The Intel Problem: CPU Efficiency & Power Consumption
video description
Date: 2023-12-20
Related videos
Comments and reviews: 20
BodySculptTV
I run my 14900K overclocked to 6.1Ghz as a daily driver, in my configuration I enabled C-States with a specific value of C0/C1. Doing so allows the 14900K to drop down all the way to 800Mhz on certain cores and not all the time. In Starfield I can average a clock speed of 5897Ghz with a consumption of about 135w, obviously that's still not as good efficiency as a 7800X3D but I don't think its as bad as the numbers you are getting for the 14900k. I have even tried this configuration on Firestorm Viewer for Second Life/Opensim and at 5897Ghz its only using 75w. It varies greatly from application to application as you mention in this video. But what good is it to have a highly efficient processor when you pair it with something like an RTX4090,which can get up to 427w of consumption in certain games. I understand the need for efficiency but I see this relating more towards a corporate server farm than a homebrew PC aiming to get the highest possible FPS, its a bit silly especially when you toss in the consumption of the high end video card. Someone that is blowing so much money for a high end processor and high end video card won't be derailed by power draw. What I am wondering is will Intel's APO yield not only higher frames but better power usage , this we will have to see.
reply
I run my 14900K overclocked to 6.1Ghz as a daily driver, in my configuration I enabled C-States with a specific value of C0/C1. Doing so allows the 14900K to drop down all the way to 800Mhz on certain cores and not all the time. In Starfield I can average a clock speed of 5897Ghz with a consumption of about 135w, obviously that's still not as good efficiency as a 7800X3D but I don't think its as bad as the numbers you are getting for the 14900k. I have even tried this configuration on Firestorm Viewer for Second Life/Opensim and at 5897Ghz its only using 75w. It varies greatly from application to application as you mention in this video. But what good is it to have a highly efficient processor when you pair it with something like an RTX4090,which can get up to 427w of consumption in certain games. I understand the need for efficiency but I see this relating more towards a corporate server farm than a homebrew PC aiming to get the highest possible FPS, its a bit silly especially when you toss in the consumption of the high end video card. Someone that is blowing so much money for a high end processor and high end video card won't be derailed by power draw. What I am wondering is will Intel's APO yield not only higher frames but better power usage , this we will have to see.
reply
marcustmusic
I knew Intel chips were less power efficient and get hotter but this was embarrassing. I can confirm this with my own testing. I don't own an AMD computer but I do have an Intel Core i3 laptop with 16GB of RAM and a 256GB SSD with a physical bigger battery vs my M1 MacBook Air with only 8GB of RAM and the 256GB SSD and playing Minecraft Java on both, the Intel laptop would get hot and the fans would kick up in the first 10 minutes, sometimes 5 minutes in, while my MacBook playing Minecraft Java non-native, didn't get warm even after an hour. And this was all on battery. I can watch 18 hours straight of YT on my MacBook and maybe 2-3 hours of YT on my Intel laptop. Again, on battery for both of them. And I had an Intel Core i7 tower PC that preformed worse.
I am curious if the Intel Core Ultra chip they introduced a little bit ago can actually preform close to AMD or Apple silicon even if Macs are not really for gaming. I've seen Dave2D's video on the Intel Core Ultra and in his video, it seems the chip was doubled even Intel's best chips in performance.
reply
I knew Intel chips were less power efficient and get hotter but this was embarrassing. I can confirm this with my own testing. I don't own an AMD computer but I do have an Intel Core i3 laptop with 16GB of RAM and a 256GB SSD with a physical bigger battery vs my M1 MacBook Air with only 8GB of RAM and the 256GB SSD and playing Minecraft Java on both, the Intel laptop would get hot and the fans would kick up in the first 10 minutes, sometimes 5 minutes in, while my MacBook playing Minecraft Java non-native, didn't get warm even after an hour. And this was all on battery. I can watch 18 hours straight of YT on my MacBook and maybe 2-3 hours of YT on my Intel laptop. Again, on battery for both of them. And I had an Intel Core i7 tower PC that preformed worse.
I am curious if the Intel Core Ultra chip they introduced a little bit ago can actually preform close to AMD or Apple silicon even if Macs are not really for gaming. I've seen Dave2D's video on the Intel Core Ultra and in his video, it seems the chip was doubled even Intel's best chips in performance.
reply
AceStrife
I really wanted to go back to Intel after getting a 3950x due to platform maturity, but their move to mobile chip design (big/little) was such a terrible fit for desktop, especially games, and the requirement to use Windows 11 sealed that coffin, as I'd rather use Linux (and have finally started learning it).
