Seriously that drove me nuts in other reviews. Anyone actually concerned about 10-20W want to know real world numbers that will actually matter. Not all core power draw with a gaming chip.
is more fun for the Pcmasterrace to tinkering with CPU like we are used to do for 20 years. These chips are tech enthusiast dream. Already on the edge CPU at default are boring.
I was somehow irritated that no one used this style till today. Its like every "simple" invention, when it's published it's so clear everyone should use it. With 20 CPUs it will be difficult, or they should publish the screenshot so we can zoom in.
Sure beats HUB just saying it sucks! If you know what you're doing, these have a ton of headroom. Wish they were a little cheaper. Also just holding out for the 9800x3d.
@@Tuhar Yeah it does. One thing though is that if X3D chips still have locked power limits, I think this data shows us that they probably won't be much faster than the current ones.
@@ronjatter yeah, if they're locked, a 5-10% increase on the x3d parts would be a little underwhelming, but if there is PBO - a 20-25% increase would be amazing.
@@Tuhar a 20-25% increase would be possible in multi core workloads, though highly unlikely for gaming as unlocking the power limit only had a minor performance advantage in gaming. Going by the data here anyway. It would be really cool to have a 16 Core x3D chip that could (almost) keep up with the 16 core non X3d chip though. That might be possible if they are unlocked this time around.
@@ronjatter Current x3d chips work on decreased clocks in comparison to standard variants. So if new gen x3d would work on same clocks due to better efficiency, they might be reasonably faster.
Seems to me that this performance / efficiency profile is purposefully designed to allow for another tier of CPU models- might be X3D, might be XT, but clearly they left performance on the table and will have a enthusiast tier with better performance (albeit at higher power) at some point.
If X-versions are this restrained, what will the non-X versions do? 65Watts? So far it doesn't look they will look promising at all, maybe this way AMD can lure us to buy more X versions rather than identical but cheaper non-X? Sure it won't affect the enthusiast that does PBO, but that will hit the mainstream plug and go user, and that is the majority.
@@rata536the 7700X with a 105W TDP and the 7700 with a 65W TDP and lower clocks made more sense, similar to how Intel’s K SKUs have a 125W TDP and the mainstream SKUs are at 65W. A 9700 and 9700X that both have the same TDP and a 200MHz clock speed difference is self-defeating on AMD’s part because consumers will just buy the cheaper non-X CPU that comes with an included cooler.
i'm really hoping we some kind of announcement saying "oh shit, we screwed up the profiles" and that these are really the non X power limits like some one else mentioned. the power restrictions really don't make any sense. there's only two reasons i could see them justifying the power restrictions, either the yields are so horrendous that there's a huge variance is CCD quality or they're fudging the temps with the sensor location change and the CCD's are running significantly hotter than it's reporting.
9:58 I think it would be interesting to see how different power limits effects the multicore performance in Cinebench. Run it for 90W, 100W, 120W .. 170W and plot the cinebench score and see how the CPU behaves. Maybe calculate the efficiency and plot that on a curve too.
Yeah I find that stuff really interesting, also gives more insight regarding the architecture. For instance my 12600k has best efficiency from 90W to 115W and gives really impressive scores if undervolted. I would love to see similar tests on some AM5 SKUs!
@@jackboudreaux588365w factory setting without PBO, single tower 120mm cooler is plenty. LOL. A high end solution like NH U12A should run virtually silent.
Thanks for testing with higher power limit! Missed that in the GN review. If you're doing sff workstation system the 9950X should be a decent choice because it should survive with an air cooler. Personally waiting for X3D of course to replace my 5800X3D. Maybe you'll have a new direct die water block then :)
7800X3D will last the same amount of years and provide similar or superior performance in most cases. And that with a lower price, while consuming less power. There's nothing in this new chip to be excited about, except maybe the OC potential.
@@itisabirdThat's only true for gaming, productivity etc the 7800x3d probably isn't what you want. I do lots of productivity work and a 13600k would be faster than a 7800x3d. I have a faulty 13900k and even having to force lower performance it's much faster. In games the 13900k is less efficient but since I'm typically at 4k or 4k DLSS quality with my 4090 I wouldn't really gain much there either with 7800x3d. AMD should have released the higher end parts first, this puts a damper on this gen right out of the gate.
Thank you for being the enthusiasts' reviewer, recognizing the headroom these CPUs have and giving it the beans. Everyone else is whining about it "only" using *significantly* less power for equivalent performance, while you took the enthusiast approach and gave it more juice. As a casual overclocker, I am very interested to see how much you can push these chips with delidding, liquid metal, custom loop, etc.
This is why we come here, Roman is the only reviewer that shows power draw numbers for all use cases he shows. *Can you share the IDLE power numbers?* The other reviewers said this was a Skylake moment for AMD (14nm+++) while brushing the efficiency gains as a nothing burger. *Proper testing between 7700X and 9700X should be done at same power draws* (7700X matching 9700X and vice versa). In the 5000s series AMD sold the CPUs around the most efficient profiles but as everybody complained that the bar was not high enough then they released the 7000 series overclocked from the factory. People complained that AMD was now playing the Intel game so they now released the 9000 series around their most efficient settings again just to get the same crap from the 5000 series. They really cannot win when most reviewers are "frame chasers" and have the memory of a gold fish. As a home server user, I prefer this approach from AMD as I can always decide whether or not to overclock / remove limits at my discretion.
@@AutieTortie Idle energy consumption is useless? wtf? Processors spend most of their life idling and some people use desktop CPUs for their homelab, where the processor will spend even more %time in idle.
@@AutieTortie Not really, especially if places are expensive for power or in SFF builds idle power matters a ton. Less fan speed, less power draw, less room cooling, etc.
summed up pretty much what i was thinking too. On the other hand GN and HU gave there reasons more than once i think why they just go with just default out of the box settings. To represent the general consumer experience who doesn't invest time in fiddling around with their hardware if i can still remember vaguely. But their conclusions does undersell this CPU imo. Anyway Der8auer ftw!
@@ericspoor8175 I agree with testing using default settings out of the box by default. But as you said, brushing off the efficiency gains made *their conclusions as ignorant at best and intentionally misleading at worst* . Why? I assume they did this because both do not want to be perceived as "Intel haters" so they felt the need to bash the 9000 series to have the perception of being neutral. It is kind of OK, as home server users are not the main demographic for GN and HU. Just look at the first reply I got in this thread =)
I much prefer the new default power profile even though it hampers the chip in all core workloads. In gaming, all core clocks are unaffected, and crucially they're super cool and calm. For 90-95% of AMD's target audience with these since CCD chips, having the 'eco' mode be default is the correct choice and a great change from Zen 4.
I want nothing less than the highest performance, however 25% perf increase for 2x the power is REALLY bad. Pretty sure you can get MUCH better performance for similar or less power if you just went with a higher core CPU.
Came here from the Gamer's Nexus review where they didnt really show performance at high wattages. This is actually a pretty impressive IPC improvement from last gen when measured against the same power level. I usually undervolt my parts for efficiency so this generation is still nice for me, even though the power limit is abit restrictive for people who dgaf about power usage. (I have roof solar and I have a strict power budget after dawn).
Steve is seriously disappointing this year - today is his new weird review with stupid mistakes. He literally destroyed the new processors, although everything is not so bad. By the way, I am also a big fan of undervolting - acoustic comfort and low temperatures are extremely important for me. In general, I am damn glad that the new AMD products are quite universal - you can find a perfectly balanced option.
@@bassyey But most people cry about not caring about high voltage, that maybe leave the processors at lower wattage and people who want higher power can overclock. But here we are, the same crowd demanding the opposite now. Cool.
@@bassyey No. He managed to make the new processors look no different from the previous generation. He didn't even see the difference in energy efficiency and operating temperatures, which became 20-30 degrees lower. I don't understand such reviews. For many, these are very important parameters.
@johnpreston8621 did you actually watch it? Because Steve points out the temp difference and how it's not really 30 degrees cooler. And repeatedly said that the efficiency is notable and the reason why it's not a complete flop.
some people are suckers for efficiency, like me, the cpus are powerful enough for anything you could ever need in a gaming pc, 60°C temps and 88W with 8 core cpu, not bad at all I think
i used to be in the 'why didn't they optimize' camp until I started salvaging used Gaming PC. Both OEM and Custom builds. It was once in a blue moon that I'd run across a PC with a H50 or Hyper 212.
Kudos for making a point about making *your test at 6000MT/s while they warranty 5200MT/s (and now5600MT/s) Techmedia needs to push back and Test at warrantied speeds.
the warranty covers your CPU if you use 6000MT/s (EXPO profile, not manual tuning). they've never refused to RMA such cases and their own recommendations show 6000MT/s. there was some drama in 2023 with Asus where they burned CPUs and the socket of mobos which had EXPO enabled because of a bad bios update. Asus tried to weasel out, but they were forced to cover the warranty.
@@mariuspuiu9555 Then why don't they put DDR5-5600 in the product specification? It's very simple: either it is covered under warranty or it's not covered under warranty. If it isn't in writing, then it's not covered under warranty. AMD will absolutely fuck you just like Intel if it turns out there is a defect in the chip they didn't know about a year down the line and the number of RMAs is too big for them to swallow. The fact their own recommendations show 6000 MT/s isn't a sign they will cover it under warranty, it's a sign that they're willing to market at 6000 MT/s, but they're not willing to officially warranty above 5600 MT/s. These "trust me bro we'll take care of you even though we're not bound to do so by the law" warranty policies are a cancer on society that needs to be eradicated.
