Help support our testing! New colorful solder mats on the store, plus copper plated mule mugs in stock! Use code THISISFINE for 10% off while it is active: store.gamersnexus.net/ Check out the latest hardware news here: th-cam.com/video/ZRJaF2ky9CE/w-d-xo.html
0:54 can i just say, i still to this day think these side bars counting down until the next slide is such an excellent touch. stood out to me when i started watching your videos years ago and i still appreciate it
@@3333218 You already have TH-cam's built-in chapters feature for that. Every topic is split into chapters that are easily accessible just by tapping on the player, so sidebars would be quite redundant.
@@SierraAAM Not really. This is a visual cue of how long the chapter takes, on screen, while you're watching. And I personally prefer watching stuff in Fullscreen.
> Lots of comments about power efficency > Steve: "Ok I will briefly address that" > one week of office sleeping later > 40ish minutes video We love you steve, no joke
Never mind FPS/Watt or frames/Joule. 1 Horsepower = 745.7 Joules per second, so you could measure in FPS/hp, which I think we can all agree sounds the most impressive.
You saved me so much money. I was going to buy Zen 5, but ended up going with a 7900x3D instead. And before someone says it, no I need more cores than I need gaming performance.
@@Electric_Bill Performs better at gaming than 7900x and only cost about $30 more. Don't see the issue right now. In the past it was terrible, but now it's a good balance. 7950x and 7950x3D are still over $500. 7900x and 7900x3D can be found for $370 and $400 respectively.
I constantly have to remind myself wtf that game was, i played it for 2 days and realized it was like somebody took Fallout 4 and took all the good parts of it out and replaced it with generic filler. What kind of bonehead thinks making a galaxy of random generic empty worlds that all functionally are the same is anything people wanna play? The funniest part of that game is that Todd really expected that game to be some ground breaking experience that changes the industry. Holy shit bethesda is dead
Between the intro and the first 7-8 minutes of testing, it was pretty obvious where the conclusion was going to end up. At about 15 minutes in, I thought "even though the comments from the opening of the video were proven "wrong" in instances, I just learned a ton and got shown a new way of thinking about things. Hopefully this doesn't discourage people from sharing their thoughts in the future." I'm not surprised at all that Steve (and his team) took a few moments around 33:30 to tell those commenters that they're appreciated. Even though I'm not surprised, that cool acknowledgement should be reciprocated. "Thanks Steve!"
half the time all TechTub is doing is acting like a 5 frames a second faster than the other guy, is some massive win that most consumers cant even see. The whole CPU and GPU testing shit from Techtub for the last 10 years is over hype click bait for views. None of it really means anything.
@@GamersNexus Do you think that if the 800 series boards have a higher stable ram speed say in the 7000mhz range that it would help improve performance for the 9000 chips at all?
@trulyskillsince07 There's a limiting factor in the infinity fabric I think, which makes 6000Mhz the point after which no performance is gained. I think maybe it was der8auer I saw point this out.
THIS is the deep dive that no one in the industry ever wanted anyone to see. Not Intel, not even AMD. THIS is the full holistic picture of why the large cache X3D CPUs are so damn good. They're efficient EVERYWHERE. It's not whether they're the power blasting 24 core kind of speed, it's that they're the better all around CPU for just about damn near everything from a perspective of this question: "What is the overall best price/performance/power CPU technology?" The answer is the monster cache CPUs. Can we have more of those across all the CPU vendor's product stacks please?
Heres the issue. The gamers have always subsidized the average user market. Most people don't need X3D or even a 7700X. I3s and Ryzen 3s are by far the biggest sellers as they are all many people need, but the way those sell is purely "What is cheapest." By charging the gamers more, AND making the low chips out of high end chip offcuts, it makes it easier to make this buisness model work. (Intel obviously does the same over a wider range... by selling i9s for more then can sell I3s and their even lower SKUs for less) The X3D line creates a problem as it takes additional money to develop X3D chips over the standard chips.Not much extra money now that AMD has done it twice, but some. If all the gamers buy the X3D line, AMD has to get more money from the initial round. And thats going to raise the price of the cheapest chips. LONG TERM, AMD probably aims to only make X3D chips, as that's clearly going to please everyone. But until they are ready to do that, they still need to release non X3D Ryzen 7 and Ryzen 9 chips to keep the lower skus cheaper. What they SHOULD do is just jump straight to APUs. Keeping their APU and CPU markets separate and releasing their CPUs first is an older strategy that appeals more to gamers, BUT leads to low level chips that really no one should buy (who really is the market for the 7500F. If I only want to spend $179 on a CPU, I am not shopping for a $300 GPU)
This is also a symptom of the fact that the DRAM speed lag is getting too extreme, especially because in the latest process nodes cache SRAM sizes are no longer shrinking, therefore it´;s harder and harder to shove more cache on chips and the ram keeps sucking
they'd be inefficient if you had two high cache CCDs and workloads that need to communicate between them, not really the case for single CCD gaming chips though. it would probably have been much better to stack a larger L4 cache on top of the IO die (kinda like RDNA), but it's hard to say what obstacles they encountered.
Seeing the blender workload show high efficiency makes total sense - one of the few AVX-512-supporting workloads, which Zen5 heavily improved upon (while not really doing much of anything for lower-width AVX2, SSE, and scalar workloads). On a technical level keeping the same-ish efficiency here _is_ kinda impressive - they managed to reduce the losses for non-AVX512 workloads far enough. Of course all that doesn't matter to most gaming workloads (today).
I don't think that has anything to do with AVX512, it's a 10% efficiency improvement over the 7700, which is about what would be expected for most workloads that aren't gaming, for example Cinebench sees similar efficiency improvement. Actual AVX512 workloads would show bigger efficiency improvements, probably around 25% when compared to the 7700, depending on the exact workload.
@@kaitoux8413 I think it’s very rare to find any pure avx512 workloads which are not synthetic. Even avx2 is rarely taken advantage of in favour of compatibility and easier code maintenance. In my experience, anything outside image/video and machine learning will have to be quite creative with their data alignment to use it efficiently.
At this rate, Steve is going to become a professor and open up Gamers Nexus University to spend entire courses proving to people how wrong they are 😁 [yes, I would enroll]
Using a method they came up with with tools that are arguably not suited for the task and motherboard that might be in a need of a bios update due to it probably running the "just make it work" version this early on.
What people should care about the power consumption was idling power consumption, since PCs are in idle most of the time. Use a bit less under load and 3x~4x more when idling is not an okay, AMD need to fix this.
YES!! I just discovered this a few weeks ago while researching parts for a home server/NAS... nobody mentions this.... I tested it with my 5900X & I couldn't get that thing to idle below 40W; whereas my new 8840HS (monolithic chip) idles at about a watt.... huge difference
One of the main reasons my 13900k has actually been quite efficient over the 1.5 year I've had it. Yea sucker draws a house worth of power during all core render workload, but I'm not running a render farm every day and most of the time the actual tasks I do during a 10h workday end up being below the inefficiency threshold where the e-cores are honestly amazing. It's barely powered on but still runs everything I need. And I still got the top-line performance for max output when I need it. Unironically I use less power now, compared to my previous 9700K which had half the rated tdp and peak power draw even lower. And yea it's been running full 10h average workdays for 1.5 years without a single crash or performance dip. Think I scored a golden sample considering all the recent meltdown issues.
@@Real_MisterSir idling consumption of mid-high end intel can go as low as 3~8 watt, that's honestly impressive. Meanwhile I'm struggling to push 3700x below 20~25 watt. Burning 25 watt for doing nothing is a pain in the ass.
PCs won't be idle most time at all if they're used in software development company... Ours are compiling all the time. After switching these peak machine from Intel to AMD, we saved 20% on electricity bill. So AMD for busy PCs, Intel for idles.
@@shootingstar7896 AMD for constant load, Intel for load intervals -but even this is way oversimplified tbh. Not all load is equal, nor constant. For some (like you guys) it is, and for me if I ran a 24/7 render farm, it would be too. But my productivity involves a multitude of tasks, most of which do not spike into the realm of inefficiency levels of load. Power consumption during photoshop vs 3d modelling vs 3d render vs film editing vs film render vs leisure time vs email and conference calls and online meets. A multitask work station often has to serve a vast array of workloads that are rarely equal.
@@CNTRI715 Its crazy how things have changed. I remembered in 2020 AMD released the 5000 series and those tech communities were stating that intels CPU would still be superior in the future. Now they they jumped ship and are defending AMD.
@@ffwast The 7700X is pretty much the same as the 5800X3D on average, so a lot of people assumed the 9700X would be on par with the 7800X3D, but cheaper. I could have told them that wouldn't be the case right off the bat, as Zen 4 is a full 70% node shrink over Zen 3, but Zen 5 is on the worst half-node shrink we've had in a long time (only 20%).
I know it's not the main topic of today's video, but here is your periodic reminder that Intel also delivers 65W cpus, in the range of non-K cpus. When doing an efficiency test comparison, it feels way more fair to include this variant rather than the overclocked K series, which will obviously fare badly at an efficiency test, same way the 7700 is way more efficient than the 7700X.
@@fral1073 Yes it was BUT STILL Intel was bashed of poor efficiency several times. I do not think that is appropriate when only the K versions were included. And comparison had power limited results for AMD, not for Intel. That makes comparing Intel efficiency here very unfair even though this focused on AMD.
@@jarnovilen5259 Well tbf Intel only cared about pushing the power as much as they can to get better scores for benchmarks, since Alder Lake we're seeing both Intel and AMD pushing their CPUs far from their efficiency curve and just now AMD is lowering their TDP to pre Zen 4 levels.
@@jarnovilen5259 Intel's K series were being compared to AMD's X series when being (rightly) bashed for their poor efficiency. That seems fair for a comparison that isn't focused on Intel versus AMD. Intel's chips get better efficiency at 65W than at 100W+ but it's still worse than AMD at the same wattage, so the rankings wouldn't change much.
