INTEL & AMD both lied! 👉REAL WORLD power consumption is MESSED UP
Vložit
- čas přidán 28. 05. 2024
- Intel & AMD have both 'lied' about their real world power consumption. Here's the Truth:
Sponsored Message:
Whokeys 25% discount code : TN20
Windows 10 Pro OEM Key ($17): biitt.ly/nEliC
Windows 10 Home OEM Key ($14): biitt.ly/e7F0u
Office 2019 Pro Plus key ($42): biitt.ly/gnDE5
Windows 10 Pro OEM + Office 2019 Pro Plus Key Pack ($61): biitt.ly/BSpz2
Buy Intel CPUs here:
🔵 Intel i5 12600k geni.us/i512600k
🔵 Intel i7 12700k geni.us/ii712700k
🔵 Intel i9 12900k geni.us/i912900k
🔵 Intel i9 12900ks geni.us/i912900KS
🔵 Intel i7 13700k Here: geni.us/i713700k
🔵 Intel i5 13600k here: geni.us/i513600k
🔵 Intel i7 13700k here: geni.us/i713700k
Buy AMD CPUs here:
🔴 Ryzen 9 7900x geni.us/7900x
🔴 Ryzen 7950x here: geni.us/7950x
🔴 Ryzen 7700x here: geni.us/7700x
🔴 Ryzen 7600x here: geni.us/7600x
🟦 Intel 12th & 13th Gen testbench:
Motherboard: geni.us/z690ProArt [Asus Z690 ProArt Creator]
RAM: geni.us/b7Sfs [Kingston Fury Beast 5200MT/s]
Cooler: geni.us/cnVpbO [Phanteks Glacier One 360MP]
Cooler Fans: geni.us/cRsu3 [Phanteks T30]
OS SSD: geni.us/eRVfwYj [Samsung 980 Pro 1TB]
Project Drive: geni.us/y75Sg [Samsung 980 Pro 2TB]
PSU: geni.us/qs5Asp6 [Deepcool PQ1000m]
GPU: geni.us/tuf3090 [Asus TUF RTX3090]
🟥 Ryzen 7000 Testbench:
Motherboard: geni.us/x670eProArt [ASUS X670e ProArt]
RAM: geni.us/b7Sfs [Kingston Fury Beast 5200MT/s]
Cooler: geni.us/StrixLCII360 [ROG Strix LC ii 360]
Cooler Fans: geni.us/cRsu3 [Phanteks T30]
OS SSD: geni.us/Hmf9MA [Teamgroup Cardea C440 1TB]
Project Drive: geni.us/2TBm480 [MSI m480 Spatium 2TB]
PSU: geni.us/yZC1Nk [Corsair hx1000]
GPU: geni.us/tuf3090 [Asus TUF RTX3090]
👇❗️PC Builds for Creators [Latest]❗️👇
➡️ $750 BB4B (Best-Bang-For-Buck) PC geni.us/750-1300BB4B
➡️ $1500 BB4B Creator PC geni.us/1500-2300BB4B
➡️ $2500 BB4B Creator PC geni.us/2500-3500BB4B
➡️ $4000 BB4B Creator PC geni.us/4000-5kBB4B
Want me to check out some tech or interested in collaborating?
📩 email me: thetechnotice@gmail.com
FB: / thetechnotice
IG: @thetechnotice
#intelVSamd #powerconsumption #efficiency
Video produced by Lauri Pesur
Edited by Sam Ruddick
-------------------------------------------------------------------------------------------------------------
GEAR USED IN THIS VIDEO:
[All 'geni.us' links below redirect to Amazon (or some other Online stores) and may earn commissions]
AUDIO:
-The Recorder + Mic I Use: geni.us/tscmdr-10l
-Shotgun Mic I Use: geni.us/T35T
-My Main Headphones [CHEAP]: geni.us/dzv7Q
-My Main Headphones 2.0 [NOT CHEAP]: geni.us/as7q9v
VIDEO:
-A-roll Camera: geni.us/NvnoA
-Lens for A cam: geni.us/J009I
-B-Roll Camera: geni.us/wVlF8J
-Top Down Cam: geni.us/56576C
-Main Bokeh Lens 4 Sony: geni.us/k09oA
-The Wide Angle beast w OS: geni.us/9KfCwy
LIGHTS:
-Main Fill Light: geni.us/qTbE
-Hair & RGB Lights I use: geni.us/TdY1
-RGB Tube Wall Light Behind me: geni.us/96gWdz8
-The Fire effect light: geni.us/GX5KA
OTHER:
-Favourite camera Monitor w LUT: geni.us/tecUD
-I get my music here:
▶Get 2 Months for FREE when joining Artlist/Artgrid below◀
Music & SFX Licensing Site: geni.us/TNartlist
Stock Footage Site: geni.us/TNartgrid
DISCLAIMER: I have to state that some of the links on this channel are affiliate links which means by clicking on them and purchasing the items I get a small commission. I'm part of an Amazon Affiliate and earn revenue from qualifying purchases. Using the provided links does NOT change the price for you.
⏲️
0:00 Intro
0:50 Sponsored Segment
1:13 I need to explain something.
1:20 1. 100% CPU utilisation vs Real Workflow
2:45 2. Why 100% Power Draw is STILL useful!
4:22 3. Idle Power Draw on Intel vs AMD
5:25 4. Which Workflow [Programs] were tested?
6:31 4.1 Why did I test these Programs?
7:39 5. How was Testing done?
9:00 Important NOTE
9:13 Test Results: Explanation
11:02 Adobe Premiere Pro
12:53 Adobe Photoshop
14:06 Adobe Lightroom Classic
15:21 The Yearly Cost of CPU Usages
17:07 IMPORTANT Note on Yearly Cost! Idle vs Under Load
18:11 Hardware vs Software monitoring
19:32 Testing the Full PC system from the wall
20:29 13900k vs 7950x From the Wall Power Efficiency
21:48 The Conclusion - Who's lied?
23:32 Best PC to build as a Creator
![DON'T BUY THIS CPU! 👉 4 Mistakes to AVOID as Creators! [More Expensive & WORSE]](http://i.ytimg.com/vi/8M_zoDKw7XA/mqdefault.jpg)
![DON'T BUY THIS CPU! 👉 4 Mistakes to AVOID as Creators! [More Expensive & WORSE]](/img/tr.png)
I've searched the entire internet for idle power usage stats and found nothing as comprehensive as this. Finally, I can make a decision thanks to you.
👍
True. I need this stat for 3D version.
What was your decision btw ?
@@Saimeigaming went wth the 13700k with a 4070ti
@@thescarletcarswhich would you choose between i5 12500h and ryzen 7 6800h for a laptop?
