Please note a small mistake in the specs table, the base clocks are switched around, the audio is correct here. The 7950X has the 4.5 GHz base and 4.3 GHz for the 9950X. That is all, as you were. Also thanks for watching :)
DISABLE PROCESSOR GPU. I think something blocks the output. May be the processor is bottle necked by integrated graphics. I don't have the 9000 processor, So can you please check the performance after disable the integrated graphics in the processor. DISABLE PROCESSOR GPU......
Zen 4 was only 8 to 10% faster pre-clock than Zen 3 It was still slower in gaming that Zen 3 with 3D V-cache Zen 4 with 3D V-cache is 20% faster than Zen 4 by itself.
@@CRAZYWIZARD It's not and never was a purely gaming focused CPU: That's the 7800x3D, hence why it does better - Because gaming is its only task. The 7950x3D is the compromise CPU for those who want the benefits of the 3D V-Cache, but also need the raw performance for productivity work, or for those heavy streamers who do both simultaneously.
@CRAZYWIZARD In most cases, there's not much in it, in some cases the 7800X3D is slower in games and it's always slower at anything else. I'm quite happy with my 7950X3D.
You know it's bad news when he's standing up. Kids get home from school.. kids are preparing for bad news about then from school their reports. No. KIDS! (INTENSIFIED DAD STARE) AMD.. (KIDS LOSE PSYCHOLOGICAL FEAR AND BRACE FOR PSUEDO CRASH INTO WALL) JUST RELEASED A SHOT PAIR OF CPUS FOR THEIR 9XXX SERIES! WHAT DO YOU HAVE TO SAY ABOUT THIS?!?! Kids: We were at school dad!! Dad: you're goddsmned right you were.. you missed the main event baby!!! It was an absolute shut show..
To be fair, they were screwed either way. Had they given us more honest claims pre-release, people would have just been shitting on those as well. In the end, the real problem is hardly marketing, it's simply that Zen 5 is an insanely underwhelming generational improvement. It obviously makes some pretty huge architectural changes, but they simply aren't the kind that will benefit the majority of users in any significant way. Zen 5 may age better if devs start optimizing more towards super wide architecture design, but nobody should be spending a lot more today for potential better relative performance four or five years down the line.
Yeah it's unbelievable that they haven't decided to "shuffle the back" yet! They have been underwhelming for decades!! And now the added untrustworthyness in the recent 5 years time should be the decisive factor for AMD's top management!
@@maynardburger There is not denying performance "gains" in gaming are underwhelming but the WORST part of Zen 5 is the performance claims, complete BS,
@@maynardburger People wouldn't be that disappointed had they been honest from the beginning and.. pricing.. prices are terrible, make zero sense, they are gonna fall off a cliff in two months.
@Celis.C a lot of zen4 products increased in price lately, not just the x3d parts, one chip increase in price by, I think it was 20%, I don't remember it that was the msrp or the actual store price.
@@concinnus i'm not very deep into this, but isn't memory-training also involving coding from AMD? It's probably not only down to the MB-vendor BIOS. I've seen other reviews and they show lower 1% with the same overall fps as the 7950X (just like here) while using other hardware so the only thing in common is the CPU. So it is a big f on the AMD-side. Microcode-error? Maybe they forgot to copy something, yeah. The delay maybe should have been longer. Or this is just a step and we get the real jump with the next gen. It's pretty confusing tbh.
@@geht-dichnix-an4183 1% lows do indeed seem to be worse than ryzen 7000, so even with the whopping 1% increase in gaming performance, it will microstutter more.
@@andersjjensen Should be, and yet...the memory compatibility has been different, causing issues for HUB itself. So clearly it's not behaving the same in reality.
Zen 4 was only 8 to 10% faster pre-clock than Zen 3 It was still slower in gaming that Zen 3 with 3D V-cache Zen 4 with 3D V-cache is 20% faster than Zen 4 by itself.
16:35 JayzTwoCents fairly recently made a video talking about the dual CCDs and the core parking issues you described here, and its not entirely true that you need to reinstall windows. He used Revo Uninstaller to completely wipe the AMD Chipset driver and reinstalled it, and everything worked fine for him. I'm assuming that is a decent way to get the dual CCD chips to work on windows without a fresh install, though its not a bad idea to have a fresh install regardless.
Both are true. AMD officially states that reinstalling windows is the best and most reliable way to undo core parking. They are aware you can remove and re-install the chipset driver, but that's more of a gamble if everything will work perfectly. It very well could be entirely fine to do the workaround method, but that's not something they would recommend. The last thing AMD wants is to recommend doing something that breaks stuff, or causes instability.
The 13th -> 14th Gen switch was a refresh. Zen 4 -> Zen 5 is a new architecture. AVX512 isn't double-pumping anymore for instance (and it shows in benchmarks which take advantage of those instructions). This is more like a sidestep into a new foundation, which they can improve upon in Zen 6. It is held back by being I/O constrained, and because games don't get a recompile when a new architecture emerges.
Anandtech: Unfortunately, the issues we experienced with the PPM Provisioning driver, despite following all of AMD's guidelines and then some, has undermined this launch. So far we've seen core parking cause quite a few issues in performance, most notably in gaming. If it was an issue limited to just our testing, that could be negated, but having checked with a few of our colleagues, it looks to be a widespread issue. Something just doesn't seem to be working in the core parking department, as we ourselves have observed random cores from the parked CCX firing up randomly and taking game threads with them, and this in itself seems in defiance of what AMD is trying to achieve by running gaming workloads within a single CCX. For now, the Ryzen 9 9950X/9900X seem like fine chips for highly threaded productivity workloads. They're just not delivering the correct behavior for games. In the meantime, we'll be continuing to look into the issue, and should AMD deliver an update that fixes Ryzen 9000's core parking behavior, we will reevaluate these chips accordingly.
@@ishaansaral502 and after 2 years; not intel s usual 1 year of refresh; and intel are also known to release filler cpus on same platform after the first cpu on a new socket; now arrow lake is coming next month... on a new socket; hopefully they ll give AMD a reality check again; bcs 1% performance after 2 years on a new zen architecture is disappointing
@@mikem9536lol no, theres no difference between any modern CPUs at 4K. Youd have to use a 4090 and go as low as a 5600 to notice a difference and the difference isnt that big. CPUs currently are mostly overkill for each GPU at their current res unless youre the bonehead that buys a 4090 for 1440p or 4070 for 1080p
@@AdiiS Remember that Intel currently has $191 billion in tangible assets, whereas AMD was $4 billion in debt when Zen 1 launched. Intel could have sustained a good deal more pounding and still bounce back.
It is very hard to believe that these processors are so bad, deep down we want it to be a problem with a driver or something, hopefully amd will do something and fast with these processors.
I absolutely expected it to be the worst Zen generation yet given the details about what was changing, but even then they are still delivering well below what I expected. Previous Zen architecture was hardly 'narrow', so going ever wider, which is basically what ALL the effort was put into for Zen 5, was simply never gonna benefit most people all that much. Software just isn't optimized to take advantage of it, and wasn't being bottlenecked by narrower architecture before.
Yeah my guess is engineering aimed to improve Infinity Fabric. But I heard rumors a year or 1+½ years ago that they had met engineering challenges. Though it wasn't specified which problems they ran into. But to me it makes sense as the large gaming performance increases has mostly all been memory performance related by either static cache supplementation or improved IF speeds. Also a good bit of core frequency and other IPC gains. Hopefully they can solve this for the Zen 6 launch, and we don't hit a slumber a la Intel's 14++++ period
Also had my 5800X3D for 2 years now and will most likely keep it for another 2 - 3 years. At that point, will buy whichever brand offers the best bang for the buck (as I've always done in the past 30 years).