Once again thanking you for including stellaris in benchmarks, initial reviews had me thinking the Clausewitz engine loved the new architecture, but seeing the simulation watts per hour makes thinks more complex
I don't think old Clauswitz engines don't love much about new architecture, but the new one in imperator Rome , CK3 , HOI4 (they get updated in recent years) and Vic3 is. The best test for the new clauswitz engine , I think it belongs to vic3 as the most cpu intensive game with this new Clauswitz engine due to the fact that the game simulate economy is extremely complicate along with with lot of data that nearly burn PDX company before.
@@LeonNadwislanski It has nothing to do with anything. You know some interstate truck drivers almost never sleep on a bed. Are they not respecting themselves?
I was one of those who thought that you were making far too little of the quoted 30% efficiency improvement. I'm disappointed by the results of your extended testing, but very thankful for the fact that you went through the extra effort. This is why this is basically the best hardware review source anywhere. I hope that you do continue efficiency testing going forwards -- even if you don't spend a huge amount of time on it in videos as you did in this one, just having the charts and conclusions is valuable.
105W-65W TDP CPUs don't really consume that much power anyway. It's like 80W at most for the 65W TDP parts, and that's at very high load. The GPU is where most of your PC's power bill goes to, especially with these CPUs. These CPUs, on average throughout the year, consume about as much as an incandescent light bulb. For mid-end AMD CPUs, it's the difference between running a 45W or a 60W light bulb at most, best case scenario. Honestly, it won't save you much at all, even if power costs are very high where you live. Even a 30% reduction in *power consumption* at high load between AMD CPUs at the low to mid end would be near irrelevant to most people. Because it's not always going to run at high load, where a 30% reduction actually saves 20-40 Watts instead of 5-15 at idle and low load - and that's if the 30% reduction even persists at idle and low, which is not at all a given. (Sometimes idle power consumption increases for some reason in an otherwise more efficient CPU, like might actually be the case with Zen 5.) It barely even matters for PSUs. It does, but only a little. It's high power consumption CPUs, like a lot of Intel CPUs, for example, where that decrease definitely matters. Both for power bills and PSU choice.
I dont get it. 7800x3d and 9700x. Dont they have the same tdp? My 7800x3d pulls lile 55 watts in games. Can you not trust hwmonitor etc? Figured it was the same.
I'll be honest, I'm curious how the next generation Intel processors are going to end up efficiency wise. Alder Lake looked fairly decent, anything after that a disaster for efficiency, with the instability problems stacked on top of that. 😬
@@DjuntasHW monitor reports what the sensors measure which could be a problem in comparisons where different products measure different parts of the CPU / GPU / Motherboard / etc. Like with AMD GPU's which only measured power draw AFTER inefficiency losses from the VRM whereas Nvidia measured power consumption BEFORE inefficiency losses from the VRM (so essentially all the power required to run your GPU). On top of that the interval at which is measured needs consideration too, because as Steve noted at the start of the video, transient spikes can be missed with longer intervals between measurements which will lead to large inaccuracies when evaluating power usage over a short period of time (which means even Nvidia's "superior measurement method" in my example isn't actually completely accurate).
@@MLWJ1993 I see. I think they should test software apps for testing. See whats accurate or whatever. Im just suprised 7800x3d is better, from my pov they should be same watts but im stupid so ehh.
@@Codyslx Defects are common with all cpu's all electronics have a defect rate. intel's is considerably higher only ARM cpu's have a lower defect rate than AMD. shit happens. not copium just facts
@SirZanZa Didn't puget release a graph showing amd having more cpu failures than the defective 13th and 14th gen cpus using no oem motherboard settings? Regardless, you don't justify it and pretend it's a good thing. The fact they are sending out completely dead cpu's is insane. Intel messed up big time with 13 and 14 but don't sit here and pretend amd cpu's were always more stable. They are only doing better now because intel decided to implode on itself.
And to watch 9800X3D blow them all out of the water in 6 months. Along with the upcoming Zen 5 chipset, updated AMD chipset drivers, and BIOS AGESA fine tuninig it'll be a big jump.
@@JFinnsOr it’ll be the 7800x3D.2 with the same tiny gains of these 9000 series chips. I think it’s kinda silly to make a firm claim either way until we see. Sincerely, AMD fan boy running a 5800x3D + 7900XTX
This is why I love GN. You take in audience feedback, address audience feedback in constructive ways, and then absolutely slaughter people with facts and supporting data. Keep up the great work
If you want gaming it's 7800X3D. If you run heavy number crunching using AVX-512, go Zen 5. If you already have a computer that turns on and does stuff, don't feel the need to upgrade. Your computer must prove itself insufficient for your needs. Or perhaps if you have plenty of money, don't mind spending and like to built new computers, go ahead and buy it. What I'm wondering at this point is what gaming would be like on 8 Intel efficiency cores or on 8 Zen 4c/5c cores.
@@DingleBerryschnappsyou are a very silly boy, and I hope you grow to realise how your myopic view hurts you someday soon before it does much lasting damage
AMD seemed to focus on enterprise workload efficiency, and made huge gains. This was to boost their Epyc sales. Gamers may have to wait for X3D or next gen for big improvements.
GN showing that data can speak by it self. Maintaining objectivity without any clickbait titles/thumbnails. And also keeping the video very informative/interesting and most of all fun to watch.
Appreciate you going back and reevaluating a result in even more depth. Looking forward to more efficiency tests going forward! Being in europe, this absolutely matters, and hopefully there will be other more efficient CPUs coming out again soon.
9:39 - "let's get started", after 10 minutes of pretty much already rebuking those commenters, explaining methods and approach, and why they're full of it. THEN following up with the coup de grace. There's medical studies with less data.
@CoffehKittehGrey Thats what ive been seeing people claim as well. Even though I bring up that AMD is marketing it for gaming and Lisa Su opening the Ryzen 9000 announcement with "excited to show us whats next for PC gaming", they defend AMD by saying that it isnt for gaming. Just a massive facepalm.
the point was that the cheaper chip was faster per watt. if theyre going to brag about efficiency, but get beat by a cheaper version of last gen, they are NOT efficient. i had high hopes for this launch, but it seems they fumbled, at least on these chips. hopefully the next bits are better.
Steve & team, Thank you so much for your continued contributions to the PC building community, especially your deep dives into the actual testing. I am so grateful to watch these videos.
I was one of the dude's pushing back on efficiency in the previous vid. I stand corrected. I'm gonna buy some pc magnets to support the hard work, appreciate it.
As the saying goes, AMD never fails to miss an opportunity Edit: not an AMD hater just that they could've slaughtered the competition at various points but released mid or overpriced stuff ☠️
Could find this video after watching it 3 days ago, it even disappeared from my history. The magically I got a notification 30 minutes ago about the video and it’s been saved now. Thanks Steve. 🤗
As someone who might be called an AMD fanboy because I've only ever built using AMD CPUs: this is why you shouldn't be a fanboy. I'll wait until the 9800X3D reviews come out, but what I'm seeing so far makes me lean towards skipping this gen (or going with the 7800X3D if I absolutely have to upgrade now).
building only systems with amd cpus does not make you what many people in comment sections would classify as "amd fanboy", and that in very simple terms is someone that will defend amd and will claim amd products are the best even when evidences may show that product X or Y may not be the best, that it may actually be worse than product Y or Z from a competitor . AMD fanboys also live in a parallel reality where AMD GPU's somehow have some sort of marketshare that makes the radeon division competitive with Nvidia.
Even I knew to skip the Bulldozer series...but felt sorry for those poor guys who bought into it hook...line...and stinker... I gave them a slight downgrade (A6400) and their computers were running faster than ever...at a $10 upgrade fee... On second thought...a real bulldozer was faster than the entire Bulldozer run....compared to their runs on my old test equipment... With the AM5 series...they're testing a double chip technology that most motherboards have problems reading... It's parking more than going full throttle...
I really like and greatly appreciate the new testing methods as well as the time, effort, and care well spent on getting the results. These results are very insightful and as simple as they can be to understand. These Results are also very useful if not the most useful results for making the best decision(s) on a person-to-person basis. It really shows great commitment to the community and their needs/demands. We all should be super thankful of all our time saved by watching a video that is a relatively tiny amount of time compared to what was put in for the results by more than one person at Gamers Nexus. I hope I speak for a vast majority of the community when I say we love you and can't thank you, Gamers Nexus, enough! Stellar Job! Really above and beyond 15/10 rating
I think the only reason why people are claiming this is because of the "TDP drop" 🙃🙄 Thanks Steve for the testing you and the rest of the team do for us
@@peterpain6625 agreed. in most tests, the 5000 chip is only marginally behind, and still doesnt really present a bottleneck, even with the 4090s, in most workloads. 5800x3d, best bang for the buck gamer probably, 7800x3d best flat out gamer, currently. i dont count 13 and 14 intels because i dont feel i can recommend those to my customers (im a builder) due to all the issues they have OUTSIDE the voltage thingy.
I really appreciate that the reaction to the comments wasn't "I'm the pro here, listen to me", but instead they validated everything in way more detail than was necessary to confirm the claims.
But still there is (always) room for improvement... He mentioned you should always check your measurement equipment, which is absolutely true. However inspecting using a power clamp is not really accurate for DC currents. It is very easy to get a significant offset using DC-capable power clamps as these use a Hall-effect sensor which is also reacting to magnetic fields close to it. Just the fact you need to zero these meters and measure in both orientations and average these values is already an indication it isn't that accurate. If you do this only once and then use the same "calibration" on all tests, then it still is a valid comparison. (though maybe not comparable to measurements made by other YT-channels.... when those will ever occur...) It is way more accurate and precise to use a programmable load to check the measurement equipment. As always, this Dutch expression applies: Measuring is knowing, as long as you know what you're measuring.
