Intel has some explaining to do...
HTML-код
- Опубликовано: 21 авг 2024
- Start configuring your next custom gaming PC with NZXT BLD at nzxt.co/JayBLD...
JTC Gaming Mats Now Available!
Get your JayzTwoCents Merch Here! - www.jayztwocen...
○○○○○○ Items featured in this video available at Amazon ○○○○○○
► Amazon US - bit.ly/1meybOF
► Amazon UK - amzn.to/Zx813L
► Amazon Canada - amzn.to/1tl6vc6
••• Follow me on your favorite Social Media! •••
Facebook: / jayztwocents
Twitter: / jayztwocents
Instagram: / jayztwocents
SUBSCRIBE! bit.ly/sub2Jayz...
I'm gonna go with "I bought a 5900x a few months back, and while my inner nerd immediately went 'WANTS!' when the test results for 12th gen Intel dropped, I don't _need_ it. Even the 5900x is total and utter overkill for my particular use." I'm glad there's finally some competition though, and I hope Intel and AMD will be trading blows for years to come as that will ultimately benefit us as consumers.
are you me?
I bought the 5900x only because Intel released garbage over and over. The cpu is not efficient at idle, not at all. Under load its very good compared to Intel, but honestly I dont use all core very much and if I do the work is done in few minutes and then its idling again at quite "high" power draw. Hwinfo reports 40-60w for the cpu doing nothing literaly, with 9900k it was under 10w idle no problem.. this is all fault of the io die on second chip that draw quite alot of power constantly even when you dont do anything it add 20w.
This has got to be the most sane and rational thought process I've ever read in a tech comment section.
Facts. Them competing will give us better cpu's :)
@@eth_saver That's just how the boost algorithm works. If you do manual all core it idles at 10W and in normal use uses only 100W max
i wonder how zen 4 will do, if it's supposed to be 5nm, it should have even better tdp, and if performance is the same or comparable then team red becomes team green, lol
Even if they'll only be more efficient and not exceed intel you'll be able to buy better gaming laptops with and cpus than intel. Because of the obvious efficiency advantage and cooling limits of course
2005 all over again huh.
Well this is not how it works. Let's say 2 processors have have same distance between the gates of transistors, since transistor distance can't be measured in 2d not but 3d, doesn't necessarily mean they are of same size
Not to mention, if it supports higher clocked ddr5
@@sirmrmcjack2167 Intel and AMD use the same amount of power in gaming though. So unless you use it for computing it doesent matter.
@2:30 "This is not for comparing idle wattages"
@14:00 "AMD bad for idle!"
You've got a good 60W of fans and lighting on the AMD system (~0.3A for RGB fans and ~1.8A for a D5 pump, let alone the extra controllers and hubs used)
Good catch! As the conclusion came I had forgotten about the 10 fans, LEDs, water cooler in the AMD build, and so had Jay.
And when he mentioned that the Intel CPU used more power under load than according to spec could be due to Intel measuring at CPU level, not outlet. Although I wouldn't put it past Intel to fudge the numbers a little in their marketing material.
I highly doubt that fans and lights can consume that much power. And for previous generations, iirc the results were similar: amd consumes more at idle, intel consumes more with load.
@@NullByte_-mm4dn My reply got deleted somehow
Yes, they use that much power. Simply go google what amperage a 140mm ARGB fan uses (usually 0.30A) a D5 pump, an ARGB fan controller, and do the math on the visible components.
AMD fanboi found
@@heyimadrian.4546 sorry, I didn't mean to scare you with math
jay at 3:00 : "Obviously we re not comparing idle here"
jay at 13:42 "Look at the two systems right now in this particular instance as idle and where intel really does shine...." then compares idle power draw :-(
Yes, compares idle of full system with water cooling, fans, LED's with only motherboard on the box 🙃
I can here to write the same thing...
Generally speaking I'm all about performance per dollar but I have to say, I'm not keen on getting that much power turned into heat in a small office on a hot summer day. I do video editing and other heavy workloads so I'm actually hitting the CPU hard for long periods of time on some days. Thankfully I don't have to worry too much because I don't have the money to upgrade anytime soon :)
For video editing, the first thing I would want upgraded, max out the RAM, then use an SSD.
While a faster CPU is great, RAM is where it's at for video editing.
Time is money though so you’re not working efficiently
@@christophervanzetta While that's a good slogan and often true it is more complicated. With the extra power burn it might be better adding a second machine to do other tasks. It all depends on the work process. In another sense time to finish might not matter that much, but could cost more doing it. If the task was so critical that a few minutes faster could save $$ then yes the extra power usage is worth it.
the way I always look at it, is that you're paying real money to draw 120 Watts of extra power, that you're then paying real money for again to have your air conditioner suck out of your house. Intel's power usage is insane.
This might be the worst "apples to apples" comparison and testing ideology I've seen in a while. A 5900X system using that many watts at idle is not normal at all. Even being water cooled etc. etc. 5900X and 5950X system at idle on average pull around 90 watts from the wall. Never seen any results this high. Just stick to the mods Jay.
Nah.
He said don't focus on the idle usage. 🤦🏻♂️ Sure it uses more wats on idle cause all the loops, rgb etc stuck onto the amd setup
@@OneDollaBill lol he says don't pay attention to the idle usage as he constantly talks about the idle usage the entire video. did you even watch the video sheep?🤦♂️
@@OneDollaBill Did you even read my comment..? I literally said "Even being water cooled etc. etc." Still wouldn't pull that much power on idle. And he spoke about idle temps and compared them to Intel through out the video regardless of him saying don't focus on them. If you don't want me to focus on them and your bad testing then don't make a video consistently talking about them. Please read slowly next time.
I have a 5950X and am happy, I just got it so any upgrades for me are out of the question or need. I game and do DAW work and ryzen is amazing at them, I get intel is faster and has better IPC now but I just don't need more than I have. I spent the last year struggling to build my system like many of us. the next upgrade will be at least 3-4 years down the line. I use the windows power limits to lower my idol state when I am just using the net, I use balanced for everything else. had these and more reasonable priced MB come out before I bought my system I would have got the 12900k but that didn't happen. I am not a fanboy of anyone. I am purely a bang for the buck for my needs kinda guy. but Motherboard price is ALWAYS a factor for me so AMD just beats them most of the time in the features I need. I have all these idiots telling me I don't understand just how bad intel is beating AMD now... and frankly that is pissing me off. I know and understand all too well, but it doesn't change the fact I have ZERO reasons to move sideways for just a few fps in most games that I don't play anyway.
Right, if you wait for better parts you'll grow old and grey. Just get the best you can when you need it.
Hey buddy what does HWMonitor register for you when choosing the Power Saver windows mode? I'm getting no more than 15w usage in all Windows tasks on my 12600k for reference. Thanks
@@drewmorg. Not OP, but my 2600X reads 20-25W right now on balanced power. On performance mode where clockspeed is full all of the time its ~35W. I know its a huge generational difference, not to mention a ton less cores, but if your usage scenario is "Windows tasks", then a older mid-to-low-end CPU should be on your shopping list. I can imagine the AMD Athlon 200GE with only 2C/4T going as low as 10W on idle for example. Ideal for the uncle who only facebooks and plays candy crush.
@@drewmorg. power saver mode will prefer the lowest energy state, which means you're running only e-cores.
@@doltBmB They all downclock heavily. You cannot game in Power Saver mostly, I got 30fps in Tarkov. It was unplayable and stuttery. I can play War Thunder without any issues it's literally flawless and I'm gaming at 15w total WHILE streaming off the GPU hardware encoder LOL
"Ryzen has a higher baseline" after explaining that the fancy sytem has tons of fans and used as a secondary RGB light to the scene 😁
Exactly this. If you want to make any kind of conclusions about baseline power draw, then at least try to make that test as apples-to-apples as possible.
2:20.
10 RGB Fans alone may draw as much as 50W from wall at full speed. Typical 120mm fan draw 4-5W from wall (after accounted for any PSU efficiency and other loss).
anyway, it was said because of E-cores, 3ghz pretty ok for some ytoob watching/office working. just being obvious
Not only this. The base line is comparing all 16 Ryzen P cores against all, which are half, 8 Intel's E cores. AMD's obviously expected to draw more power.
The lower load however is impressive because the same 16 P cores from AMD consume less power and generate less heat than only 8 P cores from Intel. That's humiliating.
Jay at the beginning: We are not comparing idle load because these are two completely different systems.
Jay at the end: starts comparing idle loads..............
Dude couldn’t even to the bare minimum for this video
Yeah it seems strange not to just pull out the 5900x and put it on a test bench as well. How else can you properly compare?