I dont know understand why people would cheer lack of progress. You should WANT progress. If you're still fine with your current CPU, you never needed to upgrade to begin with. And when you do eventually want to upgrade, you should want there to have been bigger increases. It's not like Zen 6 is gonna make up for it and be some 40% increase instead. Zen 5 will forever be a hiccup in the progress for more performance+efficiency.
They should really have prepped to make Vcache standard on all their chips by now. It would actually allow them to package it more efficiently to where clocks wouldn't even need to be as affected by putting the SRAM die underneath instead of on top. They could include more cache this way if they wanted, though that'd probably have heavy diminishing returns.
Bingo. I get the heat implications but it's time to figure out a way around you would think being the 3rd generation of x3d. If it is just one sided again it will be really boring and the 9900x3d will be just as lame as the 7900x3d. I get 8 cores is all you need for gaming but I'm not going backwards in core count from the 13700k when I do other shit with my pc.
I wonder what is the reason behind Zen 5's strange performance. Sometimes we see decent gains but other times it's slower and less efficient on a newer node. And all this after nearly 2 years. This is such an RDNA3 moment.
@@zxy7529but they did, if you look at the 400 odd benchmarks on Phoronix you'll see loads of performance gains. But most people aren't running EPYCs, Threadripper or apps that will take advantage of the changes leading to regressions or bearly more than static performance changes. And now we see the downsides of sharing the CCDs with servers because the dev time went towards those markets where the profit is made
Edit: I see the pinned comment about the audio being "correct". That sounds like it may be something else, but I'm not sure. I'll still leave this here just in case 👍: Hey Steve, I noticed in this video, when we see you standing, the audio bitrate seems a bit low and/or sounds heavily compressed. I have the Sony WH-1000XM5, thought it was my EQ, but as I went to go mess with it, I realized when the video switched to b-rolls or performance charts, the audio quality is fine. This is an amazing video none the less, and I love the work ya'll do, just figured to let you know in case there was an issue with the mic!
I mean, if yo live in a country where energy cost is absurd, like i live, it can be considered a point (specially the 9700X), but still can't justify that price increase 😂😂
@@mondodimotori funny because my pc is used for 8 hours a day for school and business. I’m lucky to play a game for at least an hour. Mind you this is done through my R5 7600💀
Thanks Steve. Would have liked to see the 7600/7700 in the cost charts as someone who is interested in value would certainly consider these CPUs. Great work and thank you guys for putting in a solid effort and not skipping the details.
Also a 5950x owner, and I always upgrade on a 4 year 2 product cycle. Never been more disappointed in a product. The fact that I could have had a 50% performance increase for the last two years, but waited for a 100% performance increase breaks my heart.
My 7950X is still perfectly fine for gaming. I game at 4K on a 144Hz monitor. I wasn’t really going to update this generation anyways. Hopefully next gen provides a bigger uplift. 16 core CCD would be great for gaming. Unified cache for more cores. Would be especially nice for X3D products.
You really need to make the leap for an X3D processor. I got a 7950X3D for $420 December of 2023 and the difference is night/day for reducing 0.1% and 1% FPS LOWS. Having additional cache helps prevent those stutters/low frame rates. It makes for a much more smooth gaming experience. Unfortunately, it seems like you are kind of required to wait for X3D Zen 5 for a meaningful upgrade unless you plan to spend $520 to get X3D v-cache and NOT losing 8-cores. Think 7950X3D are like $520 right now.
I see only couple of percentage difference between 7950X and the fastest gaming CPU in 4K gaming reviews. Reviewers love 720p and 1080p to emphasize the difference, but nobody plays at these tiny resolutions if they got a 32” 4K screen + RTX 4090. Also 7950X hits 144 fps in most games. No need to go above that on a 144Hz screen. 7950X3D is slower on code compile and I am a game developer, so I compile big C++ code bases daily. That’s another reason why I don’t see that CPU as a valid upgrade path. 9950X is 15% faster than 7950X in code compile, but gaming is only 2% faster. 9950X3D might be a reasonable update if the rumors are correct that it has higher clocks. But still, most games run either at above 144Hz already or are GPU bound at 4K. I intend to skip RTX 5000 series, so the GPU bottleneck at 4K is not going away. I am fine waiting for Zen6 X3D. Rumors say that Zen6 will have 16 core CCD. That means X3D would cover 16 cores with unified big cache. That should be awesome for games. Update worth waiting.
@@sebbbi2 Tell em brother. I own a 4080 and never ever touch 1080p, same with when I had a 3080. A CPU upgrade would do nothing for me (outside of getting an X3D) even though I have "just" a 7700X. Most people in first-world countries aren't on 1080p anymore so the fact reviewers need to use 1080p to emphasize the difference for CPUs just shows that the CPU is largely irrelevant now for gaming performance. I have no idea where the hell this 1% low talk is from because even when I had a 3080 and cranked the settings at 1440p my games just... ran fine. And obviously with my 4080 I've never seen a single stutter outside of Ghost of Tsushima after a few hours (PC port has known memory leak). I don't get how people are having stuttering issues caused by their CPU. Are people blaming shader compilation stutter and poor optimization on their CPUs? I simply don't believe it when it's described as if games will inevitably stutter no matter what because of the CPU. If games stutter, 99% of the time it's the game and not the CPU. Above 1080p even my 4080 is bottlenecking the 7700X. It would take a 5090 or an even later GPU before even Zen 4 starts to become the bottleneck. Nobody should still be on 1080p outside of e-sports kids chasing 1000fps. You can grab a used 3080 for $300 now, there's no excuse. 1440p monitors are dirt cheap. If someone says "I can't afford 1440p" then why are they looking at $300+ brand-new CPUs?
Mom I want an AMD 9950 We have a 9950 at home The 9950 at home is a first gen Phenom. (I actually had a Phenom9950 on an Nvidia Nforce chipset, was my upgrade path from an Athlon X2 5200, which gave a good performance uplift) Yes the first gen Phenoms were still mediocre against even the intel core2quads... but the higher end ones were a decent cost effective upgrade to dual core athlons.
Basically all I’m seeing here is, if you don’t need avx512 just save your money and get a 7950x on sale. The 7000 series was just such a good generation AMD is having a hard time beating it. Kind of a disappointing release. I hope the 9800X3D does better for gaming
It will certainly be better than 7800x3d. 10-20% depending. It has a reworked cache system. Though.. it gains nothing for gaming when increasing the power so we are genuinely looking at a +5% before adding vcache so even if the vcache adds say.. 25% more of an uplift than vcache did for 7800x3d.. it will only be that 10-20% depending on the game.
@@TheCompyshop It is not bandwidth starved. We've seen tests disproving this. Vcache is not going to magically unlock all this secret performance. It'll be the same bog standard increase that we get from any Vcache part. The bottleneck to better performance in Zen 5 is coming from software simply not being designed around taking advantage of a super wide architecture like this. Applications just dont need it. All the work AMD went through to make these big architecture changes just aren't gonna benefit consumers much.
Linux is alot lighter than window, most people will choose to use windows because of better software compatibility and experience in windows so that 20% z zip performance doesn't matter.
Just skip the 9000 series. It's really not worth it if you are already on AM4 or Zen 4 already. Even if you want to upgrade from AM4, the 7000 series will be much more value. The 9000 series is more of a generation in which AMD focused solely on efficiency.
I really don't understand what's going on, how could they release a chip that's exactly the same, even slower than previous gen? They obviously made improvements, case in point the uplift in IPC and single core performance, so why is there no real world improvement, save for the outliers?