@@TD-er This is an extremely misguided comment. We validate with: - A high-end Fluke current clamp - 3 other PMDs that read the same, each of which does not use a magnetic field but rather uses in-line detection - Software readings that are known to be +/- 3.0W - And a wall meter with known power from other components
@@GamersNexus Thanks for the clarification. If you're using several of those to validate, then I was indeed misinformed. I only heard you say you validated it using a clamp meter, which triggered my reply.
The comments were the reason we learned all this and did this deep-dive! They were helpful for this one. We didn't reach the conclusion in many games that I think people wanted, but we might not have dove into this without the comments!
@@GamersNexus For this exactly comment GN made: I'm relieved you had this attitude. I was one of the people that said, not that it was better, but maybe we should consider power usage. I'm tired of trying to cool cpus with 420 AIOs, it's not fun. I have a 13900k and gave up and just lowered the power level way down. i mean amd was suggesting the same perf at lower wattage which isn't true entirely). That's a bit sad. I love this kind of data-- esp how much perf for how much power i need to use. And even more your attitude about this. It's exactly why i love this channel, and have watched it and place a lot of trust in you folks. It's not just the cost. I have the 13900kf. It can be purchased for 450 easy no ram upgrade. cost for energy isn't the issue for me. It would take years to recoup.--- It's hard to cool, and costs a lot to get a cooler, and heats my room and it's hard to keep a system quiet. It's a pain. It is also not great for the environment. I have passion about this. It sucks when my stuff runs hot and i have to worry about all this all the time. Voltage whatever. It just doesn't scale. Also about time. If it helps you to know. If i don't hear for a few days I know you are up to something, plus you've even mentioned it on a community post. I have a 13900kf so won't be doing anything for a couple of generations but usually when articles/videos etc pop up fast and and are superficial you can tell depth put in is impacted; I suspect you are on to something with amd. I wonder if it actually ever works. Anyway you are certainly in a position now you don't need to be first, and are more better off focusing on being best. That's who you are.
This is actually really interesting in a way that "lol, more frames" tends to miss in regular generational review cycles. I'd love to see a similar efficiency per frame focus for gpu generations, too!
Please keep these efficiency test coming, i really appreciate it. This is definitely a decision factor for me, as i believe efficiency also has a thermal output dimension to be considered depending on the application.
Spot on, Steve! Your findings echo my own observations. The Zen 5 9700X, while powerful, doesn't deliver a consistent efficiency gain over its predecessor. Seeing it fall behind in certain gaming scenarios, sometimes by as much as 10-15%, really drives the point home. Zen 4's value proposition remains strong, especially when factoring in the price-to-performance ratio.
I swear to god if I had a penny every time Steve said some passive aggressive shit against Starfield between his benchmarks I would be a millionaire by now. That's the main reason I love Gamers Nexus
@@lucazani2730 The Starfield hate bandwagon is old and tired. A significant percentage of players didn't like or had a mixed reception to the game. We all get it.
Something else I feel like needs to be mentioned: the 7700/7900 non-x come with genuinely good stock coolers (the 7600 sadly doesn't) that can cool the processors under an all-core workload under 80c (via TechSpot's review). The 9700X doesn't have that and is more expensive than the 7700 at MSRP, and is currently more expensive than the 7900. As someone very into solar computers and mini-itx systems, I was really surprised to see why the 9700X was so celebrated by many for its efficiency when I've been building systems with the non-x 7000 parts that seem to run cooler with less wattage (and, bonus, have pretty good coolers that fit in most cases)
Is the 7600 cooler different to 7700? I am upgrading my pc this month to am5 (I want am5 for futureproofing as I’ll just upgrade the cpu in like 3 years) and was thinking of the base 7600 for being the cheapest am5 cpu and having a good stock cooler (meaning I wouldn’t need to buy one)
@@meirl5700 7600 -> Wraith Stealth Cooler 7700/7900 -> Wraith Prism Cooler(good cooler) PS: Future Proofing is not a thing, Zen5 is dud and we don't even know if Zen6 is going to be on AM5. PPS: Might as well wait for Arrow Lake reviews, that's what i'm doing, i'm currently on a 5700X3D.
@@meirl5700 the 7600 comes with Wraith Stealth cooler while the 7700 comes with Wraith Prism. Prism is objectively the better cooler among the two but the Stealth is not bad either. Better than Intel stock coolers even. The 7600 with the Stealth cooler will still get the job done since it doesn't really pull that much power to begin with unlike the X variants.
@@meirl5700 AFAIK all the stock coolers for AMD are built by coolermaster. ive run the Prism (the higher end AMD cooler) on several builds and it's a champ. but the Stealth has easily cooled heavily pushed gaming boxes without throttling problems, with PBO boosts turned on. plus, a cheap hyper 212 cooler is under 50 bucks if you want a big stack cooler, and Scythe also has a bunch of excellent big stack coolers for under that 50 dollar price. those will cool some fairly big chips as well, meaning less of a need for updated cooling later on.
Holy -in-depth testing Batman! Thank you STeve and the ENTIRE G.N. Team. you are all heroes we deserve, AND the heroes we need! This one video has absolutely solidified my decision to upgrade to a 5700X3D. one last hurrah for my AM4 system.
Rude comments aside this was legitimately a fairly confusing launch so the clarification is appreciated. It has become a good reminder that AMD (just like everyone else) plays it fast and loose with naming and therefore the names don't give all the info needed to make any assumptions or conclusions of any kind. 8-core is even more confusing on the AMD side because they use two different numbers, x700 and x800 with and without an X. Comments say they should have left the X out, but remember the 5700X is a 65W CPU. And they did make a plain 5700 but that had less L3 cache. It's definitely not as straightforward as the old days.
About wattage: Reviewers always mention power consumption in relation to power costs but another big factor for some of us is heat. My setup's power consumption/heat generation isn't as high as others (3900X + 3080 although 3 monitors totalling 100W) and the room still gets uncomfortably hot which may be fine when the AC is on but when it gets colder outside it still gets too hot inside. Unfortunately some rooms do not have windows.
Your setup is kinda on the high side of max power draw. But at least you have power limiting options. You can probably just frame cap a game. Also a lot of people don't think about this, but higher end cards are generally more efficient even though they have higher TDP. So a 4090 I bet you, would run at higher performance than other cards, even if you limited it to the sane power draw. Something to think about.
@@libertyprime9307 I do cap, and I also don't think my setup is anywhere as high as the top end (4090+14900K consumes a few hundred watts more). I just have been PC gaming long enough to remember when I used to have a dual GPU system that heated the room less
Power consumption is the same as heat. All consumed power is converted to heat (save for a few picojoules of energy used to move the electrons around).
This makes sense bc if the main improvement is the double branch prediction then in unpredictable workloads it would technically need more computations to get the same solutions as zen 4
That might still be the case but I think the bigger issue is that the new x4 decode clusters can not work in parallel on a single thread. This basically means that if a core is only using one thread the decode throughput is basically the same as Zen 4. It shouldn't be surprising then that lightly-thread apps (ie. Most desktop apps including games) don't benefit much from Zen 5.
Im really glad this test came out because it gives AMD more data points it's engineers to use when submitting changes and strengthens its position against their marketing group.
I would say their engineers need now much more than just that. Or then they have to replace them all. Some may think that the "creative" marketing was a must because of the very poorly engineered Zen 5.
I'm glad the reviews expanded. If you add 10 minutes, I'll watch it anyway; it's not like I watch any video in 1x. I'll bump the speed by 0.1 or 0.2 if it gets too long. I don't think I'm unique among this audience, so longer videos aren't necessarily bad. For people who struggle with speed-up video, get an addon that lets you have more granular control and slowly build up.
@@leeroyjenkins6103 I want to reduce the electricity used by my rig, most of the time, so efficiency is important to me. But I also want it to be able to scream when needed.
I had the exact same thoughts as those comments when watching the 9700x video; I was actually relatively stoked for Zen 5 because of it, so this video is much appreciated
36:34 Considering how many media outlets were praising Zen 5 for efficiency on launch day, you might be onto something here. With AMD's history of late review samples and that instance of even changing price the last minute (Radeon 7600), I think it's about time AMD get the NewEgg/Asus treatment. Even if it's due to incompetence and not for some cynical reason, this poor practice is taking a clear toll on the quality of the reviews and the well-being of the tech personnel: one of my fav channel's launch day review was pretty messy.
This might be the start of a new series. Steve blatantly instigates the comments section with a video and then makes more videos on correcting them. 16D chess, that's why he's Tech Jesus!
Hey Steve and GN! I really appreciate this analysis, I find it very informative and helps me communicate these findings in more plain language without the necessary, much more limited, in house testing when presenting options at work. Something I'm personally sad wasnt present was at least one mixed CCD vcache chip present in the 7000 series, as thats not something I would get the go ahead to test! I can extrapolate parallels in some of the chips tested here, but it would have been nice to be able to directly see if that non-symmetric design would have had any impacts, and maybe give us space to compare to the newer x3d chips when they come out. Thanks for your hard work, GN!
Great work with excellent methodology. I appreciate that so much effort was put in to give as unbiased results as practicable, even if it's not the results we hoped for.
Hmm, the same company that said they have greatly improved IF in zen5 over zen4 is the same company that said zen5 would be actually better than zen4, so ........
Unless AMD is having absolutely trash yields, and simply decided to drop the worst crap on the 9600X and 9700X and never bother to release non-X versions of them, then I can't see how the 9900X and 9950X will fare much better.
Personally I love Zen 5. Because it made my Zen 4 build so much cheaper to make. Plus I already had the DDR4 from previous builds, Samsung Pro PCIE 4.0 1 TB from Black Friday sale etc.
@@Lukiel666 It was was just a coincidence to get Amazon Prime Day in this window, new gen only makes old gen's price drop CONSIDERABLY, if it introduces new performance, which this didn't.