@@THEpicND An idle load CPU test should be done with 1030 GTs (20W TDP) instead of those giant power hogs.
Just glosses over 150w idle as if that’s normal - sigh
Thank you, I was viciously scrolling down to see if anyone else noticed. Now i can finally go to sleep.
6:25 i hate to be that guy, but we should take into account the different run time, then you can measure the kW h used (probably should just use Watt Minutes) what really matters for a efficiency test is the work done per power cost
You're not the only guy. I had the same thought in the comments earlier, but you put it much more succinctly.
Yea my thoughts as well, would think running at max load for a shorter duration to complete a task would be more efficient. Would love to see more tests on this with a workload being the constant and seeing how much power is consumed to complete that workload between the 2 chips.
@@vikevali doing the work in half the time for twice the power is the same efficiency, but if the idle/low usage state uses a lot more efficient then the over usage will be a lot more efficient.
I run a 5600x, 16GB 3600MT ram and a 3080 on a custom loop and I’m now intrigued as to how much the system draws when idling and watching RUclips etc. I won’t be building a new system for a long time but lower power usage when plodding about the internets is absolutely worth knowing.
This would matter if you were using the Intel in something like a render farm, where that sort of thing matters. But for everyday use, you aren't playing a game 30 seconds faster etc.
Alder Lake would need to complete the render in half the time to use the same power as Zen 3, as we already know from other reviewers that Alder Lake still consumes twice as much power as Zen 3 when under load, but Alder Lake is only about 20% faster in this particular case. It's pretty clear that AMD's Zen 3 cores are still more efficient than Intel's P-cores.
I'd stay AMD personally right now just because of the extra cost of the 12th gen motherboards and DDR5. Sure the Intel is more powerful, but it is dramatically more expensive to set up this early in it's life, at least in Canada. But I wouldn't be opposed to switching on the next gen when some costs have come down. Great video Jay, I was wondering about this very thing
12th gen cpu supports DDR4, and DDR4 boards should be the main product line for manufacturers in the near future actually.
For now, my 3700X is sufficient, so I am not upgrading. When I do upgrade, it will probably be to the next processor AMD releases with Zen 3 and 3d v-cache so I don't have to buy a new MB. That upgrade still probably won't take place until a year or more after those processors come out. A few years after that, I'll look at who leads performance at the time.
Ryzen is a bit higher in idle but you also have a lot of fans, LEDs and a pump in the mix for the AMD system. I think the earlier tests were okay but that last idle to idle comparison doesn't make a lot of sense.
Yeah that was a complete joke....
Wait what? Jay starts the vid saying idle wattage wasn't comparable, then ends the video saying the opposite. Tsk tsk Jay. You boo boo'd that one! 😜
My rig has aio, 6 corsair ql rgb fans, 4 sticks rgb ram, etc (icue). Only pulls 50w - 60w from the wall at idle (10700k). The bling doesn't change the wattage as much as you'd think. I do agree with the comments on the possible psu differences though.
Id be interested to see if he took that system and put it on a bench just how much of a difference it would make. I don't see fans and an AIO being 70W to 80W power draw though, but I could be wrong.
And both machines were pretty much at stock. A power conscious user should be able to undervolt both and get both cooler and less hungry under load.
Hey jay!
I have two systems, a 3950x with a 3090 and a 5950x with a 2080 ti, and both idle at around 80/90w. Both are watercooled in both the CPU and GPU.
Are you sure your AMD system doesn’t have something causing higher idle wattage? Possibly even the GPU in non power optimized mode?
Yeah, this seems fishy to me too. Never seen a system using nearly 200w idle.
Not to mention this 5900x score is closer to a 5800x score, mine is on the 13000's if I remember correctly.
interesting....
Swap the GPU's around please, the 3090 deserves the 5950X. 👍
@@MafiaboysWorld agreed not super big but decent bottleneck
"now we got that as comparable as we can" - Jay, while having at least 13 light sources in just one pc ;D
And a water pump. Absolutely the most biased and ridiculous test ever. He is either willfully ignorant or heavily biased and i don't know which is worse.
@@robertfullard5646 He literally addresses that first thing.
@@MyNameIsBucket Yes but then he concludes the video saying that idle draw on AMD is higher.... But he hasn't factored out the barebones intel system vs a full system with pumps and fans and who knows what else for the AMD one. His conclusion is wrong and that is either by design or ignorance and in either case he shows he cannot be trusted.
@@robertfullard5646 but intel was a thirsty boy and "lost" despite that....
To clarify, I don't want to accuse them of being biased or anything, I just think with the resources they have it would have been easy to set up two more comparable test systems and was kinda dissappointed they didnt
Intel says their CPU uses 240 watt under load, you measured SYSTEM increase of 274 watt .... that's 240 for the CPU and 34 watt for the rest of the system, like the RAM, the chipset, the data busses, etc. etc. ... I think that sounds reasonable.
also not to forget VRM losses
it looks like air coolers out out of the question for cooling these new intel chips. that is going to be a problem for a lot of ppl.
but the total power was 370W and not 274W...
@@Benman2785 lol he said increase not total power
@@666Necropsy only the top 12900K will be a real problem, plus NH-D15 or similar in capabilities air cooler will be able to cool this CPU. another option is to tinker in the bios settings and lower the clocks/voltages and as a result have better temperatures.
That CPU would add so much more heat to my room I might end up losing performance because of it lol
In future PC gaming should be seasonal activity: regular season in the winter and relaxed summer league in other time.
Long display and usb cables to put the PC just out of the room ;)
@T. Dougs if your running well over 60fps then just cap the fps to 60 and you will lower heat output alot because neither the cpu or gpu will be working as hard. unless you feel you need more than 60fps.
Last winter, I'm southern hemisphere - so may to august - I used the mining rig to heat the living room; just put it in a server box that can be rolled around and used the exhaust into a flexipipe so that it can sit in the hallway and not add noise. Except for the occasional wood fire, on the few sub 0 days, it kept most of the house pleasant. First time I made money heating the house in winter and ended up with a bunch of 3070s for free; 2021 blimin' weird year. And, no, we didn't have the same insane GPU shortages that NA had - you didn't always get to pick the model, the cheap msrp base models were very scarce, but the more upmarket ones were around.
@@yoltsbp You should rephrase that to "cap your fps at your monitors refresh rate".
Funny, the price difference between a 5950x and a 12900K is just €39 at my favorite retailer, not even close to $200, the 5900x is €160 cheaper than the 12900K.
The CPU just launched. Zen 3 only reached MSRP half a year after release.
Personally Id rather the performance with lower load temps from AMD, but that lower temp at idle is something I look forward to in future iterations of P and E cores. Very exciting times ahead!
Looking forward to the new Ryzen with 3D cashe as well if AMD can get a 10-15% improvement in game they should be back on top.. But I got enough performance from the Ryzen 7 5800x and feel it should serve me as well as my pervious Sandy bridge system from Intel 2700k. The problem with the higher power draw may show in the next 3-5 years as systems fail as you are stressing them significantly at the power drawn and heat that you put in to them not just into the cpu but all components.
---
I second this:
Especially when using all computer-electronics through an (1K watts) UPS
( including the display-monitor and internet-router and sometimes also video-game-consoles with their own TV ),
it's worse to run out of the watt-capacity under the load-spikes especially while in the middle of video-streaming/live-broadcasting.
So comparatively:
Sure, the difference currently is maybe 50/kWh in the electricity-bill if you idle-run your computer with zero-tasks non-spot for that 1K-hours/41,666...-days (0.6-days being 864-minutes).
But realistically speaking also that same watt-amount really isn't that much in the grand-scheme of things;
like for comparison a "Matsui MML50B17E"-40-litre-mini-bar/table-refrigerator needs an input power of 60-watts
( and they say on the tech-info-sticker that daily usage is 0.290kWh/24-hours ).
So energy consciousness and such wise, this idle-power-consumption difference is very negligible at least to "normal-citizen" like me
( though, I personally don't leave any other electronics running but the fully-loaded refrigerator when I leave the house ).
---
@@user-zh9kc7tw4n Wow mate, we made similar jumps... I built my 5800x coming from a 2500k that I hold for 11 years.
@@PaveMentman I have the same setup with an UPS as well.
The Intel is from an open test bench while the Ryzen System is on a full PC build, Jay did mention this at the start of the test. technically AMD had a disadvantage in terms of idle wattage because the full system was using "10 more RGB Fans" -jay
but definitely you can't go wrong with picking neither of the two, they both give awesome experience.
Sooo, Intel will help heat my room that much more in the winter? Sounds great!
My 5900x build uses around 80-94w at idle it's probably all that RGB also if you using icue that also makes the CPU spike randomly and effect's bench scores
Same. While watching this video and having other things plugged into my battery backup, it was showing 125w.