7800x3d was down on price by 40% since release when i bought it (in my country at least) so i thought that zen5 performance uplift got priced in. I feel as if i got a bargain :D
There's no question that the R7-5800X3D will be considered the GTX 1080 Ti of CPUs. At this rate, I might not need to upgrade before the release of the AM6 platform! It really surprises me just how efficient the X3D CPUs are in Cinebench as the 7800X3D and 5800X3D sit as #1 and #2 respectively. The efficiency of the 7800X3D isn't too surprising since it's Zen4 but the efficiency of the 5800X3D, being a Zen3 part, is absolutely shocking.
Zen 5 hurts everyone. Remember when AMD said AM5 would have a similar lifespan as AM4? Well, AM4 saw huge gains every generation. With Zen 5 AMD is almost trying to say AM5 supported an extra generation, when it was basically a useless generation. AM4 wasn't great because it got 4 CPU gens, but because every single one had its weight in the scale. Zen 5 is making Zen+ look better.
Steve at GN said it best. Its not a bad product, it's just more of a refresh instead of new generation. AMD has one final shot at making Zen5 mean anything with their X3D chips. If they can get even 10% average uplift over Zen4, that will at least make up for their general lineup.
That's hugely disappointing. I might still get it for the small (and economically unjustified) bump for 3D graphics creation, since I don't care as much about games, but I'll give it six months to see what happens: a price drop, or the moonshot of a magical microcode update.
The difference between the 3950 vs 5950 vs 7950 vs 9950 would have shown that the big jump from 5950 to 7950 is really the whole package. Moving from DDR4 to DDR5 and the new architecture is what gave that huge performance jump. If you look at the 3950 to 5950 performance jump, its basically the same as going from 7950 to 9950. Platform to platform the differences are small in terms of performance boost moving from the old to the new.
The multi core productivty jump from Zen 2 to Zen 3 wasn't particularly large, but the single threaded performance jump was massive, as AMD went away from 2 4-core chiplets in one CCX to 1 8-core with unified L3. That netted them about 23% uplift in gaming and Photoshop, etc.
Keep getting reminded how crazy the 7800X3D is. Chugs well behind the newer models in productivity but still somehow flooring competition in games and power efficiency.
AM5 being stuck with basically the same gaming performance for 4 years is just depressing, because it'll likely be another 2 years for Zen 6. I'm so tired of modern tech, entertainment and society in general. Everything is stagnating, but more and more often it's regressing. I miss the 90's.
AMD decided to back-port Zen 5 to TSMC N4 because N3 was getting delayed due to yield issues. I don't think it will be 2 years before Zen 5+. Zen 5c is on N3 so it will be "trivial" for AMD to release what Zen 5 always should have been. N4 is only a 20% shrink over N5, whereas N3 is a full 70% shrink over N5, just like N5 was a full 70% shrink over N7. N4 is, unfortunately, the worst half-step node we've had in decades.
I mean there's only so much you can blame the CPU for in gaming performance when it's current-gen GPUs that are bottlenecking them. X3D only gains performance in games that use the v-cache. I think those CPUs gave people the misconception that you can still gain FPS with a CPU upgrade nowadays. Dunno how you can call it depressing when my 7700X never sees above 20% usage paired with a 4080, except in very few ultra-demanding games. How is that depressing knowing you can spend $300 and have a CPU that will keep up in gaming for 5 years or more? Things used to be a lot worse.
@@JohnDoeWasntTaken You were perfectly right up until you said "above 20% usage". Total usage is almost meaningless in this context. Max core utilization is the end all be all in gaming. If a single core is pegged at 100% and the 7 others are at 0% (will never happen, but for arguments sake) you will see 12.5% CPU usage. But your point obviously still stands: At realistic settings and resolutions we are spoiled with CPU gaming performance. Vulkan and DX12 really cuts down on the amount of back-and-forth communication, between the CPU and GPU, that is needed to get the job done and honestly the industry was WAY too slow to adopt them.
Steve deserves a standing ovation. Thank you in advance. I had a feeling that this would happen. That is why I didn't get my hopes up to high, and the same might go for x3d naybe a little faster, but not by much. Again, I can and will wait for Zen-6.
im glad i sold my 7700x and got the 7800x3d when it came out. was able to sell for 300 and buy the x3d chip for 350. definitely worth 50 dollars in performance and efficiency
People seem to misunderstand that Cinebench is not some 'all encompassing' benchmark. Cinebench only benchmarks performance of Cinebench. Just cuz it shows a moderate gain in single core performance doesn't mean other applications will. Because they aren't Cinebench. Frankly, Cinebench has been hugely overused as a general benchmark, given what it's benchmarking isn't an application many people use to begin with.
@maynardburger CPUz, CPUMark and Geekbench show high single core uplifts too. Actually all single core performance benchmarks show the most consistent results I've seen so far.
The only thing I can think of is AMD engineering went to try and make the overall experience better than the 7000 series launch in regards to stability and RAM support. But honestly marketing and pricing shot this things privates clean off.
No, AMD went all-in on making Zen 5 a super wide architecture. That's why it's disappointing for consumers, cuz applications simply weren't bottlenecked by this before. Basically, these weren't really engineered for helping consumer workloads to begin with. At least not anytime in the near future. It'll fare better in datacenter and large AI farms and whatnot, but we simply weren't the priority here. Sucks, cuz it looks like the great CPU competition we'd have been having is basically gonna die out for the next couple years, given that Arrow Lake isn't expected to have huge performance improvements, either.
@@maynardburgerarrow lake actually is expected to have significant performance increases and efficiency from what ive seen. Whether Intel should be trusted with that is a whole other story.
Thanks for the straightforward video titles and thumbnails. I always know what to expect and feel confident I'm getting the truth when I check in on the details. Also, thanks for testing ACC!
Why would AMD lie about the expected performance of Zen 5? It's not like no one was going to review the CPU's once they released (or use them for that matter). AMD must've known what Zen 5's ACTUAL performance numbers were, but they lied anyway with the full knowledge that everyone was going to find out. At this point I'm convinced AMD has to be shooting themselves in the foot on purpose.
They lied because they wanted to postpone the damage. If they'd told us months ago that Zen 5 would have super underwhelming performance, people would have been shitting on it back then. It was always going to happen, but they managed to delay the backlash a while.
@@maynardburger They didn't just delay the damage, they turned damage from disappointment into damage from essentially lying to their customers. All bad press isn't created equal. AMD said their next crop of GPU's aren't going to be all that powerful compared to RDNA 3 (other than on RT), and they did so well in advance of release. Sure people are disappointed, but still looking forward to new cards. That's a much better situation vs saying the new GPU's will be 50% faster, then having everyone find out that was a lie.
Looks like they cherrypicked the games that benefitted the most from the new architecture and put those on the presentation slides. So, they didn't technically lie. There is one game on this HW Unboxed test that shows a 20% uplift because it's not memory starved by the IOD. Zen5 is a good architecture but on the desktop parts it's bottlenecked by the IOD. Epyc's Zen5 processors should show the muscle as well as 3D cache versions when they get released.
I’m really started to wonder why AMD bothered to release these. They’ve had to hear all the talk about the I tel 13/14 gens being the same now they’ve done basically the same thing whether it’s on a new architecture or not.
Except at least Intel didn't claim anything crazy with the 14th gen and said that it was just a raptor refresh since they ditched meteor lake desktop. I'm waiting for arrow lake but I'm not excited for it either as no p core increase and no HT anymore. If it is at least 10% faster in gaming I may switch just to get away from the raptor lake scare going on or I may just stick it out as my cpu is fine
Great video. Sad to see the trend for Zen5 continuing for 9950x (probably will also for 9900x). As someone that currently owns and prefers AMD I don't feel the need for an upgrade (for home use at least). Definitely there are improvements for enterprise (data center) market but for now only 7950X3D can do both gaming and work with good performance, when compared with Intel (if power draw and degradation is put aside). For gaming only the 7800X3D is still the best, even after all this time and I do hope that Zen5 X3D parts will continue that legacy, but we'll see...