I think it is great that you guys put more emphasize on efficiency testing. For me as a european, or more excatly german, with high energy prices, it is concern for me to not only build a powerful, but power efficient mashine. I already liked your testing on performance and I appreciate taking on the challenge on measuring power efficieny, which is an entire different can of worms. And GN delivers.
Someone in a previous video stated that Steve was unlocking the Louis Rossmann's ability to talk in 2x speed In this video, Steve already mastered that skill
thank you for all the efforts! for such a small team, you are really making the difference with buying choices. next$20 I save due to your statistics will be donated here!
The price of electricity interests me least of all... But how long PC will work from the 'charging station' is important for me. Cuz i live in Ukraine where regular power outages like ~12hrs in a day. I am using CTECHI ST2000 2073Wh - 13900K undervolted to 90W P-4.6Ghz/E-3.2Ghz/Ring-3.2Ghz/VCore 0.91V (i use der8auer video) rendering like 5950X with OC to ~190W (~30K CBR23), and RTX 4090 undervolted to 125W work like 3070Ti with 290W. Seems like 9950X might be better than 7950X in terms of efficiency, but... Swap DDR4 to DDR5, motherboard, CPU... And it's unclear how much I'll win at 90 watts, 10/20/30% 🤔(Working in Exterior Archviz, 3Ds Max + Corona Render, so when rendering CPU draw 90W and GPU ~25W). When the electricity is turned on i switch Intel XTU Profile and get +35% Rendering Speed on ~210W (P-5.3Ghz/E-4.3Ghz/Ring-4.5Ghz) and MSI Afterburner Profile i get +60% GPU power with 380W optimized profile. I have to switch it back and forth like this 3 times a day 😥 P.S. I apologize if I suffocated anyone with a bunch of text ))
Very true! In any field, a business without strong competition will always sit back and focus on profits. As consumers, the best scenario is always a neck and neck race with strong competitors in order to force them to excel.
The cpu performance to wattage ratio is becoming more crucial than ever, especially in countries where electricity costs are skyrocketing, such as in Europe. I'm very satisfied with the 7800x3D and don't see upgrading anytime soon.
@@Ash_G Apple just is too expensive for what you get(here in the EU). Its more popular in the US. Here in the NL people buy it for status or niche sake or they are self employed and can afford(because of tax reduction) iphones or MACs. I know because I see this in my family where a lot of family members are self employed and have iphones🙂The not self employed members have Androids.
@@degroot50 I am not telling or suggesting you buy Apple Macbook Pro. All I said was that its M series CPU represented a landmark in Watt to Performance ratio. Apple realised that 10nm process Intel, with its security and instability flaws, was too much a risk and therefore went its own way to design its own M-chips. Credit where credit is due. That's all I am saying but in more words now.
Cool to see this in-depth analysis. I thought the efficiency was a tradeoff that could make sense after the original review. So more data to disprove that notion was really helpful.
Help support our testing! New colorful solder mats on the store, plus copper plated mule mugs in stock! Use code THISISFINE for 10% off while it is active: store.gamersnexus.net/
Check out the latest hardware news here: th-cam.com/video/ZRJaF2ky9CE/w-d-xo.html
@Mans1884 Just got rid of 2 of 'em!
RTX 4070 GDDR6 (non x) review in coming? consumers might want to know if its slower and if so to make sure they dont buy those cards
first comment from a viewer in the vid is total cuckoldry defined
Again, you are comparing different classes of CPUs. The 7700 (non-X) and the 9700X are not in the same class.
@@robertlawrence9000 Ummmmm actually according to my calculations you have to compare them to understand the data and put together better stats
"UMMMMM did you test with SMT Disabled? That unlocks Zen 5's gaming performance!!!!!"
Thanks, Steve. (And I actually did think about doing that, but I thought surely no one would ask... so, thanks)
2 Steves is here bois
Thanks, Steve.
Now, you both gonna be confused. Thanks Steve
After great review on techpowerup, they have shoot themselves in a foot with a machine gun by doing this one with SMT.
0:54 can i just say, i still to this day think these side bars counting down until the next slide is such an excellent touch. stood out to me when i started watching your videos years ago and i still appreciate it
Thank you! Andrew recently did a very slight overhaul to them to make them even easier for us to use, too!
@@GamersNexus cheers, keep on truckin
I'd love to have the same idea of the sidebars but for each topic of the video as well. Not only on the slides.
@@3333218 You already have TH-cam's built-in chapters feature for that. Every topic is split into chapters that are easily accessible just by tapping on the player, so sidebars would be quite redundant.
@@SierraAAM Not really. This is a visual cue of how long the chapter takes, on screen, while you're watching. And I personally prefer watching stuff in Fullscreen.
> Lots of comments about power efficency
> Steve: "Ok I will briefly address that"
> one week of office sleeping later
> 40ish minutes video
We love you steve, no joke
Never mind FPS/Watt or frames/Joule. 1 Horsepower = 745.7 Joules per second, so you could measure in FPS/hp, which I think we can all agree sounds the most impressive.
Jeremy Clarkson just suddenly woke up without knowing why 😁
hahahaha, this is awesome.
Totally agreed. I have long wondered why more sciences don't adopt equestrian units of measure
This feels like a giraffe asteroid analogy. Anything to avoid metric.
@@zivzulander He's been continuously awake since Raptor Lake launched.
POWAHHHHHH
Steve just made a whole "PER MY LAST EMAIL" video.
I have to deal with that daily at work lol
RE: RE: RE: FWD: FWD: FWD: RE: RE:
@@kxmode REPLY ALL:
With a side of "Reply all:" 😂
I fkn hate those emails. 😂
You saved me so much money. I was going to buy Zen 5, but ended up going with a 7900x3D instead. And before someone says it, no I need more cores than I need gaming performance.
Then don't get the 3D part. That's a terrible SKU
@@Electric_Bill Performs better at gaming than 7900x and only cost about $30 more. Don't see the issue right now. In the past it was terrible, but now it's a good balance. 7950x and 7950x3D are still over $500. 7900x and 7900x3D can be found for $370 and $400 respectively.
sorry ... 7900x is not zen5 ?!?!
@@EbefrenRevo yeah that's Zen 4
@@EbefrenRevoThat would be the 9900x, which I think we will get to see tomorrow, as well as the 9950x
Steve absolutely LOVES taking super random shots at Starfield lol.
all press is good press
16 times the content!
It is unironically the least buggy Bethesda game launch on pc in a very long time lol
Starefield
I constantly have to remind myself wtf that game was, i played it for 2 days and realized it was like somebody took Fallout 4 and took all the good parts of it out and replaced it with generic filler. What kind of bonehead thinks making a galaxy of random generic empty worlds that all functionally are the same is anything people wanna play? The funniest part of that game is that Todd really expected that game to be some ground breaking experience that changes the industry. Holy shit bethesda is dead
Between the intro and the first 7-8 minutes of testing, it was pretty obvious where the conclusion was going to end up.
At about 15 minutes in, I thought "even though the comments from the opening of the video were proven "wrong" in instances, I just learned a ton and got shown a new way of thinking about things. Hopefully this doesn't discourage people from sharing their thoughts in the future."
I'm not surprised at all that Steve (and his team) took a few moments around 33:30 to tell those commenters that they're appreciated.
Even though I'm not surprised, that cool acknowledgement should be reciprocated. "Thanks Steve!"
That's awesome that you learned something! Thank you for sharing! That's always the hope with these -- we learn a lot as well!
@@GamersNexusI wonder how Linus and Tom’s Hardware got good reviews? Did they get golden chips?
@@Joe-wk9ow Linus? Ahahahahahaha.
@@Joe-wk9ow Yah... Oh and Der Bauer. Bunch of dumbasses
@@Joe-wk9ow Tom's Hardware are just a bunch of clowns these days.
7700 - $220.
9700X - $380.
Same performance, same efficiency.
Why pay more?
Also 7700 has a cooler in the box, so CPU by itself is even less
We just love pay more latest numbers lol
Bigger number better product
"B-b-but bigger number?!"
Exactly!
Looks like we're getting the 7700 non-X efficiency comparison. Let's go!
And a lot more! Super fun testing. Very tired, though.
@@GamersNexus Take it from someone who has Insomnia. Nap and Sleep when you can!
half the time all TechTub is doing is acting like a 5 frames a second faster than the other guy, is some massive win that most consumers cant even see. The whole CPU and GPU testing shit from Techtub for the last 10 years is over hype click bait for views. None of it really means anything.
@@GamersNexus Do you think that if the 800 series boards have a higher stable ram speed say in the 7000mhz range that it would help improve performance for the 9000 chips at all?
@trulyskillsince07 There's a limiting factor in the infinity fabric I think, which makes 6000Mhz the point after which no performance is gained. I think maybe it was der8auer I saw point this out.
THIS is the deep dive that no one in the industry ever wanted anyone to see. Not Intel, not even AMD. THIS is the full holistic picture of why the large cache X3D CPUs are so damn good. They're efficient EVERYWHERE. It's not whether they're the power blasting 24 core kind of speed, it's that they're the better all around CPU for just about damn near everything from a perspective of this question: "What is the overall best price/performance/power CPU technology?" The answer is the monster cache CPUs. Can we have more of those across all the CPU vendor's product stacks please?
Heres the issue.
The gamers have always subsidized the average user market. Most people don't need X3D or even a 7700X. I3s and Ryzen 3s are by far the biggest sellers as they are all many people need, but the way those sell is purely "What is cheapest." By charging the gamers more, AND making the low chips out of high end chip offcuts, it makes it easier to make this buisness model work. (Intel obviously does the same over a wider range... by selling i9s for more then can sell I3s and their even lower SKUs for less)
The X3D line creates a problem as it takes additional money to develop X3D chips over the standard chips.Not much extra money now that AMD has done it twice, but some. If all the gamers buy the X3D line, AMD has to get more money from the initial round. And thats going to raise the price of the cheapest chips.