How?RGB light leds use around 0.06 watts each.
@@BrooklynBalla It is probably more to do with all them fans. The sheer amount of LEDs would add up though, and each ring will have multiple LEDs.
@@BrooklynBalla theres an io pump running as well as atleast 7 fans youd be surprised what all that can add at the wall
@@BrooklynBalla fans use a few watts each, along with the pump that can easily eat up 20-30w. the rgb leds also can use quite a bit of power about another 1-2w total. adding fans, and water cooling increases the cooling performance but increased idle power consumption. if you want the most efficient system possible you do an NHD15 or similar tower cooler rather then watercooling.
I'm a wallet fanboy, I'll stay with AM4 for at least 5 years more. Maybe upgrade the 3700X in case a 5800X get's cheaper while still in stock, so far the price haven't moved and is not low enough to valid the small difference in performance.
I bought my 3700x at release in 2019. Skipped out on gpu since the GeForce 20 series looked like crap. Now almost 2,5 year late I'm still more than happy with my cpu and I'm still waiting on a new gpu. My 1070 is starting to get old but there is no way I'm paying €1200 for a 3070.
5800x is 299 at microcenter rn. Bought one today
Agreed, the 3700x has never yet given me a "my comp is too slow" moment, not one time.
That's exactly why I still use my I7 9700k, 5.4gghz and it runs cool off of a 120 AIO cooler.. for the amount of time I can use my pc it's still more than enough.
3700x to 5800x is a waste... wait for Zen4 atleast
You went from comparing the deltas in each system to comparing registered watt values between each system. If you're going to look at idle power draw, you need to make sure the configurations are very similar. The fully built AMD system has a number of components that will draw power at idle relative to the test bench Intel config.
It's actually pretty interesting. Im still sitting on a 6700k and have been wanting to upgrade for a while, this 12 series looks pretty appealing as I'd say 80% of the time my pc is used it is not under significant load. And if the load on the cpu is lower because of the ecores, then it should be really easy to keep temps low under that use, using even less power with lower fan rpm. The 12600k will be even more efficient than this regarding power and price and from what I have seen, gaming performace isnt hugely different on the 12900k comparably. Now my next debate is whether it's worth it to fork out the extra money for a DDR5 mobo and new ram or keep the old ram.
A benchmark of the cpus with both ddr4 and ddr5 ram would be really useful
the DDR4 and DDR5 benchmarks is something I am waiting for as well!!!
I went from from an i5 6600k (loved it) to an R5 3600 to fill the new AMD build CPU socket then onto a 5600x when they came in stock and good god the difference between the 6600k and 5600x is scary. Even the R5 3600 was an amazing CPU.
@@gr0mit948 yea I've been wanting to upgrade this CPU for like 2 years now, finally getting around to it, ordered a 12600k, gonna order a new mobo, case and a compatible cooler (that isn't my current $30 cooler master one lol) gonna be excited to see the performance gains when I get it all into a new build.
Apples to apples for Jay means comparing a closed case system full of RGB containing different PSU and a different monitor to an open air system. I thought you were not testing airflow Jay, what changed?
He only had the computers hooked up to the power meters not the monitors. But the rest is correct
different psu is not gonna change how much power a system uses in the slightest.
@@JCrook1028 it won't? Are you trollin or just clueless?
@@fgsaramago In what he was testing it won't. In the slightest. He is measuring how much power the system uses. He is measuring OUTPUT of the psu. The power supply will only put out to the system the amount of power that the system asks for! The system PULLS power, the psu does NOT PUSH power. Now if you wanted to argue that a crappy psu will draw more power from the wall because of lower efficiency you would be correct. However that is not the topic of discussion. A system will ONLY draw what power it will use, it does not care what provides said power. It will not draw more power that it uses. Fact.
So while the test in this video was really interesting, the end really bothered me. At the start of the video you said you could not compare the idle power draws because of the differences in the builds. Then at the end you say Intel uses lower idle power and should consider that if you leave your system on 24/7. But you ignored the differences in the two system builds... The conclusion was unsupported from the test. It would be interesting to see if the difference in idle power if the two systems used more similar builds. The differences in fans alone could account for 15 - 30 watts.
yup, I'd say ryzen with a test bench setup on idle would probably drawing just over 100W, at least not 150W
Agreed
Yeah, came here to say that. Wattage for 8 fans, rgb, and two water pumps could account for the difference at idle.
@@sur6e nah there is no way a fan will use a full watt. Pumps maybe 2 at most each. Rgb less than 3 total for all. So maybe 8 to10 total more for all that stuff
Its called mine conditioning, that is why I don't trust this RUclipsr anymore, but it keeps his light on so can't blame him, unfortunately new to pc will be easily fooled by this stunt.
10:55 you're measuring at the wall, remember there are losses in the voltage regulation process. Even if we assume a stellar power supply with high efficiency, the VRMs are not perfectly efficient either (that's why they get hot). You're still looking at an 89% 240/120VAC to Vcore (DC) total conversion efficiency (minus the base load power of the CPU - caches, memory controller, etc.).
This is important when comparing Intel's stated 241W to Jay's measured 274W. The numbers are right in line when you take into account the efficiency of everything.
11:33 You are measuring the power consumption of the PSU, which has not a 100 % energy efficience. The PSU draws more power as the running PC systems consumes due to this reason.
Jay man come on first you ignore or forget the power draw in your initial video, then in your follow up about power draw you say to ignore the idles yet compare them at the end. Yet fail to realise your amd rig literally has a water pump, more fans, RGB everywhere and basically no effort in trying to make it as close of a comparison in keeping some components the same.
I love your content but these last two videos seem very half arsed.
Don't get me wrong I'm pumped intel is back to competing so it's not me being a fan of amd I just think if your gonna try make a point you actually need put work in and do it properly.
Someone better check on Steve from GN. He's going to have an aneurysm when he watches this video.
It's certainly true that the systems were not equal, but the AMD system, even with the handicap, still smacked the Intel system in power draw. Because the AMD system had all that extra stuff hooked up to it, it shows that the power draw difference is potentially even greater than what we saw.
You are literally look for any reason here. You realize the amount of power draw an AIO pump vs a standalone is negligible, not to mention the 6 fans add like 12W MAYBE.
@@titan1983x My low-noise, low-RPM 140mm Noctuas draw a watt each at 1200RPM. I don't know what those fans are, but I'm guessing they're not low-RPM or low noise.
Then you have to add the pump, which could be more than 20+ watts by itself when running at speed. Again, I don't know what that pump is, but they do draw more than you seem to think they do.
Add in all the other differences between those two setups, and the only useful metric we got from this entire test was the difference between idle and full-beans usage. Intel have finally managed to beat a year-old AMD chip, but to do so it seems like the chip is chewing through almost as much power as a Threadripper going nuts on a stress test - and that's two years old at this point.
This test needs re-doing properly, with both architectures on a minimal test bench, and the current measured on the CPU rails.
@@titan1983x if he didn't compare the idles like he said at the beginning then it would have been fine since the only value that mattered really was the power draw delta during load. But he then compared the idles as a point in favour for the intel. Yes even after all the extra crap is removed intel could be lower power draw at idle, the issue is it's an unfair representation against the amd which just misinforms normie viewers.
If Intel uses twice the watts as AMD to run a task, but finishes in half the time, the watt-hours are the same in each system, so the power cost for the task is the same.
Good! It was the only review I saw until now that demonstrates clearly how P and E cores work, and there that they were actually a pratical choice.
Practical for what? I understand the methodology here, but even though Jay explained why, even he seemed to get tripped up on the results at the end. My 5900X system pulls less than 100W idle. I don't have any RGB, I run my fans at low RPMs, etc. In other words, without all the additional power draw in the Ryzen system that Jay himself noted, the idle power draw of the two systems would have likely been very similar. In other words, running just E cores gets Intel just as efficient as Ryzen, with all performance cores. As soon as the P cores kick in, Ryzen blows Intel out of the water in efficiency.
@@chrisdpratt Yes, under load ryzen is undisputed best, but idling, lets say it could do better, and that's the reason P+E seems to make sense. And it does. Also, this is their first iteration. Anyway, competition looks good.
@@ren7a8ero Well again, I think it is doing just as well as the E cores in efficiency already. Intel was very behind there, so it's good that they've made some ground. From what I've seen, though, it's not *more* efficient than Ryzen now, just as efficient in the best case scenario.
For Intel system can you decrease peak power (PL2 power limit) from Bios such that for both system delta power is same.
Now at iso power see which system provides highest absolute performance in Cinebench?