I mean, for building a first system as I am, the 9700x isn’t a bad choice at all for mixed use, and can only really get batter in the weeks to come with updates, as well as keeping in mind x870e on the horizon for faster memory. If I already had a 7700x system there’s no way I’d upgrade now, but since this is my first, I think I’m going with the Zen 5 8-core.
Gamers Nexus and Hardware Unboxed have done a great job reviewing Ryzen 9000 and debunking efficiency an consumer productivity gains and "not meant for gaming" gaslighting from AMD stans. Thanks, Steves!
Please note a small mistake in the specs table, the base clocks are switched around, the audio is correct here. The 7950X has the 4.5 GHz base and 4.3 GHz for the 9950X. That is all, as you were. Also thanks for watching :)
So even in the base frequency the 9950X sucks... damn AMD... what are you doing?
Who's buying one of the most powerful CPUs to play in 1080p ? and why not waiting for the new Motherboards? :)
DISABLE PROCESSOR GPU. I think something blocks the output. May be the processor is bottle necked by integrated graphics. I don't have the 9000 processor, So can you please check the performance after disable the integrated graphics in the processor. DISABLE PROCESSOR GPU......
how can they be so wrong. I can understand a small discrepancy but this feels like a big mistake was made somewhere
AmD SHiLl!
😂 jk great review mate.
AMD achieved 20% increase in naming versus previous generation. Marketing team just misspelled it to gaming.
Zen 4 was only 8 to 10% faster pre-clock than Zen 3 It was still slower in gaming that Zen 3 with 3D V-cache
Zen 4 with 3D V-cache is 20% faster than Zen 4 by itself.
That's genius lmfao
This isn't a gaming chip.
😂
underrated comment
As a 7950X3D owner I am glad you included the CPU in your charts. For whatever reasons many outlets completely ignore the 7950X3D, which is a shame.
Mainly cause how can it be worse than 7800x3d in gaming which shouldn't have happened
@@CRAZYWIZARD It's not and never was a purely gaming focused CPU: That's the 7800x3D, hence why it does better - Because gaming is its only task. The 7950x3D is the compromise CPU for those who want the benefits of the 3D V-Cache, but also need the raw performance for productivity work, or for those heavy streamers who do both simultaneously.
@CRAZYWIZARD In most cases, there's not much in it, in some cases the 7800X3D is slower in games and it's always slower at anything else. I'm quite happy with my 7950X3D.
Steve is still standing!!!!!
He's definitely not going to take it lying down from AMD.
@@Rose.Of.Hizaki💀💀💀💀💀
4 in a row, it's over.
Noooooo you caused spoiler bro😡😡🤬
Standing even taller, waaaah!!!!🫨
Normally people won't stand for this, but Steve won't sit for this.
Steve opened a big precedent recently, standing when the product sucks. At this rate this man is never gonna sit again in his life.
You know it's bad news when he's standing up.
Kids get home from school.. kids are preparing for bad news about then from school their reports. No.
KIDS! (INTENSIFIED DAD STARE) AMD.. (KIDS LOSE PSYCHOLOGICAL FEAR AND BRACE FOR PSUEDO CRASH INTO WALL) JUST RELEASED A SHOT PAIR OF CPUS FOR THEIR 9XXX SERIES! WHAT DO YOU HAVE TO SAY ABOUT THIS?!?!
Kids: We were at school dad!!
Dad: you're goddsmned right you were.. you missed the main event baby!!! It was an absolute shut show..
Will he stand on the desk when he is extra disappointed?
Good for his back, I guess...
Need a "Steve's sucks list"
Probably better for him overall, physically, however his knees and lower back won’t like it!
Zen 5% might have been ambitious. Zen 3% sounds more accurate.
Best comment on this video! 😂
More like Zen ±3%.
More like Zen 4.05.
Zen4+
9950x BLENDER FAST THAN 7950x 50 */*
JFC, AMD is making me not want to wait for 9800x3d. Starting to look like the 7800x3d is gonna be the next 1080ti. Undefeated til retirement.
the 1080ti of CPUs indeed
undefeated till decomposition
Ryzen 4070
x3d CPUs are for nerds
@@wallacesousuke1433 my brother you're on hardware unboxed, you're in turbo nerd hub
If Steve will continue to stand in CPU videos, we're all screwed.
It's better for the back really.
"Take a seat young Steve(walker)!" xD
@@ValenceFluxFACTS
not really just stick with the 7800x3d and your on to a winner
Zen 5 is giving so much intel 11th gen vibes.
14gen vibes
@@Hito343true, people seem to be ignoring these having stability issues and failures already
@@Hito343 Nah, unless Zen 5 cpu's start crashing but it's not that bad yet.
@@bb5307 Reviewers where getting all sorts of BSOD issues with ryzen 9000, perhaps you should watch gamer's nexus.
@@bb5307zen 5 seems to have lotsa crash, but mostly due to faulty units
according to tech youtubers at least
It's not even 5% difference to call it zen 5%😭
Meanwhile in Linux...
@@notjustforhackers4252 Meanwhile in linux 7950x also gets the boost, so ....
@@notjustforhackers4252and ps3 emu
@@flamestoyershadowkill6400 that's mostly single core though
@@notjustforhackers4252I'm sure the .5% of users that run Linux will be overjoyed.
Marketing Heads at AMD, they all need to be FIRED.
To be fair, they were screwed either way. Had they given us more honest claims pre-release, people would have just been shitting on those as well. In the end, the real problem is hardly marketing, it's simply that Zen 5 is an insanely underwhelming generational improvement. It obviously makes some pretty huge architectural changes, but they simply aren't the kind that will benefit the majority of users in any significant way. Zen 5 may age better if devs start optimizing more towards super wide architecture design, but nobody should be spending a lot more today for potential better relative performance four or five years down the line.
Yeah it's unbelievable that they haven't decided to "shuffle the back" yet! They have been underwhelming for decades!! And now the added untrustworthyness in the recent 5 years time should be the decisive factor for AMD's top management!
@@maynardburger There is not denying performance "gains" in gaming are underwhelming but the WORST part of Zen 5 is the performance claims, complete BS,
@@maynardburger People wouldn't be that disappointed had they been honest from the beginning and.. pricing.. prices are terrible, make zero sense, they are gonna fall off a cliff in two months.
@@silvio351 AMD selling snake 🐍 oil🛢
All the work just to promote old 7800x3D.
Its price rose a little bit where I live. I can see why.
@Celis.C a lot of zen4 products increased in price lately, not just the x3d parts, one chip increase in price by, I think it was 20%, I don't remember it that was the msrp or the actual store price.
Is it worth getting a 9950x if you already have a 7800x3D? I mainly just game.
@@NATEDOG001976 Better to wait for the 9800x3D reviews if you can bide your time
@@NATEDOG001976of course not
And even when it beats the 7950X the 1% lows are, well, lower? What an odd architecture.
Maybe the memory training is worse? DDR5 can be very sensitive to subtimings, and maybe they weren't all duplicated across architectures by HUB.
@@concinnus i'm not very deep into this, but isn't memory-training also involving coding from AMD? It's probably not only down to the MB-vendor BIOS.
I've seen other reviews and they show lower 1% with the same overall fps as the 7950X (just like here) while using other hardware so the only thing in common is the CPU. So it is a big f on the AMD-side. Microcode-error? Maybe they forgot to copy something, yeah. The delay maybe should have been longer.