LONG TERM, AMD probably aims to only make X3D chips, as that's clearly going to please everyone. But until they are ready to do that, they still need to release non X3D Ryzen 7 and Ryzen 9 chips to keep the lower skus cheaper.
What they SHOULD do is just jump straight to APUs. Keeping their APU and CPU markets separate and releasing their CPUs first is an older strategy that appeals more to gamers, BUT leads to low level chips that really no one should buy (who really is the market for the 7500F. If I only want to spend $179 on a CPU, I am not shopping for a $300 GPU)
The huge cache is also beneficial for pcvr.
This is also a symptom of the fact that the DRAM speed lag is getting too extreme, especially because in the latest process nodes cache SRAM sizes are no longer shrinking, therefore it´;s harder and harder to shove more cache on chips and the ram keeps sucking
they'd be inefficient if you had two high cache CCDs and workloads that need to communicate between them, not really the case for single CCD gaming chips though.
it would probably have been much better to stack a larger L4 cache on top of the IO die (kinda like RDNA), but it's hard to say what obstacles they encountered.
Gasp giving a CPU more cache makes it perform better. In other news the sky is blue.
Seeing the blender workload show high efficiency makes total sense - one of the few AVX-512-supporting workloads, which Zen5 heavily improved upon (while not really doing much of anything for lower-width AVX2, SSE, and scalar workloads). On a technical level keeping the same-ish efficiency here _is_ kinda impressive - they managed to reduce the losses for non-AVX512 workloads far enough. Of course all that doesn't matter to most gaming workloads (today).
I don't think that has anything to do with AVX512, it's a 10% efficiency improvement over the 7700, which is about what would be expected for most workloads that aren't gaming, for example Cinebench sees similar efficiency improvement. Actual AVX512 workloads would show bigger efficiency improvements, probably around 25% when compared to the 7700, depending on the exact workload.
@@kaitoux8413 I think it’s very rare to find any pure avx512 workloads which are not synthetic. Even avx2 is rarely taken advantage of in favour of compatibility and easier code maintenance. In my experience, anything outside image/video and machine learning will have to be quite creative with their data alignment to use it efficiently.
Keyboard warriors: Steve is wrong
Steve: “Here’s a 40 minute statistical analysis of why I’m not wrong”
At this rate, Steve is going to become a professor and open up Gamers Nexus University to spend entire courses proving to people how wrong they are 😁
[yes, I would enroll]
amd fanboy keyboard warriors, to be exact
@@hangender The AMD copium is funny.
@@zivzulanderGamers Nexus University? A GNU? I like it
Using a method they came up with with tools that are arguably not suited for the task and motherboard that might be in a need of a bios update due to it probably running the "just make it work" version this early on.
Steve just made a 40 minute statistical deep dive to say “well actually” to commenters. And I love it.
Remind me of HUB Steve’s “Why Reviewers Benchmark CPUs @ 1080p” video, just proving all the dumb comments wrong lol. Love it.
All your comments are ai generated
Dude actually spent a week of his life to "well actually" the community
The pettiness is awesome because he basically beats down those comments with actual methodology lol
What people should care about the power consumption was idling power consumption, since PCs are in idle most of the time. Use a bit less under load and 3x~4x more when idling is not an okay, AMD need to fix this.
YES!! I just discovered this a few weeks ago while researching parts for a home server/NAS... nobody mentions this.... I tested it with my 5900X & I couldn't get that thing to idle below 40W; whereas my new 8840HS (monolithic chip) idles at about a watt.... huge difference
One of the main reasons my 13900k has actually been quite efficient over the 1.5 year I've had it. Yea sucker draws a house worth of power during all core render workload, but I'm not running a render farm every day and most of the time the actual tasks I do during a 10h workday end up being below the inefficiency threshold where the e-cores are honestly amazing. It's barely powered on but still runs everything I need. And I still got the top-line performance for max output when I need it. Unironically I use less power now, compared to my previous 9700K which had half the rated tdp and peak power draw even lower.
And yea it's been running full 10h average workdays for 1.5 years without a single crash or performance dip. Think I scored a golden sample considering all the recent meltdown issues.
@@Real_MisterSir idling consumption of mid-high end intel can go as low as 3~8 watt, that's honestly impressive.
Meanwhile I'm struggling to push 3700x below 20~25 watt.
Burning 25 watt for doing nothing is a pain in the ass.
PCs won't be idle most time at all if they're used in software development company...
Ours are compiling all the time. After switching these peak machine from Intel to AMD, we saved 20% on electricity bill.
So AMD for busy PCs, Intel for idles.
@@shootingstar7896 AMD for constant load, Intel for load intervals -but even this is way oversimplified tbh. Not all load is equal, nor constant. For some (like you guys) it is, and for me if I ran a 24/7 render farm, it would be too. But my productivity involves a multitude of tasks, most of which do not spike into the realm of inefficiency levels of load. Power consumption during photoshop vs 3d modelling vs 3d render vs film editing vs film render vs leisure time vs email and conference calls and online meets. A multitask work station often has to serve a vast array of workloads that are rarely equal.
You mean you’re telling me the comments weren’t accurate? How am I to live my life after this glass shattering moment?
Tech communities on YT and reddit tend to skew pretty AMD biased so it's not surprising there was a large dose of copium from people after the review.
@@CNTRI715 Tech communities on YT also hype HDR, which is total scam tech.
Come on, GN also says Intel lies about stuff. We all know they are more truthful than the Pope.
How can we ever trust GN after that!
@@CNTRI715 Its crazy how things have changed. I remembered in 2020 AMD released the 5000 series and those tech communities were stating that intels CPU would still be superior in the future. Now they they jumped ship and are defending AMD.
@@raymundabenojar8613 be around long enough and you'll see it swing back forth enough it won't even be funny anymore
I saved for a 9700X. I bought a 7800x3d the day after the reviews.
So AMD managed to upsell you a few bucks :P
Why? The X3D chips were always going to be better for gaming.
@@ffwast The 7700X is pretty much the same as the 5800X3D on average, so a lot of people assumed the 9700X would be on par with the 7800X3D, but cheaper.
I could have told them that wouldn't be the case right off the bat, as Zen 4 is a full 70% node shrink over Zen 3, but Zen 5 is on the worst half-node shrink we've had in a long time (only 20%).
Well... I'm personally prefer to wait until 9800x3d (it have all chances to be good) but still a good choice.
@@andersjjensen dude got the best CPU for gaming in the entire world. I bet those extra bucks were worth as hell.
I know it's not the main topic of today's video, but here is your periodic reminder that Intel also delivers 65W cpus, in the range of non-K cpus. When doing an efficiency test comparison, it feels way more fair to include this variant rather than the overclocked K series, which will obviously fare badly at an efficiency test, same way the 7700 is way more efficient than the 7700X.
Yes, this video focused on AMD vs. AMD. Intel powerlimited or non-K CPU's are a valueable inclusion in the charts for further efficiency talks.
@@fral1073 Yes it was BUT STILL Intel was bashed of poor efficiency several times. I do not think that is appropriate when only the K versions were included. And comparison had power limited results for AMD, not for Intel. That makes comparing Intel efficiency here very unfair even though this focused on AMD.
@@jarnovilen5259 Well tbf Intel only cared about pushing the power as much as they can to get better scores for benchmarks, since Alder Lake we're seeing both Intel and AMD pushing their CPUs far from their efficiency curve and just now AMD is lowering their TDP to pre Zen 4 levels.
@@jarnovilen5259 Intel's K series were being compared to AMD's X series when being (rightly) bashed for their poor efficiency. That seems fair for a comparison that isn't focused on Intel versus AMD. Intel's chips get better efficiency at 65W than at 100W+ but it's still worse than AMD at the same wattage, so the rankings wouldn't change much.
Once again thanking you for including stellaris in benchmarks, initial reviews had me thinking the Clausewitz engine loved the new architecture, but seeing the simulation watts per hour makes thinks more complex
Absolutely! It's a great test!
Yeah, I just ordered 7800x3d yesterday, now you made me even happier with my choice, I mostly play on Clausewitz engine too, HoI4 and Vicky3
I don't think old Clauswitz engines don't love much about new architecture, but the new one in imperator Rome , CK3 , HOI4 (they get updated in recent years) and Vic3 is.
The best test for the new clauswitz engine , I think it belongs to vic3 as the most cpu intensive game with this new Clauswitz engine due to the fact that the game simulate economy is extremely complicate along with with lot of data that nearly burn PDX company before.
"Sleeping in the office, running tests." This man's committment is untouchable. Thank you so much, Steve and GN.
"Running tests" is my code-word for taking a dump at work.
I'm working at the office doing the same. Only difference is only 1 company pays me. 😭
@@LeonNadwislanski If he's happy doing it, what's the reason to hate?
@@LeonNadwislanski
It has nothing to do with anything. You know some interstate truck drivers almost never sleep on a bed. Are they not respecting themselves?
Although I'm not one of those commenters, it's still good to see this deep dive into the matter. Thanks Steve.
I was one of those who thought that you were making far too little of the quoted 30% efficiency improvement.
I'm disappointed by the results of your extended testing, but very thankful for the fact that you went through the extra effort. This is why this is basically the best hardware review source anywhere. I hope that you do continue efficiency testing going forwards -- even if you don't spend a huge amount of time on it in videos as you did in this one, just having the charts and conclusions is valuable.