Please understand that Power Vs Performance is not a straight line. Last 10% perform will consume more than 20% of the power budget.
In other words decreasing power by 20% will result in less than 10% perf impact.
It's fair to compare performance per watt only when then absolute power supplied to Soc is iso.
I think it always depends on the workload, in some weeks my computer is rendering for days, so I think that a Ryzen makes more sense, if you only occasionally need more power, the new Intel is probably the better choice. I'm curious what the new Ryzen generation will bring :)
Xd
@@_Clivey It takes twice as much power, but its not twice as fast. Hence the Ryzen would still be more effecient. You could even fit 2 5900x or 5950x into the power of the 12900k, that kinda shows. Still fast, not that efficient if you are using allcore loads.
@@_Clivey what 5800x draws 270w?
@@_Clivey As jcpOC pointed out, Alder Lake isn't finishing the render in half the time and the Ryzen system has 10 fans, a pump, and water blocks. Alder Lake indeed consumes nearly twice the power when the P-cores are loaded vs the Ryzen chip. Which is exactly why Intel went with the hybrid big core/little core design for Alder Lake, they needed E-cores to help reduce power consumption because the big P-cores are still more power hungry and inefficient compared to Zen 3 cores.
@@jcpoc7032 -- His test is VERY flawed and misleading. The reality is that at idle the cpus are also drawing power and on the Intel system the cpu was drawing way lower power at idle than the AMD system due to the E-cores. So for the 5900X system he is overestimating the power draw of the other components and by later subtracting a bigger number from the load power for the 5900X he is making it look much more efficient than it is. The fact of the matter is that the difference of the two systems in terms of power drawn by the other components (excluding the cpu) at idle is around 15W instead of 45W. Also the cpu cooler on the Intel platform under load draws more power too due to having to remove more heat, so overall there is a 10W difference between extra power which if you then consider the fact that on AMD you only had half the RAM ICs we pretty much have parity in background components' power draw under load.
So from his numbers it really is 371W Intel Vs 265W AMD i.e. 40% more total power for 35% more performance in favour of the Intel chip. Then if you are to remove the effect of the other components completely and focus only on the power drawn from the wall under load only by the cpu then you have about 285W (Intel) Vs 180W (AMD) i.e. 54% more power from the wall for 35% more performance. Considering the PSU and VRM efficiency losses you have about 230W (Intel) versus 145W(AMD) for the power drawn by the cpus themselves. Thus 58% more power for 30-35% more performance. We have to add here that he is also using an Intel motherboard that blasts the core voltage way too high for what is needed. Other tests with Cinebench R23 show the Intel system drawing 270W (and that's the total system power from the wall) with a 12900K, a Z690 ASUS Maximus Hero and a 3080 FTW3 and a 360mm cooler. So he is drawing 100W more than others when running Cinebench R23. Sure you have a 3090 but that doesn’t draw 100W more than the 3080 at load let alone at IDLE. If he repeats the test with a manually tuned voltage (leaving the rest of the settings the same) you will see that the pure cpu power draw is about 210W Intel Vs 140W AMD - that is 50% more power for 35% more performance.
Another misleading thing that comes from a conclusion of the type “well Intel is using more power thus why it wins”, is that it infers that if the 5900X were also to pull 50% more power it would also achieve the same performance. Something that is clearly not the case. If the 5900X were to draw as much power as the 12900K it would only improve its performance by 5-10%. This is a testament of the power-performance scalability of the Intel 7 node and the 12th gen architecture and that is performance can scale well with power even in these higher power ranges.
Would've been interesting to see the actual thermals of the two CPUs. I've looked at some of the other reviews where the 12900K was running at 100degrees under load whereas the 5900X was in the low 80's in the same test under the same conditions and set ups as far as was possible. Only the motherboard and CPU were different.
Under air cooling, even high end air cooler, 12900K runs hot. It can be managed with an expensive 360mm AIO.... to 80C. Air or water, Ryzen runs 20-30C cooler under full load.
The thing is, how often you run under full load. On average especially if you leave your pc on all the time, intel runs cooler. The lower power consumption on idle and browsing adds up by the hours too.
Infact, it is actually the 5ghz wall as we call it. If you cap your cores to 4.7 to 4.9ghz, you will never see it exceeding 120w on intel. Likewise when you put the ryzen on 5ghz, it will draw 300w as well. And the perf is only 5-10% so just go ahead and set the PL to 125w and even a 120mm aio will be sufficient for a 12900k.... but the irony is you dont spend that amount of money on that cpu for 125w.
Yeah Ryzen is overall cooler but at the same time, Ryzen tends to go up to 90 and sometimes 95 when it is Boosting on heavy loads too. Would love to see a thermal comparison.
depends on the cpu cooler Intel recommends the Noctua DH-15 !!!! also current cpu coolers are not specifically designed for this generation of cpu YET so the shape of the cpu and such the coldplate may not be perfect for . i would choose Intel 12th gen any day over AMD 5000 series !!! only people who would worry about wattage and this stuff are crypto moners
@@xerideaat tje 12900k price point if you cant afford an aio then there is something wrong.
Idle Power of the Ryzen system is way to high. Due to techpowerups review of the 12900K the idle power of 5900X is 51W and the 12900K even higher 56W (despite of efficency cores?).
Jay, please make a new video about this topic with compareable systems.
Hmm very interesting. Yeah it makes sense that Jay's AMD system was actually not IDLE and was doing something in the background which ruined the test. And the two systems were not identical at all, very different components.
Theres a good 60W+ of accesories on the AMD system, at the start he says its for comparing load only, then he compares idle...
*Okay so I just looked at TechPowerUp's accurate measurements of IDLE wattages:*
- Ryzen 5600X: 50 watts
- Ryzen 5800X: 49 watts
- Ryzen 5900X: 51 watts
- Ryzen 5950X: 54 watts
- Intel i9-12900K: 56 watts
So those are the true numbers and Jay is 100% wrong in his entire video.
Furthermore, the peak wattages during a maxed out CPU stress test are:
- Ryzen 5600X: 134 watts
- Ryzen 5800X: 158 watts
- Ryzen 5900X: 181 watts
- Ryzen 5950X: 194 watts
- Intel i9-12900K: 350 watts stock, 407 watts overclocked
In a normal multithreaded workload, the 12900K was at 300-330 watts, so it's not just a pig during stress tests. And the Ryzen CPUs were all at 180 watts or less.
So the Intel is an electric pig. :P
Alder Lake is actually quite efficient, it's just that Intel's turbo boost pushes WAY past any sane point in the efficiency curve on the i9 to achieve maximum performance. The numbers are greatly improved on lower CPU models, and even single core load tests on the i9 show very good efficiency.
Yeah, they basically sell factory overclocked CPUs now
Interesting. Thanks for this insight.
Truth to be told, I have Ryzen 5 3600 and for most of my use case, I rarely hit 100% CPU usage, even when I'm gaming. So, I guess I'll be quite content with my CPU for a long time. And since I have an X570 motherboard already, I think I'll still stick to 5950X since I don't need to change my motherboard, which, well, will definitely save quite a lot as I don't have to change the motherboard to accommodate Intel's new CPU (as well as new DDR5 RAM).
Don’t be fooled by cpu usage as a percentage, it’s almost meaningless. If you run a program that only uses 2 threads out of your 12 then the % usage will never be high even when the cpu is maxed out on that workload. If those 2 threads are at 100% and the other 10 are at 1% then your total cpu usage will only show around 18%.
Each thread is only 8.33% max of your CPU’s capacity.
@@oxfordsparky True. Counter intuitively you should be watching your GPU usage, if it drops below 90% and framerate is low then it means the CPU is bottlenecking the GPU and isnt capable enough
@@rohanjamadagni at 1080p unless you're playing light games you're pretty much good to go with the 3600, if you're using a 6600 XT or 3060Ti, well things can change and you should at least tune the memory and OC the CPU to close a little bit more the gap
I have a 1800x and 5800x I game mainly on my 1800x to this day.
I went from a 1700 to a 5800x after my 1700 wasn't able to play the battlefield beta comfortably, and let me tell you ryzen 5000 is fast af, definatly worth it since you don't need a new mobo and all that jazz.
I think I’m gonna wait for the new Ryzens before I upgrade the CPU. The fact that we are comparing a year old Ryzen to the brand new intel is crazy. They aren’t even that different.
Dude i just got my new intel 11800h laptop (i wanted a 5800h but oh well) and its only been a couple months with it
Essentially my standpoint. The performance boost is nice, but the fact that Intel is trading against the previous gen, and Zen 3 is still a justifiable choice, is... concerning.
I'm looking at a system rebuild, and I'm waiting to see what Raphael/Zen 4 has for us. It's hinted it's in the next few months.
