@@ciestermr1848 7800X3D is only about 20% faster than 5800X3D. While it's good i dont consider this as enough. I want at least 50% and this wont happen until Zen 6 in 2026 most likely. Assuming +20% from 7800X3D, +10% from 9800X3D and +20% from 11800X3D or whatever it ends up being called.
@@Raivo_Kits multiplicative. So even if it's 20%*10%*14% you only need 14% to reach 50% total over (our) 58x3d. I'm getting 9800x3d because i got a fancy oled 240hz monitor and I'm now cpu limited in The Finals (+-200fps). The only saving grace for zen5% is the overclocking unlocked for 9800x3d
Yeah, it's quite shocking that the 9950X is double the price of the 7800X3D (in Western Europe). Even the performance in applications besides gaming is not even remotely worth the price. I don't understand AMD. How hard is it to just keep the pricing the same as the previous generation. With performance gains of only a few percent, that's pretty reasonable.
How true. If you go AM5 the 7 series is still the best choice. AM4 old timers like me, I am thinking to upgrade to 5800x3D, maybe with the new Intel gpu, if it's not overpriced.
@@nortieroOnly upgrade to 5800x3D if you are afraid to swap your board and sell off old hardware. 5800x3D is a good chip performance wise but is a horrible value. 5700x3D is better but still would not recommend it. The 7600x is on avg 3% faster than a 5800x3D. Current pricing 5800x3D $339 7600x $195 32gb DDR5 6000/CL30 $105 That leaves $40 left towards your motherboard. $110-150 b650. So for $70 more than the cost of a 5800x3D you could be on a modern 7600x build. Now sell off your old hardware to recoup much of the cost. Also right now a very nice b650m Aorus Elite AX is $121 on Amazon.
lol the $223 AUD processor that nobody can find, nodoby is sure it ever existed beyond rumour and "mythical legend" - classic! Like one of those "rare spares", usually found in a pile of clearance parts, at computer expos a decade later. Finally! Yes! - haha... dumpster diving with CPUs.
higher cache is always going to be better, unless you want higher clock rates. Emulation might be one example where you want to put your money into higher clock rates and this cpu should deliver well.
I wouldn't be expecting huge differences, to the already existing 7800X3D, which with its 3D-Vcache, presents a great value gaming CPU right now. But you already have one! - lol - why would you even bother?
Depending on your GPU, you could stay on the 9900K forever. To be fair the 5800X3D, 7800X3D take you into a totally different world of performance so you could save money on the 7800X3D's discounts atm.
The big investment was AVX-512, it's a workstation CPU. Wait for the X3Ds for gaming. Phoronix/L1T show its legs are very long in Linux (where AVX512 is natively used across loads of different workloads/programs thanks to compiler optimizations). AVX512 is extremely demanding of memory, cache and internal busses/fabric. A large part of the bumped queue, cache, etc. are to accomodate the one-clock AVX512 extensions. So I think having no gaming *regression* is 'fine'. I get the disappointment with it being static for gaming, but the workstation performance, again avx512, is a huge leap forward, 15+ %.
I guess as long as you don't need more than 96gb ram then its fine for a workstation.. personally I am waiting, 9000 series doesn't seem all that attractive considering i need 128gb minimum for my workstation.
@@HardwareCanucks yeah my 7700x had to go -25 pbo otherwise it would crash... Just manual "OC" 5.4ghz at 1.18v... easy fps gains, stability, temps and higher low fps!
To be fair to AMD though... soooo many people freaked out over the power consumption and thermals of 7000 series... If you can get the same performance as 7000 series but at much lower power, thermals and noise, which makes them more synchronous to both ITX systems and in professional environments. Honestly, I'm happy with that. 7000 series is amazing and 9000 brought us that performance with more comfort or you can enable PBO for the extra oomph.
Same opinion, very like how the 9700x at 65w stock just didn't heat at all and still have good performance, ok it's overpriced and have same performance as last generation but I don't care.
PBO gives you like 2% more gaming performance using like 50% more power. It's not even worth using PBO. And for creators, you can find a cheap 7900X, enable Eco Mode and have better performance with less power than the 9700X with PBO.
@@AxleLotl no idea why my comments don't get posted, but use some of those pages to compare multicore performance or go see your benchmark's of choice database. Even going slower and having -15% IPC, the 7900 non-X (7900X in eco mode) has 50% more cores and threads to work.
You are wrong. The ZEN 5 efficiency lead over ZEN 4 has been tested to be not true in gaming for example. People are looking at the TDP values which DO NOT equal powerconsumption or efficiency.
AMD is aware of the Windows "bug" and Microsoft is presently working on an official update to fix the issue, therefore you will soon no longer have to use the super admin account to gain back that lost performance in games.
There must be something else going on apart from the admin account issue. Sure you have some performance gains, but that is true for AM5 as well as for AM4. That does not explain, why we see a performance uplift from AM4 to AM5 under Linux but not under Windows. There must be another issue between AM5 and Windows that hurts particular the Ryzen 9000er CPUs but hasn't been identified yet.
To add to your conclusion, I think the bigger issue is AMD's marketing of the chips in regards to gaming performance. All they had to do was be truthful about it and it wouldn't have been such a big issue.
Choosing to stay on AM4 with a 5600X3D is shaping up to be one of the better values deals I made and that is saying something. Only cost $119 after selling the bundle parts and here it is going head to head with some of the best CPUs currently available. My entire AM4 build consist of great premium parts acquired at similarly priced amazing values, a benefit of living down the street from a Micro Center, except for the Phantom Spirit cooler, that was all you HC. I am really happy to keep it gaming for a few more years.
I think the R7 7700 should be included in benchmarks as it also is a 65W TDP CPU, costs a ton less than 9700X, is way more efficient than 7700X and not that far behind. Especially the bulk versions here in Europe are cheaper than R5 7600X. And even if not, it's still quite cheap.
Please note that there has also been shown that there's a strange "bug" in windows that ca throttle performance in games for AMD Zen4 and Zen5 processors. It's mostly not all that major, but a few games showed about 10% performance improvement when run using the Administrator account. And no that's not the account you created when installing windows even though it has Admin abilities and can be used to configure just about everything in Windows. No this is the account named Administrator. In some localized versions of Windows it can be localized as I've seen on my machine. On my machine installed with settings for Sweden that account is called Administratör. So a Ö instead of an O... "But my machine doesn't have an account called Administrator!" you say. Sorry but it's there, just not activated. To see all accounts on your machine open a CMD window or a PowerShell window and type "net user" and press enter. That command lists all accounts on your computer, and one of them will be Administrator. To see the settings for that account type "net user Administrator" and press enter. This lists the data for that account and will include a line saying "Account active No" This shows that the account is deactivated. To be able to login using it you have to activate the account first. Now there are reassons that account was deactivated and that's security. By deactivating this your machine doesn't have an account that is named the same as on every other windows machine all over the world. To break a machine having an account name does help. If the bad guys know there is an account named Administrator that has every allowance there is it would be a good attack vector. So MS has disabled it by default. Now if you enable the account make sure it uses a tough password. Even if it's attacked there's no reason to make it easy to get in. The real question is why does it make a difference if you use the Administrator account instead of the user you created when installing the OS that appears to have all the admin abilities? I don't know, but according to AMD and the channel Hardware Unboxed it does make a difference. To see a video of their experiences and tests search RUclips for "Possible Windows Bug Found, Hurts Ryzen Gaming Performance". At about 1:55 in that video there are som info on what AMD said about it.
From what I have seen in certain reviews, the problem with 9000 series might be Windows itself, some have tested using Linux and these same chips run a lot better in Linux, so whether MS will fix this in windows is yet to be seen.
Are they running better in Linux or are other CPUs performing badly? From our understanding, a lot of the enhancements here were AVX-512 centric which wasn't something other Zen series accelerated with efficiency. Meanwhile a lot of Linux programs being tested do indeed use 512.
@@HardwareCanucks On Linux, native Windows game code was running better than it did in Windows 11, so it was not merely in productivity benchmarks. That's worth looking into.
here I am still rocking the R5 3600 watching all these high-end cpus drama intel 14th gen, ryzen x3d stuff, and now ryzen 9000. what a time to be alive.
I don't understand, what games struggle to run on these CPU's. Why even overclock or disable SMT at this point? People out here really chasing the dragon.
@@mrhassell Because there are some significant changes to the architecture, it could take time for a lot of software to be well optimized for it. It's possible that Zen 5 CPUs will turn out to actually be quite good for gaming, so people should keep an open mind and not write off the non X3D chips any time soon. Just the fact that Assetto Corsa already runs about 18% faster, with improved frame consistency as well, shows that this architecture has some serious potential to be better for gaming (that's also a game where frame consistency and high frame rate performance is considered to be extremely important, so serious players of that series who can afford it will definitely be upgrading to Zen 5). There will definitely be more games which will run better on Zen 5 than Zen 4 to come in the future. For now, the Ryzen 7600 is already the best value CPU (and motherboard platform) for a new desktop system for the majority of people looking to put together a new DIY desktop system, and Zen 6 will likely offer some excellent low-cost upgrade options in the future. Most people already on Zen 4 will have no need to upgrade to Zen 5, even if it does end up being closer to 15 or even 20% faster in most applications after a year or two.
