-x3d are really good for non gaming apps obviously.
I wonder if the end goal is a single large X3D cache that bridges over all CCDs so that 1 core could use all 192mb or all cores can access the same data in the cache no matter what CCD the core is on without the huge latency of going through the I/O die?
i don't know
It's possible, though that involves a larger die and thus lower yields and thus higher prices. The upside though is that with a larger die you pack significantly more cache, 256mb wouldn't be farfetched. Though diminishing returns could be a thing here.
@@kaseyboles30 they could make a long and slim cache die bridging the compute dies
Zen 6 rumours are for stacked dies so the CCDs talk to the IOD/cache via low latency method like V-cache.
This dual V-cache, will NOT run optimally without assigning application tasks to a CCD, the reason is having a lot of data in the other cache will result in a round trip via IOD to fetch it and it's possible that a thread on other CCD will then want to access the same data. On servers system managers know to keep jobs onto a single CCX, from Zen2 days even without the potential problems of swamping the IF by using both links.
@@the_retag They probably could, but that would require either support silicon or complex laser cutting on the ccd dies to get everything to fit together without overhangs. This mean an expensive increase in complexity of assymbly and the higher failure rate that goes with that and also increases costs. Now which is better than the other and what the optimal level of cache I couldn't say, This sort of thing goes beyond what they were teaching in most undergrad EE degrees when I was college 30+ years ago. Indeed on die cache was 8kb on the 486DX2 that came out around then. It was the only on die cache as you only had one core to have a cache for.
AH yes if this isnt my fav clickbait channel.
AMD came out months ago advising that it will be only on one CCD from memory for this cpu, i remember one AMD staffer quoting it in press. But we can wish.
I didn't pull the trigger for the 7700x when I saw it for $211 waiting for zen5... Now I am waiting for the 9800x3d and you are advising me to wait for the 9950x3d?
If you don’t need the extra cores just get the 9800x3d if anything happens you can sell for 90% price and buy the r9
just do it NIKE
the 9800X3D will be good enough for anything you throw at it I'm sure, unless you also do intensive workloads or rendering, but even then I'm sure the 9800X3D will be more than enough to get these done in a reasonable amount of time.
BUY! BUY! BUY!
@@chillz739Amd stopped selling the 5800x3d cpus brand new because they know for a fact you are not required to upgrade for years so they discontinued them, Greedy bastards.
The 7950x3D hasn't had scheduling problems in a very long time... literally the only thing you have to do when a new game comes out is open Game Bar settings and select this is a game and that's it from now on it'll run exactly how it's supposed to
i manually schedule my processes using process lasso and it gives me massive gains in fps in cpu intensive games, i put every process on ccd1 (non cache) and only games on ccd0 (v-cache) so they get full 100% of the cores with cache
@zimxh hell yeah I have processe lasso too it works great But Here recently I've noticed that I haven't needed to use it near as much that game bar has been doing a pretty good job with the core parking but I like to Tinker as well and process lasso is an excellent application for that
false
@sengan2475 when was the last time you had an issue with scheduling on your 7950x3d? I haven't noticed anything in a long time and I usually keep processe lasso open on my second monitor just to keep an eye on my cores and see what they're doing even on brand new games once I check the box in Game Bar no issues.
@@danielabbott9312 I also have a 7950X3D but I am not so sure. Many recent reviews still show that the 7950X3D is slower in some games vs the 7800X3D while the clocks of the vcache chiplet is a little bit higher than the 7800X3D. So that does not seem right.
lies for views..
People jumping from 7800x3D to the 9800x3D need to go to the doctor.
Yea, people don't want to jump from 5800X3D to the 9800X3D...
why? its overclockable unlike the 7800X3D - could be a game changer
I don't want them to buy so I can get one on 7th 😂
Overclocking isnt a significant improvement to ever be worth it, not ever. The only reason its advertised is ppl chasing benchmarks like its an addiction, while in real world performance it will be like 10-20fps improvement at a cost of a shorter lifespan. Which in essence is disgusting cause of e-waste and waste of money on good parts that didnt need to be rendered useless just for inflating someone's unimportant ego.
@jordan-mn6yy your intention is to keep the the CPU 10 years? Then why buy new stuff at all when all you need is second hand gear?