Then the X3D CPU's shook the industry. And Intel has just been pushing power something fierce to try and play catch-up, which seems to have become the norm nowadays with all PC hardware as Moore's law is going the way of the dodo. Pretty terrible for everyone, really; massive burden on power grids and people's electricity bills, nevermind the heat output.
Intel fast became what AMD was for a decade, and knowing their history of bullying AMD in ye olde days, I'm kinda happy to see it. Still, we need genuine competition (and these two companies keep copying eachother too much), so I hope Intel dramatically improves sooner rather than later. And that X3D CPUs become the de-facto part, deprecating non-X3D.
reply
I really wanted to go back to Intel after getting a 3950x due to platform maturity, but their move to mobile chip design (big/little) was such a terrible fit for desktop, especially games, and the requirement to use Windows 11 sealed that coffin, as I'd rather use Linux (and have finally started learning it).
Then the X3D CPU's shook the industry. And Intel has just been pushing power something fierce to try and play catch-up, which seems to have become the norm nowadays with all PC hardware as Moore's law is going the way of the dodo. Pretty terrible for everyone, really; massive burden on power grids and people's electricity bills, nevermind the heat output.
Intel fast became what AMD was for a decade, and knowing their history of bullying AMD in ye olde days, I'm kinda happy to see it. Still, we need genuine competition (and these two companies keep copying eachother too much), so I hope Intel dramatically improves sooner rather than later. And that X3D CPUs become the de-facto part, deprecating non-X3D.
reply
cmiex
I have one concern with this testing methodology: some stock motherboard settings are just dumb
While browsing on the overclocking reddit, i came across multiple instances where the motherboard dumps an insane amount of voltage into the chip while operating at stock
I know it's still stock and should be tested this way, but it results in a lot of variability depending on the motherboard brand or model/bios.
For example, a 14900k at 57 gHz goes up to 1.4-1.5 v on some boards but can operate at about 1.25-1.35 v at this frequency. This can be the difference of a couple dozen watts while gaming. (Some boards even go for too little voltage, causing crashes)
I know LLC, AC_LL, F/V-curve tuning shouldn't be a requirement to run those chips optimally! But to really compare the efficiency those settings should at least be considered, talked about, not just the power limit.
Nevertheless, it won't change the outcome that AMD is more efficient, that's for sure.
reply
I have one concern with this testing methodology: some stock motherboard settings are just dumb
While browsing on the overclocking reddit, i came across multiple instances where the motherboard dumps an insane amount of voltage into the chip while operating at stock
I know it's still stock and should be tested this way, but it results in a lot of variability depending on the motherboard brand or model/bios.
For example, a 14900k at 57 gHz goes up to 1.4-1.5 v on some boards but can operate at about 1.25-1.35 v at this frequency. This can be the difference of a couple dozen watts while gaming. (Some boards even go for too little voltage, causing crashes)
I know LLC, AC_LL, F/V-curve tuning shouldn't be a requirement to run those chips optimally! But to really compare the efficiency those settings should at least be considered, talked about, not just the power limit.
Nevertheless, it won't change the outcome that AMD is more efficient, that's for sure.
reply
OneBiOzZ
I dont care about power efficiency whatsoever on my desktop. Power is cheap here and its winter anyway.
However i care very much about it on a laptop, i very much like the ability to unplug my laptop from the wall and walk around with it especially at work. This does bring up the question of why AMD laptops are rare and ... rather disappointing. I understand that highly integrated mobile chips is a difficult challenge and something AMD struggles with but we have seen high end laptop manufacturers slap desktop chips along side chipsets and get them to function. I dont even care about x3D and a dedicated GPU, i just want a good powerful portable workstation laptop i can run simulations on and do processing without plugging it in every 20 minutes and configuring a laptop with a 7700 would be a game changer for that.
But that is a niche, however a 7800x3D desktop CPU laptop would be rather nice.
reply
I dont care about power efficiency whatsoever on my desktop. Power is cheap here and its winter anyway.
However i care very much about it on a laptop, i very much like the ability to unplug my laptop from the wall and walk around with it especially at work. This does bring up the question of why AMD laptops are rare and ... rather disappointing. I understand that highly integrated mobile chips is a difficult challenge and something AMD struggles with but we have seen high end laptop manufacturers slap desktop chips along side chipsets and get them to function. I dont even care about x3D and a dedicated GPU, i just want a good powerful portable workstation laptop i can run simulations on and do processing without plugging it in every 20 minutes and configuring a laptop with a 7700 would be a game changer for that.
