[SOLVED] Is there any difference in peak power consumption of same generation i3 and i5 having same tdp value???

Nikhilniks

Prominent
Apr 6, 2021
61
2
545
If for two different processor of same generation (ex 1115g4 @28watt tdp and 1135g7 @28watt tdp) have same tdp. Will the peak power consumption by both of these processor will be equal or different?????
 
Solution
TDP is base clocks, not turbo clocks. So an i3 at 3.1GHz will be roughly the same power draw as an i5 at 3.1GHz. But if the i3 can turbo to 4.6GHz and the i5 can only turbo to 4.2GHz, the peak power draw will be different, the i3 will pull slightly more power.

Turbo limits have nothing to do with TDP limits. So when the i3 and i5 both say 28w TDP, that's at slow clocks, the higher the turbo, the higher the power required for every 100MHz difference. And it will be well above 28w.

I9-9900k is 95w TDP.
I7-9700k is 95w TDP.
I9 can pull upto 250ish watts with locked core turbo settings (basically i9-9900KS)
I7 will pull @ 200w with the same settings.
No OC other than locking the cores at turbo speeds. @ 50w difference with the...

Karadjgne

Titan
Ambassador
TDP is base clocks only. Afaik, it based on the i5 cpus since all the architecture is the same, Intel just disables cores or unlocks hyperthreading ability, or not, depending on the designation of the cpu.

But it's based on the base speeds. So technically an i5 might use pretty close to TDP, an i3 less, and an i7 right at, but, that doesn't account for turbo as technically that's a factory applied OC.

So define peak? Peak with turbo? Peak with just base? Peak with unlocked bios/cpu?

TDP has been pretty much relegated to a basis for which cpu cooler design has something to go on, so a 28w TDP cpu should be easily kept within limits by a 40w cooling solution.

There's also a very sizable difference between temporary peak, like with PL2 turbo wattage, which lasts 28 or 56 seconds, and sustained peak which is infinite after Tau expirery.

So it's quite within the realm of possibility that a 28w TDP i5 cpu will hit 50w for 28 seconds, then drop to a continuous 28w output. Or an i3 might hit 45w for 28 seconds, then drop to 28w continuous.
 
Last edited:
  • Like
Reactions: Nikhilniks

Nikhilniks

Prominent
Apr 6, 2021
61
2
545
TDP is base clocks only. Afaik, it based on the i5 cpus since all the architecture is the same, Intel just disables cores or unlocks hyperthreading ability, or not, depending on the designation of the cpu.

But it's based on the base speeds. So technically an i5 might use pretty close to TDP, an i3 less, and an i7 right at, but, that doesn't account for turbo as technically that's a factory applied OC.

So define peak? Peak with turbo? Peak with just base? Peak with unlocked bios/cpu?

TDP has been pretty much relegated to a basis for which cpu cooler design has something to go on, so a 28w TDP cpu should be easily kept within limits by a 40w cooling solution.

There's also a very sizable difference between temporary peak, like with PL2 turbo wattage, which lasts 28 or 56 seconds, and sustained peak which is infinite after Tau expirery.

So it's quite within the realm of possibility that a 28w TDP i5 cpu will hit 50w for 28 seconds, then drop to a continuous 28w output. Or an i3 might hit 45w for 28 seconds, then drop to 28w continuous.
What i want to know is the peak power consumption at turbo will be same or not???? I have did stress test on 1115g4 @28watt tdp (i3). The CPU max power package there was around 16watt.
When i stressed 1135g7 (i5)@tdp 28watt then the CPU max power package shown there was 26watt.
And in both scenario their clock speed reached to max.
From the specs sheet of 1135g7 (i5) I have seen that PL1 as 22W for 28sec and PL2 as 51W for 2.44milisec.

And what if we limit the clock speed upto its base clock only (i.e there is no turbo boost). Now on that condation will the peak power consumption will be same or different of i3 (1115g4 base clock@3Ghz tdp@28w) and i5 (1135g7 Base clock @2.4ghz tdp@28w)?????
 

punkncat

Polypheme
Ambassador
In practice, TDP is often a made up number that the manufacturers use. Tests have shown that it is VERY rare for a CPU not to go over TDP for desktop PC.

Even if the TDP class is the "same" a chip with more cores/threads is going to draw more power than the lesser.
 
  • Like
Reactions: Nikhilniks

Nikhilniks

Prominent
Apr 6, 2021
61
2
545
The TDP 28 watt specification in theoretical.