@@osirisgolad It's not so much about in case the CPU breaks, it's more that they don't guarantee stability. DDR5-5200 WILL work, if it doesn't you RMA the CPU until you get one that handles it. DDR5-6000 SHOULD work, if it doesn't, sucks to be you -- AMD are unlikely to help.
@@osirisgolad What do you mean by below spec? Also, Intel and AMD have done this for many, many years. The 14900K supports up to DDR5-5600 even though 7200 seems to be a fairly common speed to run, and some run over 8000.
This is hands down the best review I've seen and answered many of the questions that I left others asking. Why AMD would sandbag their own product to the point of barely competing with the existing generation is very strange. Let's be real, 90% of people don't really care if 6-8 core parts are pulling another 15-20W on avg as long as performance is there and it's easy to cool. I'm suspicious this is mostly for product segmentation with X3D but I think they went a bit too far. The last question I have is memory. Looks like the IO die is effectively the same but I'm curious if there's more FCLK headroom on average compared to Zen4.
I realise its not optimal tuning, but 22k cinebench with about 160W is pretty close to an i5-13600k with the default power limits. Be interesting to see the results once the new microcode is out.
Intels microcode thing is so weird. I think there's a quality issue going on with intel because my 13700k gets 30k in R23 at 160W and 75c, but nearly incinerates itself with stock settings. I feel like Intel has pumped the limits to ensure poor quality CPUs can be sold as higher end parts, but if you have a good one its a great, cool chip.
@@chrism2964 I think you're right. Igor's Lab article "Intel Core 14th Gen Binning Results from almost 600 CPUs" found at stock settings i9s were seeing a hair over 0.1v variation on the VID between the best and worst chips. There was a similar spread on the i7 and i5 parts, but they were more closely clustered with just a few outliers.
@@chrism2964 That's probably because of via oxidation. People have mentioned that when you give it more voltage, you get around stability issue caused by that problem but while also deteriorating the CPU even further. The good chips are, imo, just the ones that didn't rust!
I think they limited power consumption for two reasons: 1. Because people see INTEL chips overheating, they'll go for AMD chips underheating (bigger gap). 2. Because it gives X3D chips more headroom to beat current X chips and previous X3D chips (bigger uplift).
I dont think there will be bigger than single digit % fps difference between 7800x3d and 9800x3d unless AMD keeps 5.5ghz Boost clocks from 9700x and sygnificaly pump up TDP from 65W to 120+W. Even than i dont expert it to be faster than 7800x3d for more than 10-15% in 1080p, 5-10% in 1440p and 1-5% in 4K.
i remember a time when power efficiency was high regarded in tech, somehow in the last 10 years people stopped giving a fuck, it's nice to see good performing parts that will give extra performance with OC for those less efficiency inclined as it should be.
Really makes me think Zen 5 is being downplayed and underrated too much. It consumes way less power, so all these other reviewers are all acting like that's all it has going for it. Why not test it with PBO!!!
I really appreciate someone stepping back from the thermal arms races which has only provided us with incremental performance benefits while escalating the power draw / heat / noise /... (I run an AMD CPU with integrated graphics: Its quiet, it keeps the power bill down, and unlike its predecessor it does not heat up the room like crazy in summer)
best video from all the reviewers. amazing that you can effectively overclock the 9700x and they sell it at a stock profile which is extremely safe and efficient. well done AMD. the other reviewers need to figure out the PBO overclocking part.
I'm not a gamer, I'm a productivity user, software development, video, etc. I'm interested in the 9700X or 9900X simply for the lower power, I don't need 2 more FPS in games that I never play.
I appreciate that you took a hard look at PBO. It is disappointing that there is little gaming uplift on the 9700X PBO over 7700X, that doesn't bode well for an eventual 9800X3D having gaming uplift over 7800X3D, unless the 9800X3D can run higher clocks. I do prefer that the out-of-box default is effectively eco-mode. It did not sit well with me that 7000-series default chased the temperature limit.
@@EtaCarinaeSC how ryzen is designed, I don't think the 9800x3d will smoke his predecessor in gaming if the keep the same cache size. If the increase it a bit (L2 or L3), it's a major win, if not ouch. I might be wrong tho and I hope so.
The most optimistic scenario is that the new architecture is memory bottlenecked at 6000 in gaming, or by Infinity Fabric, etc. The memory spec bump from 5200 to 5600 hopefully indicates that the realistic minimum has gone from 6000 to 6400. And a 9700X at 6400 might be better than 7700X at 6000. We'll see soon, when Roman or HUB do memory scaling testing, Infinity Fabric, etc. 9800X3D would presumably not care much about memory bottlenecks, as with 7800X3D.
@@evguenict Biggest drawback from X3D chips is the weak heat transfer from die to IHS because of the extra 1 stack of 3D-Vcache. If AMD can improve heat transfer , new X3D parts can clock higher. Current X3D chips cannot go higher than 1.25V Vcore because the stacked cache absolutely don't like that.
Thank you for providing an Objective empirical review of these parts Der8auer, this is highly informative. The out-of-box performance of these chips on launch is disappointing, but they have vastly improved efficiency. However when you enabled PBO, you got the performance figure uplift i was roughly expecting! PBO overclocking usually doesn't result in such a huge gain - maybe roughly 5%, but in this case it resulted in a 21% uplift. This strongly suggest to me that these chips are being limited at the firmware level by default. Awesome job as always der8auer!
It's so refreshing to see proper review and you not jumping on the hate bandwagon with rest. You are also the first to point out that while the CPU in stock is equal to 9700X, it does that at 40% less power. Kudos
Recently I had a bit of instability in my system after adding another 32GB of RAM, but as I didn't have the patience I simply used Asus' Auto OC in the BIOS. Today I noticed that the OC put my Ryzen 7 2700 (which barely pulls 80w under load) at 1.365V for just 3.4Ghz LoL
@FlavioSantos-uw1mr Hey, sometimes you have to send it. Asus says just buy an extra cpu to have on hand in case you smoke the first one lol. I have never liked all this auto OC crap people should have too dig into the settings and stress test to find the optimal clocks/voltage settings like the good old days of overclocking, but hey if people want auto OC's with way too much voltage then have at it I guess.
One of the best Ryzen 9000 review, hands down. At least you are delivering facts even about the PBO aspect of the architecture and sharing povs on the efficiency. Also, a great touch having the power draw into the graphs. Thanks for your work 👏
3700x and 5700x were also 88w ppt (65w tdp), that's just the 7700x that needed to go higher "out of the box" to look good. I'm glad they've remake the *700X segment right, a more budget firendly 8 core part with lower cooling needs, and with the *800x and the *800X3D becoming once more the higher performance/power draw parts. That seems more logic to me.
It is a branding switcheroo so they can claim these parts are releasing for a cheaper MSRP vs the 7600X and 7700X, when it is actually a price increase from 7600 and 7700. They can, in reverse order, release the stronger tuned version later on, with more Xs and Ts in the product name.
Tbh at this point it's just speculation. They had the whole delayed launch thing, maybe they royally screwed up the power profiles and this is all they can salvage in a few weeks during the delay?
@@WILLPORKERnah, even the TDP match with the leaks months before the official announcement, what I think watching other reviewers is that this is a software optimization problem, an AGESA and BIOS updates might be announced in the next months, AMD rushed the release because Qualcomm and Intel AL bluff ab that would explain why the software is still unmature
5:12 When you're saying that 7800X3D is not that strong while it's just 2 fps lower compared to 7700X while he consumes 138 watts instead, this is a misrepresentation of this CPU. Considering the power draw and the arhitectural peculiarities, 7800X3D with a bit more of a power draw can easily get head to head with 9700X which is newer and with 7700X which draws way more power compared to 7800X3D.
It feels like AMD had been blasting power at Zen4 to compete with Intel and they took Zen5 for a turn for efficiency. It would be interesting to see how 9950x perfoms efficiency-wise against 7950x with eco mode.
Good video! Yeah I think that AMD was trying to play it safe with lowering the stock power profiles. Also it kind of proves a point that with the lower power draw, they can perform just as good as last gen which some may prefer. Also this give them room to flex a bit with the higher tier chips on performance charts. I hope when the 9000 X3D chips are released we see something real impressive.
overshot last gen with their all out tactic massively undershot this gen with full out efficiency (which honestly im a fan of) but we can always count on X3D to save the day for gaming!
@@TestarossaF110Genuinely curious how 9800x3D will perform. Considering the Intel situation, will AMD release it as lackluster or actually go for the throat.
@ZioComposite I pray they go full out and just keep doing that, making those (X3D) the *actual* new CPUs gamers can and will go for, but I fear AMD doesn't know what to do and keeps being kinda stuck. time will tell I guess.
Wow a CPU that will show large gains from manual overclocking? Did I wake up in 2008? Also - this is a GREAT sign for the 9800x3d chips. If they have increased PBO headroom like rumored - this could be an awesome gen.
@@rustler08 if you go back and look at the AMD CES coverage - they kept bringing up 'additional improvements' to x3d. They wouldn't say exactly what it was, but the rumor was PBO for x3d. That could be an amazing combo if its true!