Can't wait for this CPU. Going to swap out the 7800x3d for 9950x3d. Star citizen eats all the cores
Yeah, Star Citizen is one of the very few games that justify a 16-core CPU. Have fun.
That Game is not Optimized.
RYZEN 7 7800X3D Was basically on par with the 7950X3D in Starcitizen.
I’ve got a Porsche 😬
Bless you.
You can have a cookie too
Truly sigma
noic
@@scoobpower i dont have one....
It will not change the situation too much since the inter-CCD latency will still be there. If any data has to be transferred between the ccds, it will still take way more time and way more latency.
thats the same with the non 3DVCache CPU's, and for heavily multithreaded game engines like Capcoms MT Framework and RE Engine it scales really well.
I'm not saying that thats not a problem, rather that it will scale the same way the multiccd non 3dvcache cpus scale compared to the single CCD CPUs
It makes it more predictable when it comes to performance
so for gaming the 9800x3d would be better or?
@@artorias89 well yes, there are onlx very few games than can use 8Cores / 16Threads and even fewer that also scale well with those threads.
Only Games like City Skylines 2, Monster Hunter World or Monster Hunter Wilds (the latter hasnt been released yet just a beta which is no longer available).
Those are some of those very few Titles that do scale well with high core counts and 3D V-Cache (there are more games that do prefer more L3 over slightly higher Clocks than there are games that scale with the number of threads). Altough eg MHWorld being an older Title now adays isnt hard to run and runs perfectly fine on a 6 Core CPU Zen3.
@@limeisgaming let say the those games use more than 8 cores, i thought alone because of the 2x ccds it would be worse
@@artorias89 depends uses said game a little more than 8 cores / 16 threads or is it able to spread across all the cores and threads.
for example if it were to use only 18 or 20 Threads then the added latency between the ccd's might worsen the over all performance to the point where its slower than having only 8 core / 16 threads and overloading them.
even the best scaling games do not double their performance on the cpu side when having 32 Threads compared to just 16.
As one limiting factor is when running any multithreaded program you will always have one thread that is used just to schedule the workloads across all the other threads. meaning if that one thread isnt fast enough to coordinate such a large number of tasks across such a large number of threads the it wont go any faster potentially even slower than it would with just 12, 16 or 24 Threads
(not to mention 1 or 2 threads are also being used by the OS, Linux switches automaticallly to the least utilized while windows is mor static and tends to run on thread 0)
They should start making 12/16 core chiplets.
My 7950X3D/4090 runs every game I have with ease and both temps are quite low under load. I run all with ultra settings @ 4k 144hz.
Ahh. So your copium is to post your irrelevance before the 9950X3D and 5090 ship?
Tell that to your loved ones - because no one here cares about your ol' CPU.
@@funkydmunky1 Well, since I will much more than likely have both well before you, there's no "copium" here. You seem quite narcissistic and you might wish to pursue counseling, but then again since your I.Q. is likely about or less than 90, I wouldn't waste money on it.
@@HamselGG What do you have?
@@molly_mallard probably doesn't even have the 7800x3d. Enjoy you CPU, I got the same and it's a little beast and doesn't break a sweat on anything so I'm eating good for a good while. Granted I'm running a 7900xtx, everything so bury another so no worries there
Move to 3:50 when he actually starts talking about 9950X3d.
Honestly i dont really know enough about processors to make a good descission based on various things. So my question being this. If i have the Intel i9 14900K currently and im thinking about getting the AMD 9950X3D how much more performance would i be getting? also how much more frame rate?
ALOT more on cod i know that much
I'm waiting for the 9950X3D, though I'm building an entirely new system for 4k gaming. I remember watching Jays video a while back about sanding and lapping the IHS down for improved thermals. I might try something out like that for some additional overclocking. I'm open to some options, just depends on what it's like outta the box.
from having a 7950x3d ( which died ) and then moving to a 7800x3d as a replacement I miss the extra cores when starting games up and loading also when installing the updates and games
I personally had the 3950X & moved to the 7900X as I didn't need 16 cores. If the 9900X3D has VCache on both CCD's that might be my end game for a few years!