So nice to see reviews based on actual work instead of just gaming.
You must not be watching good tech channels because the better ones do gaming and real world work loads. If you work in the enterprise sector alot of them are locked into Intel because some of the software doesn't work with AMD.
@@Sakosaga The problem is that most reputable channels use power viruses or heavy renders to measure power consumption and then base the entire review of off that which is simply misleading because many heavily multithreaded apps still don't fully utilize given cpu. And this is true even for gpus.
@@w04h even if it's the case of alot of applications that are older, may use just 1 or 2 primary cores. You're going to still end in a similar situation. Remember there was a time when Intel's turbo boost did only push 1 or 2 of it's cores depending on the platform because they took in too much power when you would try to boost all of them instead still leading to a power draw that was still higher. If you effectively remove the turbo boost function from them, sure they'll consume far less power but these CPUs are designed to be pushed hard right Dealing with massive amounts of heat. That's why the Turbo boost technology has changed so much. Also alot of reputable reviewers that I watch use super old and new testing for benchmarks and when they measure power draw they do it from the application that will measure what stretches the most from the CPU and it's idle consumption. Most large scale businesses won't be using these chips as an option and this is coming from someone who has spoken with venders such as HP , Lenovo , etc usually when your business is under a contract with them, they typically won't send you the newest hardware for a business to use because stability, compatibility, and cost. In a situation where you have people who render movies, make video games, etc. They're most likely using a platform such as AMD thread ripper or Epyc servers because of the need for more CPU power for heavy data transfers which can take an insane amount of CPU power to just transfer files. These are consumer chips and most of the sales from these things aren't going to be used for little things like typing emails and only watching CZcams, etc, those consumers would most likely find a laptop that has a decent amount of power that isn't anywhere near a 13900k's power. So I can understand the argument but in terms of real world it isn't going to be the same or used that way that's why people don't test these things like this, it's for the targeted audience of who these are for or worst case scenario of these chips.
Intel is more efficient in these productivity apps. Either due to acceleration or better optimization by intel or the app developers. So if you work and game go intel, if you only game you might want to look at amd only. In gaming on avg amd is about 30% more efficient. Cpus are pretty efficient these days so it wont matter much anyway. Laptops on the other hand amd is the only choice if you aim to use battery often.
@@Sakosagaplease name some as I really need more of these professional workloads centric channels. I know like, close to 20 channels focused on gaming sweating over 10-20 fps, but I was only able to find this one channel that focused on productivity workloads.
It would have been interesting to see ryzen 5000 cpus included in this, especially considering 12th and 13th gen are included.
I have a Ryzen 5 5600. If I remember it correctly, it idles at around 7W, sometimes under 4W if the reading from the Adrenalin software is true
@@KingMacintosh My 5700x is idling at 0.4GHz and 9W, that seems rather mild to me.
The uploader concluded, Intel is better under the assumption gaming PCs will be idle most of the time. Its called "gaming" PC. Its for gaming.
@@fynkozari9271 well your keyboard works, congratulations.
Really would have liked to see the new i5-13400 from Intel but either way great video maybe just expand the generations of both companies maybe even 10th and 11th GEN from intel
All the time I heard AMD fans saying that AMD is more efficient... turns out that AMD is only efficient if you leave your PC running at full load all the time. (Which is some people, but not most people).
Exactly...
Intel has basically always been way more efficient at low/idle load, interesting to see that is still the case today.
I remember comparing a hashwell system and amd fx system, the difference was staggering with the amd fx system drawing more power at idle than the hashwell i5 was drawing under load
Guru3D does these idle power measurements from the wall on all of their processor reviews and there's basically no difference in idle system power consumption between Intel and AMD.
@@bjarne431 Hashwell? Is that the predecessor to Kebablake?
@@michalsvihla1403 indeed, and its almost as good as kebab(lake) :-) (i know i misspelled it)
good job ! I think for CAD work, idle makes up an even bigger percentage. I was allready thinking about buiding with intel because of the DDR4 that I have, but based on this video I think I will be going with 13700K instead of 13900 K. Also needing less cooling means less noise and maybe a smaller case. I swithced to GPU rendering some years ago so more cores does very little for me.
I can reconmend the 13700k it is a beast of a CPU i pick mine up from microcenter for 370$ with coupon code. Everything runs smooth and I even undervolted it with a slight oc 5.5p 4.5e cruches through all my workloads.
If you look for a quiet build and do not travel with your PC (case is heavy) then I can recommend Fractal Define C with solid side instead of glass. Currently I run it with 9700k and noctua dh15s - it's really silent even under load (cpu is not OC'd).
Maybe a different test you can and should do with Blender is the viewport/eevvee test instead of rendering on CPU.
The viewport remaining nice and smooth while working in Blender is the number 1 thing.
That's true. Being a creator does not utilize CPU/GPU at 100% always. Thanks for this, I am looking for some irl testing of the software both in idle and stress test. Part of being a creator is thinking and that is an idle time so we should consider that.
10w vs 50w is a dramatic difference of 2-3 dollars per month if running 24/7
May make a difference if you have a company with hundreds of PC's
@@dtibor5903 This is FAR more important than that ,, this information DESTROYS the AMD SHILLS that have been LIARS
@@tilapiadave3234 😅so true
It would be useful to see the same comparison with the 65W CPUs.
I support you!
There is no point in testing applications optimized for processors from one of the manufacturers, as in this case. Adobe products are not only optimized for Intel processors, but also use a very small number of cores and threads - it is for this reason that we do not see a serious difference between 6 core and 16 core processors in task execution time. All this makes such reviews absolutely meaningless and false.
Any optimization involves completely different principles of load and processor operation - so this video is not about the truth, but about a rude and blatant lie.
@@johnpreston8621 Cope. Adobe has Quick Sync, and the point still stands. Intel overall is way more power efficient unless the processor is underload a significant amount of the time..
Trust me 7900 (65 watt) is much better then this char, only the idle will say more or less same (few watt less), all the rest will be better and for sure (without a doubt) beat intel. Did enough research. I have 7950x3d: It's much more power efficient then this video would make you extrapolate or believe.
My ryzen 7900 (non-x) has a poor idle performance. It's exactly consistent with this video. My 12700k is far better
This is the exact reason I got the 13900K, it has an amazing balance between low task efficiency, and high task brute force. When I see those 24 cores kicking in my 3D renders I love how much time it saves, but in between renders and between workloads in general, I am so thankful that I can have this beast of a CPU sitting at idle barely sipping power. Over time I reckon it will be even more efficient on a yearly basis than my old 9700K - despite being a latest gen i9 and everyone shouting "Intel power consumption bad!!!!". In reality, my experience is the exact opposite. And tbh nobody outside of enterprise studios sit with render queues 24/7 for 100s of customers waiting in line.