105W-65W TDP CPUs don't really consume that much power anyway. It's like 80W at most for the 65W TDP parts, and that's at very high load. The GPU is where most of your PC's power bill goes to, especially with these CPUs. These CPUs, on average throughout the year, consume about as much as an incandescent light bulb. For mid-end AMD CPUs, it's the difference between running a 45W or a 60W light bulb at most, best case scenario. Honestly, it won't save you much at all, even if power costs are very high where you live. Even a 30% reduction in *power consumption* at high load between AMD CPUs at the low to mid end would be near irrelevant to most people. Because it's not always going to run at high load, where a 30% reduction actually saves 20-40 Watts instead of 5-15 at idle and low load - and that's if the 30% reduction even persists at idle and low, which is not at all a given. (Sometimes idle power consumption increases for some reason in an otherwise more efficient CPU, like might actually be the case with Zen 5.) It barely even matters for PSUs. It does, but only a little. It's high power consumption CPUs, like a lot of Intel CPUs, for example, where that decrease definitely matters. Both for power bills and PSU choice.
I dont get it. 7800x3d and 9700x. Dont they have the same tdp? My 7800x3d pulls lile 55 watts in games. Can you not trust hwmonitor etc? Figured it was the same.
I'll be honest, I'm curious how the next generation Intel processors are going to end up efficiency wise. Alder Lake looked fairly decent, anything after that a disaster for efficiency, with the instability problems stacked on top of that. 😬
@@DjuntasHW monitor reports what the sensors measure which could be a problem in comparisons where different products measure different parts of the CPU / GPU / Motherboard / etc.
Like with AMD GPU's which only measured power draw AFTER inefficiency losses from the VRM whereas Nvidia measured power consumption BEFORE inefficiency losses from the VRM (so essentially all the power required to run your GPU).
On top of that the interval at which is measured needs consideration too, because as Steve noted at the start of the video, transient spikes can be missed with longer intervals between measurements which will lead to large inaccuracies when evaluating power usage over a short period of time (which means even Nvidia's "superior measurement method" in my example isn't actually completely accurate).
@@MLWJ1993 I see. I think they should test software apps for testing. See whats accurate or whatever. Im just suprised 7800x3d is better, from my pov they should be same watts but im stupid so ehh.
tbf them giving a reviewer a broken cpu just means they 100% werent checking for good samples to reviewers so atleast we got that
the copium is crazy
@@Mavis847 huh? i have litteraly not said anything about performance im confused what ur on about
@@ModrunOfficial The fact that the cpus can come on completely broken is horrible. We should not celebrate that.
@@Codyslx Defects are common with all cpu's all electronics have a defect rate. intel's is considerably higher only ARM cpu's have a lower defect rate than AMD. shit happens. not copium just facts
@SirZanZa Didn't puget release a graph showing amd having more cpu failures than the defective 13th and 14th gen cpus using no oem motherboard settings? Regardless, you don't justify it and pretend it's a good thing. The fact they are sending out completely dead cpu's is insane.
Intel messed up big time with 13 and 14 but don't sit here and pretend amd cpu's were always more stable. They are only doing better now because intel decided to implode on itself.
the 9600X/9700X launch is a 4D marketing campaign to sell more 7800X3D's?
And to watch 9800X3D blow them all out of the water in 6 months. Along with the upcoming Zen 5 chipset, updated AMD chipset drivers, and BIOS AGESA fine tuninig it'll be a big jump.
@@JFinnsOr it’ll be the 7800x3D.2 with the same tiny gains of these 9000 series chips.
I think it’s kinda silly to make a firm claim either way until we see.
Sincerely,
AMD fan boy running a 5800x3D + 7900XTX
This is why I love GN. You take in audience feedback, address audience feedback in constructive ways, and then absolutely slaughter people with facts and supporting data. Keep up the great work
If you want gaming it's 7800X3D. If you run heavy number crunching using AVX-512, go Zen 5. If you already have a computer that turns on and does stuff, don't feel the need to upgrade. Your computer must prove itself insufficient for your needs. Or perhaps if you have plenty of money, don't mind spending and like to built new computers, go ahead and buy it. What I'm wondering at this point is what gaming would be like on 8 Intel efficiency cores or on 8 Zen 4c/5c cores.
If I want gaming, 14900k.
Anything else, 13700k.
The issues they are having with those chips don't scare me.
@@DingleBerryschnappsyou are a very silly boy, and I hope you grow to realise how your myopic view hurts you someday soon before it does much lasting damage
@@DingleBerryschnapps psychofan or marketing employee?
AMD seemed to focus on enterprise workload efficiency, and made huge gains. This was to boost their Epyc sales. Gamers may have to wait for X3D or next gen for big improvements.
@@DingleBerryschnapps so if i want gaming i have to spend 600€?, why not instead get a 7800X3D for half which still beats it.
GN showing that data can speak by it self. Maintaining objectivity without any clickbait titles/thumbnails. And also keeping the video very informative/interesting and most of all fun to watch.
I love that instead of telling us we were wrong, you proved it to us in a fair and respectful GN flavoured way
That's OK! We were partly wrong on some of our assumptions too, so we all learn here!
True and this is not okay that companies dont get push back on their lies and let them do as they wish@@GamersNexus
@@GamersNexuswhen you say "we all learn here", are you including AMD too?
@@strongmokusWell yeah, but we can't forget sony and intel too, they love learning!
Yes I like this about PC Jesus, unlike some other TH-camrs attacking their audience calling them names…
Appreciate you going back and reevaluating a result in even more depth. Looking forward to more efficiency tests going forward! Being in europe, this absolutely matters, and hopefully there will be other more efficient CPUs coming out again soon.
9:39 - "let's get started", after 10 minutes of pretty much already rebuking those commenters, explaining methods and approach, and why they're full of it. THEN following up with the coup de grace. There's medical studies with less data.
Which makes me sad about those "medical studies".
People getting hung up on that ‘x’ at the end of 9700x, comparing it to the 105W 7700x instead of the 65W 7700. AMD playing some Nvidia naming games.
Not really their fault, AMD should have just left the x off.
We included both. Same efficiency challenges for Zen 5 in gaming.
@CoffehKittehGrey Thats what ive been seeing people claim as well. Even though I bring up that AMD is marketing it for gaming and Lisa Su opening the Ryzen 9000 announcement with "excited to show us whats next for PC gaming", they defend AMD by saying that it isnt for gaming. Just a massive facepalm.
the point was that the cheaper chip was faster per watt. if theyre going to brag about efficiency, but get beat by a cheaper version of last gen, they are NOT efficient. i had high hopes for this launch, but it seems they fumbled, at least on these chips. hopefully the next bits are better.
So wait they are not more efficient?
Steve & team, Thank you so much for your continued contributions to the PC building community, especially your deep dives into the actual testing. I am so grateful to watch these videos.
Commenter: "But you didn't explain why!?!?"
Steve: "hold my beer 🍺"
Amazing anyone would doubt this man 🤣
Hold my keg more like it 😅
Thats why this guy were good
Thanks Steve.
I love longform content and deep dives. Glad you are adding efficiency tests.
I was one of the dude's pushing back on efficiency in the previous vid. I stand corrected. I'm gonna buy some pc magnets to support the hard work, appreciate it.
That opening was at Louis Rossman speed 🤣
I was thinking "Project Farm".
I often listen to TH-cam vids at 1.5-2.0x speed ....no chance in hell with this one 😂
It's sped up. There are "overlaps" in the words, as in around 0:49, where he says "...power over time and while we see...", the word time is cut.
Hello Steve! Back to you Steve!
this will never NOT make me laugh.
Thanks Steve
Idle power matters a lot to me, too.
My 5900X draws 35W idle (HWINFO), which makes for 8h/day * 365days/year * 0,44€/kWh * 0,035kW = 45€/year
As the saying goes, AMD never fails to miss an opportunity
Edit: not an AMD hater just that they could've slaughtered the competition at various points but released mid or overpriced stuff ☠️
hahaha.
while i consider myself an AMD fanboi, they really are pretty good at "shoulda woulda"
The saying doesn't get any better if you repeat it 1000 times.
@@ghomerhust I have an FX 8350 and FX 9590 that agree with you.
**laughing at your comments as you wrote totally pointless**
Well that was a 40 min "I told you so". Well done. And nice work reading and addressing comments.
Could find this video after watching it 3 days ago, it even disappeared from my history. The magically I got a notification 30 minutes ago about the video and it’s been saved now. Thanks Steve. 🤗
I completely agree about red flags for short review times.
This was useful info, even if deep.
As someone who might be called an AMD fanboy because I've only ever built using AMD CPUs: this is why you shouldn't be a fanboy. I'll wait until the 9800X3D reviews come out, but what I'm seeing so far makes me lean towards skipping this gen (or going with the 7800X3D if I absolutely have to upgrade now).
building only systems with amd cpus does not make you what many people in comment sections would classify as "amd fanboy", and that in very simple terms is someone that will defend amd and will claim amd products are the best even when evidences may show that product X or Y may not be the best, that it may actually be worse than product Y or Z from a competitor . AMD fanboys also live in a parallel reality where AMD GPU's somehow have some sort of marketshare that makes the radeon division competitive with Nvidia.
if person is focused on gaming x3d parts are best buy. i have 7800x3d its been great. i came from a 5900x.
TEAM RED FOREVER!! Been with Them since 2009 & as my Girl: Kamala says: I ain't goin' back!!!
Agreed
Even I knew to skip the Bulldozer series...but felt sorry for those poor guys who bought into it hook...line...and stinker... I gave them a slight downgrade (A6400) and their computers were running faster than ever...at a $10 upgrade fee...
On second thought...a real bulldozer was faster than the entire Bulldozer run....compared to their runs on my old test equipment...
With the AM5 series...they're testing a double chip technology that most motherboards have problems reading... It's parking more than going full throttle...
I really like and greatly appreciate the new testing methods as well as the time, effort, and care well spent on getting the results. These results are very insightful and as simple as they can be to understand. These Results are also very useful if not the most useful results for making the best decision(s) on a person-to-person basis. It really shows great commitment to the community and their needs/demands. We all should be super thankful of all our time saved by watching a video that is a relatively tiny amount of time compared to what was put in for the results by more than one person at Gamers Nexus. I hope I speak for a vast majority of the community when I say we love you and can't thank you, Gamers Nexus, enough! Stellar Job! Really above and beyond 15/10 rating
I think the only reason why people are claiming this is because of the "TDP drop" 🙃🙄 Thanks Steve for the testing you and the rest of the team do for us
Looks like im sticking with the 5800x3d until the Zen5 X3Ds are released and reviewed.