@@rustler08 The rumour (and it is a bit more than unsubstantiated rumor) has it that X3D chips will be more unlocked than previously. PBO will almost definitely be usable - especially given that curve optimizer is to a degree usable on 7000 X3D chips (depends on BIOS). If they do come out almost completely unlocked (Is suspect they will have locked Temperature maximum), then yes, they will be killer CPU's.
AMD officially say the 9700X is a 65w part that boosts "up to" 5.5GHz (but under all core it only hits 4.5GHz), and the 7700X is a 105w part, with similar top clocks, so it's not really a rumor so much as basic mathematical deduction that if the chip is allowed to boost to 5.5GHz it will run faster.
This is the correct review, using PBO to unleash the CPU and there we have it up to 25% more capable when needed or just as powerfull but way more efficient when it is ok to just browse around, Awesome, Thanks der8auer
I don't wanna ding anyone looking for efficiency but I thought we already had that with the 7000 series, it's called eco mode and some folks were trying to build hype around it with the 5000 series. I think I'd like to see someone compare the 7700x in eco mode to the 9700x because I'm thinking it will be pretty underwhelming.
@@creativityfails1 I mean, seeing as zen5's default is basically 'eco mode' it matches the full performance of a 7700x while using 40% less power, the 7700x will still lose performance on its own eco mode, even if it's just 5-10% perf loss, that's still 5-10% less as compared to another cpu that is then running near same power. I'm rather interested instead at how enabling PBO is like a 21% gain for zen5 in multicore workloads especially, which then puts it way ahead of the 7700x at similar power.
I think people are way too obsessed with raw performance. Normal user doesn't need that much CPU power anyway, so having a easier (less cooling needed) and more stable build is better. I love that AMD focused on efficiency instead of destroying the CPU with too much heat and power draw. And if you need more CPU performance, you can achieve that by doing overclock.
Maybe they have some concerns about stability at "high" power and they have learnt the lesson from Intel, hence they are very conservative with the power... Just speculating.
Are people sending Zen 4s back because they're burning out? if they are, they're being pretty quiet about it. If they're not, then why would this be anything about that, given it's way more efficient. PBO lets it clock up like usual anyway. Will be interesting to see what BIOS updates bring here.
People complained about high power draw with zen 4 default settings. I think this proves people will be disappointed either way they go, but you can always change one setting and have it run the way you want before you even boot it up the first time.
Nah I just think the new gen looks more impressive like this. They basically are showing same performance at 40% power reduction, which sounds much better than 15% more performance at same power. Furthermore if you look at the 9700x in R23 with unlimited power, it scores similarly to a 14600k at a similar power budget... Thus I believe that locking them down was the best marketing choice for AMD
Thanks for a excellent review. It was thorough, multi-dimensional and full context. Is AMD making a massive play in the OEM market by keeping the power down and performance similar? It's obvious AMD is continuing it's cadence of innovation and forward looking design.
I agree with your conclusion (I even wrote on different review, that 95W TDP or something in between 65 and 95 would be great). I would prefer if AMD said that this generation isn't exactly for gamers - it won't be worse, but it won't be much better, but you will get some savings in power. As for price - it's lower than MSRP of 7000 series at launch and will likely drift down in few weeks, but gamers on budget might want to go with 7000 series. Rant: People (commenters) write that this is "AMD stagnation" comparing it to Intel... I'm sorry, but I can't agree. During Intel stagnation you were lucky to get 5% uplift with basically the same power use gen over gen and sometimes the only uplift came from increasing default power (so the actual improvements were that it moved performance and power use linearly - by again usually around 5%). Also this was never resolved with future BIOS updates. Here it is possible that 9000 CPU's will get couple percent boost compared to 7000 series, just with BIOS updates alone. Of course this isn't on reviewer to predict, this is on AMD to deliver and I don't blame anyone for reviewing CPU "as is" (it's how it should be done, with speculation being perhaps in separate chapter of the video or something), nor do I know if AMD will deliver those updates. But coming to conclusion that it is "the same" as Intel stagnation after watching 1-2 day-1 reviews is unbecoming of people. I really hope they will stop. Thankfully most reviewers didn't do stuff like that, but comments such as this are triggering me. More than they should.
Agreed. Truth be told, I'm not particularly impressed by the gains in games, but getting the same performance as the 7700X, with a decreased power consumption, _is_ to me. Also, given Zen 5 lies the groundwork for upcoming generations, I have good hope the lacking gains here will come with Zen 6 and newer.
@@JarnoOverwijk It is a bit of inverse of what they were doing earlier - but 9700X gains over 7700X in those titles that are primarily single threaded. Or with very few threads. Due to lack of power budget, you can actually have 9600X overtake 9700X in games that use more than (approx.) 5 threads. Because those 6 cores can use the same amount of energy as 8 cores. So they boost higher. So for games 9700X is a bit of a coin toss (in some games it will be worse than 7700X, but this will mostly be in games that already have high FPS). On average it gets approx. 5% over 7700X, but consumes 40% less power. In productivity workloads gains are higher on average (approx. 9%). And of course it overclocks much better. I don't think AMD rejects RMA's based on using DDR-6000 or even using curve optimizer/unlocking power limits. Though they might if you RMA the part that you used in breaking some LN2 record ;) It's going to be interesting to see 9900X (for once) and 9950X (since this part doesn't lower TDP).
People keep screaming about how these chips aren't good but the fact that it matches last gen's performance at 40% less power is nothing short of incredible.
@@medovk Some people with expensive electric bills may prefer that. It also allow for use of cheaper cooler, mobo (but not psu otherwise it would explode😂)
@@SuperLloyd84 some people really value efficiency, and if you don't value that it looks like PBO can give these the massive improvement people were looking for.
@@muhammadhanifkurnaen6689 Unless you're running 100% all-core loads 24/7, no one is upgrading from a 7700x to a 9700x to save on their electric bill. If power usage is that important, just buy a gaming laptop.
Honestly I love AMD's idea We actually need a generation that just improves efficiency, imagine rtx 4090 performance with card that pulls 200w With modern prices that would be golden
Great video. Love the fact that you talked about overcloking. I see that being a huge value to AMD. If only they would have stated that on their own. "we have an extremely efficient CPU but if you want to get the most out of it all you have to do is unlock it's power". That would have been a great PR move in my opinion. Looking forward to your overcloking videos.
@Jaywithag78 Older CPUs ran at lower clock speeds and had more robust fab processes, so they were less susceptible to damage than newer processors. That is the problem, people think they can subject the latest CPUs to the same sort of abuse and it would be ok, except it is not ok and now they cry.
dont understand why do you compare it to the 7700x and not the 7700 (non-X) which is also a 65W CPU so that would be apple to apple, everybody know that the 7000 series X version is inefficient and the non-x version is the efficient so i wanted to know to performance gain against the same TDP version
I have a feeling this will be the "highly underrated" generation. It seems like it was built and released on conflicting premises. When the gap in the commercial/professional space opens up due to Intel screwing up, it would be dumb not to capitalize and release the most efficient productivity - focused chip possible. It can still be unlocked by enthusiasts and when they inevitably saturate the market the price will come down.
Really looking forward to seeing how far this CPU generation can be pushed with direct die cooling and some effort tinkering the power and performance settings. Seems to have way more headroom than most releases.
It's time everyone grow up to the fact that power efficiency is an important factor to consider. I think that it's a huge achievement that the chips perform just as good as last gen, but have substantially improved thermals and power draw. Energy costs keep going up, it's extremely irresponsible and elitist to think that computer chips shouldn't care about these factors. The reality of how much your computer really costs is the price of the hardware + all the energy it will cost during its lifetime.
@@Neuroszima shouldn't really matter what area. We're talking about power demand under load. Unless you're maxing your system out 24/7 you shouldn't see anything close to a hit on your bill. It's a misunderstood concept by so many casual tech people.
Insulting people for wanting their CPU to run at full power is incredibly immature, especially when the power savings will have little effect on your bill. You have no right to tell anyone to grow up when you act like this.
It's basically 2 options with amd at this point. Normal X for productivity, and X3D for gaming. If you're not going for one thing, then you shouldn't bother looking at it.
In my experience, prices will drop quite a bit after either ~6 months or the release of the 9800X3D. Even the 7700X faced a reduction after a short while because it barely matched the 5800X3D so we just need to wait. This CPU looks pretty good for home server or light workstation use but we need to see the IDLE power.
@@immortalzodd7474idk, the massive price increases and binning delays makes me think that the 4nm TSMC node might not be yielding as expected. That’s going to take longer than 6 months to work out if it’s not AMD’s fault.
You can get a 16 core part for the price the 1800x launched at with inflation, or a 9900x for the price a 1700x launched for, the 8 core part is 150 dollars less than the original launch price the only thing that hasn't moved is the 6 core parts but that is due to diminishing returns in what they can remove from their margin to make the part in the first place.