You'd be better off rendering with a decent gpu. Who still uses cpu for rendering?
@@bigbro890 There are tons of render software that are cpu-specific optimized due to the accuracy that you only get from cpu rendering - so a gpu would do jackshit for me. If it was a common blender render it would be different, but that ain't me. There are plenty of reasons why one would go with either cpu or gpu rendering, thus far there isn't a "one way solves everything" solution.
In my case, Cinema 4D + Corona render only supports cpu rendering, with gpu denoising. I'd gain nothing from upgrading my 2070S to a 4090 (in terms of rendering speed).
@@Real_MisterSir There is absolutely no need for cpus accuracy for 3d renders lol. This is the dumbest thing I've heard this year so far
@@bigbro890 Do a simple test then. Take a complicated scene with complex light bounces and scattering within various transparent and opaque materials. You'll find that the light bounces will have a degree of variance that is far greater for gpu renders, than cpu renders. This is because the gpu cores calculate approximations much better than precise interactions, thus the photon tracing is randomized to a greater extend than cpu rendering. This also leads to more noise, even if the render itself is generated faster, so you will just shift the problem from one place to another, also why most gpu-based render applications use post-process denoising to a greater extend than cpu render applications require - again compromising the potential quality of the final render because denoising is inherently inferior to accurate first-time raw render calculation.
Granted, for many people the actual quality difference isn't noteworthy enough to be an issue, or sometimes not even something you'd notice at all if you don't actively search for it. Each task is different, and different render solutions cater to different needs.
On top of this there are also other advantages to cpu rendering, such as support for far greater memory capacity since it has direct access to your RAM, compared to the GPU which relies exclusively on VRAM which is still very limited even for the highest tiers of enterprise Nvidia cards, let alone consumer cards that are restricted to 24gb or less. If you render complex scenes with very high texture quality and material complexity, this too can potentially cause more issues with gpu rendering vs cpu rendering.
And yes there are other differences too. And no, I do not personally choose cpu rendering because I demand this pinnacle level of quality. I just choose it because my 3D modelling and render software of choice is cpu limited, that's all. But I do appreciate the precision and final quality potential of the renders, as do my clients.
@@Real_MisterSir Okay, I liked your comment. You obviously know what you are talking about. As you say, the differences are hard to spot. In my opinion it's not worth the insane render times on cpu.
this is a very interesting test that provokes more investigation. It would be interesting to measure actual wall power usage over an extended time (I could do this myself if I had a power meter) for a truly real-world measurement of domestic computing costs. I've noticed my own 12900 system showing a package power draw of typically something like 23W in a quiescent state. It's only when doing something like Cinebench or prime95 that the consumption figures really shoot up (along with the fan speed). Of course, HwInfo64 isn't telling the entire story of power usage, but it's a good indicator.
ETA: ah, you did the real-world wall-socket test! I commented too early. Excellent work.
Exactly all that matters is how fast they are because the time saved doing a compile or render is worth way more over a year than a couple of coffee. Probably if you get more done then you will safe money of coffees too while waiting for a task to complete 😀
Glad someone pointed it out! This is why I still prefer Intel than AMD. However, I think the low idle power of intel is because of it being monolithic while AMD is using chiplets/soc, not the best explanation but something like that.
Yep, it's due to the infinity fabric and I/O die not being able to downclock/ramp up fast enough or else you get stutters. Downside of chiplets but sort of inconsequential for desktop usage. Like it was a ~20 kwh difference for those top end chips. That's not much.
Notice how amd laptop SOCs are all monolithic though as that does matter there.
I believe you are correct. The new AMD GPU's 7900XT & 7900XTX are chiplet architecture and have high idle power draw. I do believe chiplets are the future but we have to figure out how to make them more efficient at idle.
The AMD Ryzen 7000x are just tuned very, very inefficient stock. Intel did that better, objectively. But who cares?
This video is just about clicks I realize now.
@@reptilespantoso Also 5000 series is ALOT more efficient out of the box and my 5800X3D at stock is using between 20-30 watts with some stuff running in the background like Chrome/YT and discord.
I will tune my system a bit and try to get it closer to 15 watts.
My RTX3080Ti does hover around 50 watts with spikes to 110 watts but it has a 110core and 1100mem OC, again with discord and Chrome/YT open.
That is EXACTLY it. Using chiplets means you have to boost signal levels to move them from one chiplet to another and even at idle a PC is doing some work.
But really, that's such a trivial thing I find it laughable. In all-core loads and gaming Zen 3 and Zen 4 parts like the 5800X3D, 7600X and 5600X destroy Intel parts. And for frames per watts, Techpowerup measures the 13900K at 2.13 fps/W, the 7600X at 4.47fps/W (best) and the 5800X3D at 4.1fps/W (2nd best)
This data is old. In case you don't understand this data or question it, Techpowerup does THE most detailed CPU reviews. Gamer's Nexus also does SOME power analysis but nowhere near the detail of Techpowerup.
Search "Techpowerup 13900K review".
But here's the kicker. The 7600, 7700 and 7900 are all about to lay waste to the line of Intel parts for power consumption. In fact I just checked Techpowerup and they've reviewed the 7600 and 7700 so those are now added to the charts and the 7700 is the most power efficient CPU in all-core loads and near the top for gaming efficiency.
Idle? That's your concern? Turn your PC off because the rest of the components are using as much or more than the CPU.
Have you tried lowering the power consumption of the intel 13th gen processors via bios? Like undervolting, limiting the temperature or manually putting pl2 values? Please make a video on that.
Real world testing is where it is at. Benchmarks only show a part of the picture. That being said, I think anyone would be happy with either of the two flagship chips. I'm loving the bump I got from my upgrade this week. Moved from 6th gen i7 with a 1080 to a 13th gen i9 13900k and an Asus Strix 3080ti...OMG! These two parts are working in conjunction with my Z790 Maximus Hero and the upgrade is AWESOME.
Damn that's a big upgrade!
3080Ti is the best well rounded Ampere card, good choice there. A friend build a 3080Ti system with a Ryzen 9 5900X.
Great test! I fully agree with the findings. The max power draw is only needed for deciding cooling. For the vast majority of users the CPU is 95% of the time, so the low load power draw determines the cost.
true
Except HWInfo doesnt display accurate Wattage when AMD cpus are idle. Test yourself using Ryzen Master side by side with HWInfo and you see the wattage drop when cores park on Ryzen Master, but not on HWinfo.