I've been very happy with my 5900X (sometimes I'll use more than 8 cores and my GPU isn't fast enough to make a CPU bottleneck much of a concern) so I can't really see myself upgrading unless AMD releases a 9900X3D with V-cache on both CCDs and/ or I upgrade my GPU...
I do really think it would be a huge step in the right direction if AMD does double X3D so people who like to game can get the 12+ cores without as many scheduling headaches, and the normal X (non 3D) will still be available for people who don't care about it.
Even if both ccd have vcache, there is still the cross-ccd latency issue in multithreaded workloads. Assuming they haven't found a way to unify the cache or allow both ccd direct access to the entire cache, there will still be that latency caused by one ccd accessing cache located on the other ccd. It's still faster than going to system memory, but it might be a problem for performance in games that a single ccd processor wouldn't experience. There might still be a scheduler fix that forces all of a game's threads onto a single ccd to avoid this. But then you still have the question, why not just get a 9800x3d if games are only going to use those 8 cores anyway?
There are very rarely cases that multiple threads work on shared data. Normally each thread works on its own chunk of the data. And if you pin the threads to certain cores, they won't have to move from core to core either.
@AngelicStreak there are cases where one thread generates data to be used by the other threads. It's not the most widely used strategy, but when it comes up, it could cause latency. And yeah pinning a thread to a single core/ccd is a good idea. I wonder if they'll need another software solution for that, because Windows will shuffle threads around if left on its own.
thats the same with the non 3DVCache CPU's, and for heavily multithreaded game engines like Capcoms MT Framework and RE Engine it scales really well. You could check out CPU benchmarks for the various MonsterHunter Games (except stories/stories2 that one has a 60fps hardcoded cap)
I'm not saying that thats not a problem, rather that it will scale the same way the multiccd non 3dvcache cpus scale compared to the single CCD CPUs
It makes it more predictable when it comes to performance scaling.
@@limeisgaming yeah I know it does the same thing on other multi-ccd cpus, but now I'm wondering how much benefit you might actually get from forcing windows to avoid crossing ccds when shuffling threads around.
@@kevin9218 These cases where you do data pipelining with multiple threads would actually greatly benefit from shared L3 cachem which I believe would be the case with this X3D cache. It would be shared among all 16 cores. The inter-CCD communication would happen only for registries and L1/L2 cache moves.
I wonder what is going to happen if AMD stacked Vcash both above and below the CCD!! 😮
The placement of the cache die wouldn't matter for single core performance, if you think heat is the only problem. Single core speed is lower because the voltage rail is shared with the CCD and cache, and the cache can't handle more than 1.3v. All core clocks operate with lower voltages, but the collective heat from them all is why X3D has lower all-core clocks. Lastly the scheduling issue never had anything to do with asymmetrical cache layout, because it's been happening on R9's before X3D existed. Windows normally attempts to balance light loads across the entire CPU bumping threads to the next fastest available core. It's a non-issue when all CPUs were monolithic, which is how Windows is treating R9s. The CCD opposite of where the threads spawn, will have no relevant data in it's cache whatsoever, but the scheduler can dynamically push a thread there trying to load balance. This is what happens without core parking to invalidate the opposite CCD, so adding the cache to the other CCD fixes nothing. The 7950X3D makes sense if need the cores but you're willing to split the difference of losing around 8% all-core performance with what you get from the cache in games. Without one of those standard dies being faster you'd be losing 16%+, which would do more to invalidate the 16 core than anything else. Lightly threaded productivity still gets up to 5.7Ghz clocks. Maybe 5.5Ghz max boost on a dual v-cache 9950X3D is still a reasonable compromise.
if u are just a gamer, 9800x3d is what u want
All just pure speculation.
Yup!
I plan on building a 9950X3D with the best motherboard I can find so it survives life in my rack.
Time to finally say goodbye to Threadripper.
MSI Godlike!
X870e Nova Wifi is insane value and sells out as soon as it's back in stock
I'm very happy with my 7950x3d =D Zen 5% will not take it's glory away, i waited 7 months to buy it, yet i have it 14 months already, more then a year, and my old cpu (wich i considered 3 cpu generation upgrading already but didnt) was really needing an upgrade. The 9950x3d doesnt cut it.