It's probably better to upgrade to 7800X3D and then wait for next gen. I doubt 9800X3D will be worth it unless they make some serious changes.
@@jonathancarranza6046 Nah. 5800X3d is a perfectly good cpu for gaming still. Even on the 5700X3d i'd see no reason to upgrade.
@@peterpain6625 agreed. in most tests, the 5000 chip is only marginally behind, and still doesnt really present a bottleneck, even with the 4090s, in most workloads. 5800x3d, best bang for the buck gamer probably, 7800x3d best flat out gamer, currently. i dont count 13 and 14 intels because i dont feel i can recommend those to my customers (im a builder) due to all the issues they have OUTSIDE the voltage thingy.
@@peterpain6625 i have a 5600X. i see no reason to upgrade.
I'm still happy with my 5600x.
I really appreciate that the reaction to the comments wasn't "I'm the pro here, listen to me", but instead they validated everything in way more detail than was necessary to confirm the claims.
But still there is (always) room for improvement...
He mentioned you should always check your measurement equipment, which is absolutely true.
However inspecting using a power clamp is not really accurate for DC currents.
It is very easy to get a significant offset using DC-capable power clamps as these use a Hall-effect sensor which is also reacting to magnetic fields close to it.
Just the fact you need to zero these meters and measure in both orientations and average these values is already an indication it isn't that accurate.
If you do this only once and then use the same "calibration" on all tests, then it still is a valid comparison. (though maybe not comparable to measurements made by other YT-channels.... when those will ever occur...)
It is way more accurate and precise to use a programmable load to check the measurement equipment.
As always, this Dutch expression applies:
Measuring is knowing, as long as you know what you're measuring.
@@TD-er This is an extremely misguided comment. We validate with:
- A high-end Fluke current clamp
- 3 other PMDs that read the same, each of which does not use a magnetic field but rather uses in-line detection
- Software readings that are known to be +/- 3.0W
- And a wall meter with known power from other components
@@GamersNexus Thanks for the clarification. If you're using several of those to validate, then I was indeed misinformed. I only heard you say you validated it using a clamp meter, which triggered my reply.
Steve actually clapping back
The comments were the reason we learned all this and did this deep-dive! They were helpful for this one. We didn't reach the conclusion in many games that I think people wanted, but we might not have dove into this without the comments!
GN is responding to audience feedback like normal adults, not "clapping back" 🙄.
@@junko4166 Steve clapping his cheeks back
@@GamersNexus For this exactly comment GN made: I'm relieved you had this attitude. I was one of the people that said, not that it was better, but maybe we should consider power usage. I'm tired of trying to cool cpus with 420 AIOs, it's not fun. I have a 13900k and gave up and just lowered the power level way down. i mean amd was suggesting the same perf at lower wattage which isn't true entirely). That's a bit sad.
I love this kind of data-- esp how much perf for how much power i need to use. And even more your attitude about this. It's exactly why i love this channel, and have watched it and place a lot of trust in you folks. It's not just the cost. I have the 13900kf. It can be purchased for 450 easy no ram upgrade. cost for energy isn't the issue for me. It would take years to recoup.--- It's hard to cool, and costs a lot to get a cooler, and heats my room and it's hard to keep a system quiet. It's a pain. It is also not great for the environment. I have passion about this. It sucks when my stuff runs hot and i have to worry about all this all the time. Voltage whatever. It just doesn't scale.
Also about time. If it helps you to know. If i don't hear for a few days I know you are up to something, plus you've even mentioned it on a community post. I have a 13900kf so won't be doing anything for a couple of generations but usually when articles/videos etc pop up fast and and are superficial you can tell depth put in is impacted; I suspect you are on to something with amd. I wonder if it actually ever works. Anyway you are certainly in a position now you don't need to be first, and are more better off focusing on being best. That's who you are.
The only thing Steve claps is his bikes suspension after sending it too hard.
Also Steve what are you riding these days besides the AMD bike?
This is actually really interesting in a way that "lol, more frames" tends to miss in regular generational review cycles. I'd love to see a similar efficiency per frame focus for gpu generations, too!
Please keep these efficiency test coming, i really appreciate it. This is definitely a decision factor for me, as i believe efficiency also has a thermal output dimension to be considered depending on the application.
Please for the love of all that is holy can you clean those glass blocks with all the smudgy fingerprints on hahahaha
hahahahaha. Hearting this so people can't unsee it, because now I can't.
Don't start.
Each fingerprint you clean, you find two more
Would be easier to keep applying smudges evenly across the glass until it looks like it's supposed to be textured. Gotta get that patina going.
@@GamersNexus The B-Roll with the processors on mats -- hit the edges with a toothbrush and some rubbing alcohol! Those smudges were distracting!
@@GamersNexus smear jam on it and wait
Jeeze Louise... did Steve get a side job as an advert disclaimer reader? xD
Thanks Steve!
hahaha
Nah, I watched the whole thing at double-speed and it was fine. He's still got a ways to go before he hits radio advert disclaimer speed.
=D There were in fact lots of words said
Spot on, Steve! Your findings echo my own observations. The Zen 5 9700X, while powerful, doesn't deliver a consistent efficiency gain over its predecessor. Seeing it fall behind in certain gaming scenarios, sometimes by as much as 10-15%, really drives the point home. Zen 4's value proposition remains strong, especially when factoring in the price-to-performance ratio.
1:02 "Starfield: a loading screen simulator" 😂😂😂
I swear to god if I had a penny every time Steve said some passive aggressive shit against Starfield between his benchmarks I would be a millionaire by now. That's the main reason I love Gamers Nexus
@@lucazani2730 The Starfield hate bandwagon is old and tired. A significant percentage of players didn't like or had a mixed reception to the game. We all get it.
Something else I feel like needs to be mentioned: the 7700/7900 non-x come with genuinely good stock coolers (the 7600 sadly doesn't) that can cool the processors under an all-core workload under 80c (via TechSpot's review). The 9700X doesn't have that and is more expensive than the 7700 at MSRP, and is currently more expensive than the 7900.
As someone very into solar computers and mini-itx systems, I was really surprised to see why the 9700X was so celebrated by many for its efficiency when I've been building systems with the non-x 7000 parts that seem to run cooler with less wattage (and, bonus, have pretty good coolers that fit in most cases)
Is the 7600 cooler different to 7700? I am upgrading my pc this month to am5 (I want am5 for futureproofing as I’ll just upgrade the cpu in like 3 years) and was thinking of the base 7600 for being the cheapest am5 cpu and having a good stock cooler (meaning I wouldn’t need to buy one)
@@meirl57007600 comes with the wraith stealth, same cooler as the 5600, while the 7700 comes with the wraith prism (same cooler as 3700 i think)
@@meirl5700 7600 -> Wraith Stealth Cooler
7700/7900 -> Wraith Prism Cooler(good cooler)
PS: Future Proofing is not a thing, Zen5 is dud and we don't even know if Zen6 is going to be on AM5.
PPS: Might as well wait for Arrow Lake reviews, that's what i'm doing, i'm currently on a 5700X3D.
@@meirl5700 the 7600 comes with Wraith Stealth cooler while the 7700 comes with Wraith Prism. Prism is objectively the better cooler among the two but the Stealth is not bad either. Better than Intel stock coolers even. The 7600 with the Stealth cooler will still get the job done since it doesn't really pull that much power to begin with unlike the X variants.
@@meirl5700 AFAIK all the stock coolers for AMD are built by coolermaster. ive run the Prism (the higher end AMD cooler) on several builds and it's a champ. but the Stealth has easily cooled heavily pushed gaming boxes without throttling problems, with PBO boosts turned on. plus, a cheap hyper 212 cooler is under 50 bucks if you want a big stack cooler, and Scythe also has a bunch of excellent big stack coolers for under that 50 dollar price. those will cool some fairly big chips as well, meaning less of a need for updated cooling later on.
Holy -in-depth testing Batman! Thank you STeve and the ENTIRE G.N. Team. you are all heroes we deserve, AND the heroes we need!
This one video has absolutely solidified my decision to upgrade to a 5700X3D. one last hurrah for my AM4 system.
Did someone put RedBull in Steve's coffee? I checked 3 times if I wasn't watching the video on 1.5x speed.
hum, maybe that's what sleeping in your office next to a bunch of computers that run benchmarks all night does to you
Well that means more words per minute, he's being more efficient
Rude comments aside this was legitimately a fairly confusing launch so the clarification is appreciated. It has become a good reminder that AMD (just like everyone else) plays it fast and loose with naming and therefore the names don't give all the info needed to make any assumptions or conclusions of any kind. 8-core is even more confusing on the AMD side because they use two different numbers, x700 and x800 with and without an X. Comments say they should have left the X out, but remember the 5700X is a 65W CPU. And they did make a plain 5700 but that had less L3 cache. It's definitely not as straightforward as the old days.
Once again amazing work from the crew. Thanks for your service, sirs!
You are doing a great job guys! Thanks a lot 👍
Yes, very very often checking if it's here already. Thanks Steve, back to you Steve 😀
This is an incredible piece of work, GN team! Thank you for the days of meticulous measurements.
About wattage: Reviewers always mention power consumption in relation to power costs but another big factor for some of us is heat. My setup's power consumption/heat generation isn't as high as others (3900X + 3080 although 3 monitors totalling 100W) and the room still gets uncomfortably hot which may be fine when the AC is on but when it gets colder outside it still gets too hot inside. Unfortunately some rooms do not have windows.
We've done that testing in the past: th-cam.com/video/JJUXYPZWI1w/w-d-xo.html
Your setup is kinda on the high side of max power draw.