@@dudewhersmacar7062 Zen 3D is in the next couple months, but it will still be DDR4 and AM4. Zen 4 is expected near the end of 2022.
Those E-cores have me excited for a revival of the Atom name or some similar brand. At one point there was a 16-core atom that pulled like 60W. It was meant for servers. Imagine a 16-core E-core chip. It would do what, like, 40W and would still have 16 threads.
Oh man I remember Atom netbooks.
Shitty little things but they were good for what they were. I miss my Samsung N150 Plus.
@@mandeadd these sorts of things are what I'm talking about. Imagine what a windows phone could do if it didn't have to use the ongoing dumpster fire that is windows on arm. I'm not saying I would use it, but hey, it would be better that the surface duo 2.
@@DigitalJedi cool as that sounds I dunno if ARM (or RISC ISAs in general) can be beat in the mobile space.
If Intel wants to compete with an Atom-like mobile processor it'll have to be significantly cheaper than M1 Macs to justify the purchase.
@@mandeadd Yeah I don't think that would ever happen either, it was basically just a random idea for a new atom now that netbooks are pretty much dead.
You should add around 36W ( varies from manufacturer to manufacturer ) to the Intel idle consumption for the 10 RGB case fans ( assuming there's also a rear exhaust fan covered by the monitor ) in the Ryzen system.
Yep, that high Ryzen idle power consumption is definitely because of cpu, how one could even think that lots of bright LEDs and active water pump would consume power?
He noted that at the start, but then seems to forget about it at the end when he was talking about idle.
LED's consume very little power, and even the difference between an AIO and aircooler is just going to be a few watts.
11:22 You're seeing a higher number because you're reading the power from the wall... No power conversion is 100% efficient, to get 240w you need to pull more than 240w from the wall.
Is that how that works?
i prefer amd over intel i dont give 2 shits about frame rates or any of that nonsense i just care about using less power
Yeaj i heard that too. That is why the psu has ratting like gold or silver like that. That rate show how eficient they can be, the power ftom the wall will be almost the same as power in the system,
@@billyhatcher643 Intel > AMD
It’s roughly 10-15% loss depending on the PSU quality.That still wouldn’t explain his extremely high watt usage at idle.
Now when comparing power draw you should check how much power RGB needs - especially those bright white means all RGB are on and might noticably affect power usage. My guess in this case is around 10W.
Running a 5950X I guess I won't change anytime soon.
Amount of watts drawn at load will be off too. Just remove the variable altogether and put them both on a test bench. Don't be lazy, Jay.
Or better measure the amperage at the 8pin cpu header with a clamp meter and that will be accurate more than enough.
Exactly, I found this really useless without them being equal. Pretty lazy. I love this channel, this was just really not done fairly...{I KNOW LIFE AINT FAIR and he cant afford the wiring that's in the shot, thus obviously can also not afford two equal test benches....}
@@WhatsupwSully ehh... isn't really "useless" or "lazy" as you put it so correcting you for being wrong there. Pretty accurate comparision
Were the two PSUs comparable? Ideally they should be at least the same wattage and rating (gold, platinum, etc). That will make a little difference in this same test I believe, because power consumption doesn't escalate the same for different wattage and different rated PSUs.
I enjoyed watching this comparison, but I'd like to see a future video with "as equal as it can be" open air setups. Same AIOs, PSUs, GPUs, but obviously not platforms (mobo ram and cpu).
to make a proper test they'd literally have to use the exact same psu
I've done more thorough test setups at home for writing amazon reviews. This video is a joke.
👍👍👍
I agree, this isn't as "apples to apples as one can get"
@@AdrianOkay Actually, for proper test they shouldn't use identical PSU's, because PSU efficiency is not the same on all power draws. And with 100+ watts difference in consumption that spot will vary, not by much but couple % still makes a difference.
Ideally they'd need same PSU but with different wattage.
So many flaws with this -- Not even using the same PSUs would give actual results. The wires feeding the board need to be monitored. PSUs have efficiency CURVES which means at different loads their efficiency changes. At each point along a PSU's curve, it has a different efficiency. The PSU needs removed from the equation entirely.
I wouldn't be surprised if intel is a little better at idle than AMD with the "e-cores", but this was a HORRIBLE demonstration of that with the AMD in a full build, custom loop water cooled, hella RGB fans running enclosed case vs. an open test bench with an AIO and nothing else. Especially if you want to go down the path of what "normal use" would draw, because almost no one normally uses a test bench setup.
He is extremely biased
not to mention he removed 2 sticks of ram. thats the intel fanboy in him that wants intel to win so bad
i love that cpu's are so fast now that all we have left to argue about is "how much electricity they use". something tells me people that can afford to build around these cpu's probly could care less about how much electricity they use. i've always cared about idle draw because i have several systems up 24/7 and it adds up across all of them.
over the past few years we have made some great leaps in performance. most cpu's will do the job. we need the gpu performance to keep up.
Its not the price or electricity usage im worried about.... its the new heater i will be adding to my computer room lol!
@@chrispeters8555 agree. in my bedroom i can definitely tell. upgrading to a more power hungry setup heats my room up pretty bad in summer.
If u live were i do (germany) than you will care about how much electricity it uses :D 1kwh is about 0.37usd (32 euro cent) here. Compared to the average 0.15usd that u pay in the us, it's more than 2 times the cost -.- It's like fuelprice differences :D You pay around 3.7usd for a gallon? You would pay 7.27usd for the same amount in germany
You nailed that one on the head I could care less about power draw and everything about performance.
We'll see the exact same videos when Next Gen Ryzen drops, and I expect they will surpass this Gen of Intel. Then Intel will punch back. Which means there will really be no inferior choice going forward. I like it!
unless this is literally all INTEL has since they only have the 12900K and 12600K (KF), it's very unusual for them to not have a full range of products ready to go and can't compete with anything AMD pushes out again for a significant amount of time, remember these INTEL chips were being designed and prototyped 2 years ago at the earliest, so who knows what AMD has since their 5000 series chips are now a year old and the next edition of RYZEN should be around the corner, not to mention they are doing that 5000 series change with memory and stuff, so will have to see what that does.
There is (and was) already no inferior choice when performance/dollar + availability was concerned. Each brand had its pros and cons for most cases, the only clear cut was HEDT-like use cases where AMD had dominating performance advantage.
What would actually also be decisive is the time the system needs to complete the task.
Assuming it takes half as long, but requires twice as much wattage, the total consumption should be the same again. I would find this point in the calculation quite exciting.
that's a good metric to use too.
You beat me to my comment by 3 minutes. Congratulations.
Also if you aren't always loading your CPU with high intensity tasks you are idling 40-50w lower, which also needs to be factored in. And if you are gaming they appear to be very close in actual gaming.
would be the same xD man i hate calculations xD my head ;/
@@Danthrax66 I think you are absolutely right.
Cinebench etc. is not a common scenario for everyday use.
For a real evaluation, you would have to measure two systems in everyday use (browsing, gaming, office work, etc.) and compare the consumption over the entire time.
0:28 actually no, the 5950x is better binned and uses less power than a 5900x
I was about to make a similar comment, however, I believe it still depends doesn't it? Would it not be possible to overclock a 5950X to consume more power potentially? I'm wondering if the 5900X only uses more power than a 5950X when they're both stock?
I think he was just trying to use a chip that's comparable and that most people will buy. The ryzen 9 5900x vs the ryzen 9 5950x in gaming tasks is virtually identical.
@@conza1989 If you overclock both the 5900x should still use marginally more. This is also probably why the 5950x always had a fairly low availability. Most chips aren't good enough for it.
By the way I just noticed Jay is running the 5900x with 3200MHz ram which is also hurting its performance. You really want at least 3600MHz.
@@Karavusk Yeah, you know using the 5900X is actually the worst case of scenario for Ryzen in this testing so despite what Jay said here, actually the better CPU to test with? Idk.
I've seen that most 5950X's have lower temps than 5900X's, the 5900X is the hottest of the Ryzen 5th gen chips, but it is definitely the best value of the 4 "X" chips
This was a good idea Jay.
It got me to pull out my own outlet Watt reader and hook up my Desktop tower to it that's also using a Ryzen 9 5900X with a Zotac RTX 2080 Amp. My tower has several RGB strips and case fans including the ROG Aura terminal. 4 3200 DDR4 sticks OC to 3733Mhz. It has a 280mm BeQuiet AIO, 3 SSDs, 5 HDDs, plus 1 DVD burning drive, all that packed in a Thermaltake Core V51 TG case.
At idle I was getting 177W and at full load of R23 I was getting 396W.