HWInfo is wrong
@@TheAddanzWij van wc eend adviseren wc eend
Great work. It would be interesting to measure actual wall power usage comparing an Intel based desktop versus an Intel based laptop. I know these chips are different (even with the same model numbers). I would like to know how much saving do I get using a laptop.
on windows 11 currently AMD is idling high on clocks due to a power profile glitch and I believe even latest windows 10 version is also when they removed the ryzen power profiles from OS it bugged out the power states on AMD I am seeing on avg 25w-50w higher idle power draw over when they had the ryzen optimized power profiles.
Intel's efficient cores still would use far less energy than a low clocked Zen 4 one despite the huge difference in IPC.
On Linux the idle power draw of my ryzen 7900 is equally as bad as this video demonstrates.
Did you install the latest AMD chipset drivers? They made a big difference on my system for idle power.
@@nyimakgan the latest drivers and the version before both had an impact. I’m going to fresh install 22H2 next to see how much that helps. It’s not a
I've made a lot of testing with my Ryzen 5600H Laptop to find out its best performance per Wat. Power efficiency depends on max TDP. So it was 15 W for a 50 W processor. Where it has exactly 50% of max computing power but much lower temperatures
I undervolted my i7-12700K by -120mv and power limit is set to 190 watts. I use FL Studio and it maxed out at 142 watts for a project that used 80-90% of the chip. 60 degrees as the max.
I don't know if my tuned i7 is the highest perf/watt chip on the list.
Efficency goes away when i enable AVX512, system responsiveness isn't the same too without Big-Little.
Motherboard is MSI-Z690 Pro A DDR5, Deepcool Assassin 3 , Corsair 5000D Airflow and 32GB Corsair DDR5 4800 (no XMP, just stock).
After i used stock power limits and undervolted the chip, i reduced power consumption by 35% for the same multicore perfomance on Cinebench (almost 23000 points). And iddle power draw is 10 watts with a partial overclock, no core parking, infinite turbo.
Undervolting is almost always the way to go.
How would an intel 13 gen + 4000 nvidia series gpu compare to a Max Studio in power consumption? I built a PC for video- editing in 2022 and now I need a second unit (also for video editing in Da Vinci Resolve)and I would rather to build another pc, but I worry about electricity bills, as I have already noticed a significant increment with just the one unit I use, so I'm wondering if I should go with a mac studio as a second unit. Thank you all for your help!
A PC doesn't stand a chance against Apple silicon when it comes to power draw during video editing tasks. If that's all you need it to do I'd get the studio if you're going for pure efficiency.
Great analysis. A lot of content creators consider performance per watt which obviously gives per second efficiency this could be misleading. 10:16 Power consumption over time should be considered. Example 10w for 4s is definitely better than 7w for 6s.
Not for a laptop due to thermal throttling… it’s not straight forward as you claim
Very detailed analysis. Thank you. PC World did a similar video showing the same results. I love the analogy of testing MPG at 200mph vs. average use. That was excellent.
Thanks for the data
Cheers
Rick
Curious about the figures you used to calculate the KW/Year figure (if it's in the video I missed it). Looking into cpus for a homelab so I'd have to figure out what would be the KW/year for 24/7/365 load of 30/40% on most cores to have a ballpark idea what the difference could be between a 7950x and a 13900k (and a 5950x and 7945hx and a 13900h too ideally, if I find any figures about those around)
I did some testing with the 7700X manual oc 1.18Vcore 5.2ghz. Also there's a setting in the bios that forces all the other parts of the cpu (iGPU Infinity fabric, etc) to run at full speed all the time. I turned that off. My idle is around 25-30 watts cpu package. Cinebench R23 ends up at around 19700 points using a max of 110 watts in hwinfo64. Without tweaking, these newer cpu's are power hogs.
Hello Im working full time and have no time to let my computer "idle" when i switch it on I instatly start gaming doing some work stuff. Would u conisder Amd instead of intel for example ryzen 5 7.6 instead of intel 13.5 ???
Would love to get the answer quick I'm buying pc this week❤❤
What BIOS versions were you using? There was an AGESA update that reduced Ryzen 7000 idle wattage to normal levels
This definitely needs to be expanded. Compare DDR5 and DDR4, older ryzen 5000 generations, possibly even 11th gen Intel and CPUs with lower TDPs
Oh, also the Ryzen Eco mode should be added, I love that function!
I'm having a hard time keeping my i7 13700 KF below 90c at 5.1 P and 4.1 E cores. I'm trying to get my motherboard down to 190Watt and 90c is the least I can get with acceptable impact on Cine Bench
Great work buddy… this channel is slowly becoming my go to tech channel!!! Happy new year as well…
Great video, I was looking for something like that. I'm curious where the i5 13500 would land on that comparaison. Somewhere between the 12600k and 13600k would be my guess?
Great way of presenting the data. Considering and including "idle" or "
4:55 have you checked your BIOS settings? A minimal VDDCR of 1.36V in IDLE seems unusual, this should be around 1.0V. When did you start monitoring in HWINFO? Running PS in the background isn't cosidered IDLE though.
Yeah, AMD voltages out of the box are absolute crap. I managed to get my 5700x to run all day at 4.6ghz using only 1V and 60W peak. Usually around 20-30 though
@@Phenom98 Also Intel voltages, my i7-12700K new was pulling 230 watts and 1.45 volts.
I reduced that thing from 1.45 volts to 1.2, then 230 watts to 170 for Cinebench R23.
Thanks MSI.
AMD stock voltages are very floaty. They move around a ton and rarely settle for a long time at lower voltages. Intel downclocks way more aggressively and also moves voltages quicker
@@Phenom98how
7950X são dois CCX/chiplets, eu já vi e li a respeito que algumas aplicações não conseguem usar mais de um CCX/chiplets, será que isso foi verificado no caso da aplicação usada no teste aos 20:30 até porquê se for uma aplicação que tem preferencia em single Thread será notório o melhor desempenho em determinadas aplicações dependendo de como essas foram programadas, também um dos motivos dos melhores processadores preço/benefício estarem "no meio" da lista de suas gerações...
Amd idle power is caused by the chiplets architecture. Their monolithic mobile cores don't need to move the data off chip & are far more efficient.
It has nothing to do with chiplet or monolithic. Even AMD would be wrecked by Intel on laptops.
Big-Little is used by Intel right now, AMD doesn't.
@@saricubra2867 OP is right. It is because of the chiplet architecture. The IO chip (which doesn't exist in Intel CPUs) consumes power even if the CPU cores aren't doing anything. It consumes even more power when the cores from two or more chiplets need to communicate with each other. It's one of the trade-offs of the chiplet-based architectures - the IO chip always needs to be active (hence consume power).