@@macicoinc9363 Agreed, but prices are gonna go down as I do not see general consumers (gamers) flocking to this one. It is going to sell way worse than the 7700X so it should fall in price faster than that one. We are officially in a recession/stanflataion though so things may get weird =S
When a manufacturer limits power more than expected it's almost certainly because of bad yields forcing them to requalify their bins with lower silicon quality chips. The other option is that they're trying to bait Intel to cut their processors' performance more with their microcode update by giving them a false sense of security. Then 9900X and 9950X comes out and mops up
No, it's because ARM is becoming an actual threat to X86, and the only reason power has been increasing over the years is because AMD was competing with Intel, and they had to always one up each other in benchmarks, and the only way to do this is to push the CPU beyond its most efficient point This is a very positive thing, same performance at considerably less wattage means the CPU is actually optimized, from this point forward it's going to be the same, lower power or same power consumption and increases in performance compared to the 9000 generation
@@xSyn08maybe, although I don’t really see arm usurping x86 in the desktop, workstation, and server space for a while. x86 has software inertia that arm won’t hurdle for at least a decade. If Linux can’t convince CAD/Simulation companies to port to a better OS, then Arm being in a bunch of mobile and laptop devices won’t convince them to provide support either.
@@macicoinc9363 ARM to x86 emulation on Windows is already achieving the same or even better performance than Rosetta 2 did A decade is wishful thinking when Snapdragon X Elite exists, it may be a year before we start seeing those chips sold as desktop parts For your knowledge the X Elite (X1E-84-100) is a mobile 80W part that sits between AMDs 9600X and 9700X in terms of performance From what I know x86 emulation on Windows gets you like 90%+ of the performance
Thanks @der8auer for this deep dive review, i watched GN / HUB reviews and their reviews were lacking the PBO/OC side of these CPUs... seems like they rushed their reviews... you are the no. 1 reviewer for me from now on
Finally non biased opinion (HUB says its sucks, Steve: meh). Power efficiency is huge compared to Zen 4. OC capabilities is also very nice. We need to wait for lower prices and new mobos, ofc.
I need a new rig BADLY, but I'm going to be waiting for the X870 boards to drop before making a decision. Lots of reviews from the major outlets reported BSOD, memory issues, random reboots... All the problems I had on B650, I just don't think the first-gen AM5 boards are any good.
Appreciate not bashing on it due to no performance gains @ 65tdp....which less power draw and lower temps. Great for sff builds..and still have the headroom for a little performance boost w/ oc/pbo.
Most interesting. SkatterBencher uncovered performance that appears to align with what we see here. Much more to come with all the Zen5 future hard looks.
I think reducing power for the same performance from the last gen is a smart engineering. Adding more power for more performance is a lazy angineering. Anyone can do that. Intel did that and look where they are now.
*They also used the wrong ram. They treated the 9000 CPUs like they were 7000 seties.* *The 7000 series supports an SPD ram frequency of 5,200MHz. The 9000 series supports an SPD ram frequency of 5,600MHz. Which means the sweet spot would have been around 6,800-7,200 MHz (with the proper EXPO settings, timings, and voltages).* *A 2-8% performance gain over the benchmarks done with 6,000MHz ram. (2-8% gains). What a joke!! Every benchmarker, so far, has dropped the ball and should've done their research before blindly using low-end parts.*
new ram speed might need 800 series MB. now 600 series MB looklikes 6000mhz cl30 is sweet spot. considering IFCLK 2000 supported 6000mhz. from leaks, look like 9000 series support 2400 IFCLK. haven seen anyone tested that before even tho 6400mhz seems like the new sweet spot. but getting these standard processor to 7200mhz will be very good. hoping that is possible with 800 series MB and amd can crawl back performance from there
Over the years my AM4 builds have been upgraded to quite premium quality machines. The AM4 platform is mature, incredibly stable, and comparatively capable versus the newest generation performance. The 9000X series is nowhere near compelling enough for me to leave my AM4 dream machines behind.
Only for productivity ITX. And that is probably only until the 9950X launches. Until then, as DerBauer said, the 7800X3D uses even less power, is cheaper, and performs better in gaming. I gotta say. I can't really see a niche the 9700X fills...
I was thinking the same, like in an AsRock DeskMeet X600 and equivalent. DeskMeet X600 is interesting because it has no chipsets. All I/O is directly from the CPU. It is however limited to 65W TDP, so the new 9700X would pair up nicely with it.
Give it less power with the curve and see what happens with the all-core clocks. Less power, less heat means higher boost if its stable. Combine it with an overclock should be crazy good CPU.
I just want to say I really appreciate the fact that you included the power draw in every test scenario, thank you. :)
Seriously that drove me nuts in other reviews. Anyone actually concerned about 10-20W want to know real world numbers that will actually matter. Not all core power draw with a gaming chip.
Yes! That is very important to see.
Yup without the power draw the other reviews were quite confusing or lacking. Appreciate the attention to details.
yeah, as someone who want lower wattage PC, i really like this CPU
up to 10% less performance than intel but only require a third of the power?
Me too... Best review among the others
feels like they accidentially gave the X chips the non X power profile
I am thinking the same thing. These should have launched as non-X chips.
@@PREDATEURLTthe Intel stuff isn't cooking because of the power per se, just how its all delivered over one rail.
that's X800X job, the X700 series job is to bait people buying a cut down APU product (i'm looking at you Ryzen 7 5700)
is more fun for the Pcmasterrace to tinkering with CPU like we are used to do for 20 years. These chips are tech enthusiast dream.
Already on the edge CPU at default are boring.
They didn’t want to do an Intel
Your graphs showing performance and power draw together are excellent, the best I've seen from any reviewer.
I was somehow irritated that no one used this style till today. Its like every "simple" invention, when it's published it's so clear everyone should use it.
With 20 CPUs it will be difficult, or they should publish the screenshot so we can zoom in.
Yeah that should be the standard.
There we go. I was waiting for your review.
I knew der8auer would have a play with the power limits :)
Sure beats HUB just saying it sucks! If you know what you're doing, these have a ton of headroom. Wish they were a little cheaper. Also just holding out for the 9800x3d.
@@Tuhar Yeah it does. One thing though is that if X3D chips still have locked power limits, I think this data shows us that they probably won't be much faster than the current ones.
@@ronjatter yeah, if they're locked, a 5-10% increase on the x3d parts would be a little underwhelming, but if there is PBO - a 20-25% increase would be amazing.
@@Tuhar a 20-25% increase would be possible in multi core workloads, though highly unlikely for gaming as unlocking the power limit only had a minor performance advantage in gaming. Going by the data here anyway. It would be really cool to have a 16 Core x3D chip that could (almost) keep up with the 16 core non X3d chip though. That might be possible if they are unlocked this time around.
@@ronjatter Current x3d chips work on decreased clocks in comparison to standard variants. So if new gen x3d would work on same clocks due to better efficiency, they might be reasonably faster.
Seems to me that this performance / efficiency profile is purposefully designed to allow for another tier of CPU models- might be X3D, might be XT, but clearly they left performance on the table and will have a enthusiast tier with better performance (albeit at higher power) at some point.
If X-versions are this restrained, what will the non-X versions do? 65Watts? So far it doesn't look they will look promising at all, maybe this way AMD can lure us to buy more X versions rather than identical but cheaper non-X? Sure it won't affect the enthusiast that does PBO, but that will hit the mainstream plug and go user, and that is the majority.
@@rata536all you have to do is goto. Bios., hit,swt pbo to on and lat amd's cpu manage to rest
@@rata536the 7700X with a 105W TDP and the 7700 with a 65W TDP and lower clocks made more sense, similar to how Intel’s K SKUs have a 125W TDP and the mainstream SKUs are at 65W. A 9700 and 9700X that both have the same TDP and a 200MHz clock speed difference is self-defeating on AMD’s part because consumers will just buy the cheaper non-X CPU that comes with an included cooler.
@@NadeemAhmed-nv2br Yeah people like us for sure. The mainstream user that doesn't know what bios is, not so much.
i'm really hoping we some kind of announcement saying "oh shit, we screwed up the profiles" and that these are really the non X power limits like some one else mentioned. the power restrictions really don't make any sense. there's only two reasons i could see them justifying the power restrictions, either the yields are so horrendous that there's a huge variance is CCD quality or they're fudging the temps with the sensor location change and the CCD's are running significantly hotter than it's reporting.
9:58
I think it would be interesting to see how different power limits effects the multicore performance in Cinebench.
Run it for 90W, 100W, 120W .. 170W and plot the cinebench score and see how the CPU behaves.
Maybe calculate the efficiency and plot that on a curve too.
Yeah I find that stuff really interesting, also gives more insight regarding the architecture. For instance my 12600k has best efficiency from 90W to 115W and gives really impressive scores if undervolted. I would love to see similar tests on some AM5 SKUs!
absolutely and also take it off a 360 cooler at see what happens on a 4-5 pipe single tower cooler
@@jackboudreaux588365w factory setting without PBO, single tower 120mm cooler is plenty. LOL. A high end solution like NH U12A should run virtually silent.
90w is only 2w above what it would be boosting to by default
@@WayStedYou I just rounded up from 88W. 😂
Thanks for testing with higher power limit! Missed that in the GN review.
If you're doing sff workstation system the 9950X should be a decent choice because it should survive with an air cooler.
Personally waiting for X3D of course to replace my 5800X3D. Maybe you'll have a new direct die water block then :)
One of the best reviews. Done critically and not focused only on mere fps or benchmark scores. GG
I mean, he is German.....
If it’s held back from burning I love it. I don’t need the highest numbers ever. I want a system that works correctly and will last some years lol
7800X3D will last the same amount of years and provide similar or superior performance in most cases. And that with a lower price, while consuming less power. There's nothing in this new chip to be excited about, except maybe the OC potential.