In practical terms....both CPUs should probably peak close to each other.
asking bcoz when i stressed ani3 and i5 of same generation then the power package of i5 is nearly 1.5times of i3 at its peak. And hence found same tdp has very less significance in terms of actual heat generation and power consumption for same generation of processor.
 

Nikhilniks

Prominent
Apr 6, 2021
61
2
545
In practice, TDP is often a made up number that the manufacturers use. Tests have shown that it is VERY rare for a CPU not to go over TDP for desktop PC.

Even if the TDP class is the "same" a chip with more cores/threads is going to draw more power than the lesser.
So, does it means even if the tdp is same for same genration then the more core processor will genereate more heat????
 
  • Like
Reactions: punkncat

punkncat

Polypheme
Ambassador
So, does it means even if the tdp is same for same genration then the more core processor will genereate more heat????

The thing to keep in mind here is that TDP is not a reliable number to use as the ONLY measure of heat.
There is a fellow on YouTube (and others) called Buildzoid (I think), but the guy does tremendous breakdowns of motherboard quality and manufacture, especially delving into power delivery.
So, for instance you might have a (pulling out of my butt some #'s) a 65W 4 core i3 that pulls 20A from the motherboard. Then, for instance, you have a 65W 8 core i7 that pulls 40A from the motherboard. (once again, I MADE those #'s up). The CPU with more cores, with more amperage draw is going to require more heat sink to displace heat.

On that same subject, have you ever actually looked at and weighed the various coolers that come stock with (for instance) a R3/i3 VS it's counterpart in the R7/i7?
AMD, for their part, have slightly different naming schemes for their "stock" coolers in relation to what heat they can dissipate.

It's super confusing as well because occasionally you will run across a cooler that states it's designed for "up to" a certain TDP. Then you have to sort of magic up an idea of what the REAL power draw of your CPU is. This is commonly found in more in depth review.

So TLDR; higher line CPU are going to require larger coolers.

Along that same line (again) and back to discussing Buildzoid. He will often plainly state that "such and such" board VRM would be good up to an i5 (for instance) but would have to consider something with better VRM to reliable handle the draw of an i7/i9. When he discussed OC chipsets it becomes even more important.
 
  • Like
Reactions: Nikhilniks
If for two different processor of same generation (ex 1115g4 @28watt tdp and 1135g7 @28watt tdp) have same tdp. Will the peak power consumption by both of these processor will be equal or different?????
They'll be different. TDP is simply a requirement for how much thermal energy a cooler has to dissipate for safe operation of the CPU. Intel applies a blanket TDP value to simplify things.

If you want to know peak power consumption, you're going to have to look at reviews.
 
TDP is heat dissipation (on HIS for given CPU) and is meant as a guidance for CPU cooling solution. Specified TDP is specified at sustained load, where electrical CPU parameters are kept within specifications (means, no OC whatsoever).
TDP shouldn't be used as measure of CPU power consumption, even usually it's close to that. For example, there are two CPU's having equal power consumption, however one can have lower TDP -that would be the one that has bigger heat dissipation area.

So, does it means even if the tdp is same for same genration then the more core processor will genereate more heat????
No, it won't... or let's say, differences will be relative small. Let's take a 8 core CPU having 105TDP, for example. If only one core is fully loaded, then that core can consume (say) 15W. That would make us assuming 8 cores would consume 15x8=120W. But that's not the case, because other parameters kick in (voltage/current limits) to prevent overheating -and so, every core would use much less that 15W. That also explains why 8 cores isn't 8-times faster than 1 core.
How about 6 vs 8 cores (equal TDP) at full load.. which will be hotter? Obviously 8 cores will consume somewhat more power than 6 cores and so, that will also generate more heat. But keep in mind 8 cores area is larger than 6 cores and so heat will dissipate on larger area too. Result is, same cooler can be used -that also explains why both CPU's have same TDP.
 
  • Like
Reactions: Nikhilniks

Karadjgne

Titan
Ambassador
TDP is base clocks, not turbo clocks. So an i3 at 3.1GHz will be roughly the same power draw as an i5 at 3.1GHz. But if the i3 can turbo to 4.6GHz and the i5 can only turbo to 4.2GHz, the peak power draw will be different, the i3 will pull slightly more power.

Turbo limits have nothing to do with TDP limits. So when the i3 and i5 both say 28w TDP, that's at slow clocks, the higher the turbo, the higher the power required for every 100MHz difference. And it will be well above 28w.

I9-9900k is 95w TDP.
I7-9700k is 95w TDP.
I9 can pull upto 250ish watts with locked core turbo settings (basically i9-9900KS)
I7 will pull @ 200w with the same settings.
No OC other than locking the cores at turbo speeds. @ 50w difference with the Hyperthreading, and 100-150w difference because of clock speeds per core.

