Get 14600 to behave as 14600T

I cannot see that it would cause a problem. Best way to test it is to try it and find out

Schroinxx said:

Will a 14600 or 14600K behave power-wise close to a 14600T if one set the PL1 and PL2 to the same limits, eg 35W and 92W?
Is that considered to be within the specs, or is there a risk to its data processing? It looks as the 13500 I have in the balanced mode shifts through the power settings anyway after th eload, so setting it permanent should not change that but just cut of the tp end? Will it also have anything to say to the efficiency pr cycle?
I am looking for low-power CPU for my media server for a socket 1700. Most of the time it idles or is in the 5-15%.

Intel product specifications

Intel® product specifications, features and compatibility quick reference guide and code name decoder. Compare products including processors, desktop boards, server products and networking products.

ark.intel.com

Click to expand...

I don’t see the point in doing this. If you turn on the c states and enable speedshift/core parking, the thing will consume the same amount of power as a 14600T, then if it needs to (which it shouldn’t doing what you are saying it will) it will ramp up

Maybe there could be an issue with parts that were binned into each series based on whether they were capable of performing properly at various power levels? I don't know exactly how the T and non-T models compare in that, but like if you tried to take a base processor and bring its voltage down to the level of a T series, would some individual units possibly not be able to run at such low voltage even at the lower clock speed, so they couldn't be binned as T series? (I know the base parts can auto-clock themselves down quite low, but are they all rated to be able to go as low as a T series in both voltage and clock? Is their minimum level the same?) Or are the T series more likely to just be the ones that couldn't make it to the full speeds no matter how much power was supplied?

Keljian said:

I don’t see the point in doing this. If you turn on the c states and enable speedshift/core parking, the thing will consume the same amount of power as a 14600T, then if it needs to (which it shouldn’t doing what you are saying it will) it will ramp up

Click to expand...

It might still be worthwhile in some cases to limit the clock rather than allowing it to ramp up even for a short time, but for most consumers letting the CPU ramp up and get the work done faster would be more efficient. Heat might briefly rise higher, but it should be able to cool back down quickly because the work is done faster, so unless the system is seriously constrained on cooling there shouldn't be an issue. Overall, getting the work done faster has been shown to be more efficient than keeping the power usage down but making the work take longer, within certain limits.

Changing PL1 might allow the CPU to use a little less power when it's running at the lowest clocks, but this could come down to some units benefiting and others not benefiting, as each individual CPU might not be able to drop quite as low. But it might be that the CPU could attempt to drop the voltage too low for that individual unit's capability, which might cause crashing, since that CPU will assume that since it's allowed to go so low, it's okay to do it. Similar with PL2, if the upper limit isn't high enough, the CPU could try to boost too high for the voltage it's allowed to use, which would basically be like trying to overclock too high.

I don't know about how the two series are programmed as far as whether the non-T will actually attempt to keep the same clock speed while reducing the voltage, if it will go as far as a T series. Efficiency of power per clock cycle would depend on just how much lower the voltage is able to go, but if you only get 30% lower voltage, but it takes twice as long to perform an operation due to the clock speed having dropped by half, then the overall efficiency of the operation will be much lower.

In a system that's just idling most of the time, you're probably not even going to notice any difference with everything at stock. For an individual user, with a normal system with standard cooling in a reasonable environment, the differences in the amount of power used for day to day tasks is really minimal. Even playing video isn't going to be a huge amount of power. The pushes for power efficiency on desktops (anything not relying on batteries) isn't really about individuals, it's about the mass numbers of PCs and other devices that are always running which are a significant load in aggregate.

If you have an ultra-tiny PC with a ridiculously small heatsink and fan, the ambient temperature is like 35C on a cool day, and electricity costs $0.75 per kWh, then yeah, it would probably be worth having a lower power system if you don't need high performance.

steelghost said:

I cannot see that it would cause a problem. Best way to test it is to try it and find out

Click to expand...

Pretty much this.

As others have noted there are plenty of reasons to not do this or why you might not get exactly the same results, but strictly speaking it should work fine.

It definitely won't break a 14600 to put those PL1 and 2 limits on it. You should then get no more than those figures for heat generation. How closely that will simulate the results from an actual 14600T will depend on what Intel does internally. It could be identical, it could be more efficient/faster, it could be less efficient/slower. But you'll know for sure that the maximum heat output will not exceed what you set.

If you want to be absolutely sure the chip works just like a 14600T, however, I'd recommend buying one.

edit: if you're really worried about peak heat generation, if that's the actual issue you're looking at, you might think about a 7000-series Zen chip. Those chips are quite efficient, and respond very well when you reduce their allowed power consumption. Their peak outputs are typically much lower by default, and then you can set them even lower without much impacting performance. AMD chips will frequently match a 300W Intel chip in the rough neighborhood of 100 watts.

Idle consumption on Zen chips, however, is a little worse. If you're more focused on idle power, rather than peak power (which makes perfect sense for most computers that aren't gaming boxes), then Intel is probably preferable. When you reduce their PL1 and PL2, their performance drops faster than it would with an AMD chip, but for an HTPC, that won't really matter. A 14600 is such overkill for that application that gimping its peak performance would probably never be visible, while the better Intel idle power consumption will pay off day after day after day.

Also note that PL1 and PL2 don't affect idle power at all. Neither does putting a TDP limit on an AMD chip. There's not a lot you can do to reduce idle power on any chipset, other than by having the OS go to sleep.

second edit: just as a warning, I edited this repeatedly and very heavily, so it's pretty different from the comment I originally posted.

Just apply a power limit in the BIOS. The CPU will automatically clock itself as high as it can with the available power. Each CPU has a unique voltage/frequency curve.

Based on my experiments with a 14900K, Intel CPU's respect PL1/2 pretty strictly. The CPU was being powered by a HDPlex 250W GAN PSU, along with a 4060, so limiting usage was pretty important.

The only caveat to PL limits is if the T-series has been binned to work at a lower vcore, in which case limiting a vanilla vs T series CPU, the T will perform better.

@Schroinxx Did you try it? Did it work?