@@mrhassell People are expecting supercomputers are this point. These chips are a little stronger than the previous ones at much lower energy cost. That's a big deal for anyone who just...uses their computer normally.
@@Caidezes much lower? About 7% lower power consumption at similar performance is much lower? (Comparing 7900x with 7700) And speaking about using a computer normally. Most desktop computers who are not used primarily as a gaming pc spend most of their life on idle or low working loads. AMD desktop CPUs consumes about 300-600% more power compared to Intel on such tasks. The "efficient" 9000 series included.
Doesn't seem like PBO nor SMT really check the box of "big generational gains" -- and neither should be necessary, since it never has been before. So, I'm scratching my head, like you mentioned the HWCanucks team is wondering what's going on in the outro. This is such a weird launch. I think there's something broken, maybe a bug, firmware issue, and we might figure out what it is in a month or two, but I can't imagine buying it right now when nobody knows what's going on. Over on Linux, It's a lot faster and a more normal-looking generational uplift than whatever's going on here on Windows, for who knows what reason. I have to agree it feels like they're focusing on server -- the great Linux performance almost just backs that up. But what the heck went wrong on Windows? I think it'll shake out to be a proper generational improvement on Windows, but it's absolutely nuts if that does happen... Since that'd still leave us with it having been so mixed at launch (smaller gains with some regressions right now) -- I don't remember such a weird and not-as-expected CPU launch, with such unexplainable results that totally miss the mark marketing gave us. That will stick in people's minds, they won't just forget about it like nothing happened. This is still the time AMD put out a baffling launch and nobody understood what was happening -- it's not the disappointment that stands out to me but the confusion, honestly speaking. (I also think they want to focus on both laptop and server with *perf per temperature* efficiency, and to a certain extent perf per watt efficiency, as that helps a lot on both server and laptop, where it's space-constrained and your cooler is less beefy. Maybe trying to prepare for an era of ARM really being the other big competitive choice for CPUs, even in our "bigger" compute devices. But it's not such a big deal for temperature efficiency on desktop, when we have monster air or even water coolers. Basically all of the current gen of air coolers can handle these chips at stock settings. So the gains are seen in other product segments entirely. I think laptop and server both have better margins than standalone CPU sales, and laptop/server for AMD is a business-to-business customer relationship to the laptop OEMs (businesses) or server customers (businesses), which is more return on effort for them. AMD and Nvidia both increasingly appear to prefer to be business-to-business... um... businesses than direct-to-consumer, IMO.) But back to this situation on desktop... yeah, I'm repeating some of Moore's Law is Dead's talking points here but I agreed with his video. There's something just off. Who knows what happened. Either way you slice it -- if this is the final performance, it's shocking. If there's a huge double digit uplift from a fixed bug, that also will be a second shock. Truly one of the CPU generation launches of all time. One for the history books. So dang strange. Just like MLID I could go on forever but I'll just go "what the f?-" and leave it there. (Sadly I don't have B-roll of a cute dog to share like he ended his video with.)
It's not broke. It's not meant for gaming. Start getting used to this because the technology is pretty much at its limits currently. Until they solve the internal heat problems at least. As for marketing, cigarettes never made you look cool, that midlife crises sports car won't make you more attractive to women, and you can't lose weight without dieting and exercise. A 16% IPC uplift doesn't necessarily make a CPU 16% faster. But you knew that didn't you?
It is very common in the history of computing for it to take time, often a lot of time, for new architectural features to be taken advantage of. It doesn't mean that those changes or additions to the architecture were bad, or a mistake, it's just that prioritizing better performance in the long run often means sacrificing how well a chip will perform when it's first released. It's hard to say to what extent this is true for Zen 5, but it's already been proven to actually perform significantly better in some specific applications and OS environments. Assetto Corsa already sees a 18% performance uplift, in addition to improved frame-time consistency, and this is also a game where frame-rate and frame-time consistency is extremely important and highly valued by the competitive players. That's proof that it's entirely possible that other games will also see significant performance improvements as well, and these chips have been made with only a marginally improved version of the manufacturing process that was used for Zen 4. Zen 3 was actually highly unusual in terms of how much better the performance was at launch for CPUs which were made using the same manufacturing process, same I/O chiplets, and same motherboard platform as the previous generation, especially because the previous generation was already considered very good. Zen 4 was actually much less impressive than Zen 3, if we consider that it was made with a significantly more advanced manufacturing process, more advanced I/O chiplets, and a new generation motherboard platform with a new and faster type of memory, but they also didn't actually make a lot of changes to the architecture vs. Zen 3 either, as I understand it, so that's understandable too.
@patrickweaver1105 I agree. But if you look at their slides they definitely marketed them as a strong upgrade for gamers. We are allowed to be annoyed at that deception.
@@PaulRoneClarke You're allowed to be annoyed, and you should be, mildly. You shouldn't be surprised. What annoys me is people act like it hasn't happened with every CPU, GPU, motherboard, and memory launch for the last fifty years. Even when there's a rare truly exceptional advancement it doesn't live up to the hype.
@@ich_16 I' m pretty sure manufacturers have never manufactured anything that is what the public thinks it should be at a price they think it should be. That's not how the world works.
I don't see 9700x as a shitty processor/flop in games, gives decent FPS, and it's much faster than 7700x in games, okay, but other benchmarks is a good uplift. Linux performance is AMAZING. So, I think if AMD adjust a bit the price they have yes a good processor. AMD still doing much better than Intel
Excellent vide0 Mike! If you want to tweak it further I have seen the largest gains with ram tweaking, faster freq and latencies, especially secondaries and tertiaries. Also I would take a look at win10 vs win11 performance 😊
You know what? I'm thankful that AMD has set the TDP to just 65W for 9700X, but I'm still going to go for X3D. I will wait and see what AMD has to offer in terms of X3D performance. Maybe a little bit over the 7000 series X3D chips. I could see myself setting the CPU down to 65W while still getting decent performance out of the CPUs. But at the same time, I am saddened that there is not that much of a performance increase for gaming. I'm hoping to get a new X3D CPU in order to pair it with my current 4070 non-Super GPU. Of course, Phoronix did perform nearly 400 benchmarks on those 9600X, 9700X, and 9950X and I'm very happy to see performance improvements across the board. However, speaking of server, I'm planning on building a NAS and idle power consumption is far too important to me. I'm going to go with Intel CPU instead. It will be just for backups and perhaps run lightweight virtual machines and that's about it. Hardware transcoding is not a necessity for me as I use Zidoo Z9X Pro Media Player which connects to my NFS share for watching movies.
@@mrbobgamingmemes9558 Yeah for sure. I haven't had an Intel CPU since before Zen1. I thought what AMD was doing was promising from the start so I haven't experienced Intel crazy power usage personally but I remember previously when I use to use dual GPUs and overclocked i7 6700K, that was definitely putting out a lot of heat during the summer. I'm sure you'd get similar experiences with a current Intel CPU plus a power-hungry GPU lol.
Reports I read this morning show running in full admin mode raises performance about 4%, but that's a security risk, so hopefully Microsoft can optimize fully
7:43 Right after commenting on there being no improvement to 1% lows, he says at least there are no performance regressions -- while showing the very slide that shows a 6% regression in 1% lows!
It's not compelling over Zen 4 for most people. Just hope you all keep the same energy for Arrow Lake. 4% faster than the 14900K in light loads and single thread work (games) and a big regression in MT performance. STILL using a PL2 of 250W (only 3W less than Raptor Lake) with the ability to use enough voltage and power to wipe your CPU (no lesson learned) in an unlimited mode. All for the high price of a brand new platform with no cross compatibility. And could still lose the the 7800X3D.
On AI productivity is there any improvements? With the new motherboard are we expecting an increase in performance o will not impact the numbers? Thanks for the video 👍👍
Yes, changing the price tag to $259 will fix 9700x, to get meaningful gen over gen improvements. I don’t believe it will perform better on all server load. There is still cherry picking happening when you look at benchmarks on Linux.
AMD wouldn't have approved this architecture for mass production if they didn't have good reason to believe that it will be able to offer better performance improvements than what we've seen so far, once software is optimized better to take advantage of the new architectural features. However, it has already shown some good performance improvements for certain workloads, and within various software environments. It is VERY common in the history of progress in computing for it to take time for this to happen. From a business standpoint, chip designers can't just design what they think will be the most efficient and performant chip, they have to try to strike a balance between having good performance out of the gate, and introducing new features which will help their performance in the long run. Sometimes these bets pay off, sometimes they don't, to varying degrees. AMD is in such a strong position now that they can afford to prioritize more future performance at the expense of less optimal performance in most software at launch. However, we also have to realize that their priority for this architecture is obviously what their server and enterprise customers want, with DIY and gaming customers being a fairly low priority. That being said, they still have the X3D variants, which are obviously fantastic CPUs for gaming enthusiasts. They did a poor job with PR and marketing of the 9000 series for the DIY and gaming enthusiast segment so far, that's for sure, but to think that this series is going to be a failure is very misguided.