Other then clock speed increase and some saving if CCD 1 cache goes to CCD2 cache, this cpu will NOT be faster. Amd has 5950x3d dual chiplet, they have reason not to release it. By my experience with 7950x3d if the game moves to another CCD complex, the info in cache gets removed and written in cache/ram if to small of other CCD, that alone takes latency, that he cache is insanely fast could safe it some grace, but it's still a latency penalty. Also 'double the cache' is marketing, unless scheduling (the thing you are so mad on, so dont get your hopes up) gets improved, and each CCD can seperately control different things, the cache will have lot over overlaps and thus not double in size but be more of a copy register.
If that's the price, going for the 9800x3d makes more sense, question is of course, what will be the difference be between the 9800x3d and 9950x3d then?
If they bring dual 3D cache it will cause latency, because trip across 2 CCDs is still longer than single CCD
This is why 9950X still games on first 8 cores and not on dual chiplets
I know it, because I own both 7950x3D and 9950X and they game exclusively on first chiplet
Adding 3D cache to both is a waste, as there is no scheduling for both chiplets on games
are you on windows or linux?
Linux tends to scale better in gaming in my experience with multi ccd cpus compared to windows.
-> my mothers PC has a 5600X
-> my Sisters PC has a 7600X
-> my old PC had a 5950X
-> my current has a 7800X3D
I was able to test on all of them on Linux with the same Kernel version except for the difference that the Zen4 CPUs used the Kernel build for x86-64_v4 while Zen3 CPU's used x86-64_v3 (because Zen4 has AVX512 and Zen3 doesnt, Its possible to use the v3 Kernel for Zen4 but it means you cant use AVX512, for Zen4 it doesnt make as much of a difference as it would with Zen5 since Zen4 emulates AVX512 by Concatinating the two AVX2 FMA's and Ports using a mask into 1 AVX512 FMA and Port, Zen5 on the other hand has true native AVX512 capabilities and 2 ports and FMA's per Core)
had a 5950x before i got a 7800X3d and in some games liky Cities skyline 2 and MonsterHunter world the 5950X actually outperformed the 7800X3D (In MHWorld it didnt by absolute FPS but by powerefficiency, which is kind of weird)
Ill just keep my 5800X3D for another year..
Any game trying to use more than 8 cores was always going to perform horribly no matter if the second CCD has V-Cache or not.
The 7950X3D is definitely better off using one non-3D CCD, it gets you a top binned 7700X in the same package as your 7800X3D, and background tasks were never going to use V-Cache anyway.
The only reason to not carry this strategy forward to Zen5 is because the basic Zen5 cores have no advantage over Zen4 so the 9950X3D really would be pointless if they don't put 3D V-Cache on both CCDs.
The issue that exists with dual CCDs in games is more specifically a thing because games aren't needing more than 16 threads. It would be less problematic for games that do use more. Right now without core parking Windows is treating the CPU as if it's monolithic and will bump a thread to any available core to load balance across the entire CPU. None of the data it expects is in the L3 in the other CCD, but it checks for it, fails to find it, and then has to fetch it. Core parking stops the random excursion of threads to a core devoid of relevant data. If you have a game using more than 8c/16t, then the other CCD is constantly filled and there less idle space for threads to as arbitrarily get bumped around. StarCitizen for example is capable of using around 20 threads and is fine on Ryzen 9 CPUs. It needed a patch to specifically ignore Intel's E-cores because the huge difference in IPC and clocks caused really bad constant stuttering.
The 7950X3D is about 8% slower in all-core productivity, but up to 20% faster in games than the 7950X. The all-core drop is basically linear to the clock drop of 1 CCD. The 16 core 3D wouldn't make sense to lose 16%+ by slowing down both CCDs. The scheduling issue still exists which is why standard Zen 5 R9s get the core parking driver. So a dual v-cache 9950x3D would still be no benefit for gamers, while making the entire package slower for nearly everything else. With higher overall clocks it would be less slow than if they did it with Zen 4, so maybe that makes it more acceptable. You don't go for the 16 core if you literally only care about gaming, but you would want the least hit to performance everywhere else you'd get the cores for.
@@blkspade23 The one gaming application for 16 core CPUs is Shader Compilation, there's been at least a few games where I've heard about people waiting 10+ minutes for a Zen3 CPU to finish compilation where my 7950X3D crushes the same workload in under a minute.
