Hmm.... there seems to be different results in different tests.
1. From this one here which igor_kavinski linked to previously, I get the impression that 7950X gets 51.46 and 7950X3D gets 49.90 (so about equally as fast) when both are running at their default power setting (i.e. 7950X3D @ 120W and 7950X @ 170W):
AMD Ryzen 9 7950X3D Linux Performance Review - Phoronix
www.phoronix.com
View attachment 83957
So if the power setting for 7950X is changed from 170W to 142W, I'd assume it'll have lower performance than 7950X3D at 120W?
2. But then there is the test which you linked to, where 7950X and 7950X3D gets roughly the same performance when operating at the same TDP (e.g. performance index 93 vs 91 when both @ 142W, and 80 vs 82 when both @ 88 W). Extracted from this diagram:
Ryzen 9 7950X3D im Test gegen 5800X3D und 13900K(S): Leistung in Anwendungen, Einrichtung und Fazit
AMD Ryzen 9 7950X3D im Test: Leistung in Anwendungen, Einrichtung und Fazit / Leistung und Effizienz in Anwendungenwww.computerbase.de
Just so that I understand this correctly: When 7950X and 7950X3D use the same power setting (e.g. both at 120W), their performance is almost the same (within ~2 index points according to the Computerbase article for 88W and 142W). In that case, what's the point of paying more for 7950X3D? Is all that extra cache useless, or only beneficial for some specific gaming use cases?
It seems some here may be confusing power limits and TDP (thermal design power). The two are NOT the same. AMD uses PPT (package power tracking) to describe power limits. Take the TDP and multiply it by 1.35 to calculate PPT.7950X3D seems to consume less total energy for completing a task though (even when 7950X3D @ 120W is compared to 7950X power limited to 105W):
Ryzen 7950X3D review: An expensive but incredibly efficient 16-core CPU
AMD's second run at a 3D V-Cache CPU is easier to recommend than the first.arstechnica.com
It seems some here may be confusing power limits and TDP (thermal design power). The two are NOT the same. AMD uses PPT (package power tracking) to describe power limits. Take the TDP and multiply it by 1.35 to calculate PPT.
As mentioned in this post, 7950X with ASROCK TAICHI X670E with DDR5-5200 (JEDEC timings. don't overclock) would give you total system power consumption of 62W at idle. It may go maybe a watt or two lower with Hynix P31 2TB SSD (most power efficient PCIe 3.0 SSD). I'm not sure if the PCIe 4.0 P41 consumes more power than the P31.Is there any specific Motherboard + RAM + AMD Zen4 CPU + SSD combination you know will have a total idle system power consumption of ~20W?
Ok, but 62W at idle seems high considering that eek2121 mentioned in a previous post that he knew about systems with 20W total system power consumption at idle. That's why I wondered if he could provide some example(s) of specific parts for a "Motherboard + RAM + AMD Zen4 CPU + SSD" system which would only consume ~20W at idle.As mentioned in this post, 7950X with ASROCK TAICHI X670E with DDR5-5200 (JEDEC timings. don't overclock) would give you total system power consumption of 62W at idle. It may go maybe a watt or two lower with Hynix P31 2TB SSD (most power efficient PCIe 3.0 SSD). I'm not sure if the PCIe 4.0 P41 consumes more power than the P31.
In that case, you would be limiting yourself to a single CCD CPU like 7700X or 7800X3D.That's why I wondered if he could provide some example(s) of specific parts for a "Motherboard + RAM + AMD Zen4 CPU + SSD" system which would only consume 20W at idle.
That s the most reasonable for whom doesnt need a 16C throughput, the 7700/7700X/7800X3D IOD consume significantly less that the one in the 7950X, beside a 7700 can use a cheap 620 chipset MB wich is surely less power hungry than any other AM5 chipset based MB, there s also some 620 MB that support all X3D.In that case, you would be limiting yourself to a single CCD CPU like 7700X or 7800X3D.
This A620 mobo supports the 7950X: https://www.anandtech.com/show/1892...board-supports-ryzen-r9-7950x-r9-7950x3d-cpusThat is unless the former assumes A620 chipset (which cannot be used with 7950X since A620 is restricted to 65W TDP CPUs)
According to the chart below the idle power consumption difference is not that big between single vs dual CCD:
E.g. AMD 7700 is 21W and 7950X is 24W. So only 3W difference.
So if it's possible to build a system using 7700 that idles at 20W, it should be possible to build one with 7950X that idles at ~23W? That is unless the former assumes A620 chipset (which cannot be used with 7950X since A620 is restricted to 65W TDP CPUs), and the power consumption difference between A620 and B650 is big?
This A620 mobo supports the 7950X: https://www.anandtech.com/show/1892...board-supports-ryzen-r9-7950x-r9-7950x3d-cpus
And with this 1.1V future RAM kit, you can have both high speed of DDR5-6000 and lower power consumption: https://www.anandtech.com/show/18988/teamgroup-unveils-jedec-spec-ddr5-6400-kits-faster-11v-memory
At computerbase they measured 12.2/13.9W for the 7700X and 13.7/16.3W for the 7950X at stock settings with Asus and Gigabyte MBs respectively, the 7950X with DDR6000 jump to 19.4/ 21.2W in these same MBs.
So RAM speed overclock has the bigger effect and MB is less important, at least for thoses two MBs and when it comes to CPU idle power, dunno for the rest of the MB power.
That being said with half the core count the 7700(X) doesnt need as much RAM BW as the 7950X.
AMD Ryzen 7000 im Test: So schnell sind 7950X und 7700X: Leistungsaufnahme und Effizienz
Ryzen 9 7950X & Ryzen 7 7700X im Test: Leistungsaufnahme und Effizienz / TDP und PPT steigen deutlichwww.computerbase.de
But it's not just the chipset that differs in that test, it's also different motherboards. So it's hard to know what idle power consumption difference can be attributed to the A620 vs B650 chipset specifically.B650 consumes less than A620 here.