@@syncmonism Microcenter is selling 9700x with free 32GB ddr5 6000 cl32 ram at $360. The ram is retailing at $109 at microcenter at this moment. They discounted 9700x by $109 just after one week. They got the messages.
they are targetting different power envelope and temperature range with zen 6 with almost same performance, it is mainly server chip , 4 and 6nm of zen 6 will replace 5 and 6nm of zen 5 in long term everywhere, yes they are trying to sell to gamers through shady marketing but it's still better than 13th and 14th gen which are oxidising which changes everything
Hi HwC, would you do tests limiting the 7700X at 90W and see the performance difference? It would justify whether the 9000 series improves efficiency or is a 7000 series refresh.
Isn't there that setting in the PBO options that letçs you increase the maximum clock by up to 200mhz? Would be curious to see what that does to the single core performance.
I guess i will stay with the 7700x for a vit and see how things pan out. But I do think as drivers and stuff like that are optimized things will get better in a few months
Not sure X3D will add much this time. If the underlying architecture is the same as the none X3D then I expect the performance of ZEN5's X3D to be just a couple of percent better than ZEN4's. Why would it not be? What special sauce could they add? More cache? Higher clocks? Higher power draw out of the box? I now expect this cycle's X3D to be a real disappointment.
Performance will likely improve over time as more software becomes better optimized for the new architectural features. This is very common with new architectural features. You shouldn't judge a chip by its performance at day one, just as you shouldn't assume that it will get better (or how long that will take). Nobody should buy Zen 5 based on the assumption that performance will improve in any of their preferred applications in a timely manner, but people should also keep an open mind. Zen 4 had great performance out of the gate, but it didn't have much in the way of changes to the architecture over Zen 3, and the performance gains were mostly to do with the fact that it was made using a SIGNIFICANTLY more advanced manufacturing process, in addition to using newer and more advanced I/O chiplets, and a whole new motherboard platform with a new and faster memory standard. Zen 3's performance uplift over Zen 2 from day one was actually highly unusual, in that there were big performance improvements on chips which used the same manufacturing process as the previous generation. Significant performance improvements happen when advances in hardware development come together with advances in software development. The hardware, the compiler(s) used, the OS, the application, etc.. all have to fit well together, and often the software won't get optimized until long after certain architectural features first appear in mass produced chips.
The real winner in cpu race is arm with apple m4 as champion It has 27 percent single core improvement over m3 and beat all arm and x86 CPU in single core score!!!
Perfectly fine CPU. The 9950X is going into my new Linux build machine. Replacing the old one that has a 5950X. Can't care less if it's not well supported on winblows. That said if you run a 7950X right now skip this one as it is and was utterly pointless to buy every new CPU generation.
Still on my 3700X and happy as ever. This is now a three generations older CPU that is still relevant. All these later generations are incremental upgrades as well, with two latest ones mostly benefiting from DDR5. Definitely not worth spending outwards of $1000 for a new MB, CPU, RAM and PSU to upgrade from AM4.
Yup, still great and honestly if all you needed it for was gaming then the 5800x 3D and GPU upgrade would be more cost effective especially if you play with higher resolutions.
I never under stand you people that upgrade the PSU or think you need to upgrade it when updating to a new platform. Nothing on the MB has changed that needs a new PSU. Power draws are not changed enough to require a brand new PSU, at least with Ryzen.
@@NyteStalker89 They left it broad but most people upgrade PSU's mostly for the increased wattage requirements for components especially if their first build was already at the threshold. Some people just want to spend money, no need to understand them lol
Did they do the same(PBO max / SMT off) with 7700x? but regardless the difference is minimal and if you consider the price difference zen4 at the moment is a no brainer for gamer on am5
New generations should beat previous out of the box. No over clocking or tweaking needed. They also bumped up the prices so they don't have to lower last Gen prices. Sleezy moves AMD
Doesn't the 9700x read temps from a different, cooler, area. Meaning running at the same temp is far worse. Isn't default temp limit also supposed to be 85° for that reason?
Not necessarily. There's an array of temperature sensors scattered throughout the die area. The temperatures derived from this are used to determine boost algorithms. AMD has optimized the location of several of those sensors to increase overall accuracy.
@@HardwareCanucks ok I'm just going off what I heard in other reviews. I've heard the temp target is supposed to be 85° on these tho and it's kind of glitched rn and keeps reverting to 7000 series target.
All the reviews I've seen so far seem to shout that there's something broken about the zen 5, but All I see is improved branch prediction, more alus, and some efficiency improvements. Those things don't really effect gaming much. Gaming is all about memory bandwidth. And BTW, Zen4 already had AVX512 support. SO, I'm wondering how is Zen5 supposed to be broken? This isn't even the gaming variant, so what were people expecting? I don't get it?
First, people want to see benchmark results that come at least close to what AMD promised (16% IPC gains on average). Second, people want to see Windows results in the same ballpark as Linux results. Then they will conclude that there is nothing broken with Zen 5.
People were expecting what the AMD marketing team lied to them about regarding gaming performance uplift prior to the release of this chip. It's pretty simple really. You don't need to be their pro bono attorney, you'd lose.
Isn't a gaming variant? This is the first time mainline CPUs have been released without gaming gains. Ya know these aren't server or even workstation chips right? Also based on what we know so far, there's no reason to expect the X3D versions of these chips to outperform last gen X3D. Lastly, even if they did, you'd be confining "gaming variant" CPUs to the high end of the price spectrum, which has never been the case before. Budget and mid tier CPUs have had gen over gen increases just like the high end. Except for now. Now something is wrong. Even AMD thinks so, since they promised so much more.
It's misleading, look at hardware unboxed and gamers nexus power usage stats. It's not really more efficient for gaming. Very cherry picked claims unfortunately
Curve optimizer is a very important part of PBO, why not explore it as well? There is also a new Curve Shaper that adds to optimizer, but it is a bit complicated.
Note that those are used to primarily undervolt. This is a 65W chip that's not being held back by thermals so there's zero need for it, even with PBO on.
@zbrkesbris5987 Jayztwocents played with the auto overclock/optimizer and got consistent crashes to where they just gave up and didn't include those results. It's likely that the chips are so new that they're optimizing for standard use cases first, and then focusing on overclocking stability once they've nailed the baseline driver support. Though I do agree, if you could optimize the thermals and power draw while maintaining stability and lift the performance that would be of pretty significant benefit. Then again, unfortunately the 13600K seems to beat the 9700X by an overall average of around 9% for $130 USD less, and is an older platform meaning additional cost-savings across the board. Was really hoping to upgrade my 5900X, but looks like it'll be a few more years to go.
@@HardwareCanucks there is also +200 mhz max frequency adjustment for boost algorithm, which in conjunction with more juice and undervolting might allow higher all-core boost.
This is what MOST reviewer get wrong, they lack the basic knowledge of physics. TDP DOES NOT equal power consumption or heat produced. Power consumption is NOT constant and most certainly not TDP. It varies all the time dynamically and here ZEN 4 is even better. Efficiency in very fluctuating loads like gaming ZEN 4 has considerably better efficiency. The TDP numbers DO NOT MATTER in gaming. It is the real power consumption (which is not measured in watts, it is measured with time, on watthours etc.) that counts. And ZEN 5 does not shine here compared to ZEN 4.
Nope the current 9000 series is cooked. So many people have tried different ways to improve its gaming performance but could only manage single-digit improvements. Meanwhile, the 7000, 5000, and 3000 series dunked on their predecessors without breaking a sweat. No one was attempting to turn off SMT, enable PBO, use a Windows admin account, or anything like that because the performance bump was already there. What we should be investigating is AMD for how they put misleading improvements in their marketing that didn't materialize. IMO that's why the response has been so negative: it failed to meet expectations.
The decrease in power consumption is worth something too. Zen 4 had the Eco Mode, which brought similar power savings at the cost of 10% less performance. But obviously most people prefer performance over efficiency and the CPU makers follow demand, at the expense of making PCs less efficient. . Personally, I got a 65W Zen 3 model partly to avoid that. Also partly because it was cheap, admittedly.
que terrilbe wnnnn! nadieeeeeeeeeeeeeeeeeee de todos los canales son realmente enfaticos con los ryzen 9000. Solo veo comparativas en gaming. Estos procesadores son SUPER EFICIENTES! no consumen nada En performance ganan por poco en stock. Activando el PBO en 9600 y 9700 son joya y rinden mejor e incluso con menos temperatura. PËRO!!!!! estoy chato de ver videos en donde compraran el performance en juevgos con el 9950X o el 9900X. ESOS DOS PROCESADORES no son para juegos!! son para trabajaos profesional! me desiluciona un poco que no se esta estudiando el proecuto como se debe y por salir con un video en la fecha se hace lo mismo de siempre en reviews. Le tengo mass cariño que la cresta al canal. Pero deben mejnorar eso !! y eso sucede con todos TODOS LOS CANALES! Un abrazo a todos!"