@@bgtubber Is not the I/O, it's the Zen 4 core's power consumption vs the Gracemont cores from Alder Lake and Raptor Lake. AMD will use Big-Little for Zen 5.
@@saricubra2867 Again, this has nothing to do with the power consumption of the cores because most of the cores are _parked_ (inactive) when a CPU is in idle and they consume pretty much 0 Watts each. The IO chiplet though HAS to be active at all times even if no cores are active, so this brings up the power consumption of AMD chips in Idle (but not by much compared to Intel). Nothing to do with _core_ power draw since in idle almost all cores are parked (0 Watts consumption).
@@bgtubber The iddle power consumption in the video doesn't look like parked cores. I reach 10 watts on my i7-12700K at iddle without core parking, partially overclocked AND and an adaptative undervolt at the same time.
amazing video! just what i was looking for, would you do another video for the 7000 series NON-X? just to keep a track if the newest AMD cpu's are more power efficient this time as they are claiming :O
And compare to the locked 13th gen cpus as well!
Interesting results, especially since they seem to be in contrast with results from most other reviews, and with physics (5nm technology ought to be more efficient than 10nm). I did understand that Ryzen's performance/watt improves a lot when running them in eco mode, e.g. when running the 7950x in 105W or even 65W instead of its default 170W mode.
Yeah, imagine when Intel gets on the same node size! Can't wait for them to make something in the leadership position again power performance wise!
@@maxjames00077kinda necroposting but intel is pretty damn close to tsmc 5nm with intel 7(former 10nm and the reason they renamed it is obvious haha)
The faster cpu should have paid for itself in less than a day even when you factor electricity for the whole year.
There's a point in there...
I like this. Im a truck driver and a gamer and lookong to build a 1080/1440 gaming computer while limiting power for my truck. Definitely still looking at a 12600kf /3070.
Great combo! I just got my 13600k and love it but 12th gen still amazing too
Look into something like the i5 12400T, T sku is more power efficient.
Tech notice didn't include any x3D chips, which give better performance per watts, in my observations. Is there any comparison with x3d chips ?
Also interesting about pugetbench scores. I really wonder what is the amount of included tests that are part of 99% of general usage of an overage photoshop user. Could be that there are some included tests that in real life are commonly used by only 1% of photoshop users, but on the other hand make the results of intel look nicer?
By the way, techpowerup's photoshop custom test puts 7700X above 13900k in performance.
You test shown so much real life usage of processor, thanks a ton for this man & do answer my below question please:
So what would you recommend for productivity on premiere & davinci
I5 12600k or 13600k with MSI B660m mortar wifi & Ak620 cooler (ain't gonna overclock it) paired with 3060 or 3060ti
buy what you can afford. but beware when buying a b660 as it will probably require a bios update before you can plug a 13th gen on it. and you will need a 12th gen cpu to update the bios. some motherboard do have a feature to update bios without the need for a cpu. do some research first.
@@sensible4170 yess 🙌🏼
I wonder how much difference it makes to do a set amount of work instead of a set amount of hours. So converting 10,000 photos, exporting 4 hrs of video a day etc. The faster processors should take less time whilst consuming more energy whilst working. :)
Very interesting content! This makes me doubt on my choice on building an pc for cad/cam software with an AMD 7 7700 processor. What processor would you advice for design and manufacturing software (Inventor and Fusion360)?
Why? Because you can save 10$ on energy bills a year?!
@@harrison00xXx no, it’s going to be my first build and im still researching on things.
my goal is to make the pc as quite as possible. For that reason I want to use hardware that doesn’t generates an lot of head, so i don’t have to run the case fans on an high rpm. Less head = less noise are my thoughts behind it.
@@BT-st4ho Fans? I mean i have 2 fans in my notebook, but unless i drive it really hard they are not noticable and mostly just go in idle speed of 1600 rpm, otherwise they are just OFF.
Rarely, when video upscaling the fans are getting noticable, but still more silent than the average Desktop.
No good performing x86 platform will ever be very silent (or efficient) in the near future, if you want it as silent as possible you need to switch to Apple silicon Macs.
I bought over a year ago a 16" M1 Pro and i cant say anything bad except that its not good for gaming but i cant care less since i have a dedicated gaming notebook and an older gaming PC.
@@harrison00xXx yess i almost pulled the trigger on an MacBook PRO 16 M2. I love their power efficiency compared with an windows. Problem is that most CAM software an cnc controlling software don’t run on macos. That’s why i have to go with X86. I recently received an R9 7900 (65/88w tdp) and an Noctua U12A cpu cooler. I think the acoustics should be very low in my fractal design north with 5 noctua case fans. If I ever need more power from the cpu this should be no problem.
If I ever buying an laptop, it’s going to be an macbook for sure
@@BT-st4ho yes, intel is cooler and has less noise at idle. It also has better ram control . Its actually hard to find a suitable ddr5 that does not take mins on start screen loading if it’s on am5 system.
Here after the Gamers Nexus video today.
The first time I saw the same conclusion was at PcWorld. At that time there was the 12th gen, but the same seems to be valid for the 13th gen, it is even better actually. The same I would like for laptops, power is more important in a laptop.
This is awesome content, it's rare to see these type of calculations. I would like to see the X3D chips as the 3DV-cache helps with power efficiency. But keep up the good work! :)
Amazing comparison! I just wanted to know what power plan were you using during this test and have you tried using AMD power saving plan to see if it changes idl result ?
Can you measure power draw ar the wall instead of relying solely on what software reports?
Did you NOT watch the video???????????????????????????????????????????????
@@tilapiadave3234 obviously you didn't watch the video, because he specifically mentioned he used software only. The issue with software reporting is that it depends on the calibration of the hardware sensors, and different manufacturers may produce very different results for the same measurements.
Power draw in all-core, 100% workloads is quite scary on my 14700 KF (over 300W peak), but I totally agree that this is hardly a measure of how the CPU will perform in the real world. Unless all you do all day is run CPU benchmarks like Cinebench or Blender, I suppose.. :)
Power limits on my board are indeed set to "no limits" at the board's factory settings and that's the whole problem. Also doesn't help that MSI have labeled the three PL-presets in the BIOS in a sightly misleading way. Even the lowest preset will set PL1 and PL2 to 253 W and it's labeled "boxed cooler". I'd like to see any Intel stock-cooler keep up with this CPU at those PL-settings, lol. The "no limits" preset is called "water cooler" or something similar, implying that you can run your CPU without power limits on a 240mm (or even a 120?) AIO. Which I could with my much less power-hungry 12600K but definitely NOT with the 14700KF.
All that said: After having read up on the whole power-limit thing, I adjusted PLs, max current and voltage offset manually and the chip will now run right up to the limit of thermal throttling while drawing ~240W max and giving me the CB23 results you'd expect from an i7. All that on a 240 AiO.