@@itisabirdThat's only true for gaming, productivity etc the 7800x3d probably isn't what you want. I do lots of productivity work and a 13600k would be faster than a 7800x3d. I have a faulty 13900k and even having to force lower performance it's much faster. In games the 13900k is less efficient but since I'm typically at 4k or 4k DLSS quality with my 4090 I wouldn't really gain much there either with 7800x3d.
AMD should have released the higher end parts first, this puts a damper on this gen right out of the gate.
@@innopriest Nope, it does not.
@@innopriestdlls. Quality is. 1440p native so you would still gain at least 5%
@@itisabird Just wait for the X3D variants, those might be more exciting this time around.
Thank you for being the enthusiasts' reviewer, recognizing the headroom these CPUs have and giving it the beans. Everyone else is whining about it "only" using *significantly* less power for equivalent performance, while you took the enthusiast approach and gave it more juice. As a casual overclocker, I am very interested to see how much you can push these chips with delidding, liquid metal, custom loop, etc.
This is why we come here, Roman is the only reviewer that shows power draw numbers for all use cases he shows.
*Can you share the IDLE power numbers?*
The other reviewers said this was a Skylake moment for AMD (14nm+++) while brushing the efficiency gains as a nothing burger.
*Proper testing between 7700X and 9700X should be done at same power draws* (7700X matching 9700X and vice versa).
In the 5000s series AMD sold the CPUs around the most efficient profiles but as everybody complained that the bar was not high enough then they released the 7000 series overclocked from the factory. People complained that AMD was now playing the Intel game so they now released the 9000 series around their most efficient settings again just to get the same crap from the 5000 series.
They really cannot win when most reviewers are "frame chasers" and have the memory of a gold fish.
As a home server user, I prefer this approach from AMD as I can always decide whether or not to overclock / remove limits at my discretion.
Idle power draw figures are pretty useless.
@@AutieTortie Idle energy consumption is useless? wtf?
Processors spend most of their life idling and some people use desktop CPUs for their homelab, where the processor will spend even more %time in idle.
@@AutieTortie Not really, especially if places are expensive for power or in SFF builds idle power matters a ton. Less fan speed, less power draw, less room cooling, etc.
summed up pretty much what i was thinking too. On the other hand GN and HU gave there reasons more than once i think why they just go with just default out of the box settings. To represent the general consumer experience who doesn't invest time in fiddling around with their hardware if i can still remember vaguely. But their conclusions does undersell this CPU imo. Anyway Der8auer ftw!
@@ericspoor8175 I agree with testing using default settings out of the box by default.
But as you said, brushing off the efficiency gains made *their conclusions as ignorant at best and intentionally misleading at worst* .
Why? I assume they did this because both do not want to be perceived as "Intel haters" so they felt the need to bash the 9000 series to have the perception of being neutral.
It is kind of OK, as home server users are not the main demographic for GN and HU.
Just look at the first reply I got in this thread =)
Showing power draw and fps together is very helpful
I much prefer the new default power profile even though it hampers the chip in all core workloads. In gaming, all core clocks are unaffected, and crucially they're super cool and calm. For 90-95% of AMD's target audience with these since CCD chips, having the 'eco' mode be default is the correct choice and a great change from Zen 4.
It should be better tuned less than 5% uplift is sad. Also pricing isn't great (as always).
yeah but for gaming it just makes zero sense. (full stop).
It looks like with the 105w default boost it would get a nice jump in multi core and then not fall behind in gaming with a few extra watts
I want nothing less than the highest performance, however 25% perf increase for 2x the power is REALLY bad. Pretty sure you can get MUCH better performance for similar or less power if you just went with a higher core CPU.
I want 115W. It can throttle on lesser loads, that's where efficiency is supposed to come in.
Came here from the Gamer's Nexus review where they didnt really show performance at high wattages. This is actually a pretty impressive IPC improvement from last gen when measured against the same power level. I usually undervolt my parts for efficiency so this generation is still nice for me, even though the power limit is abit restrictive for people who dgaf about power usage. (I have roof solar and I have a strict power budget after dawn).
Steve is seriously disappointing this year - today is his new weird review with stupid mistakes. He literally destroyed the new processors, although everything is not so bad. By the way, I am also a big fan of undervolting - acoustic comfort and low temperatures are extremely important for me. In general, I am damn glad that the new AMD products are quite universal - you can find a perfectly balanced option.
Wow. I was wondering why HUB and GN reviews were like that. Maybe they tested on configurations most people will use?
@@bassyey But most people cry about not caring about high voltage, that maybe leave the processors at lower wattage and people who want higher power can overclock. But here we are, the same crowd demanding the opposite now. Cool.
@@bassyey No. He managed to make the new processors look no different from the previous generation. He didn't even see the difference in energy efficiency and operating temperatures, which became 20-30 degrees lower. I don't understand such reviews. For many, these are very important parameters.
@johnpreston8621 did you actually watch it? Because Steve points out the temp difference and how it's not really 30 degrees cooler. And repeatedly said that the efficiency is notable and the reason why it's not a complete flop.
Glad you unlocked PBO! I get why GN doesn't do it, but this shows the full potential. Although I think there is big value in a efficient cpu.
some people are suckers for efficiency, like me, the cpus are powerful enough for anything you could ever need in a gaming pc, 60°C temps and 88W with 8 core cpu, not bad at all I think
He didnt try Curve Optimizer on BIOS.
i used to be in the 'why didn't they optimize' camp until I started salvaging used Gaming PC. Both OEM and Custom builds. It was once in a blue moon that I'd run across a PC with a H50 or Hyper 212.
@@m8x425 I'm not sure what you are trying to say, sorry I'm not a native speaker. Could you maybe explain it to me?
Best review of Ryzen 7 9700X.
Kudos for making a point about making *your test at 6000MT/s while they warranty 5200MT/s (and now5600MT/s)
Techmedia needs to push back and Test at warrantied speeds.
the warranty covers your CPU if you use 6000MT/s (EXPO profile, not manual tuning). they've never refused to RMA such cases and their own recommendations show 6000MT/s.
there was some drama in 2023 with Asus where they burned CPUs and the socket of mobos which had EXPO enabled because of a bad bios update. Asus tried to weasel out, but they were forced to cover the warranty.
@@mariuspuiu9555 Then why don't they put DDR5-5600 in the product specification? It's very simple: either it is covered under warranty or it's not covered under warranty. If it isn't in writing, then it's not covered under warranty. AMD will absolutely fuck you just like Intel if it turns out there is a defect in the chip they didn't know about a year down the line and the number of RMAs is too big for them to swallow. The fact their own recommendations show 6000 MT/s isn't a sign they will cover it under warranty, it's a sign that they're willing to market at 6000 MT/s, but they're not willing to officially warranty above 5600 MT/s. These "trust me bro we'll take care of you even though we're not bound to do so by the law" warranty policies are a cancer on society that needs to be eradicated.
@@osirisgolad It's not so much about in case the CPU breaks, it's more that they don't guarantee stability. DDR5-5200 WILL work, if it doesn't you RMA the CPU until you get one that handles it. DDR5-6000 SHOULD work, if it doesn't, sucks to be you -- AMD are unlikely to help.
@@exscape Anything below spec is broken, and as Intel is showing, they can start breaking en masse a year later as well.
@@osirisgolad What do you mean by below spec? Also, Intel and AMD have done this for many, many years. The 14900K supports up to DDR5-5600 even though 7200 seems to be a fairly common speed to run, and some run over 8000.
This is hands down the best review I've seen and answered many of the questions that I left others asking. Why AMD would sandbag their own product to the point of barely competing with the existing generation is very strange.
Let's be real, 90% of people don't really care if 6-8 core parts are pulling another 15-20W on avg as long as performance is there and it's easy to cool. I'm suspicious this is mostly for product segmentation with X3D but I think they went a bit too far.
The last question I have is memory. Looks like the IO die is effectively the same but I'm curious if there's more FCLK headroom on average compared to Zen4.
So basically ryzen 9 is just a efficiency boost over 7.
BUT has absolutely crazy overclocking and custom cooling potential. Interesting!
Way faster in gaming too the game he chosed didn't shot any improvements unfortunately . But i went with zen 4 x3d anyways and saved a tone of money
now we can use Ryzen as space heater during the winter. Did I leave the Ryzen on?
It is a Zen 4 refresh. Which means it's a flop.
@@teethashaquan5395 nope but you have the right to feel that way when 15-16% ipc is not enough crazy world we live in
@@teethashaquan5395its not though? Lay off the drugs bro. Its a completely new architecture. 16% IPC improvement is refresh for you?
I realise its not optimal tuning, but 22k cinebench with about 160W is pretty close to an i5-13600k with the default power limits. Be interesting to see the results once the new microcode is out.
Intels microcode thing is so weird. I think there's a quality issue going on with intel because my 13700k gets 30k in R23 at 160W and 75c, but nearly incinerates itself with stock settings. I feel like Intel has pumped the limits to ensure poor quality CPUs can be sold as higher end parts, but if you have a good one its a great, cool chip.
@@chrism2964 I think you're right. Igor's Lab article "Intel Core 14th Gen Binning Results from almost 600 CPUs" found at stock settings i9s were seeing a hair over 0.1v variation on the VID between the best and worst chips. There was a similar spread on the i7 and i5 parts, but they were more closely clustered with just a few outliers.