Wile you using PBO, why than you not using Curve Optimazer to ? It will lower the temps and power usage, and let the CPU clock ever on more higher frequency 🤨
Was with you all the way to the end until you mentioned Zen5 and servers, working in datacenters and solution design, this is so so so so far off target its crazy. Zen5 has been totally targetted at the laptop/desktop space, as businesses will see the benefit of being able to almost halve their power consumption at the laptop/desktop. Efficiency matters in the datacenter, but most AI hosts are usually running 8-16 core chips as processing power is nigh on irrelevant, its more the pcie lanes that Epyc offers for the price vs Intel. Generally speaking datacenter customers haven't forgiven AMD for their previous to Zen architecture being so bad, and they're still very much trying to get a foothold in that space against Intel
HardwareUnboxed Says something. Hardware Canucks says the same thing. I’m convinced. Ryzen 5 uses little power but gives you very little performance gain. Intel 14th Gen gives you modest gains but uses even more power than 13th. Which one is worse? I don’t think anyone would be having this conversation right now if AMD didn’t price these processors so stupidly. Basically go and buy a 7800X3D.
If PBO could offer performance bumps, and SMT off could offer performance bumps... I wonder what would happen if you did PBO + SMT Off (Maniacal Laughter) *Reality* Oh look 0.5% overall improvement
Maybe because I'm not in the enterprise server world, but why would companies use a desktop cpu in their servers when Epyc and even theeadripper exist? I know there are those new Epycs (Ryzen) with certified ECC, but i can't imagine servers using a 13600k on the Intel side.
Of course they haven't. My local shaman tells me that AMD would likely keep on releasing new CPU for this chipset, from this CPU generation till at least 2027.
CPU performs as well as last generation, noticeable improvements of AVX512 specific loads over the 7700x, otherwise, it's the same, pretty much identical. Over volt and disable core parking to gain anything, of course enabling EXPO / PBO ETC.. Exactly the same fiasco we just witnessed with Intel, all over again? 🤣
That can potentially change over time as software becomes better optimized, but time well tell. People should keep an open mind. Zen 3 was actually highly unusual, in that it was a CPU architecture which was made on the same manufacturing process, with the same I/O chiplets, on the same motherboard platform as the previous generation, and it offered big performance improvements over previous gen chips in pretty much all types of applications from day one, and without using more power.
Yes. 9000 series can be "fixed", even though it's not broken in the first place. When the x800 motherboards come out with better stability support for higher Mhz RAM. When the software patches for the 9000 series get released. You know, the same software patches that the 7000 series received; except the 7000 series CPU had legitimate and authentic problems, particularly when it came to running hot, which was declared as "normal".
Don't expect that about the ram. It's the memory controler in the IO die of these cpus that's the limiting factor here. It hasn't had any changes from the Ryzen 7000.
The "fix" for gamers will probably come in the form of an X3D version. Right now, those are so far ahead that I doubt overclocking of the 9000 series will compensate for that.
Intel SMT has always been weak. Because of that it was often smart to turn off if you have enough cores like 6. AMD SMT has always been strong. And has been made even stronger in Zen5. A lot of the chip is duplicated for SMT. This performance regression with SMT on can only be explained by a Windows scheduler fuck up. Leave SMT on and it will be fixed in windows with updates. 95C is not the maximum temperature for Zen4 and Zen5 but the ideal temperature. If you see your max clocks go down from a 5.7Ghz to lower you might need better cooling on your cpu. But if you reach maximum expected boost clocks, you know you are not running too hot.
@@jankees4037 Don't listen to your feelings. Listen to the AMD engineers. Your CPU is designed to run at 95C forever. The higher temperature your CPU is the easier it is to cool.
It is very clear that the BUS in the chip of the 9000 series is too small to process it all. Increasing power, and then core frequency should make gaming faster, which they don't. So the 9000 series behave actually very weird on games. Which was never the case with 7000-5000-3000 series. So AMD messed up with their whole chip design. Also with the parking stuff on the top models. They could have tackled that better. It shouldn't matter for gaming that 16 cores are on it. One shouldn't have to park half of them. That is latency failure of the 9000 series.
My all core overclock 13700K also runs like a dream and beats most AMD's even for a years old chip. Intel is so gigantically fast when you change many settings yourself in the BIOS. Don't wait for microcodes from Intel, all not needed.
If you want better gaming performance, why not upgrade GPU instead? These CPUs is clearly for production workload, any gain in gaming is just cherry on top.
That is simply not true about it being tune for "Sever market." You can't fix an internal problem in Zen 5. The dual predictors pipelines & decoders just don't work like how their engineer explained. only one is being used when running a single thread. At times the get pickup to use for newer instructions like AVX512 & a few other, that's' why all those other applications show large gains vs gaming. It just uses the same amount of IPC that Zen 4 has & uses the increased load, stored, & data entries of the cpu so you get IPC that's similar to Zen 4. What's crazy in your graphs is the Zen 4 V-cache cpu is so much faster than normal zen 4. people just don't get how large it is it's 20% on average for the basically the same cpu design. Buldars gate 3 it's 43 to 46% faster! for V-cache. cyberpunK 2077 it's 43% faster! for V-cache.
Pretty much AMD forgot quality control on the 9700X. Why would you bring out such power limited processor, when you already advertise in announcements to offer better than the 7700X. Big failure on AMD by not being consequent in the lineup.
AMD with their X3D chips are gaming chips period. AMD made a mistake to position their 9000 series chips as better gaming chips than the 7800X3D which is very false. Every review that are gaming first tanked because of how the chips where marketed. Anything AMD Ryzen 9***X is purely for productivity and not gaming. Simply wait for the 9800X3D, but kudos to AMD with the new chips. They're Great for us who are not into gaming.
My legendary 5800X3D will last me until AM6.
Sell upgrade 7800x3d gg
@@ciestermr1848 7800X3D is only about 20% faster than 5800X3D. While it's good i dont consider this as enough. I want at least 50% and this wont happen until Zen 6 in 2026 most likely. Assuming +20% from 7800X3D, +10% from 9800X3D and +20% from 11800X3D or whatever it ends up being called.
@@Raivo_Kits multiplicative. So even if it's 20%*10%*14% you only need 14% to reach 50% total over (our) 58x3d. I'm getting 9800x3d because i got a fancy oled 240hz monitor and I'm now cpu limited in The Finals (+-200fps). The only saving grace for zen5% is the overclocking unlocked for 9800x3d
5800X3D and 7800X3D are only very slow in productivity workloads. Gaming yes. But productivity no, they get blown away even by a mere 13600K.
@@jankees4037 That's what 7950X3D exists for. If you need both productivity with 32 threads and X3D for gaming then 7950X3D is the obvious choice.
The 5800x 3D & 7800x 3D basically stand out especially if your primary concern is gaming.
Yeah, it's quite shocking that the 9950X is double the price of the 7800X3D (in Western Europe). Even the performance in applications besides gaming is not even remotely worth the price. I don't understand AMD. How hard is it to just keep the pricing the same as the previous generation. With performance gains of only a few percent, that's pretty reasonable.
How true. If you go AM5 the 7 series is still the best choice. AM4 old timers like me, I am thinking to upgrade to 5800x3D, maybe with the new Intel gpu, if it's not overpriced.
@@nortieroOnly upgrade to 5800x3D if you are afraid to swap your board and sell off old hardware. 5800x3D is a good chip performance wise but is a horrible value. 5700x3D is better but still would not recommend it. The 7600x is on avg 3% faster than a 5800x3D.
Current pricing
5800x3D $339
7600x $195
32gb DDR5 6000/CL30 $105
That leaves $40 left towards your motherboard. $110-150 b650.
So for $70 more than the cost of a 5800x3D you could be on a modern 7600x build. Now sell off your old hardware to recoup much of the cost. Also right now a very nice b650m Aorus Elite AX is $121 on Amazon.
but i guess stellaris players are better with 9700x?
A lot of folks are also recommending the 5700 X3D....
I really like how you manage your graphs and the art style is great
Thanks!
Gaming: Ryzen 5600X3D (Good luck finding that one), 5700X3D, 5800X3D and 7800X3D.
5500X3D rumored too.
I'd remove the 5700x3d because I've seen benchmarks where the 5600x3d is faster due to clock speeds and if you need 8 cores grab the 5800x3d
@@dnice42 I wouldn't remove it due to availability.
@@jamesdavies686 I'm actually hoping it gets released
lol the $223 AUD processor that nobody can find, nodoby is sure it ever existed beyond rumour and "mythical legend" - classic! Like one of those "rare spares", usually found in a pile of clearance parts, at computer expos a decade later. Finally! Yes! - haha... dumpster diving with CPUs.
Maybe it is best to wait for a 9800X3D. My trusty i9 9900K can hang on a bit longer.
higher cache is always going to be better, unless you want higher clock rates. Emulation might be one example where you want to put your money into higher clock rates and this cpu should deliver well.
This was solid intel back then mine is still awesome also. Ring bus will be missed
Yeah, I can’t justify moving past my 9900 K
I wouldn't be expecting huge differences, to the already existing 7800X3D, which with its 3D-Vcache, presents a great value gaming CPU right now.
But you already have one! - lol - why would you even bother?
Depending on your GPU, you could stay on the 9900K forever. To be fair the 5800X3D, 7800X3D take you into a totally different world of performance so you could save money on the 7800X3D's discounts atm.
The big investment was AVX-512, it's a workstation CPU. Wait for the X3Ds for gaming.