Or this is just a step and we get the real jump with the next gen.
It's pretty confusing tbh.
@@geht-dichnix-an4183 1% lows do indeed seem to be worse than ryzen 7000, so even with the whopping 1% increase in gaming performance, it will microstutter more.
@@concinnus It's the exact same I/O die as Zen 4, so the memory controller is the exact same.... so the timings should be the exact same.
@@andersjjensen Should be, and yet...the memory compatibility has been different, causing issues for HUB itself. So clearly it's not behaving the same in reality.
Zen 0.5%
😂
Google gives a translate prompt for this 🤔
Zen 4 was only 8 to 10% faster pre-clock than Zen 3 It was still slower in gaming that Zen 3 with 3D V-cache
Zen 4 with 3D V-cache is 20% faster than Zen 4 by itself.
@@userxyz783 apparently "Zen" means "It was" in... some language. I'd love to know which one.
Zen 5% was a bit too optimistic 😅
16:35
JayzTwoCents fairly recently made a video talking about the dual CCDs and the core parking issues you described here, and its not entirely true that you need to reinstall windows. He used Revo Uninstaller to completely wipe the AMD Chipset driver and reinstalled it, and everything worked fine for him. I'm assuming that is a decent way to get the dual CCD chips to work on windows without a fresh install, though its not a bad idea to have a fresh install regardless.
Both are true. AMD officially states that reinstalling windows is the best and most reliable way to undo core parking. They are aware you can remove and re-install the chipset driver, but that's more of a gamble if everything will work perfectly. It very well could be entirely fine to do the workaround method, but that's not something they would recommend. The last thing AMD wants is to recommend doing something that breaks stuff, or causes instability.
Isn't Jay the guy who deleted criticizing comments?
Please let Steve be sitting,
Please let Steve be sitting,
Please let St... bummer.
intel is in hot water and amd decided to scam buyers lol
Best buddies
Agreed
How is it a scam? Seriously it isn't
@@jdrok5026 Did you see AMD's misleading marketing?
Yeah, scam may be the wrong word here but it's pretty close in my opinion.
This dude tested with slower ram and likely a dozen other variables, Then there is the lame duck windows scheduler. @@dawienel1142
I think this is the AMD Ryzen 14th Gen Edition
it's actually worse when you think about how AMD use newer technology and architecture unlike just a clock boost of 14th gen
And Intel at least did not try to make an “innovation” fuss, calling 14th gen a refresh of 13th gen
nah its more like intel 11th gen, where those cpu's regressed
And Intel actually had results, the i7 14th gen actually beat the i9 13th gen, and the 14900k is just insane@@kernel_inpage_error
The 13th -> 14th Gen switch was a refresh. Zen 4 -> Zen 5 is a new architecture. AVX512 isn't double-pumping anymore for instance (and it shows in benchmarks which take advantage of those instructions). This is more like a sidestep into a new foundation, which they can improve upon in Zen 6. It is held back by being I/O constrained, and because games don't get a recompile when a new architecture emerges.
5700X3D=150€
B550=100€
32gb DDR4 3200mhz=60€
310€ best cost per frame in Europe
-gb- *GB*
GUYS
STEVE IS STILL STANDING
Steve is not going to take it lying down from AMD.
hemorrhoids 🍑💥🫴 he can't sit
Anandtech:
Unfortunately, the issues we experienced with the PPM Provisioning driver, despite following all of AMD's guidelines and then some, has undermined this launch. So far we've seen core parking cause quite a few issues in performance, most notably in gaming. If it was an issue limited to just our testing, that could be negated, but having checked with a few of our colleagues, it looks to be a widespread issue. Something just doesn't seem to be working in the core parking department, as we ourselves have observed random cores from the parked CCX firing up randomly and taking game threads with them, and this in itself seems in defiance of what AMD is trying to achieve by running gaming workloads within a single CCX.
For now, the Ryzen 9 9950X/9900X seem like fine chips for highly threaded productivity workloads. They're just not delivering the correct behavior for games. In the meantime, we'll be continuing to look into the issue, and should AMD deliver an update that fixes Ryzen 9000's core parking behavior, we will reevaluate these chips accordingly.
This sort of a middling uplift is what I'd expect from a refresh (a la Zen+) - *but not from a step forward architecturally*
Zen + was a pretty big uplift for a refresh
It's pathetic that for a new gen, it effectively feels like an Intel refresh. Not a good look.
@@ishaansaral502 and after 2 years; not intel s usual 1 year of refresh; and intel are also known to release filler cpus on same platform after the first cpu on a new socket; now arrow lake is coming next month... on a new socket; hopefully they ll give AMD a reality check again; bcs 1% performance after 2 years on a new zen architecture is disappointing
even a refresh shouldnt perform worse than its predecassor. The fact that Zen5 performs WORSE than Zen4 in some scenarious is borderline unacceptable
It sure seems like compiler changes are needed here when you look at what has changed architecturally.
Just built my 7600X PC back in January so I think I'm good to wait to upgrade for a while.
7800x3d jus chillin at the top.
For those gaming at 1080p or lower😂
@@joelfernando1 ?
@@joelfernando1 Huh?
@@joelfernando1 Or 4k.
@@mikem9536lol no, theres no difference between any modern CPUs at 4K. Youd have to use a 4090 and go as low as a 5600 to notice a difference and the difference isnt that big. CPUs currently are mostly overkill for each GPU at their current res unless youre the bonehead that buys a 4090 for 1440p or 4070 for 1080p
Ahh, but did you test the games with a 6600XT?
i believe it was even worse, only 6600 without XT lol
AMD had a chance to really stick it to Intel when they're down and has failed miserably.
Thank God they didn't now we can finally have some competition.
Yeah they could have full leadership but now they are letting intel 15th gen win
AMD is such a nice guy, whenever Intel or Nvidia is down in trenches AMD always makes sure to go in with them.
@@AdiiS Remember that Intel currently has $191 billion in tangible assets, whereas AMD was $4 billion in debt when Zen 1 launched. Intel could have sustained a good deal more pounding and still bounce back.
The problem with your post is Zen 5 was designed over a year ago, so they didn't know about any of this.
I bought a 7950X3D for 400USD a week ago when 9700X/9600X dropped. Cannot be more satisfied with it right now.
Isn't this false advertising by AMD???
it is
Once again AMD marketing department giving us false expectations.
yeah , AMD never lose a chance to STFU and take the easy win over Intel. I would be very mad at the Ryzen markething if I was in Lisa Su shoes.
@@matttiaz7576 fym? Lisa Su is the one deciding this type of shit as well
Unless they had all the settings listed, we won’t be able to prove it in court
It is very hard to believe that these processors are so bad, deep down we want it to be a problem with a driver or something, hopefully amd will do something and fast with these processors.
I absolutely expected it to be the worst Zen generation yet given the details about what was changing, but even then they are still delivering well below what I expected. Previous Zen architecture was hardly 'narrow', so going ever wider, which is basically what ALL the effort was put into for Zen 5, was simply never gonna benefit most people all that much. Software just isn't optimized to take advantage of it, and wasn't being bottlenecked by narrower architecture before.
Yeah my guess is engineering aimed to improve Infinity Fabric. But I heard rumors a year or 1+½ years ago that they had met engineering challenges. Though it wasn't specified which problems they ran into. But to me it makes sense as the large gaming performance increases has mostly all been memory performance related by either static cache supplementation or improved IF speeds. Also a good bit of core frequency and other IPC gains.
Hopefully they can solve this for the Zen 6 launch, and we don't hit a slumber a la Intel's 14++++ period
Perfect. Can stick to my 5800x3d for the next 2-3 years. Waiting for the next Zen architecture.