In the real world, the CPU never draws more than 40 or 50 W in more demanding games, sometimes peaking as high as 70 to 100W in certain circumstances. At idle and/or in desktop/browsing mode, it's very frugal and hardly ever jumps over 10W. True idle with just a sensor monitoring app open it'll sit at around 3W all the time.
The AMD minimum clock was 3500mhz on one core, while on Intel it was 1100/800mhz on p/e cores...There's definitely something wrong with the AMD power management, either in windows and or in bios....
BAD Engineering ,,,, but so many MNAY millions believe those AMD shill liars
Exactly
This is a surprise and a very cool discovery. Congratulations, great work !!
Out of curiosity, which power mode have you set in config ? Also, have you tried AMD eco mode ?
Both systems tested in balanced power mode
AMD Eco mode measurements please? Especially with the 7900X, which is the most efficient of all at highest load, according to my calculations.
Two years ago I built a pc for my son with a i5-10400f. I choose this instead of Ryzen 3600 because of the lower temperatures in games and better (in game) performance at the same time. His pc is really dead silent with a FUMA 2 cpu cooler and Nitro 5700xt card at 1440p gaming and most important, it is problems free. I recently bought a 5800x3d for my pc to replace my Ryzen 2700 and now I am in process of RMA it because of whea errors following by restarts. If I could afford it I would replace my system for a 13600k/13700k but, unfortunately I cannot do this right now...
" because of whea errors following by restarts."
I had the exact same issue with my 5950x, only new cpu sample fixed it. Its actually very well known issue and whats bizarre that even with 5800x3d launch amd couldnt manage to fix that. Im going to use my 5950x for a long time, but when the time comes for an upgrade im going intel 100%
Sorry I am a newb to this, what software he is using to monitor CPU wattage usage?
Hwinfo64
I think a better way of measuring power consumption for low power draws is using a current clamp or a power meter, since even in HWinfo64, it tells you lower down that there's a power measurement deviation for ryzen of up to 200% (this is just what I've seen on 5950x), with it regularly being around 150% at around 50W displayed.
but yes, that aside, I do agree with you that amd tends to suffer on desktop for idle power consumption
for a current clamp, they can be had for less than $50 and work by measuring the current passing through the wire, you would just need to clamp it sround the 4 (or 8) 12V wires for the eps plugs
For Ryzen 5000 series it really depends on the CPU and utilization for workload power draw. I can't say fully yet for 7000 series.
But per core power utilization I get more from my 5950x than my 5900x or my 5800x. I think that has to do with the binning.
Both my 5900x and 5800x use a decent more watts per core.
My 5600x is pretty close to the 5950x per core. But the point of that thing is to max frequencies. So a lot gets wasted going those last few hundred.
This is the highest idle measurement I have ever seen on an AMD CPU package. Weird...
Great content mate, happy new year with even more great stuff from you!
can you disable efficiency cores and rerun the tests? I have doubts if they really help and if they are responsible for the economy. thx _o/
Probably this was what Intel was showcasing on their 13th gen keynote when they announced those. This was the type of test that I was looking for when they announced it and it's kinda surprising. I'm pretty sure 14th gen will further improve those E cores efficiency. Hopefully, this applies to laptops too since that's what E cores even in smartphones use. Seems like Intel is slowly tuning its CPU capabilities properly.
Intel's e-cores are similar to mobile Qualcomm arm e-cores as they are in order execution but just like qualcomm are not nearly as efficient as apple out of order efficiency cores in performance per watt.
I doubt it's about efficiency cores, at least for idling. The 12400 idles at 5watts (without e-cores) when the 5600 idles in double digits.
The E-cores is not really "power efficient", but "die space efficient" which allows intel to cram tons of it and catch up to AMD in multi threading performance. My personal test using 13500 shows that Ecores draws 60% less power but 62% slower than Pcores on cinebenchr23.
155watts, 21000 cinebenchR23
99watts, 13800 cinebench R23 ecore disabled
@@PMARC14 Intel is on 10nm tho while Apple is using TSMC's 5nm. Let's see what Intel has once they are on the same node!
@@maxjames00077 well I am saying this based on the snapdragon 8 gen 2 on the same process gen at Apple. It shows that yes apple gets a big lead on process node from its arm competitors, on the same node they are held back by their in order efficiency design (consumes less power but for far less work). I think intel and AMD on a similar mode as Apple could be far stronger, but Apple also makes larger silicon chips to get their efficiency (even discluding accelerators)
I measured it myself. My server with an 11500 uses about 60 watts idle. My Workstation with an Ryzen 3600 uses 100 wats idle. Kinda sucks honestly. Thought about replacing it because of this.
Do you have any experience with the Ryzen 7 5700x CPU? If so, how does it run with the RTX 3060 ti for video editing? Any info you can provide would help me out big time! Thanks for providing great information.
you won't have any issues with that CPU.
@@mariuspuiu9555 Thanks. I was thinking of upgrading to a i7 12700k and 3070 but for $600 more than my current system. I mostly do 4k braw editing on Premiere while starting to learn Davinici. Just so many options in computers, and I don't know what's the best bang for the buck.
@@lochnkey5216 you don't need to be on the bleeding edge to do your work properly. enjoy your very capable and cheaper system. put that money towards more ram and a good GPU (Davinci Resolve uses the GPU a lot, especially the studio version)
Hello. Please advise. at the moment, I have a 10400 and a 6700xt (undervolt). I was assembling a PC to consume less energy, since I live in Ukraine. Should I change to 12400 or 7600 or 7500f? and which of them will consume less? thank you
12400 + get some nvidia gpu for lower power consumption undervolted 4060 goes below 100w
After building a 5th Gen AMD system, I saw that AMD Idle power draw is quite high as well... for a Ryzen 5 5600, I was idling at 80W total system power draw measured from the wall. I was able to reduce power consumption by up to 20W with lowering the FCLK, reducing SOC Voltage to 0.8V, leaving the ram at stock frequency as well as having a -0.1V offset on the CPU ... that's quite a lot of power used at idle.
Did you used PBO2 curve optimizer or an adaptative undervolt?
@@saricubra2867 I believe it's adaptive Undervolt. not sure what Asus Bios settings did I changed. Im still new to the AMD platform.
@@LeafInTea -0.1volts offset looks like an adaptative undervolt to me.
Well done. It has been hard to find the idle power consumptions for the AMD parts. Even on the old Intel parts, idle power consumption in HW info is often in the single digits (9-11w on i7-6700 while typing this comment). I rarely put my computer to sleep as they rarely wake up properly and often leave them on to quickly resume work the next day. The power of the 7950x would be nice for encoding but the high idle power is a concern as that is 90% of the use case.