But that is a niche, however a 7800x3D desktop CPU laptop would be rather nice.
reply
Xeonzs
I really like the 23:00 chart, I'm in the EU, NL, and my main rig has a 14700k, it's shocking that I can save 100-150 euro a year assuming I game 8 hours a day, changing to AM5 and getting a 7800X3D the CPU would pay itself back in energy savings in just 2 years and 4 months, that's absolutely bonkers.
I already regretted getting a intel 1700 mobo cause the e-cores cause so many shenanigans with games (worse frame times, worse cpu scheduling) the only reason I really got it was cause of DDR5 and the bigger L1 and L2 cache even compared to AM5, but I've been thinking of selling that machine for a while now and going AM5 instead, to also future-proof myself.
reply
I really like the 23:00 chart, I'm in the EU, NL, and my main rig has a 14700k, it's shocking that I can save 100-150 euro a year assuming I game 8 hours a day, changing to AM5 and getting a 7800X3D the CPU would pay itself back in energy savings in just 2 years and 4 months, that's absolutely bonkers.
I already regretted getting a intel 1700 mobo cause the e-cores cause so many shenanigans with games (worse frame times, worse cpu scheduling) the only reason I really got it was cause of DDR5 and the bigger L1 and L2 cache even compared to AM5, but I've been thinking of selling that machine for a while now and going AM5 instead, to also future-proof myself.
reply
deRNmEpRrMm
It wasn't really mentioned here, or I missed it, but I think it's important to note that the CPU power consumption of a locked frame rate might theoretically not always be the same.
For example a VSync locked frame rate, compared to a Nvidia control panel locked frame rate, compared to an inbuilt FPS limiter in the game(-engine) could theoretically be different. By how much I don't know.
It's also important to note that some parts of a game, like the physics, run at a fixed update rate of about 60-200 Hz depending on game and what the developers chose for it. This won't impact the results but is good to know as an audience member I'd say.
reply
It wasn't really mentioned here, or I missed it, but I think it's important to note that the CPU power consumption of a locked frame rate might theoretically not always be the same.
For example a VSync locked frame rate, compared to a Nvidia control panel locked frame rate, compared to an inbuilt FPS limiter in the game(-engine) could theoretically be different. By how much I don't know.
It's also important to note that some parts of a game, like the physics, run at a fixed update rate of about 60-200 Hz depending on game and what the developers chose for it. This won't impact the results but is good to know as an audience member I'd say.
reply
HackedGlitch265
So at this point, Intel's losing on most fronts.
But would it be inaccurate to say this could be due to the node size used to produce the CPUS?
Unless I'm mistaken, AMD is using a much more advanced and smaller node than Intel is, which typically helps in terms of energy consumption?
If AMD's running 5nm, but Intel is using, IDK, 7nm, theoretically Intel could regain the lead in efficiency by gaining node size parity with AMD? Obviously not something we could test, and if Intel still keeps chugging all the power on the chips, it's unlikely to change, BUT this still seems like Intel's greatest weakness at this point.
reply
So at this point, Intel's losing on most fronts.
But would it be inaccurate to say this could be due to the node size used to produce the CPUS?
Unless I'm mistaken, AMD is using a much more advanced and smaller node than Intel is, which typically helps in terms of energy consumption?
If AMD's running 5nm, but Intel is using, IDK, 7nm, theoretically Intel could regain the lead in efficiency by gaining node size parity with AMD? Obviously not something we could test, and if Intel still keeps chugging all the power on the chips, it's unlikely to change, BUT this still seems like Intel's greatest weakness at this point.
reply
Leo99929
Every excess watt is one more we need to provide carbon neutral generation for in order to avoid burning fossil fuels which threaten humanities continued enjoyable existence. You don't need to get the efficient CPU for the money savings, or because your specific area is clean electricity powered. Doing it because it lowers total demand is enough. These are the tiny individual choices that add up to save the species. It really is that simple. Use less; lower demand; less fossil fuels burned directly or indirectly; less emissions; less extreme death and discomfort for humanity. Efficiency matters.
reply
Every excess watt is one more we need to provide carbon neutral generation for in order to avoid burning fossil fuels which threaten humanities continued enjoyable existence. You don't need to get the efficient CPU for the money savings, or because your specific area is clean electricity powered. Doing it because it lowers total demand is enough. These are the tiny individual choices that add up to save the species. It really is that simple. Use less; lower demand; less fossil fuels burned directly or indirectly; less emissions; less extreme death and discomfort for humanity. Efficiency matters.
reply
nt78stonewobble
Regarding gaming. Personally I don't really care so much whether the current most popular competitive online game gets 340 or 360 fps.