It's a trade-off if you go with Intel. You do get the lowest idle watts but pay with insanely high all core power consumption, unless you rein it in with power limits but then performance also gets reduced.
Try Asrock, they seem to have Zen 4 figured out. Also important to try to keep vSOC <1.20V. It’s directly related to vSOC.Is there any specific Motherboard + RAM + AMD Zen4 CPU + SSD combination you know will have a total idle system power consumption of ~20W?
Does 13900K also have a "sweetspot" in that power limit range (i.e. much lower power consumption but without much lower performance)?
A monolithic CPU always consume lower idle power, same with AMD APUs.I noticed Intel 12700K was only consuming 6.7W at idle in that same test according to computerbase. That's quite impressive. The CPUs from AMD were ~200-300% higher than that (13.5W to 21.2W to be exact).
I wonder what Intel is doing differently to get the idle CPU power consumption so low?
Yes, that could also be an option. But how would 13900K perform against 7950X / 7950X3D if all are power limited to e.g. 120W TDP? E.g. with regards to multicore performance and perf/watt. Does 13900K also have a "sweetspot" in that power limit range (i.e. much lower power consumption but without much lower performance)
A short summary would be that Intel's typical desktop chips are shared with mobile, whereas AMD's typical desktop chips are shared with servers. AMD does have mobile chips on desktop as well (those with G suffix) but they don't have as many cores and no V-cache.I wonder what Intel is doing differently to get the idle CPU power consumption so low?
Hmm.... e.g. the Asrock Taichi X670E idles at 62W according to the post below, which is quite high. The system consisted of AMD 7950X, 32GB DDR5-5200, nVidia RTX 3080 FE, and SilverStone DA850 PSU. The idle power consumption due to the GPU should be deducted from the 62W to make a fair comparison. But I don't think it'll reach anywhere near 20W total system idle power consumption anyway.Try Asrock, they seem to have Zen 4 figured out. Also important to try to keep vSOC <1.20V. It’s directly related to vSOC.
As mentioned in this post, 7950X with ASROCK TAICHI X670E with DDR5-5200 (JEDEC timings. don't overclock) would give you total system power consumption of 62W at idle. It may go maybe a watt or two lower with Hynix P31 2TB SSD (most power efficient PCIe 3.0 SSD). I'm not sure if the PCIe 4.0 P41 consumes more power than the P31.
A short summary would be that Intel's typical desktop chips are shared with mobile, whereas AMD's typical desktop chips are shared with servers. AMD does have mobile chips on desktop as well (those with G suffix) but they don't have as many cores and no V-cache.
I haven’t looked to see how they got those power figures but the only way to really isolate the CPU power usage is by measuring it at the EPS. Also, If it’s using DDR5-5200 it’s not going to be representative of a real-world scenario running DDR5-6000.
View attachment 83955
Here, the 7950X idle is only 3 watts higher than 13900K. So I guess yes, it depends on mobo. So if you go with the mobo in this review, 7950X should be your choice. 13900K is again the worst choice ever. 633W watts just for a CPU intensive workload! And the CBR23 figure is more than 100W extra! Yabba dabba YIKES!
As long as it's a chiplet arch, it'll never have the same idle power usage. As explained above, it's not really something that has an effect in real world scenarios.Ok, that reasoning kinda makes sense.
But is it known what the actual technical root causes are for the lower idle power consumption on the Intel desktop CPUs? In what way are they designed differently (e.g. more quickly ramps down frequency when idle, has better power gating for CPU blocks, or something else)?
And is there any reason AMD could not use similar technology to bring down the idle power consumption on desktop AMD CPUs, or would that result in some drawbacks (which e.g. could affect the server CPUs with which the CPU arch is shared negatively)?
With a system consisting of MB + CPU + RAM + SSD, the total idle power consumption will not be anything near 70-84W (for a system without GPU, tuned for low idle power consumption). Some in this thread are even claiming ~20W is possible. And if we're at such low numbers, the difference between 6W vs 20W for the CPU idle power consumption alone will be substantial.Unless you turn everything off you need to measure total system power consumption including peripherals and monitor, not just the CPU. The difference between 6W and 20W idel CPU might seem a lot, but if total idle is 70W and 84W then the difference is not that much.
So do you mean that the main/only reason for the lower idle power consumption on desktop Intel vs AMD CPUs is that AMD uses a chiplet design? From previous posts, there was still shown a difference between Intel CPUs vs single CCD/chiplet AMD CPUs.As long as it's a chiplet arch, it'll never have the same idle power usage. As explained above, it's not really something that has an effect in real world scenarios.
Yes, the best way to measure the effect it has would be to look at mobile chips since that's an environment where both the Intel and AMD platforms are optimized for low power / battery life.So do you mean that the main/only reason for the lower idle power consumption on desktop Intel vs AMD CPUs is that AMD uses a chiplet design? From previous posts, there was still shown a difference between Intel CPUs vs single CCD/chiplet AMD CPUs.
Also, what do you mean that it does not have any effect in real world scenarios?
But do you mean the AMD chiplet design has a disadvantage w.r.t. idle power consumption even when a single chiplet/CCD is used? What would be the reason for that?Yes, the best way to measure the effect it has would be to look at mobile chips since that's an environment where both the Intel and AMD platforms are optimized for low power / battery life.
Jarrod's Tech review compared a 13980HX against a 7945HX in the same chassis, the 13980HX got 263 minutes of battery life while the 7945HX managed 158 minutes. The 7945HX was markedly more efficient under load, but for basic tasks such as playing a youtube video the Intel platform got 66% more battery life.