@@chrism2964 That's probably because of via oxidation. People have mentioned that when you give it more voltage, you get around stability issue caused by that problem but while also deteriorating the CPU even further. The good chips are, imo, just the ones that didn't rust!
I think they limited power consumption for two reasons:
1. Because people see INTEL chips overheating, they'll go for AMD chips underheating (bigger gap).
2. Because it gives X3D chips more headroom to beat current X chips and previous X3D chips (bigger uplift).
That and people complained about the high power draw and high temp of the 7000 series X chips when they first launched.
I dont think there will be bigger than single digit % fps difference between 7800x3d and 9800x3d unless AMD keeps 5.5ghz Boost clocks from 9700x and sygnificaly pump up TDP from 65W to 120+W. Even than i dont expert it to be faster than 7800x3d for more than 10-15% in 1080p, 5-10% in 1440p and 1-5% in 4K.
@@johnrehak they can maintain the higher clocks with lower wattage as shown here so why wouldnt the clocks be higher
i remember a time when power efficiency was high regarded in tech, somehow in the last 10 years people stopped giving a fuck, it's nice to see good performing parts that will give extra performance with OC for those less efficiency inclined as it should be.
Really makes me think Zen 5 is being downplayed and underrated too much. It consumes way less power, so all these other reviewers are all acting like that's all it has going for it. Why not test it with PBO!!!
@@jackofthecoke because you do multiple tests and most people dont even turn PBO on in their boards
They could've still upgraded to 75-85W, still low for today's standards while the average consumer still benefits a lot from it
I really appreciate someone stepping back from the thermal arms races which has only provided us with incremental performance benefits while escalating the power draw / heat / noise /... (I run an AMD CPU with integrated graphics: Its quiet, it keeps the power bill down, and unlike its predecessor it does not heat up the room like crazy in summer)
Efficiency is still important in mobile. Look at Apple's M series of chips and the stir they made.
best video from all the reviewers. amazing that you can effectively overclock the 9700x and they sell it at a stock profile which is extremely safe and efficient. well done AMD. the other reviewers need to figure out the PBO overclocking part.
I'm not a gamer, I'm a productivity user, software development, video, etc. I'm interested in the 9700X or 9900X simply for the lower power, I don't need 2 more FPS in games that I never play.
It still shouldn't require tweaking PBO limits to get it. It should be in a power profile end users can access.
I appreciate that you took a hard look at PBO.
It is disappointing that there is little gaming uplift on the 9700X PBO over 7700X, that doesn't bode well for an eventual 9800X3D having gaming uplift over 7800X3D, unless the 9800X3D can run higher clocks.
I do prefer that the out-of-box default is effectively eco-mode. It did not sit well with me that 7000-series default chased the temperature limit.
i think 9800X3D will be unlocked for OC so it will smoke 7800X3d
@@EtaCarinaeSC how ryzen is designed, I don't think the 9800x3d will smoke his predecessor in gaming if the keep the same cache size. If the increase it a bit (L2 or L3), it's a major win, if not ouch.
I might be wrong tho and I hope so.
The most optimistic scenario is that the new architecture is memory bottlenecked at 6000 in gaming, or by Infinity Fabric, etc. The memory spec bump from 5200 to 5600 hopefully indicates that the realistic minimum has gone from 6000 to 6400. And a 9700X at 6400 might be better than 7700X at 6000. We'll see soon, when Roman or HUB do memory scaling testing, Infinity Fabric, etc. 9800X3D would presumably not care much about memory bottlenecks, as with 7800X3D.
@@evguenict guess time will tell, will come back here after its release.
@@evguenict Biggest drawback from X3D chips is the weak heat transfer from die to IHS because of the extra 1 stack of 3D-Vcache. If AMD can improve heat transfer , new X3D parts can clock higher. Current X3D chips cannot go higher than 1.25V Vcore because the stacked cache absolutely don't like that.
Impressive that the 8 core part is getting the same multicore Cinebench R23 score as my 16 core 3950x with just a small overclock.
Thank you for providing an Objective empirical review of these parts Der8auer, this is highly informative.
The out-of-box performance of these chips on launch is disappointing, but they have vastly improved efficiency. However when you enabled PBO, you got the performance figure uplift i was roughly expecting! PBO overclocking usually doesn't result in such a huge gain - maybe roughly 5%, but in this case it resulted in a 21% uplift. This strongly suggest to me that these chips are being limited at the firmware level by default.
Awesome job as always der8auer!
It's so refreshing to see proper review and you not jumping on the hate bandwagon with rest. You are also the first to point out that while the CPU in stock is equal to 9700X, it does that at 40% less power. Kudos
Not enough power.... Asus, hold my beer. 😂
Multi core enhancement TM as usual lol
Asus will add the "Intel Specs" option for the Zen 5 CPUs.
Recently I had a bit of instability in my system after adding another 32GB of RAM, but as I didn't have the patience I simply used Asus' Auto OC in the BIOS.
Today I noticed that the OC put my Ryzen 7 2700 (which barely pulls 80w under load) at 1.365V for just 3.4Ghz LoL
Stock voltage for the 2700 is ~1.05 by the way it runs at 3.2 GHz base and goes to ~1.09V with boost
@FlavioSantos-uw1mr Hey, sometimes you have to send it. Asus says just buy an extra cpu to have on hand in case you smoke the first one lol. I have never liked all this auto OC crap people should have too dig into the settings and stress test to find the optimal clocks/voltage settings like the good old days of overclocking, but hey if people want auto OC's with way too much voltage then have at it I guess.
Thank you for providing an informative and detailed review without feelings (like most others did). Based on your review these are picks for me.
be interesting to see how the CPU reacts after some proper PBO tuning
The core for the x3D chips is ready. It's amazing.
I have been planning on getting the 9700X, thank you for showing us what it can do when the power limits are cranked up! Thumbs up.
One of the best Ryzen 9000 review, hands down. At least you are delivering facts even about the PBO aspect of the architecture and sharing povs on the efficiency. Also, a great touch having the power draw into the graphs. Thanks for your work 👏
3700x and 5700x were also 88w ppt (65w tdp), that's just the 7700x that needed to go higher "out of the box" to look good. I'm glad they've remake the *700X segment right, a more budget firendly 8 core part with lower cooling needs, and with the *800x and the *800X3D becoming once more the higher performance/power draw parts. That seems more logic to me.
My thoughts exactly. Depending on how the 9800x3d is tuned, we may or may not see a 9800x if the market calls for it.
By far the best review of the 9700x thank you very much
Thanks a lot :)
It is a branding switcheroo so they can claim these parts are releasing for a cheaper MSRP vs the 7600X and 7700X, when it is actually a price increase from 7600 and 7700. They can, in reverse order, release the stronger tuned version later on, with more Xs and Ts in the product name.
but this also causes disappointing reviews 🤔
Tbh at this point it's just speculation. They had the whole delayed launch thing, maybe they royally screwed up the power profiles and this is all they can salvage in a few weeks during the delay?
@@WILLPORKERnah, even the TDP match with the leaks months before the official announcement, what I think watching other reviewers is that this is a software optimization problem, an AGESA and BIOS updates might be announced in the next months, AMD rushed the release because Qualcomm and Intel AL bluff ab that would explain why the software is still unmature
5:12 When you're saying that 7800X3D is not that strong while it's just 2 fps lower compared to 7700X while he consumes 138 watts instead, this is a misrepresentation of this CPU.
Considering the power draw and the arhitectural peculiarities, 7800X3D with a bit more of a power draw can easily get head to head with 9700X which is newer and with 7700X which draws way more power compared to 7800X3D.
It feels like AMD had been blasting power at Zen4 to compete with Intel and they took Zen5 for a turn for efficiency.
It would be interesting to see how 9950x perfoms efficiency-wise against 7950x with eco mode.
Well, can't really complain that much if there's real efficiency gain, I guess.
Gamers can complain
Good video! Yeah I think that AMD was trying to play it safe with lowering the stock power profiles. Also it kind of proves a point that with the lower power draw, they can perform just as good as last gen which some may prefer. Also this give them room to flex a bit with the higher tier chips on performance charts. I hope when the 9000 X3D chips are released we see something real impressive.
Kinda seems like the main vice 9700x's got is overclocking possibility. Can't wait for the delid vid!
Brilliant review, thanks a lot for sharing these results!
Including Intel 12 gen CPUs is completely fair for the reference.
Thank you for being the only reviewer to be posting metrics at 1440p or higher!!
AMD doing an inverse-intel and killing anything this CPU might have going on by restricting power too much lol
Some people love efficiency
They can release a 9800x if they feel like it personnaly i always bought the 7xx variant they're always the more fun to oc R1700 R7 3700x R7 5700x
overshot last gen with their all out tactic massively undershot this gen with full out efficiency (which honestly im a fan of) but we can always count on X3D to save the day for gaming!
@@TestarossaF110Genuinely curious how 9800x3D will perform. Considering the Intel situation, will AMD release it as lackluster or actually go for the throat.
@ZioComposite I pray they go full out and just keep doing that, making those (X3D) the *actual* new CPUs gamers can and will go for, but I fear AMD doesn't know what to do and keeps being kinda stuck. time will tell I guess.
Always touching on the topics that matter to me most
Wow a CPU that will show large gains from manual overclocking? Did I wake up in 2008?
Also - this is a GREAT sign for the 9800x3d chips. If they have increased PBO headroom like rumored - this could be an awesome gen.
