AMD Ryzen 8000 "Granite Ridge" Desktop CPUs Could Utilize Same IO Die as Ryzen 7000

[AusWolf]

The above graph is using high performance power plan. In which case yes, Intel draws around 18 to 20w. Balanced power plan drops that to below 3. Amd sadly doesn't drop any further cause of the IO die. If you had the cpus in question, you would know.

Why would anyone test with the high performance power plan? I'm not saying that they don't, but if they do, I'll call bullshit on those graphs. Even on AMD, balanced, and power saver can shave a couple of W off.

 

[Deleted member 232534]

Yes, mobile amd is great, but the desktop isn't. It's not efficient at work either. Unless your work is rendering 24/7 and you have a 7950x, amd is less efficient.

Case in point, my work involves lots of excels, words pdfs and browser tabs. My intel does that at 5 to 15w, no amd cpu can get anywhere near that level.

all i can do is to repeat for the 3rd/4th time - already 7900 is more efficient than your setup. you've spend $1000 and set it up to match ryzen 7700 running on $85 4 vrm board

Idle power consumption varies based on the memory configuration on both Intel and AMD platforms. Hence why you see some variance:

View attachment 311096



And I assume you think the above provided graph is incorrect as well? What evidence do you have that would disprove the results of multiple professional reviews?

yes i think this graphic is wrong for absolute idle and you can set lga1700 to run 15 watt less than am5.

 

[R0H1T]

Idle power consumption varies a lot based on your power plan as well, if you allow the cores to downclock to 1% you can save a lot & if you allow more cores to be parked regularly you save even more! But as usual some like to believe this is such a big deal on AMD, hint ~ it kinda saves you more time on race to idle when running a task

This is why unless you understand why power consumption varies so much from one system to another it's pointless bickering about it, heck using hwinfo to poll at 200ms will spike your idle power consumption a lot more!

 

[Nhonho]

AMD should make their CPUs on a single die only (a SoC); thus, they would perform better because they would have lower communication latencies between all the internal components of the CPU and between the CPU and the RAM memory. A CPU with 2 or more dies greatly increases all latencies.

It would be enough to make a central SoC with 8 cores and add chiplets for CPUs with more than 8 cores.

 

[JustBenching]

Why would anyone test with the high performance power plan? I'm not saying that they don't, but if they do, I'll call bullshit on those graphs. Even on AMD, balanced, and power saver can shave a couple of W off.

I think that's the default windows power plan, no?

all i can do is to repeat for the 3rd/4th time - already 7900 is more efficient than your setup. you've spend $1000 and set it up to match ryzen 7700 running on $85 4 vrm board

View attachment 311100



yes i think this graphic is wrong for absolute idle and you can set lga1700 to run 15 watt less than am5.

According to the graph you just posted, the 13900k is the most efficient cpu, lol. Also the 13600k at 88w is faster than the 7700x at 125

 

[AusWolf]

I think that's the default windows power plan, no?

I think balanced is the default, which also gives you this slider:

The other two plans don't.

Using balanced with the above "best energy savings" option is kind of a must with Ryzen, imo.

 

[close]

With balanced power plan intel 12 and 13th sit at 3w or less. Browsing the web watching YouTube and all that less than 10w.

But guru3d is measuring system power draw, so different mobos

For most use cases (idle probably being somewhat an exception*) there's a lot of practical value in measuring system power draw as long as you try to keep the baseline system as relevant as possible across generations and manufacturers. In the end you need the entire system to operate and that's the baseline of your power consumption. Knowing CPU-specific figures is useful but mostly academic if the MoBo/chipset it has to come with intrinsically gobbles up a lot of power. Especially on platforms like Intel's, with a very short upgrade path of 2 CPU gens at most, so little room to optimize/lower the MoBo's power envelope from one to the next.

Of course, getting *relevant* system power consumption numbers is a whole different ball of yarn even if you make only the absolute minimum number of changes between 2 systems.

*CPU only power numbers are indeed most useful for idle because they affect disproportionately the people who build systems which run non-stop and idle most of that time. They'll target a low power version of every component so having per-component power envelope is super useful.

 

[JustBenching]

I think balanced is the default, which also gives you this slider:
View attachment 311101
The other two plans don't.

Using balanced with the above "best energy savings" option is kind of a must with Ryzen, imo.