Phoronix/L1T show its legs are very long in Linux (where AVX512 is natively used across loads of different workloads/programs thanks to compiler optimizations). AVX512 is extremely demanding of memory, cache and internal busses/fabric. A large part of the bumped queue, cache, etc. are to accomodate the one-clock AVX512 extensions.
So I think having no gaming *regression* is 'fine'. I get the disappointment with it being static for gaming, but the workstation performance, again avx512, is a huge leap forward, 15+ %.
I guess as long as you don't need more than 96gb ram then its fine for a workstation.. personally I am waiting, 9000 series doesn't seem all that attractive considering i need 128gb minimum for my workstation.
@@emory2025then why not get threadripper or something better?
What are you babbling about? AM5 supports 192GB RAM. And even ECC on select boards.
Intels excuses are coming back…
😂
I wished you included benchmarks for 7700x with PBO to compare apples to apples..
Exactly.
Us too. Sadly, our 7700X didn't pass our validation tests with PBO enabled. BUT we have a 7700 PBO vs 9700X PBO comparison coming up.
@@HardwareCanucks Thanks !
@HardwareCanucks The SMT disabling was interesting.
Any possibility that the 7800X3D could see an improvement with this feature?
@@HardwareCanucks yeah my 7700x had to go -25 pbo otherwise it would crash...
Just manual "OC" 5.4ghz at 1.18v... easy fps gains, stability, temps and higher low fps!
To be fair to AMD though... soooo many people freaked out over the power consumption and thermals of 7000 series...
If you can get the same performance as 7000 series but at much lower power, thermals and noise, which makes them more synchronous to both ITX systems and in professional environments.
Honestly, I'm happy with that. 7000 series is amazing and 9000 brought us that performance with more comfort or you can enable PBO for the extra oomph.
Same opinion, very like how the 9700x at 65w stock just didn't heat at all and still have good performance, ok it's overpriced and have same performance as last generation but I don't care.
PBO gives you like 2% more gaming performance using like 50% more power. It's not even worth using PBO. And for creators, you can find a cheap 7900X, enable Eco Mode and have better performance with less power than the 9700X with PBO.
@@Just_An_Ignacio And you have the data to back that ECO 7900X claim?
@@AxleLotl no idea why my comments don't get posted, but use some of those pages to compare multicore performance or go see your benchmark's of choice database. Even going slower and having -15% IPC, the 7900 non-X (7900X in eco mode) has 50% more cores and threads to work.
You are wrong. The ZEN 5 efficiency lead over ZEN 4 has been tested to be not true in gaming for example. People are looking at the TDP values which DO NOT equal powerconsumption or efficiency.
AMD is aware of the Windows "bug" and Microsoft is presently working on an official update to fix the issue, therefore you will soon no longer have to use the super admin account to gain back that lost performance in games.
There must be something else going on apart from the admin account issue. Sure you have some performance gains, but that is true for AM5 as well as for AM4. That does not explain, why we see a performance uplift from AM4 to AM5 under Linux but not under Windows. There must be another issue between AM5 and Windows that hurts particular the Ryzen 9000er CPUs but hasn't been identified yet.
Zen 4 gains like 10% performance too, so even that doesn't help Zen5 to look good unfortunately.
Yes, not expecting it to compete with X3D chips when it's not an X3D chip.
The 7700X matching the 5800X3D in the distance: 👀
To add to your conclusion, I think the bigger issue is AMD's marketing of the chips in regards to gaming performance. All they had to do was be truthful about it and it wouldn't have been such a big issue.
Well researched and explained. Thank you!
Choosing to stay on AM4 with a 5600X3D is shaping up to be one of the better values deals I made and that is saying something. Only cost $119 after selling the bundle parts and here it is going head to head with some of the best CPUs currently available. My entire AM4 build consist of great premium parts acquired at similarly priced amazing values, a benefit of living down the street from a Micro Center, except for the Phantom Spirit cooler, that was all you HC. I am really happy to keep it gaming for a few more years.
5800x3d showing the world AM4 will forever hold its place in the socket hall of fame
I think the R7 7700 should be included in benchmarks as it also is a 65W TDP CPU, costs a ton less than 9700X, is way more efficient than 7700X and not that far behind. Especially the bulk versions here in Europe are cheaper than R5 7600X. And even if not, it's still quite cheap.
We honestly didn't care about efficiency here. We threw it into the wind.
Please note that there has also been shown that there's a strange "bug" in windows that ca throttle performance in games for AMD Zen4 and Zen5 processors. It's mostly not all that major, but a few games showed about 10% performance improvement when run using the Administrator account. And no that's not the account you created when installing windows even though it has Admin abilities and can be used to configure just about everything in Windows. No this is the account named Administrator. In some localized versions of Windows it can be localized as I've seen on my machine. On my machine installed with settings for Sweden that account is called Administratör. So a Ö instead of an O...
"But my machine doesn't have an account called Administrator!" you say. Sorry but it's there, just not activated. To see all accounts on your machine open a CMD window or a PowerShell window and type "net user" and press enter. That command lists all accounts on your computer, and one of them will be Administrator. To see the settings for that account type "net user Administrator" and press enter. This lists the data for that account and will include a line saying "Account active No" This shows that the account is deactivated. To be able to login using it you have to activate the account first.
Now there are reassons that account was deactivated and that's security. By deactivating this your machine doesn't have an account that is named the same as on every other windows machine all over the world. To break a machine having an account name does help. If the bad guys know there is an account named Administrator that has every allowance there is it would be a good attack vector. So MS has disabled it by default.
Now if you enable the account make sure it uses a tough password. Even if it's attacked there's no reason to make it easy to get in.
The real question is why does it make a difference if you use the Administrator account instead of the user you created when installing the OS that appears to have all the admin abilities? I don't know, but according to AMD and the channel Hardware Unboxed it does make a difference.
To see a video of their experiences and tests search RUclips for "Possible Windows Bug Found, Hurts Ryzen Gaming Performance". At about 1:55 in that video there are som info on what AMD said about it.
From what I have seen in certain reviews, the problem with 9000 series might be Windows itself, some have tested using Linux and these same chips run a lot better in Linux, so whether MS will fix this in windows is yet to be seen.
Are they running better in Linux or are other CPUs performing badly? From our understanding, a lot of the enhancements here were AVX-512 centric which wasn't something other Zen series accelerated with efficiency. Meanwhile a lot of Linux programs being tested do indeed use 512.
@@HardwareCanucks On Linux, native Windows game code was running better than it did in Windows 11, so it was not merely in productivity benchmarks. That's worth looking into.
@@HardwareCanucks The Phoronix and Level 1 Tech Linux reviews showed better results on just about every benchmark.
Most of them don't use AVX 512.
here I am still rocking the R5 3600 watching all these high-end cpus drama intel 14th gen, ryzen x3d stuff, and now ryzen 9000. what a time to be alive.
I don't understand, what games struggle to run on these CPU's. Why even overclock or disable SMT at this point? People out here really chasing the dragon.
None. They run fine. It's just for some reason, people were expecting more?
@@mrhassell Because there are some significant changes to the architecture, it could take time for a lot of software to be well optimized for it. It's possible that Zen 5 CPUs will turn out to actually be quite good for gaming, so people should keep an open mind and not write off the non X3D chips any time soon. Just the fact that Assetto Corsa already runs about 18% faster, with improved frame consistency as well, shows that this architecture has some serious potential to be better for gaming (that's also a game where frame consistency and high frame rate performance is considered to be extremely important, so serious players of that series who can afford it will definitely be upgrading to Zen 5). There will definitely be more games which will run better on Zen 5 than Zen 4 to come in the future.
For now, the Ryzen 7600 is already the best value CPU (and motherboard platform) for a new desktop system for the majority of people looking to put together a new DIY desktop system, and Zen 6 will likely offer some excellent low-cost upgrade options in the future. Most people already on Zen 4 will have no need to upgrade to Zen 5, even if it does end up being closer to 15 or even 20% faster in most applications after a year or two.
@@mrhassell People are expecting supercomputers are this point. These chips are a little stronger than the previous ones at much lower energy cost. That's a big deal for anyone who just...uses their computer normally.
@@Caidezes much lower? About 7% lower power consumption at similar performance is much lower? (Comparing 7900x with 7700)
And speaking about using a computer normally. Most desktop computers who are not used primarily as a gaming pc spend most of their life on idle or low working loads. AMD desktop CPUs consumes about 300-600% more power compared to Intel on such tasks. The "efficient" 9000 series included.
Right? I have a 4090 and a 12700k and have no desire to upgrade my cpu.
Looking forward to the 9950x3d
I'm glad I recently built my new PC using the 7800X3D. Looks like it won't be topped on gaming performance for quite a while.
More to the point, what are those lovely speakers in the background? 😁
Doesn't seem like PBO nor SMT really check the box of "big generational gains" -- and neither should be necessary, since it never has been before. So, I'm scratching my head, like you mentioned the HWCanucks team is wondering what's going on in the outro. This is such a weird launch. I think there's something broken, maybe a bug, firmware issue, and we might figure out what it is in a month or two, but I can't imagine buying it right now when nobody knows what's going on.