I'd rather know how a cpu/gpu handles eg. late game Stellaris or other strategy game lag, Ark survival evolved, Conan exiles, 7 days to die giant bases, hosting servers for such games or heavily modded games like that.
So to me, minimum FPS in heavy use scenarios is more important and whether there are weird bottle necks, that make a system run much worse.
I find it hard to extrapolate that relevant information, from more generic bench marks.
reply
Regarding gaming. Personally I don't really care so much whether the current most popular competitive online game gets 340 or 360 fps.
I'd rather know how a cpu/gpu handles eg. late game Stellaris or other strategy game lag, Ark survival evolved, Conan exiles, 7 days to die giant bases, hosting servers for such games or heavily modded games like that.
So to me, minimum FPS in heavy use scenarios is more important and whether there are weird bottle necks, that make a system run much worse.
I find it hard to extrapolate that relevant information, from more generic bench marks.
reply
gamersnexus
Steve, you should really look into Igor's Lab's fan testing. Not just the Alphacool Apex but there seem to be more misleading info in previous fan tests. For example , 60 mm radiator results in higher airflow than 25 and 40 mm rads??! Not just in the Apex fan test... I think they have big big problems with their equipment and somebody hint hint should look into it. Der8auer have been criticizing them but we need an overall view of the scope of this and who is better at that than GN. Can we even trust Igors lab at all, for any fan tests, or ANY test whatsoever??
reply
Steve, you should really look into Igor's Lab's fan testing. Not just the Alphacool Apex but there seem to be more misleading info in previous fan tests. For example , 60 mm radiator results in higher airflow than 25 and 40 mm rads??! Not just in the Apex fan test... I think they have big big problems with their equipment and somebody hint hint should look into it. Der8auer have been criticizing them but we need an overall view of the scope of this and who is better at that than GN. Can we even trust Igors lab at all, for any fan tests, or ANY test whatsoever??
reply
JikoMuskato
Yeah, greetings from Germany: power consumption really is relevant here. And not just because more power consumption also means you need better (=more expensive) cooling solutions to keep it under control (so while the 7800X3D could be cooled by a mid-class tower cooler you need a really beefy water cooler to keep the high-end Intel CPUs under control when run without a power limit). No, not just that but you could actually pay a lot for your power you could use for new hardware instead.
reply
Yeah, greetings from Germany: power consumption really is relevant here. And not just because more power consumption also means you need better (=more expensive) cooling solutions to keep it under control (so while the 7800X3D could be cooled by a mid-class tower cooler you need a really beefy water cooler to keep the high-end Intel CPUs under control when run without a power limit). No, not just that but you could actually pay a lot for your power you could use for new hardware instead.
reply
c.m.z7213
Is there a gamer here who cares about power consumption?
For players, the only important thing is overclocking and high frame rate
CPU power consumption is ridiculous for someone who plays from morning till night
A power supply 1200w platinum!! I don't think these things are important for this power supply
I can't tell myself how good the processor is, it consumes 86 watts and its performance is good
Don't spend time on these things and don't make fun of yourself!
reply
Is there a gamer here who cares about power consumption?
For players, the only important thing is overclocking and high frame rate
CPU power consumption is ridiculous for someone who plays from morning till night
A power supply 1200w platinum!! I don't think these things are important for this power supply
I can't tell myself how good the processor is, it consumes 86 watts and its performance is good
Don't spend time on these things and don't make fun of yourself!
reply
heyitsjel
Great video; no doubt AMD are currently more efficient than Intel. However, I think it would be worth measuring TOTAL system power draw from the wall - keeping things like GPU/RAM/HDD constant. A CPU can't run independently of it's motherboard, so I think both should be measured to get a true idea of each platform . Also, it's worth looking at IDLE power consumption for each platform as well - as most users' systems spend a large portion of their days at IDLE, or close to.
reply
Great video; no doubt AMD are currently more efficient than Intel. However, I think it would be worth measuring TOTAL system power draw from the wall - keeping things like GPU/RAM/HDD constant. A CPU can't run independently of it's motherboard, so I think both should be measured to get a true idea of each platform . Also, it's worth looking at IDLE power consumption for each platform as well - as most users' systems spend a large portion of their days at IDLE, or close to.
reply
nach1113
is Efficiency even worth for Desktop? Average user couldnt care less for desktop power effiency, raw performance is much more important.