How about to run both architectures in ECO mode?
Are the 13th gen processors under volt'd or running stock?
Stock
For general use, which cpu produces less heat?
I think I will stick with my Threadripper and maybe add an Arc GPU. I also don't limit myself to just Adobe products, which is clearly optimized for Intel processors.
Isn't about being optimized for Intel processors, they only have better IPC and single core perfomance than your Threadripper.
I use NVENC on the Premier timeline.
Once you use workstation it's hard to return to desktops. Extra PCI express lanes buss, bandwidth, memory capacity, storage throughput.
If you don't need it then desktop is fine irrespective of the brand.
I am used to idle on my pc a lot and seeing usage of 10W while watching youtube or just some light tasks its crazy good.. Intels 13 and 12gen cpus are just really amazing, I am in love with this thing
1.6 to 1.8 Watts on Energy-profile 'enery-saving' for my 13600k with browser and vid open
otherwise 20
get the switch for ya taskbars guys - powerplanswitcher - microsoft store
I’m a gamer and I still love to follow your channel. You talk about a lot of issues that are not covered anywhere else and offer a lot of details regarding testing. Keep up the great work!
gaming and many other applications types have hard spikes in utilization, it is still very very important to get a psu that can handle your full CPU.
Very nice the only thing missing is metrics when undervolted/power limited as it makes quite a big difference between the generations. It puts Intel and AMD on par in workload and you can get 90% of performance at 50% of the power often times. The new generation is quite a bit more efficient as it’s supposed to with that in effect. Idle is still significant as an issue though I hope AMD finds a way to deal maybe using efficiency cores in the future as well as they are already increasing their expertise in chiplets design. That nd gaming benchmarks for power really unfortunate Intel removed AVX 512 instructions because immigration and future gaming is supposed to benefit a lot from that as it etches closer towards simulation
Intel CPU's perform even better when undervolted, I don't see the point of your comment.
Just measure the WATT OUTPUT comming from your wall going into your PC and do THE SAME workflow on every CPU and see how much WATT they ACTUALLY use per WORKFLOW.
idk why everyone says AMD consumes less. Got 7800X3D/ 3090/ 64GB DDR5 with X670E Mainboard and it consumes much more than my previous build (11600k/3090/64GB DDDR4) in idle. Ingame the 7800 does consume way less than intel, but idle tdp (155w with 2 1440p monitors) is wild in my current setup.
Dunno what you are doing but 155 watt is insanely much for your setup. I use 104 watt idle average with 7950x3d (uses more idle, about 8-10 watt), 1080 gtx, single monitor, X670E motherboard and 64gb DDR5. You are either overlocking or disabled some sleep state bios settings, causing massive idle power.
This is with 5 SSD's and 1 harddrive.
Yeah I hate that they use rendering for benchmarks. I want to know how the chip is going to work in the view port, I didn't understand this and the system I built is good, but I would have changed to a better chip. I use blender to design but I'm not rendering that much, I spend more time designing. So what I learned later is single core performance is what I should have been paying attention too. I got an AMD 5800x, rtx 3080 12gb, 64 gb ram, m.2 drive, which isn't bad, but I can bog it down with a lot of modifiers. one good thing is I can still upgrade to a 5950x. I'm just waiting for the prices to drop some more. I think with all that I've learned in the past year I would have gotten an intel chip, they have better single core performance.
and how about the new ryzen 7700, 7900, without the x ? its only 65 tdp?
I don’t think my PC is idle very much. I turn it off when I’m not gaming.
Is main B760i suitable for intel 13700k?
I don't understand these results. Many reviews are showing idle and single-threaded power consumption of AMD and Intel systems to be on a similar level. Some are measuring total system power draw, some CPU draw only.
TechPowerUp review of the Ryzen 7600 shows several applications running below 20 W when measuring CPU power draw. Guru3D shows 73 W system power draw in idle (with a 3090) and 67 W for a 13600K.
Who is lying, or doing their testing wrong?
They are doing the testing wrong.
Yeah if these results are anywhere close to reality, all i can say is my i7 4790k from 2014 is as efficient as modern CPUs....
Especially the idle power draw he fked up pretty hard with his measurements, and i also dont trust the other data he collected since something is pretty off as well.
This is a really interesting test, but I'm trying to decide between the 13600k, 13700k and 13900k. I'm a full time photographer and I'm trying to understand the difference in power consumption and therefore electricity cost would be over a typical year between the 3 CPUs? Is it significant or inconsequential?
You're not going to make a dent to your electricity bill for photography since it's mainly burst CPU usage (culling & photo adjustments). You would have to export ten of thousands of photos daily to even consider power consumption. Which software do you use and are you a high volume photo exporter?
@@Rocky-AR Mostly LR and Photoshop. I don't do high volume work, so most of my time is spent doing compositing and detailed retouching work in PS. I kind of came to a similar conclusion after watching some other videos and doing some maths on it. I bought the i9 13900k in the end.
@@mdhazeldine 13900k isnt a bad choice. Since its a full time job, processing photos will require a lot of cores and high productivity. 13600k is more power efficient though.
I'm a little confused. Doesn't Idle mean the computer is not being used at all; meaning the computer is not running any programs? Your "idle" power consumption shows a program running in the background, which means it wouldn't be idle. It would be running in the background.
No, "idle" can include non intensive apps being run or left open even. Reading a website for example or having an empty scene with blender running, and nothing being rendered. Think of it like starting a car and while its in drive, without pushing down on the gas pedal, the car still moves forward. You can still move from point A to point B. That's kind of what "idle" refers to in the context of the CPU.
so in the end, if we factor gaming into the mix, intel or amd will be more power efficient? both in desktop & laptop. in uk, 5 quid difference isn't a lot annually, but in 3rd world countries, it's a big difference. considering these days we have tons of power hungry features emeging like ray tracing, high refresh rate monitors etc.
This is what I thought as well. Even in gaming, it was really easy to add a negative offset voltage on my 12700k and remove around 50-70w during gaming, and incredible idle power usage.
Meanwhile, everyone was harping about Ryzen's power efficiency, but that is only on full load which is usually only in a short time. Still, there is something about getting almost the same performance at almost half the power usage with the 7950x with the 105 eco mode. At the end of the day, both platforms are really good this generation and choose whichever you think will fit your needs!