Over on Linux, It's a lot faster and a more normal-looking generational uplift than whatever's going on here on Windows, for who knows what reason.
I have to agree it feels like they're focusing on server -- the great Linux performance almost just backs that up. But what the heck went wrong on Windows? I think it'll shake out to be a proper generational improvement on Windows, but it's absolutely nuts if that does happen... Since that'd still leave us with it having been so mixed at launch (smaller gains with some regressions right now) -- I don't remember such a weird and not-as-expected CPU launch, with such unexplainable results that totally miss the mark marketing gave us. That will stick in people's minds, they won't just forget about it like nothing happened. This is still the time AMD put out a baffling launch and nobody understood what was happening -- it's not the disappointment that stands out to me but the confusion, honestly speaking.
(I also think they want to focus on both laptop and server with *perf per temperature* efficiency, and to a certain extent perf per watt efficiency, as that helps a lot on both server and laptop, where it's space-constrained and your cooler is less beefy. Maybe trying to prepare for an era of ARM really being the other big competitive choice for CPUs, even in our "bigger" compute devices. But it's not such a big deal for temperature efficiency on desktop, when we have monster air or even water coolers. Basically all of the current gen of air coolers can handle these chips at stock settings. So the gains are seen in other product segments entirely. I think laptop and server both have better margins than standalone CPU sales, and laptop/server for AMD is a business-to-business customer relationship to the laptop OEMs (businesses) or server customers (businesses), which is more return on effort for them. AMD and Nvidia both increasingly appear to prefer to be business-to-business... um... businesses than direct-to-consumer, IMO.)
But back to this situation on desktop... yeah, I'm repeating some of Moore's Law is Dead's talking points here but I agreed with his video. There's something just off. Who knows what happened. Either way you slice it -- if this is the final performance, it's shocking. If there's a huge double digit uplift from a fixed bug, that also will be a second shock. Truly one of the CPU generation launches of all time. One for the history books. So dang strange. Just like MLID I could go on forever but I'll just go "what the f?-" and leave it there. (Sadly I don't have B-roll of a cute dog to share like he ended his video with.)
It's not broke. It's not meant for gaming. Start getting used to this because the technology is pretty much at its limits currently. Until they solve the internal heat problems at least. As for marketing, cigarettes never made you look cool, that midlife crises sports car won't make you more attractive to women, and you can't lose weight without dieting and exercise. A 16% IPC uplift doesn't necessarily make a CPU 16% faster. But you knew that didn't you?
It is very common in the history of computing for it to take time, often a lot of time, for new architectural features to be taken advantage of. It doesn't mean that those changes or additions to the architecture were bad, or a mistake, it's just that prioritizing better performance in the long run often means sacrificing how well a chip will perform when it's first released. It's hard to say to what extent this is true for Zen 5, but it's already been proven to actually perform significantly better in some specific applications and OS environments.
Assetto Corsa already sees a 18% performance uplift, in addition to improved frame-time consistency, and this is also a game where frame-rate and frame-time consistency is extremely important and highly valued by the competitive players. That's proof that it's entirely possible that other games will also see significant performance improvements as well, and these chips have been made with only a marginally improved version of the manufacturing process that was used for Zen 4.
Zen 3 was actually highly unusual in terms of how much better the performance was at launch for CPUs which were made using the same manufacturing process, same I/O chiplets, and same motherboard platform as the previous generation, especially because the previous generation was already considered very good. Zen 4 was actually much less impressive than Zen 3, if we consider that it was made with a significantly more advanced manufacturing process, more advanced I/O chiplets, and a new generation motherboard platform with a new and faster type of memory, but they also didn't actually make a lot of changes to the architecture vs. Zen 3 either, as I understand it, so that's understandable too.
its considered broken cuz this cpus should be the non X
simple
@patrickweaver1105 I agree. But if you look at their slides they definitely marketed them as a strong upgrade for gamers. We are allowed to be annoyed at that deception.
@@PaulRoneClarke You're allowed to be annoyed, and you should be, mildly. You shouldn't be surprised. What annoys me is people act like it hasn't happened with every CPU, GPU, motherboard, and memory launch for the last fifty years. Even when there's a rare truly exceptional advancement it doesn't live up to the hype.
@@ich_16 I' m pretty sure manufacturers have never manufactured anything that is what the public thinks it should be at a price they think it should be. That's not how the world works.
I don't see 9700x as a shitty processor/flop in games, gives decent FPS, and it's much faster than 7700x in games, okay, but other benchmarks is a good uplift. Linux performance is AMAZING. So, I think if AMD adjust a bit the price they have yes a good processor. AMD still doing much better than Intel
Excellent vide0 Mike! If you want to tweak it further I have seen the largest gains with ram tweaking, faster freq and latencies, especially secondaries and tertiaries. Also I would take a look at win10 vs win11 performance 😊
I really like the 1080p/1440p chart style.
You know what? I'm thankful that AMD has set the TDP to just 65W for 9700X, but I'm still going to go for X3D. I will wait and see what AMD has to offer in terms of X3D performance. Maybe a little bit over the 7000 series X3D chips. I could see myself setting the CPU down to 65W while still getting decent performance out of the CPUs. But at the same time, I am saddened that there is not that much of a performance increase for gaming. I'm hoping to get a new X3D CPU in order to pair it with my current 4070 non-Super GPU.
Of course, Phoronix did perform nearly 400 benchmarks on those 9600X, 9700X, and 9950X and I'm very happy to see performance improvements across the board. However, speaking of server, I'm planning on building a NAS and idle power consumption is far too important to me. I'm going to go with Intel CPU instead. It will be just for backups and perhaps run lightweight virtual machines and that's about it. Hardware transcoding is not a necessity for me as I use Zidoo Z9X Pro Media Player which connects to my NFS share for watching movies.
Looking at what the Core i7 is pulling for power compared to the AMD parts is hilarious 🤣
This is why i would not consider intel except their lower end cpu maybe , because of awfull performance per watt . And it might be unusable on summer
@@mrbobgamingmemes9558 Yeah for sure. I haven't had an Intel CPU since before Zen1. I thought what AMD was doing was promising from the start so I haven't experienced Intel crazy power usage personally but I remember previously when I use to use dual GPUs and overclocked i7 6700K, that was definitely putting out a lot of heat during the summer. I'm sure you'd get similar experiences with a current Intel CPU plus a power-hungry GPU lol.
Reports I read this morning show running in full admin mode raises performance about 4%, but that's a security risk, so hopefully Microsoft can optimize fully
Zen 4 gets the same increases with an elevated Admin account.
7:43 Right after commenting on there being no improvement to 1% lows, he says at least there are no performance regressions -- while showing the very slide that shows a 6% regression in 1% lows!
It's not compelling over Zen 4 for most people. Just hope you all keep the same energy for Arrow Lake. 4% faster than the 14900K in light loads and single thread work (games) and a big regression in MT performance. STILL using a PL2 of 250W (only 3W less than Raptor Lake) with the ability to use enough voltage and power to wipe your CPU (no lesson learned) in an unlimited mode. All for the high price of a brand new platform with no cross compatibility. And could still lose the the 7800X3D.
Intel users looking at 150w: rookie numbers 😂
Haha, power consumption go brrr
Yet you can cool it. AMDs IHS design was so dumb on AM5. Should have just made a new mounting system 😅.
On AI productivity is there any improvements? With the new motherboard are we expecting an increase in performance o will not impact the numbers? Thanks for the video 👍👍
Yes, changing the price tag to $259 will fix 9700x, to get meaningful gen over gen improvements.
I don’t believe it will perform better on all server load. There is still cherry picking happening when you look at benchmarks on Linux.
AMD wouldn't have approved this architecture for mass production if they didn't have good reason to believe that it will be able to offer better performance improvements than what we've seen so far, once software is optimized better to take advantage of the new architectural features. However, it has already shown some good performance improvements for certain workloads, and within various software environments.
It is VERY common in the history of progress in computing for it to take time for this to happen.
From a business standpoint, chip designers can't just design what they think will be the most efficient and performant chip, they have to try to strike a balance between having good performance out of the gate, and introducing new features which will help their performance in the long run. Sometimes these bets pay off, sometimes they don't, to varying degrees. AMD is in such a strong position now that they can afford to prioritize more future performance at the expense of less optimal performance in most software at launch. However, we also have to realize that their priority for this architecture is obviously what their server and enterprise customers want, with DIY and gaming customers being a fairly low priority. That being said, they still have the X3D variants, which are obviously fantastic CPUs for gaming enthusiasts.
They did a poor job with PR and marketing of the 9000 series for the DIY and gaming enthusiast segment so far, that's for sure, but to think that this series is going to be a failure is very misguided.
@@syncmonism Microcenter is selling 9700x with free 32GB ddr5 6000 cl32 ram at $360. The ram is retailing at $109 at microcenter at this moment. They discounted 9700x by $109 just after one week. They got the messages.
they are targetting different power envelope and temperature range with zen 6 with almost same performance, it is mainly server chip , 4 and 6nm of zen 6 will replace 5 and 6nm of zen 5 in long term everywhere, yes they are trying to sell to gamers through shady marketing but it's still better than 13th and 14th gen which are oxidising which changes everything
Hi HwC, would you do tests limiting the 7700X at 90W and see the performance difference? It would justify whether the 9000 series improves efficiency or is a 7000 series refresh.