I really don't remember what's the default, but I've noticed that on high power plan consumption goes up to 15 - 20w on idle, while on balanced it is below 3w

 

[Onyx Turbine]

No matter the comparisson done (which is very tedious work) is that intel cpu are more efficient in idle and amd wins when gaming with it.
But for a user like me who does most of time stream videos, office work power and leave pc on, idle and basic task energy consumption mattes a lot,
if power scales up when doing a demanding game that is ok, same with a car if you go drive really sporty in this case a triple a game cyberpunk
you now thing need to work

 

[evernessince]

No matter the comparisson done (which is very tedious work) is that intel cpu are more efficient in idle and amd wins when gaming with it.
But for a user like me who does most of time stream videos, office work power and leave pc on, idle and basic task energy consumption mattes a lot,
if power scales up when doing a demanding game that is ok, same with a car if you go drive really sporty in this case a triple a game cyberpunk
you now thing need to work

I just showed two reviews proving otherwise. If someone has proof to the contrary, please link.

 

[JustBenching]

I just showed two reviews proving otherwise. If someone has proof to the contrary, please link.

Sure, here you go, im on a discord call while streaming 2 videos from twitch and youtube and a bunch of tabs on the browser, 5w power draw, 8w average.

Spoiler

 

[AusWolf]

No matter the comparisson done (which is very tedious work) is that intel cpu are more efficient in idle and amd wins when gaming with it.
But for a user like me who does most of time stream videos, office work power and leave pc on, idle and basic task energy consumption mattes a lot,
if power scales up when doing a demanding game that is ok, same with a car if you go drive really sporty in this case a triple a game cyberpunk
you now thing need to work

I just ran a calculation... if your CPU eats 10 Watts more in idle, and you pay £0.3 per kWh like I do, and you run your PC idling 24/7 (never turning it off, and never playing any games), that means you pay £2.23 more per month on electricity. You mitigate this by turning your PC off, and playing games where AMD is way more efficient (not just by 10 W). This idle power consumption matter is way overblown.

Edit: Another fun fact... if your CPU eats 10 W more in idle than another one, but 100 W less in gaming, it means that by playing a game for 1 hour, you compensated for 10 hours of idle power.

 

[JustBenching]

This idle power consumption matter is way overblown.

The power consumption matter in general is way overblown ifyou are talking about power bills. There will be not a single difference in your power bills no matter what cpu you choose.

Think about it, even for gaming where you claim 100 w difference. The tests are done with a 4090. When the GPU alone draws up to 450w - or 500w for the non reference models, that's 550w from the wall just for the GPU alone. Let's say another 100 for the rest of the system besides the CPU, you are already hitting 660w. Add the monitor, which can get up to 100w if you are using a 40" oled like me, now you are close to 800w. Whether your CPU draws 70, 100 or 130 won't make much difference at this point.

Besides, you can power limit any CPU for gaming to whatever you feel comfortable with. On the other hand you CANNOT power limit a ryzen CPU to consume less power at idle and light loads, it will constantly use 5 to 10 times more power than Intel for simple tasks like browsing excel youtube etc.

 

[AusWolf]

The power consumption matter in general is way overblown ifyou are talking about power bills. There will be not a single difference in your power bills no matter what cpu you choose.

Think about it, even for gaming where you claim 100 w difference. The tests are done with a 4090. When the GPU alone draws up to 450w - or 500w for the non reference models, that's 550w from the wall just for the GPU alone. Let's say another 100 for the rest of the system besides the CPU, you are already hitting 660w. Add the monitor, which can get up to 100w if you are using a 40" oled like me, now you are close to 800w. Whether your CPU draws 70, 100 or 130 won't make much difference at this point.

Besides, you can power limit any CPU for gaming to whatever you feel comfortable with. On the other hand you CANNOT power limit a ryzen CPU to consume less power at idle and light loads, it will constantly use 5 to 10 times more power than Intel for simple tasks like browsing excel youtube etc.

Sure, but that "5 times more" means around 15 Watts. If you're not worried of your PC eating 100-200 W more in games, then why would you be worried about 10-15 W in idle?

 

[Cipher908]

But does anyone know what causes a CPU to use more power when idle if the configurations are as close to identical?

 

[AusWolf]

But does anyone know what causes a CPU to use more power when idle if the configurations are as close to identical?

The OS power plan, power saving features being disabled, background programs, memory controller clocks, voltages, and in case of Ryzen, the infinity fabric.

 

[Eskimonster]

Those idle graphs can only be used with a grain of salt.
Power settings in your pc/lap changes the wattage to differ a lot from them graphs.