Also had my 5800X3D for 2 years now and will most likely keep it for another 2 - 3 years. At that point, will buy whichever brand offers the best bang for the buck (as I've always done in the past 30 years).
I'll then have to make the choice of whether to go for Zen 6 X3D or wait for AM6.
hopefully it will have more than 6% performance uplift. 5800x3d still going strong
With a 7950X for productivity I can say the same.
I dont know understand why people would cheer lack of progress. You should WANT progress. If you're still fine with your current CPU, you never needed to upgrade to begin with. And when you do eventually want to upgrade, you should want there to have been bigger increases. It's not like Zen 6 is gonna make up for it and be some 40% increase instead. Zen 5 will forever be a hiccup in the progress for more performance+efficiency.
I feel like the only way AMD could make Zen 5 interesting for gamers would be to release the 9950X3D with 2 v-cache enabled dies
They should really have prepped to make Vcache standard on all their chips by now. It would actually allow them to package it more efficiently to where clocks wouldn't even need to be as affected by putting the SRAM die underneath instead of on top. They could include more cache this way if they wanted, though that'd probably have heavy diminishing returns.
Bingo. I get the heat implications but it's time to figure out a way around you would think being the 3rd generation of x3d. If it is just one sided again it will be really boring and the 9900x3d will be just as lame as the 7900x3d. I get 8 cores is all you need for gaming but I'm not going backwards in core count from the 13700k when I do other shit with my pc.
I wonder what is the reason behind Zen 5's strange performance. Sometimes we see decent gains but other times it's slower and less efficient on a newer node. And all this after nearly 2 years. This is such an RDNA3 moment.
Looks like the architecture changes didn‘t work out the way they hoped?
Maybe some prevention of side channels attacks while keeping performance?
Newer node? Isn't it the same node as zen 4?
@@zxy7529but they did, if you look at the 400 odd benchmarks on Phoronix you'll see loads of performance gains. But most people aren't running EPYCs, Threadripper or apps that will take advantage of the changes leading to regressions or bearly more than static performance changes.
And now we see the downsides of sharing the CCDs with servers because the dev time went towards those markets where the profit is made
@@samh5886 Zen 4 is TSMC N5 and Zen 5 is N4P apparently.
Edit: I see the pinned comment about the audio being "correct". That sounds like it may be something else, but I'm not sure. I'll still leave this here just in case 👍:
Hey Steve, I noticed in this video, when we see you standing, the audio bitrate seems a bit low and/or sounds heavily compressed. I have the Sony WH-1000XM5, thought it was my EQ, but as I went to go mess with it, I realized when the video switched to b-rolls or performance charts, the audio quality is fine.
This is an amazing video none the less, and I love the work ya'll do, just figured to let you know in case there was an issue with the mic!
Some Stan in the comments “But Zen 5 will save you 5 watts of total power for an extra $100🙌” LMFAO🤣🤣🤣
I mean, if yo live in a country where energy cost is absurd, like i live, it can be considered a point (specially the 9700X), but still can't justify that price increase 😂😂
@@HansTheilLuiz 😂😂😂
@@HansTheilLuiz ahahahahahahaahha
@@LCC.Gaming Try and run your CPU at 100% for 8 or plus hours a day.
Oh yeah, a gamer can't comprehend this use case.
@@mondodimotori funny because my pc is used for 8 hours a day for school and business. I’m lucky to play a game for at least an hour. Mind you this is done through my R5 7600💀
Thanks Steve.
Would have liked to see the 7600/7700 in the cost charts as someone who is interested in value would certainly consider these CPUs. Great work and thank you guys for putting in a solid effort and not skipping the details.
While that is true, people looking at the 7600/7700 would not be considering the 9950X or 7950X for that matter.
Oh no.... He's standing there...
MENACINGLY
@@kaznika6584 GET OUT OF THERE SPONGEBOB!!!!!!!
He realized...
As a 5950x owner that 7950x is looking real tempting, especially if they drop the price more since the 9750x has released...
Also a 5950x owner, and I always upgrade on a 4 year 2 product cycle. Never been more disappointed in a product. The fact that I could have had a 50% performance increase for the last two years, but waited for a 100% performance increase breaks my heart.
@@shoppster300same lol, just proves how much more the 5950x is worthy of admiration
Thank you AMD for saving me $649
I bought it. New build entirely tho.
Or use $649 to buy an untested Threadripper 7960X if you can find one
My 7950X is still perfectly fine for gaming. I game at 4K on a 144Hz monitor. I wasn’t really going to update this generation anyways.
Hopefully next gen provides a bigger uplift. 16 core CCD would be great for gaming. Unified cache for more cores. Would be especially nice for X3D products.
You really need to make the leap for an X3D processor. I got a 7950X3D for $420 December of 2023 and the difference is night/day for reducing 0.1% and 1% FPS LOWS.
Having additional cache helps prevent those stutters/low frame rates. It makes for a much more smooth gaming experience.
Unfortunately, it seems like you are kind of required to wait for X3D Zen 5 for a meaningful upgrade unless you plan to spend $520 to get X3D v-cache and NOT losing 8-cores.
Think 7950X3D are like $520 right now.
I see only couple of percentage difference between 7950X and the fastest gaming CPU in 4K gaming reviews.
Reviewers love 720p and 1080p to emphasize the difference, but nobody plays at these tiny resolutions if they got a 32” 4K screen + RTX 4090.
Also 7950X hits 144 fps in most games. No need to go above that on a 144Hz screen.
7950X3D is slower on code compile and I am a game developer, so I compile big C++ code bases daily. That’s another reason why I don’t see that CPU as a valid upgrade path.
9950X is 15% faster than 7950X in code compile, but gaming is only 2% faster. 9950X3D might be a reasonable update if the rumors are correct that it has higher clocks. But still, most games run either at above 144Hz already or are GPU bound at 4K. I intend to skip RTX 5000 series, so the GPU bottleneck at 4K is not going away.
I am fine waiting for Zen6 X3D. Rumors say that Zen6 will have 16 core CCD. That means X3D would cover 16 cores with unified big cache. That should be awesome for games. Update worth waiting.
@@sebbbi2 Tell em brother. I own a 4080 and never ever touch 1080p, same with when I had a 3080. A CPU upgrade would do nothing for me (outside of getting an X3D) even though I have "just" a 7700X. Most people in first-world countries aren't on 1080p anymore so the fact reviewers need to use 1080p to emphasize the difference for CPUs just shows that the CPU is largely irrelevant now for gaming performance.
I have no idea where the hell this 1% low talk is from because even when I had a 3080 and cranked the settings at 1440p my games just... ran fine. And obviously with my 4080 I've never seen a single stutter outside of Ghost of Tsushima after a few hours (PC port has known memory leak). I don't get how people are having stuttering issues caused by their CPU. Are people blaming shader compilation stutter and poor optimization on their CPUs? I simply don't believe it when it's described as if games will inevitably stutter no matter what because of the CPU. If games stutter, 99% of the time it's the game and not the CPU.
Above 1080p even my 4080 is bottlenecking the 7700X. It would take a 5090 or an even later GPU before even Zen 4 starts to become the bottleneck. Nobody should still be on 1080p outside of e-sports kids chasing 1000fps. You can grab a used 3080 for $300 now, there's no excuse. 1440p monitors are dirt cheap. If someone says "I can't afford 1440p" then why are they looking at $300+ brand-new CPUs?
Mom I want an AMD 9950
We have a 9950 at home
The 9950 at home is a first gen Phenom.
(I actually had a Phenom9950 on an Nvidia Nforce chipset, was my upgrade path from an Athlon X2 5200, which gave a good performance uplift)
Yes the first gen Phenoms were still mediocre against even the intel core2quads... but the higher end ones were a decent cost effective upgrade to dual core athlons.