What's the chart look like for PBO on and SMT off??
I'm shocked that Starfield is that multithreaded. I assumed SMT wouldn't help it.
Maybe another deceptive video that is actually an ad from you could fix it?
Would be nice to see how performance ist affected when you vary the PPT Limit in steps of 10W from 88W-142W to see if there is a sweet spot.
The sweet spot is (IMO) what AMD set it to.
Isn't there that setting in the PBO options that letçs you increase the maximum clock by up to 200mhz? Would be curious to see what that does to the single core performance.
Did you try turning it off, and then back on.
I guess i will stay with the 7700x for a vit and see how things pan out. But I do think as drivers and stuff like that are optimized things will get better in a few months
Really want to see 9950x with PBO now
Not sure X3D will add much this time. If the underlying architecture is the same as the none X3D then I expect the performance of ZEN5's X3D to be just a couple of percent better than ZEN4's.
Why would it not be? What special sauce could they add? More cache? Higher clocks? Higher power draw out of the box?
I now expect this cycle's X3D to be a real disappointment.
Performance will likely improve over time as more software becomes better optimized for the new architectural features. This is very common with new architectural features. You shouldn't judge a chip by its performance at day one, just as you shouldn't assume that it will get better (or how long that will take). Nobody should buy Zen 5 based on the assumption that performance will improve in any of their preferred applications in a timely manner, but people should also keep an open mind.
Zen 4 had great performance out of the gate, but it didn't have much in the way of changes to the architecture over Zen 3, and the performance gains were mostly to do with the fact that it was made using a SIGNIFICANTLY more advanced manufacturing process, in addition to using newer and more advanced I/O chiplets, and a whole new motherboard platform with a new and faster memory standard.
Zen 3's performance uplift over Zen 2 from day one was actually highly unusual, in that there were big performance improvements on chips which used the same manufacturing process as the previous generation. Significant performance improvements happen when advances in hardware development come together with advances in software development. The hardware, the compiler(s) used, the OS, the application, etc.. all have to fit well together, and often the software won't get optimized until long after certain architectural features first appear in mass produced chips.
The real winner in cpu race is arm with apple m4 as champion
It has 27 percent single core improvement over m3 and beat all arm and x86 CPU in single core score!!!
Perfectly fine CPU. The 9950X is going into my new Linux build machine. Replacing the old one that has a 5950X. Can't care less if it's not well supported on winblows. That said if you run a 7950X right now skip this one as it is and was utterly pointless to buy every new CPU generation.
Is that the emperor of mankind in the background?
Still on my 3700X and happy as ever. This is now a three generations older CPU that is still relevant. All these later generations are incremental upgrades as well, with two latest ones mostly benefiting from DDR5. Definitely not worth spending outwards of $1000 for a new MB, CPU, RAM and PSU to upgrade from AM4.
Yup, still great and honestly if all you needed it for was gaming then the 5800x 3D and GPU upgrade would be more cost effective especially if you play with higher resolutions.
I never under stand you people that upgrade the PSU or think you need to upgrade it when updating to a new platform. Nothing on the MB has changed that needs a new PSU. Power draws are not changed enough to require a brand new PSU, at least with Ryzen.
@@NyteStalker89 They left it broad but most people upgrade PSU's mostly for the increased wattage requirements for components especially if their first build was already at the threshold. Some people just want to spend money, no need to understand them lol
Focus your efforts on the GPU and you'll be fine.
now we need a test where you test differenr ram speeds
We're doing that diligently since one of our chips already seems to have a weak memory controller.
Did they do the same(PBO max / SMT off) with 7700x? but regardless the difference is minimal and if you consider the price difference zen4 at the moment is a no brainer for gamer on am5
New generations should beat previous out of the box. No over clocking or tweaking needed. They also bumped up the prices so they don't have to lower last Gen prices. Sleezy moves AMD
Doesn't the 9700x read temps from a different, cooler, area. Meaning running at the same temp is far worse. Isn't default temp limit also supposed to be 85° for that reason?
Not necessarily. There's an array of temperature sensors scattered throughout the die area. The temperatures derived from this are used to determine boost algorithms. AMD has optimized the location of several of those sensors to increase overall accuracy.
@@HardwareCanucks ok I'm just going off what I heard in other reviews. I've heard the temp target is supposed to be 85° on these tho and it's kind of glitched rn and keeps reverting to 7000 series target.
All the reviews I've seen so far seem to shout that there's something broken about the zen 5, but All I see is improved branch prediction, more alus, and some efficiency improvements. Those things don't really effect gaming much. Gaming is all about memory bandwidth. And BTW, Zen4 already had AVX512 support. SO, I'm wondering how is Zen5 supposed to be broken? This isn't even the gaming variant, so what were people expecting? I don't get it?
Zen 4 had a 2x256 AVX functionality not full bore 1x512, and you are right there seems to be something wrong with Zen 5 at launch.
First, people want to see benchmark results that come at least close to what AMD promised (16% IPC gains on average). Second, people want to see Windows results in the same ballpark as Linux results. Then they will conclude that there is nothing broken with Zen 5.
People were expecting what the AMD marketing team lied to them about regarding gaming performance uplift prior to the release of this chip. It's pretty simple really. You don't need to be their pro bono attorney, you'd lose.
Isn't a gaming variant? This is the first time mainline CPUs have been released without gaming gains. Ya know these aren't server or even workstation chips right? Also based on what we know so far, there's no reason to expect the X3D versions of these chips to outperform last gen X3D. Lastly, even if they did, you'd be confining "gaming variant" CPUs to the high end of the price spectrum, which has never been the case before. Budget and mid tier CPUs have had gen over gen increases just like the high end. Except for now. Now something is wrong. Even AMD thinks so, since they promised so much more.
So Gameers 5800x3D , 7800x3D , I'm not a gamer (well its A tertiary option) so 7900 (NON-X) & the 7950x3D seem to be the best options for AMD
Which speakers are you using on the table behind?
They are Edifier 1280T
Do you have any rational explanation as to why we in Canada have to pay as much $200.00 more per Component than in the U.S. ???
Why no 7700x PBO?
You said you're going level the playing field
Because that doesn't level the playing field. ;)
@@HardwareCanucks what do you mean?
I'll stick with my direct die 7800x3d. If I was building itx 10ish liters right now, I'd def get a zen5 cpu. That 65w tdp is nuts for the performance.
7700 non-x is also 65w tdp and its identical.
the 7800x3d is literally more efficient than the zen 5 chips lmao
It's the same as 7700. AMD just renamed their non X CPU to one with an X in the name. Pure marketing.
why are you direct die cooling a CPU that can be easily cooled even with a single tower air cooler???
It's misleading, look at hardware unboxed and gamers nexus power usage stats. It's not really more efficient for gaming. Very cherry picked claims unfortunately
for SMT off you need only 12 cores chips,with pbo enable and curve otimizer -15 all cores
Curve optimizer is a very important part of PBO, why not explore it as well? There is also a new Curve Shaper that adds to optimizer, but it is a bit complicated.
Note that those are used to primarily undervolt. This is a 65W chip that's not being held back by thermals so there's zero need for it, even with PBO on.
@zbrkesbris5987 Jayztwocents played with the auto overclock/optimizer and got consistent crashes to where they just gave up and didn't include those results. It's likely that the chips are so new that they're optimizing for standard use cases first, and then focusing on overclocking stability once they've nailed the baseline driver support. Though I do agree, if you could optimize the thermals and power draw while maintaining stability and lift the performance that would be of pretty significant benefit.
Then again, unfortunately the 13600K seems to beat the 9700X by an overall average of around 9% for $130 USD less, and is an older platform meaning additional cost-savings across the board. Was really hoping to upgrade my 5900X, but looks like it'll be a few more years to go.
@@HardwareCanucks there is also +200 mhz max frequency adjustment for boost algorithm, which in conjunction with more juice and undervolting might allow higher all-core boost.
The idea is:
Stock is power capped
PBO is voltage capped (1.35v)
PBO+undervolt could be frequency capped...
This is what MOST reviewer get wrong, they lack the basic knowledge of physics. TDP DOES NOT equal power consumption or heat produced. Power consumption is NOT constant and most certainly not TDP. It varies all the time dynamically and here ZEN 4 is even better. Efficiency in very fluctuating loads like gaming ZEN 4 has considerably better efficiency. The TDP numbers DO NOT MATTER in gaming. It is the real power consumption (which is not measured in watts, it is measured with time, on watthours etc.) that counts. And ZEN 5 does not shine here compared to ZEN 4.
PBO will work best for the coming MOBO (zen 5)
super helpful
SMT on +PBO should be checked as well
Nope the current 9000 series is cooked. So many people have tried different ways to improve its gaming performance but could only manage single-digit improvements. Meanwhile, the 7000, 5000, and 3000 series dunked on their predecessors without breaking a sweat. No one was attempting to turn off SMT, enable PBO, use a Windows admin account, or anything like that because the performance bump was already there. What we should be investigating is AMD for how they put misleading improvements in their marketing that didn't materialize. IMO that's why the response has been so negative: it failed to meet expectations.