2008... immediately had my sub-100eur E4300 1.8GHz at 3.6GHz in mind. honestly miss those times
I highly doubt you'll be able to do much with PBO in an X3D chip
@@rustler08 if you go back and look at the AMD CES coverage - they kept bringing up 'additional improvements' to x3d. They wouldn't say exactly what it was, but the rumor was PBO for x3d.
That could be an amazing combo if its true!
@@rustler08 The rumour (and it is a bit more than unsubstantiated rumor) has it that X3D chips will be more unlocked than previously. PBO will almost definitely be usable - especially given that curve optimizer is to a degree usable on 7000 X3D chips (depends on BIOS). If they do come out almost completely unlocked (Is suspect they will have locked Temperature maximum), then yes, they will be killer CPU's.
AMD officially say the 9700X is a 65w part that boosts "up to" 5.5GHz (but under all core it only hits 4.5GHz), and the 7700X is a 105w part, with similar top clocks, so it's not really a rumor so much as basic mathematical deduction that if the chip is allowed to boost to 5.5GHz it will run faster.
Thanks for doing this! It was exactly what I was thinking about.
I wonder how Curve Optimizer will affect performance and temperatures with this CPU.
This is the correct review, using PBO to unleash the CPU and there we have it up to 25% more capable when needed or just as powerfull but way more efficient when it is ok to just browse around, Awesome, Thanks der8auer
Getting the same performance at a lower power limit generation after generation is considered a win in my book. Number go down is also a good thing.
It's only consuming 10% less power as compared to previous generation. That's garbage.
You're paying 30% more for 3% performance, dude, wake up.
Except it costs more…
I don't wanna ding anyone looking for efficiency but I thought we already had that with the 7000 series, it's called eco mode and some folks were trying to build hype around it with the 5000 series.
I think I'd like to see someone compare the 7700x in eco mode to the 9700x because I'm thinking it will be pretty underwhelming.
@@creativityfails1 Not just that, but 7600 and 7700 had that "ECO" mode on from default, Zen 5 outside servers/Linux is just trash.
@@creativityfails1 I mean, seeing as zen5's default is basically 'eco mode' it matches the full performance of a 7700x while using 40% less power, the 7700x will still lose performance on its own eco mode, even if it's just 5-10% perf loss, that's still 5-10% less as compared to another cpu that is then running near same power.
I'm rather interested instead at how enabling PBO is like a 21% gain for zen5 in multicore workloads especially, which then puts it way ahead of the 7700x at similar power.
9800x3d will be awesome with the new thermal improvement. X3D cpus always suffered from heat accumulation despite drawing little power.
I think people are way too obsessed with raw performance. Normal user doesn't need that much CPU power anyway, so having a easier (less cooling needed) and more stable build is better. I love that AMD focused on efficiency instead of destroying the CPU with too much heat and power draw.
And if you need more CPU performance, you can achieve that by doing overclock.
best review ive seen. you earned a new sub!
Maybe they have some concerns about stability at "high" power and they have learnt the lesson from Intel, hence they are very conservative with the power... Just speculating.
Yes, perhaps a lesson learned from the Intel current problem. I see it as a defensive strategy against possible RMA's if something unexpected happens.
Are people sending Zen 4s back because they're burning out? if they are, they're being pretty quiet about it. If they're not, then why would this be anything about that, given it's way more efficient. PBO lets it clock up like usual anyway. Will be interesting to see what BIOS updates bring here.
People complained about high power draw with zen 4 default settings. I think this proves people will be disappointed either way they go, but you can always change one setting and have it run the way you want before you even boot it up the first time.
Nah I just think the new gen looks more impressive like this. They basically are showing same performance at 40% power reduction, which sounds much better than 15% more performance at same power.
Furthermore if you look at the 9700x in R23 with unlimited power, it scores similarly to a 14600k at a similar power budget...
Thus I believe that locking them down was the best marketing choice for AMD
this, AMD is obv scared of PBO stability
Thanks for a excellent review. It was thorough, multi-dimensional and full context. Is AMD making a massive play in the OEM market by keeping the power down and performance similar? It's obvious AMD is continuing it's cadence of innovation and forward looking design.
I agree with your conclusion (I even wrote on different review, that 95W TDP or something in between 65 and 95 would be great). I would prefer if AMD said that this generation isn't exactly for gamers - it won't be worse, but it won't be much better, but you will get some savings in power. As for price - it's lower than MSRP of 7000 series at launch and will likely drift down in few weeks, but gamers on budget might want to go with 7000 series.
Rant:
People (commenters) write that this is "AMD stagnation" comparing it to Intel... I'm sorry, but I can't agree. During Intel stagnation you were lucky to get 5% uplift with basically the same power use gen over gen and sometimes the only uplift came from increasing default power (so the actual improvements were that it moved performance and power use linearly - by again usually around 5%). Also this was never resolved with future BIOS updates. Here it is possible that 9000 CPU's will get couple percent boost compared to 7000 series, just with BIOS updates alone. Of course this isn't on reviewer to predict, this is on AMD to deliver and I don't blame anyone for reviewing CPU "as is" (it's how it should be done, with speculation being perhaps in separate chapter of the video or something), nor do I know if AMD will deliver those updates. But coming to conclusion that it is "the same" as Intel stagnation after watching 1-2 day-1 reviews is unbecoming of people. I really hope they will stop. Thankfully most reviewers didn't do stuff like that, but comments such as this are triggering me. More than they should.
Agreed. Truth be told, I'm not particularly impressed by the gains in games, but getting the same performance as the 7700X, with a decreased power consumption, _is_ to me. Also, given Zen 5 lies the groundwork for upcoming generations, I have good hope the lacking gains here will come with Zen 6 and newer.
@@JarnoOverwijk It is a bit of inverse of what they were doing earlier - but 9700X gains over 7700X in those titles that are primarily single threaded. Or with very few threads. Due to lack of power budget, you can actually have 9600X overtake 9700X in games that use more than (approx.) 5 threads. Because those 6 cores can use the same amount of energy as 8 cores. So they boost higher.
So for games 9700X is a bit of a coin toss (in some games it will be worse than 7700X, but this will mostly be in games that already have high FPS). On average it gets approx. 5% over 7700X, but consumes 40% less power.
In productivity workloads gains are higher on average (approx. 9%). And of course it overclocks much better. I don't think AMD rejects RMA's based on using DDR-6000 or even using curve optimizer/unlocking power limits. Though they might if you RMA the part that you used in breaking some LN2 record ;)
It's going to be interesting to see 9900X (for once) and 9950X (since this part doesn't lower TDP).
I doubt 95 would make much difference vs 88w that it is already boosting to, but 105 would be enough without tipping it off efficiency curve
@@WayStedYou It is operating on 65W TDP. Which gives it 88W of PPT. So I'm talking about 125W PPT.
Also I meant that best point would be between 65 and 95 probably. Remember TDP is different measure than PPT.
I appreciate the balanced review without being clickbaity
Hilarious to see the 4090 being bigger than the 360 AIO kit!
The 4090 cooler is cooling 450 watts, and the 9700X cooler is cooling 170 watts with PBO overclock.
@@exileutThe 4090 cooler’s were made to cool 650watts. Which is why they are so massive.
@@exileut 170w in a die that is 1/5th the size so the heat is far more dense
Sounds like a Good Value with PBO or Overclocking, great video! Very enlightening.
Intel had crazy high power draw, AMD now has crazy low power draw !
Soon: ”More power is better!”
😂😂😂
For user bencmark…
😂
These days being extreme is all that matters :/
Thank you for actually benching with PBO as well!!! Was very frustrated that most reviewers barely even mentioned it
Thanks!!
@@der8auer-en Can you test curve shaper with negative offsets once you have time?
People keep screaming about how these chips aren't good but the fact that it matches last gen's performance at 40% less power is nothing short of incredible.
yeah but you wouldnt upgrade from last gen just for that
Who cares about that though? I'd much rather they made use of that efficiency by increasing tdp and clocks for more performance.
@@medovk Some people with expensive electric bills may prefer that. It also allow for use of cheaper cooler, mobo (but not psu otherwise it would explode😂)
@@SuperLloyd84 some people really value efficiency, and if you don't value that it looks like PBO can give these the massive improvement people were looking for.
@@muhammadhanifkurnaen6689 Unless you're running 100% all-core loads 24/7, no one is upgrading from a 7700x to a 9700x to save on their electric bill. If power usage is that important, just buy a gaming laptop.
Great vid, thanks!
Honestly I love AMD's idea
We actually need a generation that just improves efficiency, imagine rtx 4090 performance with card that pulls 200w
With modern prices that would be golden
They didnt improve anything xD Check out computerbase review and gaming performance as the same W... its the same
@@gaav87 Did you even watch the video?
@@duccc Read/watch other reviews Its litteraly same power-draw in gaming xD
This is exactly what invidia did with the 40 series and took a lot of flack for it. My 4070 does what my 3080 did at 200w lower power.
@@gaav87 you didnt get over 5min mark in this video did you?
Great video. Love the fact that you talked about overcloking. I see that being a huge value to AMD. If only they would have stated that on their own. "we have an extremely efficient CPU but if you want to get the most out of it all you have to do is unlock it's power". That would have been a great PR move in my opinion. Looking forward to your overcloking videos.
Overclock it so it degrades and burns out after 3 months?