I should also mention that I do optimize my CPU so it's not running at bone stock performance. (I do Precision Boost Overdrive from Ryzen Master (PBO2), Plus Core Optimizer configuration in the BIOS with the PBO there set to Manual/Motherboard max values for PPT/TDC/EDP for MSI's flagship X470.
My R23 Score was 23258, and I ran it again at the load was about the same. *So a 219W power difference* based on my configuration.
Glancing at the read out of the outlet meter now as the system is idle with R23 running but not benching shows 175.9W idle
I clicked Start again on R23 and saw it jump to 400W then back to 396W at full load and when R23 completed, my new score is 23334, it's back to idling at 175.9 now.
This was a good test.
I will stay with AMD for 1 or 2 Gens, to see what will come out of this new Type of CPU, since i don't plan on Updating to win11 until maybe the end of 2022.
So i'll be fine with my 5900X and 5950X in my Workstations.
Still have to say, really impressive Performance have Intel there
10:55 i do not think the intel spec sheet is total system draw, i would expect the to be the VRM output power usage
Would love to see another test where duration would be considered, like how many loops of Cinebench can the Intel chip do in an hour then have the AMD chip do the same number of loops and then compute for the power drawn for the duration of the loops between the 2.
@@n1t21r3 Yes Intel CPU is definitely less efficient than AMD counterpart but of course difference is less when you take time into account, would love to see that test.
Also in the video Jay was measuring power at the wall, PSU is at most 96% efficient, VRMs that are supplying power to the CPU also have losses (VRMs probably around 93-95% and if you actually plug in these numbers CPU power draw was actually 244W which is give or take what is in the spreadsheet), also since Intel CPU is drawing more power and losses scale in % it will all add up to make it even more exaggerated.
Yeah that's what I was trying to show above... people need to look at the wattage drawn *over a time period* for the amount of work/tasks done. :)
The Handbrake Encode shows that total energy drawn for the same amount of work done, was more or less the same (Intel marginally less; but the additional RGB on the AMD system is probably a partial contributor to the AMD's higher power consumption).
@@XPACT3 I would bet that this kind of testing would show the intel CPU might burst draw more but lets include say running the machine for 24 hours and this 1 hour of 100% load is in the middle of it ... I'd bet value for money is with intel here on power draw.
I had this debate about AMD being on top when ryzen 3 CPU's landed, tons of my discord community were like "intel is done, they can't beat this" ... I reservedly said ... "it's not like intel to sit down and take a beating without a fight, I'm betting once whatever their current fab issues are are resolved, the fight is on.". Apparently I was on to something!
There's also another factor to consider ...
Quality of the circuit, any flaws due to manufacturing process will cause inefficiencies in the CPU, and intel is hard pushing for these new 12VO power supplies to help reduce losses to only that which are actually needed by having the VRM's handle stepping to other voltages as needed, my bet is the next step for intel is to reduce the number of different voltages across the entire system needs somehow.
@@paulward8087 I bet that 13th gen will be a big step up in terms of efficiency (people need to keep in mind that this is first generation of LGA1700, basically it is new architecture on an old production process, Intel will need one or two generations to really nail it, same as AMD with Ryzen) of course we will see what AMD has to offer for their next generation which I am pretty sure will be very competitive, also another thing to consider with these new Intel CPUs is that heat output and power draw are probably measurably influenced by new memory controller which now needs to handle 5000+ MHz memory kits...
@@paulward8087 On a power supply note, well 3.3V and 5V are definitely needed for a lot of circuitry on the board but it is easy to step down from 12V on the motherboard directly but would rise their cost tho. So it is definitely possible to have only 12V power supply but they will still have to step it down for every component that requires it I don't really see how that will affect efficiency of the system since main power users are CPU and GPU which already accept only 12V and then through VRMs step it down to 1v+ range for the core. In theory PSUs would be a bit cheaper to produce but I highly doubt manufacturers would actually change their price, on the other hand motherboards would definitely go up in price, they wouldn't miss on that. Only useful thing from that idea would be "ease of use" since everything would be able to use same connectors from the PSU. Also SSDs and HDDs would need to get power from the motherboard now so new connection type would be required instead of SATA.
">1s" is "more than 1 second" (that's why it's called "sustained")
The main reason the Ryzen chip consumes more power at idle is because it has three dies, essentialy tripling the idle-state operations and therefore how much power it consumes.
This is easy to see when you notice that Ryzen 5000 on mobile(Which is a single die) is much more efficient at idle than any intel chip would ever dream about , even tho the 10900k is still more efficient than the 5900x at idle(So it's not a generational thing either)
Clearly the Intel CPUs are impressive, but I hope amd gets to lead the market for a couple of years to make Intel run for a while.
I'd rather just have true competition
I follow value for money. I'm still rocking a ryzen 3600, rx5600xt but I have been hoping to upgrade for over 1 year now. But I refuse to pay current prices. 3600+5600xt still great for me however so I'm in no rush.
Then wait for Zen 3 3D (early next year) at least - the huge L3 cache will give gaming performance a nice big boost. Hopefully CPU prices will drop with Intel having proper competition.
(I have no hope for GPU pricing becoming sane anytime soon, even with Intel entering the GPU market. The chip shortage is expected to last into 2023.)
I disagree with your conclusions. The overall energy consumption of the 12900 will be far less than the 5900. Lower idle, more power to finish a task faster and get back to idle. The 12900 finished the task, dropped back to 54 watts while the 5900 was only about halfway done. I think those meters will show kwh. I bet if you had zeroed them at the beginning of the video and checked kwh at the end the 12900 wins easily.
I like where you are going with this but I think you need more clear tests. 1# Record each system in a test bed setup for Peak draw and Total used. Boot PC > Idle PC 10 minutes > run all your cpu tests 3x > Idle system 10 minutes > power off PC. 2# Setup task scheduler to open set web pages and apps a handful of times > repeat every hour > Shut off PC after 8 hrs. Report those numbers. That should give you a more realistic set of results to go off of. What a high performance workload might look like and what an average office PC would look like. I know that would take at least 2 days to complete but it would give viewers a full understanding.
I really think the 5950x should be the one compared to the 12900k. Cost wise I still think it is cheaper @ $750 then the 12900k @ $650 (both current prices at newegg) once you figure in the probably more expensive z690 motherboard and the defiantly more expensive ddr5 ram. I am sure the 12900k will still be some faster but I believe it will be closer at least in some tasks and still cost less.
I get 26000 on cinebench with my 5950x, so pretty close in scores, seing as the 12900k seems to get around 27700
newegg gouges new hardware --- I've been a customer of theirs for nearly 2 decades and they've really gone to $hit.
Agree! Not sure why Jay thinks he should compare a 5900X to the 12900K when literally EVERYONE else is comparing it to the 5950X. Losing my faith in Jay's technical ability.
Calculating the electricical work for cinebench (watt *time) would give a better comparison for the efficiency
I'd love to see a look at where thermal throttling may come into play with the 12th gen vs ryzen in CPU processes. If you do same cooling on the two, is intel faster but only for "x" minutes, and then Ryzen pulls even or passes after intel hits high temps due to the extra power? Is throttled intel slower than ryzen? or just the same speed?
just swallow AMD is beat accept it ...lol no one cares about anything but FPS
@@guydodge9845 I didn't say they weren't. But as power=heat, it's an aspect I'm surprised to see unaddressed either way. If the power use really is a factor only in cost to run, that would be good to know, or if it will require a moderately better cooling solution, that's good to know too for when you're specing out a system. Or if you want to do a SFF build if it is kicking out noticeably more heat it may make the Intel less feasible.
Christ you ask for additional information and fanboys gotta fanboy and get on your junk for asking totally reasonable questions...
That was really a cool experiment. Can you also test power draw while gaming with a couple of titles to demonstrate the power consumption? Coz the performance is relatively similar with intel having a small lead in most titles. Most folks here are gamers I guess than content creators. Even for content creators I don't think they keep doing rendering all the time
You could improve the idle consumption by turning of the LEDs. It could make a difference of more than 10 W. In my case i have an AMD 2700x and a GTX 660Ti and my consumption is less than 90 W at idle with an Saesonic TX-650. If i turn on my LEDs (Just the MB and a few Strips) the consumption goes up around 11 W.
And yes it is an 660Ti not a 1660Ti ...
But that would also reduce load consumption, so the delta doesn't change.
The whole idle use comparison is so BS in his setup.
The watercooling case uses 7 more RGB fans.
And the water pump is couple watts more powerful than the AIO's one.
Jay and his lazy compare test...
He could literally build 8 cardboard box setups with AIO's with the amount of parts he has in his warehouse.
But Jay be lazy and do these easy quick money video's...