But at least you have power limiting options. You can probably just frame cap a game.
Also a lot of people don't think about this, but higher end cards are generally more efficient even though they have higher TDP.
So a 4090 I bet you, would run at higher performance than other cards, even if you limited it to the sane power draw. Something to think about.
@@GamersNexus I know, I'm just talking in general
@@libertyprime9307 I do cap, and I also don't think my setup is anywhere as high as the top end (4090+14900K consumes a few hundred watts more). I just have been PC gaming long enough to remember when I used to have a dual GPU system that heated the room less
Power consumption is the same as heat. All consumed power is converted to heat (save for a few picojoules of energy used to move the electrons around).
This is more entertaining than the Intel debacle because we're nerding out heavily and no one's chips are being fried
Appreciate the deep dive and great data from direct testing.
This makes sense bc if the main improvement is the double branch prediction then in unpredictable workloads it would technically need more computations to get the same solutions as zen 4
It's cool tech, but goddamn are mispredictions gonna hurt
It also lacks mico ops fusion to macro.
But i see issues with using a dual predictor. Like overlaping predicitions
@@kevinerbs2778 not to mention how some of the biggest cpu vulnerabilities have come from branch prediction? AMD was ballsy here for sure
That makes sense! Good explanation. Thanks for posting.
That might still be the case but I think the bigger issue is that the new x4 decode clusters can not work in parallel on a single thread. This basically means that if a core is only using one thread the decode throughput is basically the same as Zen 4. It shouldn't be surprising then that lightly-thread apps (ie. Most desktop apps including games) don't benefit much from Zen 5.
Im really glad this test came out because it gives AMD more data points it's engineers to use when submitting changes and strengthens its position against their marketing group.
Xdddddd
I would say their engineers need now much more than just that. Or then they have to replace them all. Some may think that the "creative" marketing was a must because of the very poorly engineered Zen 5.
Thank you very much for this video. I really care about this data and so far I didn't find anyone else making these tests
Actually had to check my playback speed. Steve talks fast when he's got a mountain of data to deliver.
As disappointing as zen5 is, AMD should have priced these CPUs properly... Same current prices as the 7600, 7700, 7900 and 7950x, I didn't miss any X
I'm glad the reviews expanded. If you add 10 minutes, I'll watch it anyway; it's not like I watch any video in 1x. I'll bump the speed by 0.1 or 0.2 if it gets too long. I don't think I'm unique among this audience, so longer videos aren't necessarily bad. For people who struggle with speed-up video, get an addon that lets you have more granular control and slowly build up.
I don't think people are as interested in efficiency as they are just wanting to justify the new AMD CPUs.
It's as if they're a bunch of corporate marketing bots sometimes.
@@leeroyjenkins6103 I want to reduce the electricity used by my rig, most of the time, so efficiency is important to me.
But I also want it to be able to scream when needed.
Got'em
I had the exact same thoughts as those comments when watching the 9700x video; I was actually relatively stoked for Zen 5 because of it, so this video is much appreciated
36:34 Considering how many media outlets were praising Zen 5 for efficiency on launch day, you might be onto something here. With AMD's history of late review samples and that instance of even changing price the last minute (Radeon 7600), I think it's about time AMD get the NewEgg/Asus treatment. Even if it's due to incompetence and not for some cynical reason, this poor practice is taking a clear toll on the quality of the reviews and the well-being of the tech personnel: one of my fav channel's launch day review was pretty messy.
This might be the start of a new series. Steve blatantly instigates the comments section with a video and then makes more videos on correcting them.
16D chess, that's why he's Tech Jesus!
It's the infinite content loop 😂
Hey Steve and GN! I really appreciate this analysis, I find it very informative and helps me communicate these findings in more plain language without the necessary, much more limited, in house testing when presenting options at work.
Something I'm personally sad wasnt present was at least one mixed CCD vcache chip present in the 7000 series, as thats not something I would get the go ahead to test! I can extrapolate parallels in some of the chips tested here, but it would have been nice to be able to directly see if that non-symmetric design would have had any impacts, and maybe give us space to compare to the newer x3d chips when they come out.
Thanks for your hard work, GN!
Steve: *Facts*
Idiots: *Ugahbugah*
Steve: *Facts: Extended directors cut [HFR]*
[4K]
HFR stands for High Facts Rate
@@HupfderFloh Thats even better xD
Addition:
Aka F-U edition
You mean all of gods PC knowledge is NOT located within these comments…
Great work GN!
Great work with excellent methodology. I appreciate that so much effort was put in to give as unbiased results as practicable, even if it's not the results we hoped for.
great video ! still happy with my p3 and 500mhz turbo button ;D
TRS-80 Color Computer 2 with a 900Hz 6809E FTW
@@marsMayflower I still remember going to Radio Shack and being so excited when my mom bought me the upgrade to double it to 4k RAM
oh no, 9950x will probably have the same fate.. Thought we might get a 7950x with improved efficiency/less temp. Great video, as always!
Hmm, the same company that said they have greatly improved IF in zen5 over zen4 is the same company that said zen5 would be actually better than zen4, so ........
Unless AMD is having absolutely trash yields, and simply decided to drop the worst crap on the 9600X and 9700X and never bother to release non-X versions of them, then I can't see how the 9900X and 9950X will fare much better.
Or else it will be mildly beastly and AMD didn't sink all its engineering expertise into an 8 core mid-range chip? We will see... =D
@@marks9233 hopefully!
This is an incredibly well made video, thanks for all the effort!!
Personally I love Zen 5. Because it made my Zen 4 build so much cheaper to make. Plus I already had the DDR4 from previous builds, Samsung Pro PCIE 4.0 1 TB from Black Friday sale etc.
Zen4 prices were already low before Zen5 was even announced, it didn't change much after the announcement.
@@MaxIronsThird True. But dropped even more Amazon Prime day after Zen 5.
Zen4 uses DDR5...
@@Lukiel666 It was was just a coincidence to get Amazon Prime Day in this window, new gen only makes old gen's price drop CONSIDERABLY, if it introduces new performance, which this didn't.
@@frankk1358 Noted. I was thinking of pcie 4 vs 5. So much to learn. I get confused.
Y’all really thought Steve won’t gone hit y’all with the science huh…. Crazy
I think it is great that you guys put more emphasize on efficiency testing. For me as a european, or more excatly german, with high energy prices, it is concern for me to not only build a powerful, but power efficient mashine. I already liked your testing on performance and I appreciate taking on the challenge on measuring power efficieny, which is an entire different can of worms. And GN delivers.
Someone in a previous video stated that Steve was unlocking the Louis Rossmann's ability to talk in 2x speed
In this video, Steve already mastered that skill
But, but, the 9700x is 2000 more than the 7900x!
thank you for all the efforts! for such a small team, you are really making the difference with buying choices. next$20 I save due to your statistics will be donated here!
Steve is always watching.
Fanboys - "blah, blah, blah"
Steve - "hold my charts"
God, GN is honestly leading the consumer tech youtube game. Every single video is just made with such high quality care and research.
The price of electricity interests me least of all... But how long PC will work from the 'charging station' is important for me. Cuz i live in Ukraine where regular power outages like ~12hrs in a day. I am using CTECHI ST2000 2073Wh - 13900K undervolted to 90W P-4.6Ghz/E-3.2Ghz/Ring-3.2Ghz/VCore 0.91V (i use der8auer video) rendering like 5950X with OC to ~190W (~30K CBR23), and RTX 4090 undervolted to 125W work like 3070Ti with 290W. Seems like 9950X might be better than 7950X in terms of efficiency, but... Swap DDR4 to DDR5, motherboard, CPU... And it's unclear how much I'll win at 90 watts, 10/20/30% 🤔(Working in Exterior Archviz, 3Ds Max + Corona Render, so when rendering CPU draw 90W and GPU ~25W). When the electricity is turned on i switch Intel XTU Profile and get +35% Rendering Speed on ~210W (P-5.3Ghz/E-4.3Ghz/Ring-4.5Ghz) and MSI Afterburner Profile i get +60% GPU power with 380W optimized profile. I have to switch it back and forth like this 3 times a day 😥
P.S. I apologize if I suffocated anyone with a bunch of text ))
🤗❤️🇺🇦
Stay safe
Remember, kids- just because Intel done screwed up good, doesn't mean AMD is your infallible friend.
Very true! In any field, a business without strong competition will always sit back and focus on profits. As consumers, the best scenario is always a neck and neck race with strong competitors in order to force them to excel.
Based comment couldn't be more honest.
Thank you so much for always doing this level of in-depth reviewing! Stellar information as always!!!
some really good data.
How practical would it be to scatter-plot CPUs with power on the X axis and performance on the Y axis?
The cpu performance to wattage ratio is becoming more crucial than ever, especially in countries where electricity costs are skyrocketing, such as in Europe. I'm very satisfied with the 7800x3D and don't see upgrading anytime soon.
You are so right. 🥲 Nobody seems to realise or care why Apple M-series was a true CPU breakthrough in decades of IntelAmdWindows fail.
@@Ash_G Apple just is too expensive for what you get(here in the EU). Its more popular in the US. Here in the NL people buy it for status or niche sake or they are self employed and can afford(because of tax reduction) iphones or MACs. I know because I see this in my family where a lot of family members are self employed and have iphones🙂The not self employed members have Androids.
@@degroot50 I am not telling or suggesting you buy Apple Macbook Pro. All I said was that its M series CPU represented a landmark in Watt to Performance ratio.
Apple realised that 10nm process Intel, with its security and instability flaws, was too much a risk and therefore went its own way to design its own M-chips. Credit where credit is due. That's all I am saying but in more words now.
Cool to see this in-depth analysis. I thought the efficiency was a tradeoff that could make sense after the original review. So more data to disprove that notion was really helpful.
looking forwards to tomorrow's review of 9950x