@@Tugela60 I have never had a CPU burnup by overcloking. Maybe had some stability issues but if you dial them in they are rock solid.
@Jaywithag78 Older CPUs ran at lower clock speeds and had more robust fab processes, so they were less susceptible to damage than newer processors. That is the problem, people think they can subject the latest CPUs to the same sort of abuse and it would be ok, except it is not ok and now they cry.
dont understand why do you compare it to the 7700x and not the 7700 (non-X) which is also a 65W CPU so that would be apple to apple, everybody know that the 7000 series X version is inefficient and the non-x version is the efficient so i wanted to know to performance gain against the same TDP version
Because the name suggests that it's a replacement for 7700x, not the 7700.
@@ALph4cro who cares what the name suggests? AMD naming has been a shitshow for years
Good review - as always :) Cheers!
I have a feeling this will be the "highly underrated" generation.
It seems like it was built and released on conflicting premises. When the gap in the commercial/professional space opens up due to Intel screwing up, it would be dumb not to capitalize and release the most efficient productivity - focused chip possible. It can still be unlocked by enthusiasts and when they inevitably saturate the market the price will come down.
Really looking forward to seeing how far this CPU generation can be pushed with direct die cooling and some effort tinkering the power and performance settings. Seems to have way more headroom than most releases.
It's time everyone grow up to the fact that power efficiency is an important factor to consider. I think that it's a huge achievement that the chips perform just as good as last gen, but have substantially improved thermals and power draw. Energy costs keep going up, it's extremely irresponsible and elitist to think that computer chips shouldn't care about these factors. The reality of how much your computer really costs is the price of the hardware + all the energy it will cost during its lifetime.
It's really not. A gaming PC costs $20/yr to run.
@@emericanchaos in your area
@@Neuroszima shouldn't really matter what area. We're talking about power demand under load. Unless you're maxing your system out 24/7 you shouldn't see anything close to a hit on your bill. It's a misunderstood concept by so many casual tech people.
@@emericanchaos third world countries, laptop users and data center admins:
Insulting people for wanting their CPU to run at full power is incredibly immature, especially when the power savings will have little effect on your bill. You have no right to tell anyone to grow up when you act like this.
Thanks Der8auer for being transparent and giving a real look at how the CPU perform rather testing it at tdp and saying it's a bad CPU.
It's basically 2 options with amd at this point. Normal X for productivity, and X3D for gaming. If you're not going for one thing, then you shouldn't bother looking at it.
Yay! more exotic content incoming for Ryzen 9000 series, no complaints!
I love the power efficiency, i just wish it was priced less
In my experience, prices will drop quite a bit after either ~6 months or the release of the 9800X3D.
Even the 7700X faced a reduction after a short while because it barely matched the 5800X3D so we just need to wait.
This CPU looks pretty good for home server or light workstation use but we need to see the IDLE power.
@@immortalzodd7474idk, the massive price increases and binning delays makes me think that the 4nm TSMC node might not be yielding as expected. That’s going to take longer than 6 months to work out if it’s not AMD’s fault.
You can get a 16 core part for the price the 1800x launched at with inflation, or a 9900x for the price a 1700x launched for, the 8 core part is 150 dollars less than the original launch price the only thing that hasn't moved is the 6 core parts but that is due to diminishing returns in what they can remove from their margin to make the part in the first place.
@@macicoinc9363 Agreed, but prices are gonna go down as I do not see general consumers (gamers) flocking to this one.
It is going to sell way worse than the 7700X so it should fall in price faster than that one.
We are officially in a recession/stanflataion though so things may get weird =S
Considering that, this gen is able to achieve that performance as the predecessor with 2/3-1/2 with factory power limit is impressive to me.
Micro code ain't gonna fix the ring bus. 💀
Best review, great work!
But you have to void warranty just to get okayish performance lift from zen 4... lol
So glad you relaxed the power limits to actually show what this chip can do
The most accurate 9700X review on RUclips.
When a manufacturer limits power more than expected it's almost certainly because of bad yields forcing them to requalify their bins with lower silicon quality chips. The other option is that they're trying to bait Intel to cut their processors' performance more with their microcode update by giving them a false sense of security. Then 9900X and 9950X comes out and mops up
No, it's because ARM is becoming an actual threat to X86, and the only reason power has been increasing over the years is because AMD was competing with Intel, and they had to always one up each other in benchmarks, and the only way to do this is to push the CPU beyond its most efficient point
This is a very positive thing, same performance at considerably less wattage means the CPU is actually optimized, from this point forward it's going to be the same, lower power or same power consumption and increases in performance compared to the 9000 generation
@@xSyn08maybe, although I don’t really see arm usurping x86 in the desktop, workstation, and server space for a while. x86 has software inertia that arm won’t hurdle for at least a decade.
If Linux can’t convince CAD/Simulation companies to port to a better OS, then Arm being in a bunch of mobile and laptop devices won’t convince them to provide support either.
and yet it can easily boost to double the power draw as shown here
@@macicoinc9363 ARM to x86 emulation on Windows is already achieving the same or even better performance than Rosetta 2 did
A decade is wishful thinking when Snapdragon X Elite exists, it may be a year before we start seeing those chips sold as desktop parts
For your knowledge the X Elite (X1E-84-100) is a mobile 80W part that sits between AMDs 9600X and 9700X in terms of performance
From what I know x86 emulation on Windows gets you like 90%+ of the performance
Thanks @der8auer for this deep dive review, i watched GN / HUB reviews and their reviews were lacking the PBO/OC side of these CPUs... seems like they rushed their reviews... you are the no. 1 reviewer for me from now on
HUB did used PBO for gaming test, a 2-4% increase
@@lokeung0807 people don't buy CPUs only for gaming
@@HoretzYT I know, but their main focus is gaming, for all CPU, especially for HUB
Finally non biased opinion (HUB says its sucks, Steve: meh). Power efficiency is huge compared to Zen 4. OC capabilities is also very nice. We need to wait for lower prices and new mobos, ofc.
I need a new rig BADLY, but I'm going to be waiting for the X870 boards to drop before making a decision. Lots of reviews from the major outlets reported BSOD, memory issues, random reboots... All the problems I had on B650, I just don't think the first-gen AM5 boards are any good.
@@JPDuffy How badly? Owner of 3770K (since 2012) here, lol.
@@FroschYankee ouch
@@FroschYankee i7-5930K. You win this one. Hahaha.
Appreciate not bashing on it due to no performance gains @ 65tdp....which less power draw and lower temps. Great for sff builds..and still have the headroom for a little performance boost w/ oc/pbo.
This is the only worthwhile review out of every techtuber, lol. What a mess.
Yeah, even GN is a joke this time.
For servers this is big. Thanks for the great video.
Sees GN review. "Yawn Chip."
Sees Der8auer remove power limits. "Yowsers CPU!"
It's still a yawn chip compared to previous generations that worked out of the box.
Most interesting. SkatterBencher uncovered performance that appears to align with what we see here. Much more to come with all the Zen5 future hard looks.
as I said on GN Steve channel - this should be just called 9700. Basta ! 😎
And also priced acordingly...
They are going to release 9800X, just like 3800X was 105W version of 65W 3700X
The only proper review for the 9700x on YT.
I think reducing power for the same performance from the last gen is a smart engineering. Adding more power for more performance is a lazy angineering. Anyone can do that. Intel did that and look where they are now.
Finally a review that talks the truth instead of just bashing it
*They also used the wrong ram. They treated the 9000 CPUs like they were 7000 seties.*
*The 7000 series supports an SPD ram frequency of 5,200MHz. The 9000 series supports an SPD ram frequency of 5,600MHz. Which means the sweet spot would have been around 6,800-7,200 MHz (with the proper EXPO settings, timings, and voltages).*
*A 2-8% performance gain over the benchmarks done with 6,000MHz ram. (2-8% gains). What a joke!! Every benchmarker, so far, has dropped the ball and should've done their research before blindly using low-end parts.*
You benchmark a processor at its spec settings, not overclocked settings.
new ram speed might need 800 series MB. now 600 series MB looklikes 6000mhz cl30 is sweet spot. considering IFCLK 2000 supported 6000mhz. from leaks, look like 9000 series support 2400 IFCLK. haven seen anyone tested that before even tho 6400mhz seems like the new sweet spot. but getting these standard processor to 7200mhz will be very good. hoping that is possible with 800 series MB and amd can crawl back performance from there
Thanks Mr. Bauer
Over the years my AM4 builds have been upgraded to quite premium quality machines. The AM4 platform is mature, incredibly stable, and comparatively capable versus the newest generation performance. The 9000X series is nowhere near compelling enough for me to leave my AM4 dream machines behind.
Great review Roman, thx.
Crazy good CPU's for ITX.
Only for productivity ITX. And that is probably only until the 9950X launches. Until then, as DerBauer said, the 7800X3D uses even less power, is cheaper, and performs better in gaming.
I gotta say. I can't really see a niche the 9700X fills...
I was thinking the same, like in an AsRock DeskMeet X600 and equivalent.
DeskMeet X600 is interesting because it has no chipsets. All I/O is directly from the CPU.
It is however limited to 65W TDP, so the new 9700X would pair up nicely with it.
Give it less power with the curve and see what happens with the all-core clocks. Less power, less heat means higher boost if its stable. Combine it with an overclock should be crazy good CPU.