 

[JustBenching]

Sure, but that "5 times more" means around 15 Watts. If you're not worried of your PC eating 100-200 W more in games, then why would you be worried about 10-15 W in idle?

Im worried about both, but my pc isn't eating 100-200w more in games, which is my point. The 4090 is limited to 320w and the CPU never exceeds 70w in 4k that im playing.

 

[AusWolf]

Im worried about both, but my pc isn't eating 100-200w more in games, which is my point. The 4090 is limited to 320w and the CPU never exceeds 70w in 4k that im playing.

Then you're already saving way more power than an average AMD user would by shaving 10 W off the CPU idle.

My point is that having a several hundred W GPU and then worrying about one's CPU eating 10 W more in idle is just silly. Sure, most home PCs run idle most of the time, but you need 10 hours of idle time to make up for a 100 W difference in GPU (or any other) power. A lot of people don't seem to be bothered by 2-3-400 W GPUs these days, but 10 W more idle CPU power is a catastrophe for some reason.

 

[JustBenching]

Then you're already saving way more power than an average AMD user would by shaving 10 W off the CPU idle.

My point is that having a several hundred W GPU and then worrying about one's CPU eating 10 W more in idle is just silly. Sure, most home PCs run idle most of the time, but you need 10 hours of idle time to make up for a 100 W difference in GPU (or any other) power. A lot of people don't seem to be bothered by 2-3-400 W GPUs these days, but 10 W more idle CPU power is a catastrophe for some reason.

Idle isn't that important I agree, but high idle power also translates to high low load power. Going by tpup numbers, simple tasks have the 7950x not only slower than the 13900k, but also using 30% more power. Check the st mp3 encoding graph from their 13900k review. The difference between the 13700k and the 7900x is even bigger, with the amd part consuming double the power.

 

[AusWolf]

Idle isn't that important I agree, but high idle power also translates to high low load power. Going by tpup numbers, simple tasks have the 7950x not only slower than the 13900k, but also using 30% more power. Check the st mp3 encoding graph from their 13900k review. The difference between the 13700k and the 7900x is even bigger, with the amd part consuming double the power.

No. Just... no.

Here's my 7800X3D while playing Kingdom Come: Deliverance (average CPU usage between 20-25%) - bone stock, EXPO on, PBO on auto:

Also:

Idle, low load and high load power consumption always need to be looked at separately.

 

[JustBenching]

No. Just... no.

Here's my 7800X3D while playing Kingdom Come: Deliverance (average CPU usage between 20-25%) - bone stock, EXPO on, PBO on auto:
View attachment 311248

Also:
View attachment 311247

Idle, low load and high load power consumption always need to be looked at separately.

What framerate are you playing at? Tried the game recently, cpu is at 50-60w at 1080p with a 4090 completely cpu bound. I have a video somerhwdre I can link if you are interested.

 

[R0H1T]

Here's my 7800X3D while playing Kingdom Come: Deliverance (average CPU usage between 20-25%) - bone stock, EXPO on, PBO on auto:

Check you clocks, his power plan settings probably allow downclocks to 10%(?) & you can get close to those idle clocks if you allow that as well ~

This is on an Intel laptop but you can change your power plan setting to get lower idle power probably matching Intel in just a few steps!

 

[AusWolf]

Check you clocks, his power plan settings probably allow downclocks to 10%(?) & you can get close to those idle clocks if you allow that as well ~

This is on an Intel laptop but you can change your power plan setting to get lower idle power probably matching Intel in just a few steps!

My power settings are fine, as far as I know. I've got Steam, GOG Galaxy, the EA app and Wargaming.net running in the background, which keep the CPU alive every now and then. HWinfo doesn't pick up on Ryzen's fast clock changes.

What framerate are you playing at? Tried the game recently, cpu is at 50-60w at 1080p with a 4090 completely cpu bound. I have a video somerhwdre I can link if you are interested.

I lock all my games at 60 FPS. I've just done a quick run with an unlocked frame rate for a fair result: it reached around 110-130 FPS, and the CPU ate 55 W max.

I also recommend looking at the TPU review diagram showing every single Intel CPU except for 12th gen Core i5 eating more power in gaming than the 7950X3D.

 

[R0H1T]

The point is those clocks are not normal, Intel nor AMD downclock to base clocks on desktop under HP or balanced schemes. Unless something changed under ADL or RPL drastically they would not go that low. I don't have a P+E core setup on MSDT so can't say that for sure. So like I said in the post before that you can definitely go lower on AMD but that'd cost you some responsiveness & probably a few benchmark points!