Josh, for the builds that you've done with the 6700T - what is the typical power consumption (e.g. gaming) versus the 65w TDP counterparts?
Great question. TL;DR heat is what you want to manage. As even on the Dell 330W the build below runs fine from a power draw perspective. Every watt counts.
I don't have the greatest metrics, as for most of the systems with 45w CPUs I was using the 285 and 380 ITX systems which can soak alot of juice. For most users there isn't a HUGE difference between the two, but some games that 20W difference can give you an edge--personally just for heat management I think it nice. I do have lots of good metrics for for following system, as I seem to always be building one for a customer:
(highest sustained point at least 1 second long, micropeaks not measured)
65W CPU + z170 system and Asus GTX 970 (5 systems averaged)
Web browsing: 48W
Youtube: 56W
Prime 95: 107W
Valley Bench : 215W
Valley + Prime95: 255W
Most Gaming: 220W
Mech Warrior Online: 245W
45W CPU with GTX970:
Mech Warrior Online: 222W
That's pretty much what I have for wattage measurements.
However I
DID measure system performance between 45W CPUs and 65W CPUs for REAL LIFE gaming, and I found that there is barely a difference
even in MWO, which uses more of the CPU power. The biggest difference came on the Forest Colony 2.0 map with lots of stupid particle effects and physics, where I could gain up to 7 FPS between the 6700-T stock vs a 5820K @4.2Ghz (all features enabled).
Now that is impressive to me...
I'm a content creator so I like CPU power more than GPU power myself, but it is impressive to me that you can buy a 45w CPU and have it on par with a 145w CPU (base clock!).
CONSIDER HEAT
All of the custom builds I have done the customer has requested 970s (earlier this year it was the 380). The system will get ridiculously hot without air circulating in the room and touching the system. Here is a chart I made of the current system I have in my office I am building for a customer. Please note that all these were taken in the vertical orientation with the black bezel pointed down (heat difference is almost 25% less if you flip it around the other way!)
Ambient was measured via my thermostat on the other side of the room, meaning where the system was obviously would be much warmer.
The point of this run wasn't to be scientific, but rather to see at what point the system would crash or throttle to a stop. Neither happened, but you can see how much throttling can happen with no room cooling!
What is interesting though is how little throttling impacts games. I played Overwatch on Sunday for like...5 hours. I followed a similar regimine and took heat measurements between matches. For two hours I played without air conditioning and the room got to 81F and the S4 Chassis front bezel (which can act as a heatsink for the HDPLEX if you butt it against without the velcro which is my standard practice now) reached 51C!!!!!! However the game ran perfectly smooth at 2560x1440 with AA and vsync enabled with zero system lag spikes, something I would have expected seeing a 25% throttle in heaven under
lighter conditions. It is worth noting that RealTemp put the CPU at 55C at the hottest point, which is unexpected but very nice. With a 85w CPU I would be seeing 70s-mid 80s depending on the game, which could also spill over to the GPU...
SIDE NOTE: This was done with a Ribbon Cable that was NOT mine (for testing purposes.) I ran the test one more time right after the last one in the graph and it scored 4730 at 102FPS! I am very happy with the performance of my cables, to say the least!
So you reach a point where you want to decide one of three things.
1. Go with a 750ti/950 and a 45w/65w CPU and dont have throttling but lower base scores
2. Go with a really fast build like the one above, and have bursts of raw power and then throttle down to lower GPU speeds
3. Air condition your room and have no throttling, but still a hot system!
This is why I test these builds for half a week before shipping. I need to know they can withstand these really hot temps and remain stable. My goal here is not to convince anyone that these temperatures are ideal...they aren't! But you need to make tradeoffs somewhere and some people choose power over heat management.
One final note about heat. I hear that people experience throttling with their 950PROs, but I've not found it to be an issue with the S4 Mini, at least in any metric that I have measured. Possibly sustained transfer of video files for hours at a time, but I've even tested that for an hour and when the system reaches these stupid hot temps it still runs strong:
