TDP refers to the temperature difference between the center of mass and the engine's cooling system.
TDP refers to the temperature difference between the center of mass and the engine's cooling system.
A few years back I chose to replace my Intel i5 4440 with a Xeon E3 1280V3. Intel said the i5 has an 84w TDP while the Xeon is around 82w. I thought it would work, but the Xeon actually overheated a bit. For those not knowing, this CPU is Haswell and functions like a non-K i7 4770 at 3.6ghz base with a 4.0ghz boost. The i5 runs at 3.1ghz base and 3.3ghz boost. I didn’t keep it this way for long; instead, I added an NH D15S cooler. My main takeaway was that TDP doesn’t really compare the two chips. Turning off the hub and lowering clock speeds didn’t help much. The temps stayed high even after changes. Buying a 8t cooler later made more sense since I planned to run all threads. In 2018, this upgrade let me use the machine for gaming and editing for another year. When I retired it, the cooling would move to a new system, likely a 5600 for server tasks.
TDP stands for Thermal Design Power. It's a standardized figure (though AMD and Intel calculate it differently) that indicates the cooling requirements for a processor at standard speeds. It serves as a reference for cooler manufacturers rather than a precise measure of power consumption or turbo performance needs. The concept doesn't cover maximum draw under high loads or detailed cooling specs. For more details on AMD's perspective, see this article: https://www.gamersnexus.net/guides/3525-...r-opinions
I usually use a big salt chunk the size of a planet for TDP adjustments, but... When evaluating two CPUs on the same socket, I assume they should perform similarly. Observing a 1151 CPU next to another 1151 CPU with identical TDP, I’d anticipate similar idle temperatures, possibly even matching. My expectations were that lowering the core/HT settings to align them would result in comparable heat output, yet... My i5’s peak temps during load matched the Xeon’s idle temps. I also thought adjusting those settings would bring their thermal behavior closer, but it didn’t happen.
You also tried the test without HTA? Yeah, that’s just the silicon lottery—some chips heat up more than others. I’ve got a bunch of old Pentium E5000 and E5400, plus one Golden E5400. There are definitely differences even when clocks and voltages match. The E5400 gets hotter than an average E8000, but my E5800 stays so cold that Prime95 can barely run at 2.2GHz with a 0.9V boosted core and just a fan on top. No heatsink, just the fan itself. It kept temps in the mid-70s, but I didn’t do much more testing because it was more of a fun experiment with a CPU that didn’t need a heatsink.
Yeah, HT off, I even set the base clock down to 3.0. I’m curious about possible problems—maybe a bad mount, uneven thermal paste, or those Intel plastic push pins in the box coolers are really tricky to fit. The D15S is doing great though. I tried the low RPM adapter and it still works smoothly, though those coolers are built for much more powerful CPUs.
TDP mainly reflects theoretical limits rather than real-world usage. Both CPUs and GPUs provide a single figure, and most—particularly with auto-turbo enabled—exceed that number. You can adjust settings to match the listed values, but this restricts performance. For optimal cooling, prioritize higher power output.