AMD's 105w mode consumes 140+ w, so as 65w mode consumes 88w, so it's closer 2/3 of max power usage than half
@@aldisskudra3687 You are right, I was doing a quick google search to find out the max power draw the chip did at stock lol
Most of the issues with zen4 are based on their stock clocks and voltages, once you do a PBO curve or outright lower the non boost clock speeds, most of the idle power consumption numbers go down by a lot, that's pretty much what the zen4 non X SKU's are, idle clocks are almost or are a whole GHz lower than X variants, but turbo speeds are around 200mhz lower or less, zen3 didn't have this issues either out of the box, so the video is mostly flawed since with intel you can do a PL2 to 250W or underclock a zen4 chip and both will consume the same power at idle, but zen4 will keep the crown with higher loads
where would the 7900 non X be on the chart on 9:25 ?
Hello, what's a good program to use to monitor wattage?
HWInfo64
@@theTechNotice Thank you very much!
@@Garshock1 some feedback say it's less accurate for AMD devices. (also potentially because amd motherboard chipset could use also more juice) Use a wattmeter at the wall too, it's affordable.
Interesting! My 12400F idles at 5-10W while my 5600X idles at 15 to 22W, in gaming the 5600X generally uses 5W more, but also performs a bit better. They run tweaked and OCed ram (3700cl14 g1 and 3800cl14 and curve optimizer) and voltages on Ryzen has been tuned lower. 12400F can't be tuned due to locked voltages, but performs good anyways. In idle 12400F drops to 0.65-0.7v, 1-1.1v at load, 5600X drops to 0.85v idle and 1.15-1.22v during load.
What are you using to measure? 5600x idle package W is about 30-40W
@@Arcona hwmon64. Using wallplug measuring actual powerdraw from wall, idle consumption is 55-60W for B550m MB, 3x120mm fans always spinning, 1tb m.2, 3060ti tuf. You may be using high powerplan so it doesnt go to low power state in idle? Mine runs voltage at 850-870mv, clockspeed 200-3600MHz in idle using balanced powerplan.
@@Taraquin83 use hwinfo cpu package power will give accurate read
@@Arcona I did that, in idle it uses 15 to 19W cpu package power.
@@Taraquin83 that's odd. The IO die makes it use more idle than that regardless of core clock. Maybe the undervolt also effects the IO die consumption
how is the power consumption in games?
Since games are more intensive, in general amd consumes less than intel
is it possible to find a i9 13900T ?
I wonder whether the new AMD APUs (8000G) will do better with idle power consumption
Hey, Get an digital PSU like HXi series from Corsair. With digital PSUs you can monitor each cable or socket power consumption.
Apart from power consumption do high temperatures affect life of processor in long run ???
It depends what you mean by hi temps. Anything up to 80 degrees C is no issues whatsover but even at 90 degrees C the CPU should run fine for the life of the CPU as it will never be running at max load 24/7. I have always overclocked and now undervolt as well but during my older days with the likes of a 4790K heavily overclocked, I had no issues over the 5 year run...Good cooling is important and I have been using AIO's for many years now and they do keep the CPU's even at max load/speed Cinebemch 23 at around 70 to 80c on my current 13700K.
@@vMaxHeadroom ok thank you for answer.... i m gonna buy new pc in few days so this was the only thing i was concerned about because i want igpu of intel ........ But seeing high temperature thing about INTEL was confusing me ....
Can the AE crowd request some testing benchmarks as well?
Most of the rendering workload falls on the CPU and Ram for motion design/animation. :)
My 13700K is using 10-15W watching this video. :)
As a German, I can tell you that the fuel consumption of a car at full throttle and top speed absolutely matters for every day life.
:)
I wonder how the efficiency compares between Windows 11 and 10 because apparently the E cores of the newest Intel generation were designed around Windows 11 but not 10. I'd really love to get the benefits of the idle energy efficiency but Win11 is still too awful to be worth switching to.
Unfortunately there's lots of unknowns here and I wish there wasn't.
1. If you used some testing workflows with a fixed duration(?) then depending on the type of test, there could be a problem: The faster CPU might complete more work in the same amount of time than the slower CPU and as expected use more power.
2. guru3d checked total system power consumption of the system measured directly from the wall and the idle system draw with the 7800x3D was 75W while the most efficient modern CPU from Intel, the 12600k, was at 76W. So either there is something wrong with your AMD system or the HWINFO metrics can't be trusted.
Also to quote techpowerup on the subject:
"We're not using the software sensors inside the processor, as these can be quite inaccurate and will vary between manufacturers."
3. Your AMD idle power draw results seem to be insanely high. Something seems wrong when one compares it with other reputable sites' results.
Came to post something similar after CZcams wanted me to watch this. I trust Guru3Ds total system idle power consumption numbers over these.
CPU idle power usage differences are useless if the system draws the same power from the wall anyway.
Are C-states on and all for both sides?
Great video and I would agree for people that solely uses the software tools tested in this video.
That said, if you tune both systems to like 65 watt then the AMD architecture may be much more efficient than intel overall, I would classify applications that use the intel iGP as an outlier since its not a feature that can be used across the board and if this software is all you use then you know your answer already.
(Also in some applications the Intel iGP encoding is not as high quality as just using software encoding)
Both of these architectures are running far outside of their efficiency range so if this is at all important to you then tune your system, undervolt and limit to 65watt and then we see which architecture is the most efficient, also add more software if possible.
Intel clearly has better power management software between applications and at idle so this gives them a massive advantage here.
For myself and others we need to move away from blanket statements like "this architecture is more efficient" or "which ever CPU is better" since these days it REALLY depends on what workloads one has, for how long and with which settings/power tuning.
That's why I'm waiting for Ryzen 7900 (non-X) benchmarks...
AMD consumes 30-40w for moving cursor even on desktop no matter how much you tune. AMD's tuning algorithm is not nearly half as good as intel. speaking from testing two systems.
@@freaky425 i do not own any CPU of the Ryzen 7000 series yet, but my 5800X consumes 5W on idle using Eco mode, and 10W when I keep moving the mouse cursor. My GPU is 6700 XT auto-undervolted, and it's around 8W at idle (right now).
@@SirWolf2018 I actually tested my 5800X3D and RTX3080ti as well.
I'm not using ECO mode and my idle tempts are 20-30 watts (with some stuff open in the background like chrome and discord) for the CPU.
I will test with absolutely nothing open and confirm.
On the GPU it's around 50 watts with chrome and YT open, with spikes of 110 watts at times.
Sadly it seems like the power plans don't affect this at all on my side and the clocks bounce around a lot between 3.3GHz and 4.45GHz.
The 5800X3D is stock with XMP on a Asus B550 Prime board, the 3080ti has a 110core and 1100mem OC.
Might be something wrong with the 7000 series to get 50w idle temps which should trigger a bit of an investigation if you ask me.
@@SirWolf2018 That's some crazy Efficiency if true lol.🤣 (In an all AMD system as well)