Intel 'T' CPUs

[Bob Hope]

Is an i7-7700 underclocked to 2.9GHz the same as an i7-7700T ?

 

[LTC8K6]

Is an i7-7700 underclocked to 2.9GHz the same as an i7-7700T ?

Yes, pretty much.

 

[Topweasel]

Is an i7-7700 underclocked to 2.9GHz the same as an i7-7700T ?

Underclocked and undervolted. Though like a Ryzen 1700 and other examples of the same chips with a lot less power usage under less clocks. These chips usually binned specifically because of it's power usage. So it possible that you might need to run the 7700 at slightly more power for stability.

 

[kwalkingcraze]

My understanding is that the "T" Intel processors are designed for passive, fanless cooling.

 

[Azuma Hazuki]

Not exactly, though at 35W this is certainly feasible with the right chassis. They're simply designed to produce less heat by default; think of them as permanently cTDP-down'd versions of the standard non-T CPUs.

One thing I've been wondering: underclocking and undervolting a non-T CPU produces more or less the same effect +/- a few watts. But...if you put one of those in a heat-limited chassis, how do you get it to "live" long enough to set the relevant settings? Especially since they seem to go all-guns-blazing in BIOS/UEFI firmware menus.

 

[coercitiv]

Underclocked and undervolted.

Do you have data on that, a comparison of a few chips at least?

My own little experiment with limiting TDP on 6600K showed it to be on par if not better than it's T counterpart.

Quick update on our T SKU discussion: got some time to test in Cinebench with a 35W package power hard limit. Max clocks were set to 4Ghz, voltage was stock.

• Running CB15 on 4 threads resulted in a score of 595, the i5 6600K averaged about 3.55Ghz while compelting the bechmark.
• Running a custom 3 threaded bench resulted in a score of 496 and an average clock of 3.85Ghz.
• Prime 95 Large FFT load resulted in sustained 2.9Ghz. (28.8 avg)

For reference, the 6600T max 4 thread turbo is 3.3Ghz and the max 3 thread turbo is 3.4Ghz. Base clock is 2.7Ghz. The T CPU may well still use less power than the standard SKU in the loads above, but simply clocking higher than the T CPU at the same 35W TDP outlines the flexibility of using standard CPU in combination with TDP limits to obtain higher performance in a DIY SFF system.

 

[Topweasel]

Do you have data on that, a comparison of a few chips at least?

My own little experiment with limiting TDP on 6600K showed it to be on par if not better than it's T counterpart.

Sorry I don't get what you are asking? It should be fairly obvious that trying to maintain clocks when undervolting isn't going to work on every die unless all dies clocked exactly the same. For that reason all T processors are going to be binned to run at at it's given wattage at it's given clockspeed. So it should be entirely possible (but not a given), that other dies for leakage reasons and others wouldn't run at the same voltage at the same clock speed.


I do understand your test managed better clocks with 35 watts in a K chip while setting a hard limit. But 35w is a temperature bracket not a hard setting on power limit. Without trying to clock the K at exactly the same settings as the T while being able to measure exact power usage we won't know.

 

[coercitiv]

Sorry I don't get what you are asking?

I'm asking if you know for a fact that T chips are binned for lower voltage @ same speed than standard SKUs.

But 35w is a temperature bracket not a hard setting on power limit.

Intel may have a more complex definition of TDP in relation with TCASE, but in low wattage locked SKUs (15W, 35W, 45W) you can rest assured there is also a hard limit on power usage equal in value to TDP. They will all throttle to that power value after the Turbo Time limit has expired, no matter the core temp.

For example my 4700HQ with 47W TDP may indeed consume close to 60W for a limited period of time (~30 seconds IIRC), but after that, no matter how good the temps are, it will throttle to reach and maintain 47W continuous power draw. No exceptions.

Without trying to clock the K at exactly the same settings as the T while being able to measure exact power usage we won't know.

So you reckon a 7700T running at max clocks under a heavy load (eg: CB15) will consume significantly less than 35W? It's strange you are ready to use "fairly obvious" deductions when your claims are of concern, but in my case accurate measurements for both platforms are required.

 

[Topweasel]

I'm asking if you know for a fact that T chips are binned for lower voltage @ same speed than standard SKUs.

No, its why I used a non-difinitive statement like "usually". The 1700 is an example where people were finding that the chip while generally not overclocking better and could run at better voltages than the 1700x and 1800x at it's given clocks. The general theory being that the 1700 is actual the premium binned chip as it had less leakage but allowed it to use significantly less power considering 8c CPU. Less Leakage also usually lower clock and voltage limits before degradation.

Intel may have a more complex definition of TDP in relation with TCASE, but in low wattage locked SKUs (15W, 35W, 45W) you can rest assured there is also a hard limit on power usage equal in value to TDP. They will all throttle to that power value after the Turbo Time limit has expired, no matter the core temp.

For example my 4700HQ with 47W TDP may indeed consume close to 60W for a limited period of time (~30 seconds IIRC), but after that, no matter how good the temps are, it will throttle to reach and maintain 47W continuous power draw. No exceptions.

This can go back and forth between CPU's. A mobile CPU like that is all about eaking out it's max performance while still maintaining it's cooling limits. Where on the other hand we have seen tests that where CPU's like even the 7700k and the 1700x where they don't hit their "TDP" because TDP isn't power usage. Not in the sense that people want it to be. The fact that the CPU can ever exceed the TDP in the case of the high performance mobile quad you just specced out shows that. Generally (another non definitive word) Intel or AMD will target a power usage and bin and select chips capable of a common denominator in performance at the given power limit and sell from there. There can be a lot of cross groupings based on process and demand. Probably a decent amount of chips even within the sku that could do the same performance at better power usage, or clock better at its current power usage. From there there are probably a lot of 7700T that can clock to 7700 levels and 7700's that have as good if not better power management as a 7700T at lower clocks. The point is binning is real.

So you reckon a 7700T running at max clocks under a heavy load (eg: CB15) will consume significantly less than 35W? It's strange you are ready to use "fairly obvious" deductions when your claims are of concern, but in my case accurate measurements for both platforms are required.

I couldn't care less about your requirements. You gave a single example where you set a power limit and let the chip clocks itself. Not running the chip at setting limited by a T CPU. So you should have to get actual power usage of all three configs (T, non T at T settings, and your non T at power wall settings). I just covered the most understood knowledge of Die binning. That's the obvious part. That binning is real. If you are questioning that, then I have nothing to say.

 

[coercitiv]

No, its why I used a non-difinitive statement like "usually"

There, it wasn't that hard.

I couldn't care less about your requirements.

Well then, don't let me take any more of your time.

 

[Topweasel]

And ignored god this has been terrible month.