I would like to keep that capability when I upgrade to Zen5, but just getting 9800X3D comes with peace of mind that it cannot go wrong.
@@blkspade23 thanks bro
Hmm, get a CPU that will still be bottlenecked by a 4090 and likely a 5090 at 4K, or wait two and a half months to get a CPU that will still be bottlenecked by the 4090 and likely a 5090 that will cost more? Decisions, decisions...
How about: Get a 9800X3D, which will be more than relevant for at least a few years, and then at the end of the AM5 lifecycle, slot in the last X3D that is released that will work with it?
I have been flipping back and forth on my current computer build. Originally I planned on AMD, but was put off by high thermals. Turned to Intel, just to find their chips actually degrading with physical damage to the cpu. Back to AMD and ran into instability, latency, and core parking. Took a hard look at new Intel Core ultra 9 285k for cooler thermals-but appears to be a dead-end for socket and cpu line. So I am waiting once more. I am praying that AMD comes through with two large caches. I plan to have this new computer for the long haul and being able to upgrade the cpu sometime down the road sounds rather enticing.
when will the 9950X3D release?
dual 3d v-cache ccd is what I'm waiting for 😍
I hope Microsoft fixes the windows 11 "BUG" that slows down amd cpus. i want my 7800x3D speed back!
@@erdwinjc I think they did, in 24H2.
Install win 10
Honestly if they do put the 9950x3d dual ccd dual v cache and make it over clock capable??? You're talking a cpu they can produce and sell at whim until someone else makes something better. They can put more time into RnD before their next line because nothing would compare to that for workload or gaming at that point. They would have 5+years to develop. I'd replace all my chips to that. I run a team of esports gaming and streaming pcs. Please make my workload a little easier I'm beGGing you AMD
I am undecided whether to get the ryzen 7 9800x3d or the 9 9950x3d. I use the PC for gaming and when I have time I do some streaming
I'm happy with my 7800X3D but who knows If I get the itch, I might upgrade to Zen 6 :)
As someone who's had the 7950X3D since day one, the challenges are there with the CPU, but once you get it to sing, it's awesome. I have customized my entire pc to work around those cores. Things that are meant for nonX3D cores run in the background so the X3D cores run through the game flawlessly. I think it comes down to how big of a nerd you are for gaming. I love AMD tho, I will never go Intel for CPUs
YES dual vcacheX3D let's go
The 9950x3d will be better than the 9800x3d amd claims
Source
Trust me bro
@@tutuuo you really think they will make the same mistake twice? It will be better.
Why even get any x3D cpu if they have these issues. I would think the X versions would do just as well.
I'd considering upgrading from my 7800X3D if all of these numbers ended up being close to the actual product. That's a big overall performance gain. I could probably recoup half the cost by selling my 7800X3D.
I've seen other sites that for gaming the difference is about 8%....also depending on you gpu you might be maxed already...about 18% difference in other benchmarks/encoding. I would stay where you are. just my 2 cents
Companies love ppl who want to upgrade every single new iteration, for sales. But in reality doing that is just trolling everyone else who doesnt and waits a proper long while then ends up with inflated prices cause ppl want to get the newest even when they dont need it.
I don’t think we will see X3D on both CCD’s but here’s hoping.
I wonder if this cache beneath the die can be back ported to AM4 for a 5950x3d.
192 MB of cache hot damn that is nice.
That would explain the later release if cache on both ccd
I might have to skip the 9800X3D and go straight to the 9950X3D for Microsoft Flight Simulator 2024. These chips ought to do well on CUDIMM 9000MT/s RAM then the crappy 6400MT/s RAM .
Yes, 9950x3d , if price isn't stupid.
i just want the 9950x3d to be as good in games or slightly better than 9800x3d and ill get it. Need to replace my 14900k and just use it to render and crap in the background secondary cpu.
If ur on top tier the only replacement necessary currently is how to think. Consuming products unnecessarily is not going to fill the void where thoughts are supposed to go
the 9000 series is now same fast as at linux? how microsoft fix it? (x3D is sometimes worst ever in linux)
The 950X anything series from AMD should be the Top of The Top for consumer cpus
I have a 9950x and still Im gonna get a 9950x3D
Wait for Zen 5! Wait for Zen 5 X3D! Wait for the highest end Zen 5 X3D!
Wait for AM69
I use the 3950x... I can wait a little longer :)
Real journalist
Yes they should put X3D on both chips.
I don't think AMD would not have at least 1 cpu with X3D-Vcahe on all CCD's now. If AMD doesn't have 1 cpu with Vache on all CCD's now ; Good bye AMD hello Intel. This is AMD's BIG chance to recover from the earlier initial release of the 9k series and really put Intel further into the ground within a Pine-Box.
5 more days until I buy 9800X3D
Are you the "monitor guy"? You sound a lot like him!
GPU Today AMD and Nvidia is up at me Seeking Alpha comment spot; % channel available, sales trend, % share. mb.
As much as I hope it’s true I don’t think they will add 3D cache to both CCDs because they will literally render their own 9950x useless, no one will buy it if they can a 9950x3D with the cache on both sides. I just don’t see AMD shooting themselves in the foot like that unless they make the 9950x3D VERY expensive as to not destroy the standard 9950x sales. But hey I guess when your only competition is yourself you get to ponder these questions lol.
This makes no sense whatsoever. The 9950X would still be the best choice for workstation tasks which don't benefit from extra cache, regardless of the 9950X3D has 1 CCD with Vcache or 2. The 9950X will be cheaper and will be able to clock faster across all cores. The 7950X has a similar advantage over the 7950X3D in workstation tasks.
@ this new x3d version will clock higher with the re design of how the 3d cache is implemented. The gap in clock speed between the 9950x and 9950x3d will not be as far apart as they were on last gen Ryzen 7000 thus making them much closer in workstation tasks then they were before. Did you even watch the video ?
Maybe in 4 or 5 years. Not liking the direction computers are heading they are already drawing power like a microwave oven at this point. I'm sitting at 850 watts on AM4 platform and not interested in going above that.
@@Toutvids low iq comment, am5 uses less power than am4
Zen 5 X3D is shaping up to be a beast in gaming and productivity. Especially in gaming. Zen 5 X3D is said to have significantly higher clock speeds than Zen 4 X3D, so that will substantially boost gaming performance. Gaming performance tends to scale well with higher clocks. A reason why Zen 5 X3D is expected to have considerably higher clocks is because the 3D V cache will be stacked below the CCDs, unlike the prior generations, which had 3D V cache stacked above the CCDs. This will significantly improve thermal temperatures and should address heat dissipation issues. All of this should allow the 3D V cache cores to clock just as high as the non-3D V cache cores. 3D V cache and high clock speeds massively boost gaming, and getting that in one package in Zen 5 X3D has me hyped. Zen 5 X3D also has an unlocked multipler for overclocking. This all looks really promising. Hopefully, they won't cost more than the Zen 4 X3D CPUs did.
I want dual socket 128 core EPYC 9785X with 3D V Cache and overclockable with 24 TB *(48 DIMMS on the motherboard ) 24 channel (12 channel per socket ) DDR5 6000 MHz ECC CL30 support with CXL support up to 2 petabytes of additional DDR5 CXL 6000 MHz ECC CL30 RAM and 8 RDNA5 Workstation flagship GPUs each having 384 GB of 48 Gbps GDDR7X VRAM (3 GCDs per GPU, 8192 SPs per GCD) and 64X Sabrent Rocket 5 8TB PCIE5 M2 NVME SSDs on 2 double sided M2 adapter cards each side having 16 M2 slots. 24 Kioxia Enterprise 128 TB PCIE5.0 SSDs all in E-ATX form factor without watercooling and with Dynatron CPU fans and 6X 8K 240 Hz monitors which can convert into 4K 480Hz and 1080p 960Hz on demand via software. all powered by 16X 2500 Watt FSP PSU. my case should support 8X 3 slot GPUs, 24 2.5 inch SSD bays and 64 140mm fans which run at 18450 RPM each and have zero decibels.
Ryzen 7 5800X3D still the best!
4:00 slide its an 'R7' 9800X3D, not an R9.