@@N0N0111 How is it lazy? He's checking idle first, and it produces more idle. Sure, that's all the fans and shit.
But then he checks under load, and it produces less under load. So how the fuck does that matter? All it shows is that intel uses a shitton more wattage.
@@Geodomus You are the kinda of people that drink the Kool-Aid way to often cause you want it to simple. But this is not even simple, this is over the border presenting AMD idles are that high wrongly.
Each RGB fan uses about 2 Watts minimum - cause they run 0.30A and a range of 6V - 13.2V
He has 8 MORE RGB fans on the case watercooling setup, he easily could turn all the RGB on the fans off in software with couple clicks.(lazy1) The iCUE Commander PRO slurps power as well, but that would be a couple watts.(lazy2)
I bet that whole RGB fan setup +iCUE Commander PRO is slurping North of 30Watts!!! You can take that off the idle power usage and Max power usage.(lazy3)
And the comparison between the AOI's power and the full watercooling pump power will be couple Watts too.(lazy4)
And we could not see if the intel system was using RGB Ram or not. Did he even give those details???, i don't feel loading this shitty test video up again.
And i forgot he used 3200MHz on the AMD system.
And used DDR5 4000MHz on the intel system. (lazy3000)
How is he not giving the intel system more praise .. righttt....
That's what I was discussing with a friend. That the new Intel CPUs look great in paper, but in order to run a 12900k you need to take into account the cost of a 360mm AIO because anything less than that will make it throttle, and then a more powerful PSU Since between that CPU and a high end graphics card you can easily use about 560W without taking into account board, HDDs, SSDs, fans, any other PCIe device, RAM, etc., so you need at least a 750W PSU, and that would be the minimum.
Also the boards are more expensive because they must be able to handle 250W flowing to the CPU.
@Ivar Donado the PSU should be 1K watts. So full load is 50% the peak efficiency for a Platinum or Gold+ rated psu. The costs for a GOOD motherboard that may only be good for 1 generation is also a consideration. I got my X570 two years ago. When Zen 3D is released that will be my next CPU for two years. DDR5 is way too expensive. A cheap Intel board is just that CHEAP! It won't enable much if any features like GOOD vrms.
Man... I wouldn't trust anything less than 850. Even with my 10900k/3080 I wouldn't. I would likely go 1k at least with a 12900k because I have a lot of extra junk going on.
For me it will be great I have an i5-4670k at 4.7 MHz and I am in urgent need of an upgrade I will skip DDR4 and I already have an 850W power supply, but I'll wait a while for prices to drop
most >GOOD< PSU can handle over 100% power drawn without any trouble. The only reason to go for more than 700w PSU is if you have two GPUs or use a ton of HDDs and other stuff
The higher performance and E-core option of Intel gen 12 is definitely a good idea. But I'll just be going with Ryzen for the backwards compatibility so I can try playing Sniper Elite 1 at > 200FPS.
I think a more fair comparison is to turn PBO on for the 5950x so the playing field is more equal in terms of power draw.
You're crippling AMD by limiting it to 140watts while Intel is pulling twice the wattage. AMD would likely have a stronger showing in a handful of benchmarks if PBO is allowed.
Highlights how Intel is using sneaky tactics to try to boost their scores.
I'm not an amd fanboy, i just try to shop with wallet conciousness... And in that regard, I really had to swallow hard seeing intel pricing for both mobo and cpu... I know a 5950x is more expensive, but you at least get to buy a mobo prior, even like a year ago or so, or even more... and still have the ability to buy the cpu later... intel sadly requires you to both buy a cpu and a mobo at once... so that cost goes up to at least 800 euros or more, in one go... while amd at least has the option to spred the cost and all..
I'm expecting a new socket for next gen AMD.
@@jetah50 Zen 4 will be on a new socket, AM5. There is a new Zen 3 chip coming out early next year that will still be on AM4, however.
cpu mobo and new ddr5 sticks - thats a lot of investment for bleeding edge tech
But if you look at the whole thing: even x470 or even b450 with some luck, can still get 5000 going. But let's take b550 or x570. the backwards and forwards compatibility makes this a pheaseable long term investment, not to mention, ddr5 is a freaking wallet hole ... sort of... like putting all together, you're nearing 1000 euros for only mobo, cpu and ram... just nuts... Pritty sure you could get to 1000 euros for amd too, but you might be able to add an ssd if you do it smart... besides, there's always, buy mobo a year ago, then buy cpu later.. If amd comes out with a new socket next year, I'm really hoping it equals am4 bigtime...
i5 + mobo + 16gig dd5 ram was about 650 euros for me - still alot of money. Ryzen 5600x + mobo + 16 gig ddr4 could have been around 500. I rather have the better performance and upgrade paths.
I think it really depends on what you already have: Got a Am4 board already? Don´t get 12gen. Got a decent DDR4 Kit? Consider a "cheap" z590 for 200 Euros - works just as well with 12th gen.
Having a i5 4460 from 2013 with ddr 3 like me- well thats different. Can´t use any of it - thats fine, it worked great for 7 years :)
Holy GPU, Jay! You have subtitles! Thank you!
Let's all remember the main reason of these hybrid architectures and that's exactly why smartphones and consoles use it.
LOWER POWER CONSUMPTION and LOWER HEAT GENERATION.
Now, with that in mind, let's read the benchmarks again.
It's a lot like the bulldozer days when AMD had to push a ton of power into the chip to get the performance they wanted, except in this case Intel doing it results in actually performing better. Nice to have this competition back!
For this test the one thing that needs to be the exact same at least is the power supply as a difference in efficiency of it will skew the results in power consumption between the two systems on load 🤷♂
it should be the same powersupply and the same motherboard, and the same ram and the same boot drive and the only difference should be the CPU... but that's not possible. it does however show the important part... Intel draws a shit tone of power on load... lots more then AMD.
You'd only be looking at 10% difference with the worst-case scenario of PSU efficiencies.
@@RognisVornel With each discrepancy in the components the end result is less reliable. The impact of different case, cooler and GPU in total power draw is pretty obvious, so I was just pointing out to one part that is more likely to be neglected in this scenario of high power demand which is the PSU. efficiency.
@@RognisVornel take those 10% and add everything else making a difference 😉
@@fgsaramago How would 10% of the load wattage change anything? 110 would turn into 84 or whatever. Intel's chip would still be power hungry in production workloads.
Nah, editing this. Let's say Jay had the best PSUon the AMD and worst on Intel. That's still over 50 watts higher load. What does that change?
For CPU rendering (and other cpu tasks) for long periods of time, i think the Intel would take too much power compared to the AMD to be worth it, Normal use, yeah.
Mobo prices also matter, since we all know Intel ones are way more expensive than AMD and people tend to overlook that when they talk about price/performance
Also, DDR5 premium pricing.
The power draw on the AMD is lower than shown, keep in mind its a full build. You would have to calculate and subtract all fan wattage, water pump(s) wattage and any other accessories such as led strips to get actual load wattage. I'm thinking maybe somewhere around 30+ watts lower, maybe.
The CPU pricing in Germany is kinda wonky over here the 5950X is only 50$ more expensive than the 12900k ....
I'm waiting for AMD's upcoming line to compare it to.
For sure, I may upgrade then too, from my R5 to an R7.
If I would buy a new system right now, I would look at the whole platform cost (mobo + cpu + ram). I don't think Intel comes out cheaper. The cpu price itself is great but you need to look at the whole picture. If both were overall the same I would prefer performance (Intel) since you rarely have these high end cpus fully loaded, power cosumption is not a dealbreaker.
A high-end Z690 motherboard and DDR-5 RAM would be fairly expensive, as anyone jumping to Alder Lake now will pay the early adopter tax. I'd recommend that most people wait for DDR-5 to improve timings and frequencies for lower latency to make the jump worth the cost.
Alder Lake works with DDR4 too and the difference in DDR4/5 performance is currently negligible (will change in the future, but then you will have to buy new sticks anyway).
@@zuruumi9849 That is correct, on lower-tier MBs you can run DDR-4. But when you are ready to jump to DDR-5, then you would have to buy new RAM sticks and a new MB, if you jump to Alder Lake now with DDR-4. So there are very few cases where it would make sense to do this.
@@zuruumi9849 But you still need a new MB and that can eat the price difference. Intel boards are not cheap. And I think when you decide to go intel it would be stupid to choose a MB that is not able to use the newest supported RAM technology.
If I would need a system I would try to wait until the new AMD system are available. That what pisses me of the most with intel beside the power needed is that you need to use Windows 11 for full support and that would kill my Dual Boot with Linux because of the shit TPM 2.0 wanted by Win 11.