Not even remotely close to identical wattage, even with identical TDP, since effectively the i9 and i7 are the same exact 8core cpu, one just has hyperthreading unlocked.
 
Last edited:
  • Like
Reactions: Nikhilniks
Solution

Nikhilniks

Prominent
Apr 6, 2021
61
2
545
TDP is heat dissipation (on HIS for given CPU) and is meant as a guidance for CPU cooling solution. Specified TDP is specified at sustained load, where electrical CPU parameters are kept within specifications (means, no OC whatsoever).
TDP shouldn't be used as measure of CPU power consumption, even usually it's close to that. For example, there are two CPU's having equal power consumption, however one can have lower TDP -that would be the one that has bigger heat dissipation area.


No, it won't... or let's say, differences will be relative small. Let's take a 8 core CPU having 105TDP, for example. If only one core is fully loaded, then that core can consume (say) 15W. That would make us assuming 8 cores would consume 15x8=120W. But that's not the case, because other parameters kick in (voltage/current limits) to prevent overheating -and so, every core would use much less that 15W. That also explains why 8 cores isn't 8-times faster than 1 core.
How about 6 vs 8 cores (equal TDP) at full load.. which will be hotter? Obviously 8 cores will consume somewhat more power than 6 cores and so, that will also generate more heat. But keep in mind 8 cores area is larger than 6 cores and so heat will dissipate on larger area too. Result is, same cooler can be used -that also explains why both CPU's have same TDP.
A very satisfactory answer.
But tell me what do you mean by "8 core area is larger"???
Are you indicting towards processor dimension??? If so, then i3 processor having 2 cores and i5 processor having 4 cores have same processor dimension.
 
A very satisfactory answer.
But tell me what do you mean by "8 core area is larger"???
Are you indicting towards processor dimension??? If so, then i3 processor having 2 cores and i5 processor having 4 cores have same processor dimension.
It actually depends on the particular processor itself. For example, Ryzen 5 and 7s share the same CPU die, the difference is two cores got disabled on the Ryzen 5 due to manufacturing defects (usually, or AMD just disabled them to have inventory of Ryzen 5s). However, because two cores aren't working in the Ryzen 5, it's not generating as much heat.
 
  • Like
Reactions: Nikhilniks

Karadjgne

Titan
Ambassador
Same architecture. Same die. The difference lies in which cores are active.

The i9 9900k has 8 cores. The i5 9400f has 8 cores. With the 9900k, all 8 cores are activated. With the 9400f, only 6 best cores are activated, 2 cores are electrically disabled by Intel. So while the die is the same, the actual die surface area the same, there's 8 heat producing cores under an i9 IHS but only 6 heat producing cores under the i5 IHS.

If power used was the same, that power is concentrated more with 6core use, but spread out more with 8cores.

Picture a cpu using 120w. If 8 core, that's 15w per core power. If using a 6 core, that's 20w per core power, concentrated in a smaller surface area since the 6 cores only take up 3/4 of the surface area of the full 8 cores.

That can affect cooler design, especially at lower TDP ratings, you'd see multiple coolers on i3's or FX4 series would have copper cores, where you'd think the i5's and FX8 should, but didn't. Same power draw between all the various cpu classes, but more concentrated power in the lower core models. So more concentrated heat output.

125w FX4 is the same cpu as 125w FX8, but all that 125w is concentrated in ½ the surface area of the die, not spread out with all 8 cores. Didn't require a larger capacity cooler, but did require a more efficient/effective cooler.
 
  • Like
Reactions: Nikhilniks
Are you indicting towards processor dimension??? If so, then i3 processor having 2 cores and i5 processor having 4 cores have same processor dimension.
They have the same IHS size, but chip (die underneath) is bigger in case of 4 cores. As @hotaru.hino pointed out: 2 core and 4 core die can have same size, but if 2 cores are disabled (on 4 core die), then that unused area doesn't count. Thermally, 2 core die is 2-times smaller than 4 core die, even they (can) have same physical dimensions.
 
  • Like
Reactions: Nikhilniks

Nikhilniks

Prominent
Apr 6, 2021
61
2
545
They have the same IHS size, but chip (die underneath) is bigger in case of 4 cores. As @hotaru.hino pointed out: 2 core and 4 core die can have same size, but if 2 cores are disabled (on 4 core die), then that unused area doesn't count. Thermally, 2 core die is 2-times smaller than 4 core die, even they (can) have same physical dimensions.
Thanks
This is a very deeper information for me regarding my doubts.