Might.
thanks for making avideo telling us how you feel and think (no facts)
I wish they had it with a NPU i would buy it but I want a npu
Amd stop think about money, make amd gamers cpu great again, no excuses Intel have to leave a lead!! Best cpu!!! Best gpu's!!!! Best value!!!! AMD!!!
glaze, i lost nnn to you
Bro wtf😭😭😭😭
Im buying the 9800x3d and if 9950x3d is good then I will upgrade... I am not just gonna wait on rumors lol ... B.S video
Even if this fake news about 9950x3d is true, switching from 9800x3d to that is by definition NOT an upgrade. I wanna see a blind test of you telling the difference between a 50fps boost on a 120+ avg fps range b4 i believe the claim that it should be considered an upgrade. And vsync has to be enabled cause counting disgusting screen tearing in non vsync is easy to tell apart
😧 Unbelievable
Instead of dual X3D CCDs, I'd love to see AMD start with asymmetrical compute on desktop with a 10 core Zen5C CCD paired with the 9800X3D.
Yeah me too i really want to see these 2 Configs on Desktop, I mean they already did 4+8 on Laptop so where are them hybrid CPUs for Desktop ?
Dual CCD Zen5 + Zen5c
6 + 12
as well as a CPU that has a single CCD but with 16 zen5c Cores just like the Biggest Epyc Server CPU has
@@limeisgaming Oooh I didn't know they were already making Zen5c chiplets...
I'll take 8 core Zen5X3D with 16 core Zen5c.
The only thing extra cores gets me is reduced shader compilation time, but I hate shader compilation so this configuration would definitely win as the best all around gaming CPU.
@@budthecyborg4575 its the ryzen ai 9 370 hx for laptops
i don't appreciate clickbait showing on my recommended videos so i just came in to block your channel
lol 7 7800x3d is the king.
On both obvs...
Waiting for AM6
@Irnwkr28 am5 supported till 2027 might be a while. Suggested by friends that there might be a am5 like intel that supports ddr6 in 2026.
@deadpool790 Cool
might as well wait for the 6090!🤣
What does it matter when all the games are sh 9t
only 36% more bullshit tales&leaks!!similar was with normal zen5 leaks and that Amd shit benches on Ces
9950X3D Dual v-cache CCDs + 5090 ... YES !!
It's the 9900x3d that will blow everything away, it's same insane as 9950x3d cache but cheaper.
Cheaper but not cheap. They will be 500+ not long after launch. I got a 7950x3d cheaper than that
Fake mustaches
Did your chart really just say the 9950x outperforms 7950x3d in gaming? 😂
Please stop that stupid move 🤣
Theres no logic to putting vcache on both CCDs when it will go to waste due to cross CCD latency penalties. Even the current 9950X parks one of the CCDs to avoid this issue.
Your reasoning doesn't make any sense for this very reason, the current non vcache 9950X already requires AMD software parks one of the CCD during gaming.
PC hardware lol. Profit ball!
Thurd
Will most likely be the last cpu generation where AMD are leading in gaming. Intel are going to catch up very quickly, which isnt a surprise. AMD are the massive underdogs to intel at the end of the day. They cant even make their own cpus 😂
I don't believe that, in the short term, AMD have been better since Zen3 and Intel with the same architecture for 10 years, making compromises on processors even adding fake "efficiency cores" that are useless on desktops. It does not compete with TR or Epycs.
What's this based on. Intel have been failing hard. Their latest gen is a flop as well
@mojojojo6292 just look up backside power delivery. Intel has been developing 2nm 20A backside power delivery cpus which have superior power delivery and efficiency. TSMC who make AMD cpus are 2 to 3 years behind in the technology. Intel are going to have a huge lead in the cpu world. Yes intel cpus had a flaw. But this was to do something with the current 10nm older architecture node (lg 1700 socket V) which was naturally inefficient compared to TSMC's current 5nm node, and they pushed it to the limit. Intel are currently using TSMC chips with a 3nm node compared to AMD ryzen 9's 4nm, which is why intel ultra cpus are now more efficient, but performance is on par.
Arrow lake is on tsmc idiot? Who cant make their own chips? Fucking fanboy
@@prussell890and its not more efficient than zen 5
I don’t understand why your talking about it when no one has tested it were do u get your numbers from that’s what I don’t understand 😂
Intel, oh Intel... How thee hath falleth...
It will be 2 percent fast let’s be real mate 😂
I call bullshit and are lairs 😂