18:53 I looked everywhere for this test, thank you!
Basically all I’m seeing here is, if you don’t need avx512 just save your money and get a 7950x on sale. The 7000 series was just such a good generation AMD is having a hard time beating it. Kind of a disappointing release. I hope the 9800X3D does better for gaming
will be with unlocked multiplier, if that has any worth
@@EtaCarinaeSC I’m thinking the zen5 architecture is bandwidth starved, so in theory 3D cache should allow it to stretch its legs
It will certainly be better than 7800x3d. 10-20% depending. It has a reworked cache system. Though.. it gains nothing for gaming when increasing the power so we are genuinely looking at a +5% before adding vcache so even if the vcache adds say.. 25% more of an uplift than vcache did for 7800x3d.. it will only be that 10-20% depending on the game.
😔
@@TheCompyshop It is not bandwidth starved. We've seen tests disproving this. Vcache is not going to magically unlock all this secret performance. It'll be the same bog standard increase that we get from any Vcache part. The bottleneck to better performance in Zen 5 is coming from software simply not being designed around taking advantage of a super wide architecture like this. Applications just dont need it. All the work AMD went through to make these big architecture changes just aren't gonna benefit consumers much.
I am confused, 7-Zip on linux is faster about 20%... while here on Windows we have a regression ???
Linux is alot lighter than window, most people will choose to use windows because of better software compatibility and experience in windows so that 20% z zip performance doesn't matter.
I guess google facebook and big corpo love zen 5, as it's optimized for server tasks
but for consumers, it's 5% thats it
I wonder how the X3D versions will look. Glad I got the 7800X3D.
If it's not X3D I won't buy it!
Just skip the 9000 series. It's really not worth it if you are already on AM4 or Zen 4 already. Even if you want to upgrade from AM4, the 7000 series will be much more value. The 9000 series is more of a generation in which AMD focused solely on efficiency.
@@Hardcore_Remixer- as in the video, 9950x consuming the same or more power than 7950x
the x3d chips are gonna be just as disappointing. what do you expect to change?
@@Hardcore_Remixer did you watch gamers nexus? 9000 series sucks at power efficiency as well. AMD planned this with intel, i have no doubts.
@@Amfibios Not expecting anything, just making a statement.
Bring on the 9950X3D thanks!
I’ll take that primarily due to the lower power draw but I’m also really keen on the AVX-512 improvement.
I think it's a bios issue
Because considering the IPC uplift seeing a performance regression is highly irregular
The IPC uplift was a lie man. That's the point here. AMD lied to people.
I really don't understand what's going on, how could they release a chip that's exactly the same, even slower than previous gen? They obviously made improvements, case in point the uplift in IPC and single core performance, so why is there no real world improvement, save for the outliers?
Those who chose to go with the 7800X3D are laughing right now 😂
Just got mine like a month ago, did it before motherboard prices where expected to go up for the zen 5 launch 😅
The 1080Ti of CPUs
7800x3D + b650 bundles might be the best value purchase in the market
As someone who just got a 7950x for work/light gaming I feel the same
7800x3d was down on price by 40% since release when i bought it (in my country at least) so i thought that zen5 performance uplift got priced in.
I feel as if i got a bargain :D
There's no question that the R7-5800X3D will be considered the GTX 1080 Ti of CPUs. At this rate, I might not need to upgrade before the release of the AM6 platform!
It really surprises me just how efficient the X3D CPUs are in Cinebench as the 7800X3D and 5800X3D sit as #1 and #2 respectively. The efficiency of the 7800X3D isn't too surprising since it's Zen4 but the efficiency of the 5800X3D, being a Zen3 part, is absolutely shocking.
9950X = 7950X
more like =
Bigger number better.
Better number higher.
Best summary. 😂
@@pikameenumbawan5280
@@animationgaming8539, programmer spotted. 😁
Zen 5 hurts everyone. Remember when AMD said AM5 would have a similar lifespan as AM4? Well, AM4 saw huge gains every generation. With Zen 5 AMD is almost trying to say AM5 supported an extra generation, when it was basically a useless generation. AM4 wasn't great because it got 4 CPU gens, but because every single one had its weight in the scale. Zen 5 is making Zen+ look better.
Still standing 😭🙏
😭
Steve at GN said it best.
Its not a bad product, it's just more of a refresh instead of new generation.
AMD has one final shot at making Zen5 mean anything with their X3D chips. If they can get even 10% average uplift over Zen4, that will at least make up for their general lineup.
That's hugely disappointing. I might still get it for the small (and economically unjustified) bump for 3D graphics creation, since I don't care as much about games, but I'll give it six months to see what happens: a price drop, or the moonshot of a magical microcode update.
It's astonishing when Intel drops the ball; AMD goes, 'nah, hold my beer,' instead of driving it home.
The difference between the 3950 vs 5950 vs 7950 vs 9950 would have shown that the big jump from 5950 to 7950 is really the whole package. Moving from DDR4 to DDR5 and the new architecture is what gave that huge performance jump. If you look at the 3950 to 5950 performance jump, its basically the same as going from 7950 to 9950. Platform to platform the differences are small in terms of performance boost moving from the old to the new.
The multi core productivty jump from Zen 2 to Zen 3 wasn't particularly large, but the single threaded performance jump was massive, as AMD went away from 2 4-core chiplets in one CCX to 1 8-core with unified L3. That netted them about 23% uplift in gaming and Photoshop, etc.
Well structured script, talking points and review. Addressing many of the methodology concerns about the 9600X / 9700X review...same conclusions hold.
9000 series definetly Zen 4,5 = Zen 4 + 0,5% performance increase
Keep getting reminded how crazy the 7800X3D is. Chugs well behind the newer models in productivity but still somehow flooring competition in games and power efficiency.
AM5 being stuck with basically the same gaming performance for 4 years is just depressing, because it'll likely be another 2 years for Zen 6. I'm so tired of modern tech, entertainment and society in general. Everything is stagnating, but more and more often it's regressing. I miss the 90's.
AMD decided to back-port Zen 5 to TSMC N4 because N3 was getting delayed due to yield issues. I don't think it will be 2 years before Zen 5+. Zen 5c is on N3 so it will be "trivial" for AMD to release what Zen 5 always should have been. N4 is only a 20% shrink over N5, whereas N3 is a full 70% shrink over N5, just like N5 was a full 70% shrink over N7. N4 is, unfortunately, the worst half-step node we've had in decades.
@@andersjjensenRocket Lake 2.0
@@saricubra2867 Sorta-ish, yes. Intel sacrificed two cores on the top model for a 5% single core boost (on average, there were regressions too).
I mean there's only so much you can blame the CPU for in gaming performance when it's current-gen GPUs that are bottlenecking them. X3D only gains performance in games that use the v-cache. I think those CPUs gave people the misconception that you can still gain FPS with a CPU upgrade nowadays. Dunno how you can call it depressing when my 7700X never sees above 20% usage paired with a 4080, except in very few ultra-demanding games. How is that depressing knowing you can spend $300 and have a CPU that will keep up in gaming for 5 years or more? Things used to be a lot worse.
@@JohnDoeWasntTaken You were perfectly right up until you said "above 20% usage". Total usage is almost meaningless in this context. Max core utilization is the end all be all in gaming. If a single core is pegged at 100% and the 7 others are at 0% (will never happen, but for arguments sake) you will see 12.5% CPU usage. But your point obviously still stands: At realistic settings and resolutions we are spoiled with CPU gaming performance. Vulkan and DX12 really cuts down on the amount of back-and-forth communication, between the CPU and GPU, that is needed to get the job done and honestly the industry was WAY too slow to adopt them.