@@Andrew-ep9tg I do agree that waiting is the best idea (either for cheaper DDR5 or even a new AMD chip). But if you need to buy a new PC right now and DDR5 is over the budget you don't necessarily need to go AMD.
2 things:
1) Thank you for doing a power consumption comparison. Would it be possible to do a side-by-side temperature comparison as well with the same load and cooler, please?
2) Overclocking Ryzen™: I've done some experimenting with my 5600X and it is true that using Ryzen™ Master "Auto OC" or overclocking beyond the max boost clock of 4.6 GHz, in general, will hurt performance. Using only Cinebench R20 as a testing method I discovered that the best scenario for my CPU, in particular, is locking all of the cores at 4.6 GHz (max boost) and the voltage at the highest stable level, which in my case appears to be 1.225 V (I had it at 1.2 V for months but the system would crash every now and then with one particular application and now it doesn't). My numbers are consistently higher with that "overclock" than they are at stock settings and the stock settings perform better than when "Auto OC" is enabled. It's technically not an overclock because it's not passing the max boost threshold, but tweaking the speeds and voltages just a little bit seems to have given me much better all-core performance.
I have yet to overclock the highest-performing core beyond 4.6 GHz because I like everything the way it is now.
I'm folding my laundry, and that background hiss made me think my dryer was about to blow up on me. XD
Jay you have to remember PSU efficiency. 272 out of 247 = roughly 110% This falls right in line with ~80+ gold.
Actually while idling the PSU is less efficient (close to 80%). Under load the efficiency of a typical gold rated PSU comes close to 90%. So it's actually slightly worse.
Good call, that was one of my first thoughts.
Unless you have such a massive PSU to run your 3090 that under CPU tests you never hit peak efficiency. (he does use 1600w PSU's sometimes)
My EVGA 1600 T2 is 90% efficient minimum at 10% to 100% load range. 50% load is when it is at its most efficient, 94%.
The vast majority of PSU's hit their peak efficiency at 50% of the rated load.
I'm more interested in seeing a comparison between Alder Lake and Zen 3 at the same TDP.
Yes but it is possible to restrict the Intel CPU that much ? A 5950x has a TDP of 105W. That ist less than Intels TDP of 125W and less than half of the MTP of 250W (unlimited over 300W)
The same TDP but this value is just indicative
the difference in idle power draw comes from the node they are using, tsmc or amd cant go lower then ~0.7 volt on a pair of cores for sleepmode ( in amd cpu´s there are allways 2 cores conected to each other, same logic applys to overclocking and boosting, you have allways to overclock 2 cores ) , going lower then that deactivates the cores and cant be enabled anymore until restart, where intel can go as low as 0.3-0.4 volt per core before it stops working
thanks for pointing out the cables i cant stop looking at them
i stay with ryzen and wait for the 3d cache stuff
Me too especially if they will release the AM5 socket
@@franzescodimitra8815 3d cache will be before AM5/Zen4.
IMO AMD are setting themselves up for another FX disaster. All major chip manufacturers have switched to the PE design (Samsung, Google, Intel, Apple, Qualcomm).
AMD it's going to be stuck between a rock and a hard place in a couple of years constantly trying to balance their single core between energy efficiency and performance.
You already seen the preview of it happen here when the Intel performance cores kicked in and outperformed and then the Intel efficiency cores were overly efficient over Ryzen. That balancing act AMD has to continue to do for AM5 all the while Intel continues to develop each core individually making them better at their specific functions RIP AMD AM5.
AM5 first gen may keep pace but after that it'll be all dust in the wind. Esp. once the process node difference shrinks down. Remember AMD is getting their performance because they have continually shrunk down to smaller nodes. So PE chip combined with Intel leaving 12, AMD might as well declare bankruptcy.
Can we get a "I digress...." coffee mug?
Thanks for showing this Jay.
in NZ the 5900x is like 850 nzd 5950x 1240n zd 12900k 1100 nzd
I would personally compare the 5950x as if you include platform costs and needing a better cooler the prices would be comparable however prices are different in every region.
the logic for the power consumption made sense but I have seen 5950x,s using less power online due to better binning.
edit not criticising your video you showed everything and your logic was sound
Twice the power, Twice the speed, Half the life?
Yeah the 12600k is awesome, best value for performance with power draw being decent, 12900k is too much for me, ryzen just so damn efficent with there architecture
yeah but we still have to see amd respons, if the new zen 4d have the same performance they will win cause they will absolutely have less power usage
WAIT FOR 12400/12500 if value matters for you.
@@djandry223 100% I feel intel closed the gap for this holidays but ryzen will widen it again!
@@KryssN1 yeah bro funny how intel lower 6 core line up is the best value in there whole series haha
AMD is bringing 3d v-cache with a Zen 3 refresh in early 2022, which should provide roughly a 15% performance uplift for games and cache sensitive applications. Definitely awesome for current AM4 owners.
Intel : mo powah baby!
Voltage: nervous sweating
Do I detect another Donut Media fan among us?
@@daydroid7951 yes sir
J no offense but I don't think it's fair to run this comparison on windows 11 considering Steve from gamer's Nexus who I consider to have much more scientific and insider knowledge To the performance and accuracy of windows 11 tests tests refuses to do tests on windows 11 because he feels that it is not 100% A fair comparison therefore I feel it is a bad idea for you to be posting this video saying it is an Apple's to Apple's comparison
Using cinebench for measuring power efficiency is not a wise idea. It is a heavily synthetic workload and rarely do we run workloads that demand such sustained power draw. The best way to test this is to use an energy meter, run both the systems through identical workloads (a balanced test suite that on average tries to keep everything moderately busy, random time gaps in between) and measure the total energy consumed (watt-hours) after say, 8 hours. That number tells you which one you are paying more power for.
Fantastic, completely unbiased comparison. Personally, I'll stay with AMD until there's a compelling reason to switch - their culture of constant innovation is something I'd like to see in other companies - but Intel seems to finally be getting its shit together.
likewise. Also amd's support for linux also earns my money.
@@xmetrix your the only person who ive ever heard of that uses linux
dejavu moment when 1st gen Ryzen came out. Only thing it was the other way round with staying with Intel and AMD getting their shit together.
@@JLawlietK I don't think so, AMD has never been the type to sit on their asses after achieving dominance. Leaks for their future architectures already show some crazy increase in performance and efficiency. We're in for a long battle between these brand and we only benefit as consumers.
@@mechapope9168 A dude in Vancouver named Linus has been using linux. He has posted recent videos about it. As a big time windows guy, he's...conflicted about it. It works fine, but his kids hate him.
So misleading, small bench against a fully RGB fan system with tubes and all of that white light is also super watt consuming too, crazy.
When talking about idle power I think this setup is not comparable, just as you say in the beginning of video. I just don't understand what you say at end of the video when you use the idle wattage to say the Intel processor uses less... obviously the fans, waterpump and other components should be the same to compare idle power usage and this is not the case.
Idle use cases would be funny because you would utilize power settings to reduce them even further, the GPU to Display turns off your monitor when you have that function enabled and can cut down the wattage from the GPU side for that. Probably cuts down like 10~50 watts on both systems if the standby mode kicks in and the display turns off the screen.
Jay, how about the math associated with taking longer for doing the same job. It sounds like Intel used more power for a shorter period of time versus the AMD. How about calculating the kilowatt hours used to complete the same task?
Would be interested to see these comparisons as well.
This is a very good point. Might be equal or better. Intel has a pretty low idle even when adding a few more watts for the other stuff. Probably harder to cool tho under full load. But if its quick it might not saturate the cooler before it finishes the task, depending on the task ofcourse.
It's using double the power isn't it? Unless it's doing things twice as fast, those numbers won't make Intel look much better. Would be an interesting calculation to see though.
Technically at the end of the cine bench score with his performance per watt metric, he does show that AMD is more power efficient.
Basically how much it pulls for a given task. Reminds me of my Core 2 back in the day. Could give it a 40% oc with default voltage. Increased power draw by around 20% but performance by 40% making it more efficient in the end.
You should redo the tests, and measure the load off the EPS cables to get a true power reading.
Unfortunately that's not good enough because part of that power goes to the GPU through the PCIe slot, the fans through the fan headers and the chipset (or whatever they call it these days), which also has different power draw between AMD and Intel.
@@Steamrick The 12v EPS cable(s) (the 8 pin(s) (Or 8+4 pins) feed the CPU directly, so exact power draw can be measured with a meter.
The 242w to 270w difference on Intel is probably due to measuring power draw at the processor VS power draw at the wall, that takes in account the psu efficiency being around 80-90%. A psu that supplies 700w might pull 800w from the wall to do so
It would be good to see how much power they actually used to complete those tasks. Another words (watts X time to complete each task).