Doing this kind of analysis on Laptops does way more sense. For desktop its a waste of time and negligible for average joe.
If you trully want to do something good for the environment, average joe's footprint is worthless in comparison to corporate contamination, so start your efforts there. Don't let them throw that ball on joe.
reply
is Efficiency even worth for Desktop? Average user couldnt care less for desktop power effiency, raw performance is much more important.
Doing this kind of analysis on Laptops does way more sense. For desktop its a waste of time and negligible for average joe.
If you trully want to do something good for the environment, average joe's footprint is worthless in comparison to corporate contamination, so start your efforts there. Don't let them throw that ball on joe.
reply
libaf5471
Amd and intel seem to operate from an entirely different design philosophy. Zen has gotten progressively more efficient over time, even within the same socket and during the + architectures (zen1->z1+, zen2->z3 -> jump to zen 4). Intel tends to start at some kind of efficiency target and then spends time squeezing every single watt out of the architecture until it stops scaling. Imo only one of these philosophies has a future and it's not intel's.
reply
Amd and intel seem to operate from an entirely different design philosophy. Zen has gotten progressively more efficient over time, even within the same socket and during the + architectures (zen1->z1+, zen2->z3 -> jump to zen 4). Intel tends to start at some kind of efficiency target and then spends time squeezing every single watt out of the architecture until it stops scaling. Imo only one of these philosophies has a future and it's not intel's.
reply
stewenw4120
First. I would really appreciate when these efficency charts become a regular for future reviews and even older CPUs. Efficency and raw power are both important figures but eypecially in europe eff. has become more relevant. For games to test on i would prefer Cyberpunk and Baldurs Gate 3 because they will still be relevant in the next years and u have one very demanding and one well optimised title.
reply
First. I would really appreciate when these efficency charts become a regular for future reviews and even older CPUs. Efficency and raw power are both important figures but eypecially in europe eff. has become more relevant. For games to test on i would prefer Cyberpunk and Baldurs Gate 3 because they will still be relevant in the next years and u have one very demanding and one well optimised title.
reply
KingsKirch
the way i see it more consumption also means more heat and therefore more noise, which is a thing almost everybody cares about to at least some degree, so there is definitely more to it than just the cost of increased power consumption itself, which as long as you look at a single entity is ridiculous to worry about if you're rocking a 650 CPU (my opinion as German paying well >0,4 kw/h)
reply
the way i see it more consumption also means more heat and therefore more noise, which is a thing almost everybody cares about to at least some degree, so there is definitely more to it than just the cost of increased power consumption itself, which as long as you look at a single entity is ridiculous to worry about if you're rocking a 650 CPU (my opinion as German paying well >0,4 kw/h)
reply
gamersnexus
If I can make a humble suggestion: you're referring to the voltage/frequency curve a few times in this video. It would be interesting to actually plot the curve, then find the sweet spot for the CPUs (7800X3D vs. 14700K would probably be most interesting) on that curve and compare the performance at that particular point. This would put a figure on the maximum achievable efficiency.
reply
If I can make a humble suggestion: you're referring to the voltage/frequency curve a few times in this video. It would be interesting to actually plot the curve, then find the sweet spot for the CPUs (7800X3D vs. 14700K would probably be most interesting) on that curve and compare the performance at that particular point. This would put a figure on the maximum achievable efficiency.
reply
gamersnexus
I'm legitimately amused seeing the i3 12100F breaking AMD levels of power efficiency. Seems like that, and the i5 13400F, are the last great Intel CPUs for general use. Truly an amazing value option if you're really scraping every dime you find inside the couch, would actually recommend for 3rd world gamers out there. Intel's really been falling off hard on their high-end products.
reply
I'm legitimately amused seeing the i3 12100F breaking AMD levels of power efficiency. Seems like that, and the i5 13400F, are the last great Intel CPUs for general use. Truly an amazing value option if you're really scraping every dime you find inside the couch, would actually recommend for 3rd world gamers out there. Intel's really been falling off hard on their high-end products.
reply
Add a review, comment
Other channel videos