" it failed to meet expectations."
The most sensible comment I've seen so far
The decrease in power consumption is worth something too. Zen 4 had the Eco Mode, which brought similar power savings at the cost of 10% less performance. But obviously most people prefer performance over efficiency and the CPU makers follow demand, at the expense of making PCs less efficient. .
Personally, I got a 65W Zen 3 model partly to avoid that. Also partly because it was cheap, admittedly.
que terrilbe wnnnn! nadieeeeeeeeeeeeeeeeeee de todos los canales son realmente enfaticos con los ryzen 9000. Solo veo comparativas en gaming. Estos procesadores son SUPER EFICIENTES! no consumen nada En performance ganan por poco en stock. Activando el PBO en 9600 y 9700 son joya y rinden mejor e incluso con menos temperatura. PËRO!!!!! estoy chato de ver videos en donde compraran el performance en juevgos con el 9950X o el 9900X. ESOS DOS PROCESADORES no son para juegos!! son para trabajaos profesional!
me desiluciona un poco que no se esta estudiando el proecuto como se debe y por salir con un video en la fecha se hace lo mismo de siempre en reviews. Le tengo mass cariño que la cresta al canal. Pero deben mejnorar eso !! y eso sucede con todos TODOS LOS CANALES!
Un abrazo a todos!"
This has nothing to do with the content of this, but I have to say this video looks really good. Clear, poppy, contrasty.
Thanks
What’s the problem… i think it is already fast enough with more power efficiency.
Thanks Canucks
Me just retiring a 3770k to play more, does not ring any bells.
Wile you using PBO, why than you not using Curve Optimazer to ? It will lower the temps and power usage, and let the CPU clock ever on more higher frequency 🤨
That might have been the case for 105W CPUs but it isn't the case here with a 65W one.
Was with you all the way to the end until you mentioned Zen5 and servers, working in datacenters and solution design, this is so so so so far off target its crazy. Zen5 has been totally targetted at the laptop/desktop space, as businesses will see the benefit of being able to almost halve their power consumption at the laptop/desktop. Efficiency matters in the datacenter, but most AI hosts are usually running 8-16 core chips as processing power is nigh on irrelevant, its more the pcie lanes that Epyc offers for the price vs Intel. Generally speaking datacenter customers haven't forgiven AMD for their previous to Zen architecture being so bad, and they're still very much trying to get a foothold in that space against Intel
HardwareUnboxed Says something. Hardware Canucks says the same thing. I’m convinced.
Ryzen 5 uses little power but gives you very little performance gain. Intel 14th Gen gives you modest gains but uses even more power than 13th. Which one is worse?
I don’t think anyone would be having this conversation right now if AMD didn’t price these processors so stupidly.
Basically go and buy a 7800X3D.
What about pbo + smt off?
Some chips of the prev gen benefit from undervolting could it also be the case in this gen ?
If PBO could offer performance bumps, and SMT off could offer performance bumps... I wonder what would happen if you did PBO + SMT Off (Maniacal Laughter)
*Reality*
Oh look 0.5% overall improvement
Maybe because I'm not in the enterprise server world, but why would companies use a desktop cpu in their servers when Epyc and even theeadripper exist? I know there are those new Epycs (Ryzen) with certified ECC, but i can't imagine servers using a 13600k on the Intel side.
They havent released all models yet though right?
Of course they haven't. My local shaman tells me that AMD would likely keep on releasing new CPU for this chipset, from this CPU generation till at least 2027.
CPU performs as well as last generation, noticeable improvements of AVX512 specific loads over the 7700x, otherwise, it's the same, pretty much identical.
Over volt and disable core parking to gain anything, of course enabling EXPO / PBO ETC.. Exactly the same fiasco we just witnessed with Intel, all over again? 🤣
That can potentially change over time as software becomes better optimized, but time well tell. People should keep an open mind.
Zen 3 was actually highly unusual, in that it was a CPU architecture which was made on the same manufacturing process, with the same I/O chiplets, on the same motherboard platform as the previous generation, and it offered big performance improvements over previous gen chips in pretty much all types of applications from day one, and without using more power.
What would happen if you under volt it then turn on pbo like with the 5000 series 🤔
zen 6 can fix zen 5
Yes. 9000 series can be "fixed", even though it's not broken in the first place.
When the x800 motherboards come out with better stability support for higher Mhz RAM. When the software patches for the 9000 series get released. You know, the same software patches that the 7000 series received; except the 7000 series CPU had legitimate and authentic problems, particularly when it came to running hot, which was declared as "normal".
Don't expect that about the ram. It's the memory controler in the IO die of these cpus that's the limiting factor here. It hasn't had any changes from the Ryzen 7000.
That's an empty promise. I'd rather be running slightly lower clocked memory with tight timings than high memory frequencies with crap timings.
The "fix" for gamers will probably come in the form of an X3D version. Right now, those are so far ahead that I doubt overclocking of the 9000 series will compensate for that.
Intel SMT has always been weak. Because of that it was often smart to turn off if you have enough cores like 6.
AMD SMT has always been strong. And has been made even stronger in Zen5. A lot of the chip is duplicated for SMT. This performance regression with SMT on can only be explained by a Windows scheduler fuck up. Leave SMT on and it will be fixed in windows with updates.
95C is not the maximum temperature for Zen4 and Zen5 but the ideal temperature. If you see your max clocks go down from a 5.7Ghz to lower you might need better cooling on your cpu. But if you reach maximum expected boost clocks, you know you are not running too hot.
95 degrees is not the ideal temperature. It is AMD running hot like Intel runs hot at 100 degrees. Most AMD fans don't like 95 degrees on their CPU's.
@@jankees4037 Don't listen to your feelings. Listen to the AMD engineers. Your CPU is designed to run at 95C forever. The higher temperature your CPU is the easier it is to cool.
@@Ferdinand208 "The higher temperature your CPU is the easier it is to cool." Wait. What?
Fanboy tears 😂 Best thumbnail ever
It is very clear that the BUS in the chip of the 9000 series is too small to process it all. Increasing power, and then core frequency should make gaming faster, which they don't. So the 9000 series behave actually very weird on games. Which was never the case with 7000-5000-3000 series. So AMD messed up with their whole chip design. Also with the parking stuff on the top models. They could have tackled that better. It shouldn't matter for gaming that 16 cores are on it. One shouldn't have to park half of them. That is latency failure of the 9000 series.
Hey, we have a BUS expert here...
Yeah I think ill stick with my 13600k. Costs way less and beats it in production workloads. Also have mine undervolted so wont have any issues.
My all core overclock 13700K also runs like a dream and beats most AMD's even for a years old chip. Intel is so gigantically fast when you change many settings yourself in the BIOS. Don't wait for microcodes from Intel, all not needed.
If you want better gaming performance, why not upgrade GPU instead? These CPUs is clearly for production workload, any gain in gaming is just cherry on top.
That's what I said in the review
yes, but what about this reddish forhead ?
Well if you run Linux you get a very good performance on the new 9000 series, so everything is pointing toward Microsoft nothing else.
Isnt the predecessor for the 9700x the 7700 non X?
Nope. That is just the same TDP.
Is JIUSHARK JF13K Diamond cooler good for modern platforms
Ryzen 2000 series all over again. Pointless release. Hopefully next gen has the leap 3000 series or 5000 series had.
That is simply not true about it being tune for "Sever market."
You can't fix an internal problem in Zen 5. The dual predictors pipelines & decoders just don't work like how their engineer explained. only one is being used when running a single thread. At times the get pickup to use for newer instructions like AVX512 & a few other, that's' why all those other applications show large gains vs gaming. It just uses the same amount of IPC that Zen 4 has & uses the increased load, stored, & data entries of the cpu so you get IPC that's similar to Zen 4.
What's crazy in your graphs is the Zen 4 V-cache cpu is so much faster than normal zen 4. people just don't get how large it is it's 20% on average for the basically the same cpu design.
Buldars gate 3 it's 43 to 46% faster! for V-cache.
cyberpunK 2077 it's 43% faster! for V-cache.
No one uses a 9700X for a server. There is better to be gotten then.
@@jankees4037 I was talking about the Design of Zen 5 was for "Server"
So it's pretty much in Eco Mode out of the box.
Pretty much AMD forgot quality control on the 9700X. Why would you bring out such power limited processor, when you already advertise in announcements to offer better than the 7700X. Big failure on AMD by not being consequent in the lineup.
if the non gaming chips are matching the x3d ones from last gen, i am interested to see how the new x3d ones perform
Windows updates , driver & bios updates , x870e chipset..
Nope, nope, nope and definitely nope.
Only X3D chips are solely aimed at gamers
AMD with their X3D chips are gaming chips period. AMD made a mistake to position their 9000 series chips as better gaming chips than the 7800X3D which is very false. Every review that are gaming first tanked because of how the chips where marketed. Anything AMD Ryzen 9***X is purely for productivity and not gaming. Simply wait for the 9800X3D, but kudos to AMD with the new chips. They're Great for us who are not into gaming.