The Zen 5 stands for 5% performance improvement.
This really should have been called Zen4+ because it's the same mediocre improvement Ryzen 2000 had over Ryzen 1000.
You’re the only one to include the 5950X in tests, that’s nice 👍
Good news: AMD is outstanding; Bad news: Steve is doing the standing
Steve deserves a standing ovation. Thank you in advance.
I had a feeling that this would happen. That is why I didn't get my hopes up to high, and the same might go for x3d naybe a little faster, but not by much. Again, I can and will wait for Zen-6.
HES JUST STANDING THERE..... MENACINGLY!!!
im glad i sold my 7700x and got the 7800x3d when it came out. was able to sell for 300 and buy the x3d chip for 350. definitely worth 50 dollars in performance and efficiency
Rocket Zen Moment
Did I saw Performance Regression while Drawing more power there
Im so happy with my 7950x. Undervolted it and together with PBO it can hit 6GHz. All while beeing fully stable .
There are good gains in single core performance. Why doesn't it lead to better gaming performance? Clocks seem to be about the same.
Memory.
People seem to misunderstand that Cinebench is not some 'all encompassing' benchmark. Cinebench only benchmarks performance of Cinebench. Just cuz it shows a moderate gain in single core performance doesn't mean other applications will. Because they aren't Cinebench. Frankly, Cinebench has been hugely overused as a general benchmark, given what it's benchmarking isn't an application many people use to begin with.
@@maynardburger CPUz, CPUMark and Geekbench show high single core uplifts too.
@@maynardburger CPUz, CPUMark and Geekbench show high uplifts too.
@maynardburger CPUz, CPUMark and Geekbench show high single core uplifts too. Actually all single core performance benchmarks show the most consistent results I've seen so far.
17:52 Interesting change in the white balance
The only thing I can think of is AMD engineering went to try and make the overall experience better than the 7000 series launch in regards to stability and RAM support.
But honestly marketing and pricing shot this things privates clean off.
Pew Pew 🏀🏈💥💥🔥🔥🔥
No, AMD went all-in on making Zen 5 a super wide architecture. That's why it's disappointing for consumers, cuz applications simply weren't bottlenecked by this before. Basically, these weren't really engineered for helping consumer workloads to begin with. At least not anytime in the near future. It'll fare better in datacenter and large AI farms and whatnot, but we simply weren't the priority here. Sucks, cuz it looks like the great CPU competition we'd have been having is basically gonna die out for the next couple years, given that Arrow Lake isn't expected to have huge performance improvements, either.
@@maynardburgerarrow lake actually is expected to have significant performance increases and efficiency from what ive seen. Whether Intel should be trusted with that is a whole other story.
Lol the stability on these so far has been wayyyy worse than Zen 4 though. This is just a total failure of a generation
@@maynardburgerbut these are consumer cpus for consumer market and not epyc, threadripper or even xeon
This release is reminiscent of the 2000 series vs 1000 series.
The 9000 series doesn't suck, the 7000 series was just a generation ahead of its time!!
/s
Too soon. To get us started on the copium. Not after two long years.
So happy I chose AMD for the first time in my life (over 25years of PC gaming) and went with the 7800X3D this time round.
Hi Steve could you test more with cities skyline 2 for cpu testing?
Steve... please sit down soon. We're concerned.
I'll sit for the video tonight then ❣️
Just bought 5700x3d, will skip the whole AM5 platform
jump to 7000 was a very good buy, coming from Intel.
I'll be like you in a few weeks
Am6 here come, maybe we'll see lol. Cuzz it's looking bleak out here😂😂
Thanks for the straightforward video titles and thumbnails. I always know what to expect and feel confident I'm getting the truth when I check in on the details. Also, thanks for testing ACC!
Why would AMD lie about the expected performance of Zen 5? It's not like no one was going to review the CPU's once they released (or use them for that matter). AMD must've known what Zen 5's ACTUAL performance numbers were, but they lied anyway with the full knowledge that everyone was going to find out. At this point I'm convinced AMD has to be shooting themselves in the foot on purpose.
They lied because they wanted to postpone the damage. If they'd told us months ago that Zen 5 would have super underwhelming performance, people would have been shitting on it back then. It was always going to happen, but they managed to delay the backlash a while.
@@maynardburger They didn't just delay the damage, they turned damage from disappointment into damage from essentially lying to their customers. All bad press isn't created equal.
AMD said their next crop of GPU's aren't going to be all that powerful compared to RDNA 3 (other than on RT), and they did so well in advance of release. Sure people are disappointed, but still looking forward to new cards. That's a much better situation vs saying the new GPU's will be 50% faster, then having everyone find out that was a lie.
Looks like they cherrypicked the games that benefitted the most from the new architecture and put those on the presentation slides. So, they didn't technically lie. There is one game on this HW Unboxed test that shows a 20% uplift because it's not memory starved by the IOD. Zen5 is a good architecture but on the desktop parts it's bottlenecked by the IOD.
Epyc's Zen5 processors should show the muscle as well as 3D cache versions when they get released.
They should have called the r9-9950x the r9-9950avx.
That's about the only reason to get this cpu over older gens.
I’m really started to wonder why AMD bothered to release these. They’ve had to hear all the talk about the I tel 13/14 gens being the same now they’ve done basically the same thing whether it’s on a new architecture or not.
Except at least Intel didn't claim anything crazy with the 14th gen and said that it was just a raptor refresh since they ditched meteor lake desktop. I'm waiting for arrow lake but I'm not excited for it either as no p core increase and no HT anymore. If it is at least 10% faster in gaming I may switch just to get away from the raptor lake scare going on or I may just stick it out as my cpu is fine
Great video. Sad to see the trend for Zen5 continuing for 9950x (probably will also for 9900x). As someone that currently owns and prefers AMD I don't feel the need for an upgrade (for home use at least). Definitely there are improvements for enterprise (data center) market but for now only 7950X3D can do both gaming and work with good performance, when compared with Intel (if power draw and degradation is put aside). For gaming only the 7800X3D is still the best, even after all this time and I do hope that Zen5 X3D parts will continue that legacy, but we'll see...
Just a heads up - The microphone sounded underwater for the Welcome, Conclusion and Final Thoughts chapters. Not sure what went wrong there.
Thanks for the review and tests. It's good that the test includes older generations such as 12900k
Alder Lake and Raptor Lake are extremely similar on IPC. Zen 4 is still slower than both to this day minus the X3D chips
mild refresh. nothing criminal
AMD's fault for their incompetent marketing of these chips
I mean, for building a first system as I am, the 9700x isn’t a bad choice at all for mixed use, and can only really get batter in the weeks to come with updates, as well as keeping in mind x870e on the horizon for faster memory. If I already had a 7700x system there’s no way I’d upgrade now, but since this is my first, I think I’m going with the Zen 5 8-core.
I have a 7600x, i only will upgrande to the max for am5.
Friendship ended with Lisa Su. Now Kim Jung Un is my best friend.
This has convinced me to stay on the AM4 platform, I'll take a look again at Zen 6.
everyone knew whats up when they saw steve stood up
9:03 commentary mistake here.
AMDs 14nm+++ moment
It makes Haswell and Skylake impressive by comparison. And those gains used to be considered minimal over Sandy Bridge back in the day.
Gamers Nexus and Hardware Unboxed have done a great job reviewing Ryzen 9000 and debunking efficiency an consumer productivity gains and "not meant for gaming" gaslighting from AMD stans. Thanks, Steves!
as someone who stopped gaming, the 9950x seems like the best for my production workload on my workstation
They should call it Zen 4+1%
That Single Core Perf and the Efficiency are Straight BEASTS! Amazing work @AMD