Is Intel up to its old tricks regarding the Ryzen launch??

[CAT-THE-FIFTH]

I can predict what some review sites will do - they will try to "sell" that the AMD memory controller is "weak" by trying to show how high the Core i7 7700K one can be overclocked more and then ignore the X99 ones. The thing is Intel only officially supports 2400MHZ DDR4 anyway on KL- AMD is really no different in that regard. Last time I checked a 2400MHZ memory controller being pushed to 3400MHZ(AMD tweeted this) is not weak.

The same goes with clockspeeds - they will quietly ignore that the X99 CPUs don't overclock that high. They will push the narrative to the Core i7 7700K instead of the Core i7 6850K and Core i7 6900K,which the R7 1800X is targeted towards at close to £500.

Other areas where they will try to warp things - use CPU power consumption and not system wide power consumption. Ryzen is an SOC,so it will consume more power as a CPU I expect,but probably lower as an entire platform.

[Xlucine]

How would they measure CPU consumption? For graphics cards you can mess around with risers, but CPU sockets are a bit more complex

[aidanjt]

How would they measure CPU consumption? For graphics cards you can mess around with risers, but CPU sockets are a bit more complex

It's not that difficult, you can measure the heat output of the CPU under various workloads and calculate how much energy it consumes from there.

[prwilson]

We definitely need more competition in the cpu and graphics card markets, drive down these insane prices a bit.

[scaryjim]

How would they measure CPU consumption? ...

... you can measure the heat output of the CPU under various workloads ...

Talk about needlessly complicated and error-prone...! Measuring heat output accurately isn't exactly straightforward, and a CPU is not a 100% efficient heat pump.

However, you can fairly easily follow the motherboard traces and work out which power pins the CPU is drawing its power from (hint: on a modern motherboard is virtually all comes from the CPU power socket). And sticking a passthrough connector onto a 4-pin, 8-pin, or even the 24-pin sockets on a motherboard is a doddle. You can then, completely accurately, measure the amount of DC power being drawn through the various conductors.

Or if you want to increase the complexity slightly, you attach sensors at the 'fets for the CPU power phases.

Basically, there are loads of ways to directly measure CPU power draw that are way more accurate than trying to back-calculate from the waste heat generated by CPU load....

[Xlucine]

If you're using a pass-through for the motherboard cables then you'll get RAM, southbridge, etc power draw as well. Traces would work, but sods law sez that they're probably buried in the motherboard layers. Basically, I'm prepared to eat hat if anyone manages to separate CPU power draw from motherboard power draw to made the ryzen SOC look bad

[scaryjim]

If you're using a pass-through for the motherboard cables then you'll get RAM, southbridge, etc power draw as well. ...

You can follow the traces to check which conductors power what. AFAIK RAM power is mostly drawn from the 3.3v lines, and many other peripherals use the 5V lines. I believe that in most motherboards practically all CPU power is drawn from the 4 or 8 pin CPU 12v connector. Don't forget that, for the 24pin motherboard connector, you can attach meters to individual pins to determine the current/voltage passing through them. So it should actually be pretty easy to isolate just those connectors that power the CPU power phases. A detailed tech manual for the motherboard in question might even give you all that info without having to try to follow mobo traces....

As it is, the power draw for a couple of USB ports and an NVMe slot shouldn't be so dramatic as to make the power draw from Ryzen significantly worse than any extant Intel platform, and most reviewers use at-wall system draw anyway - either on its own or in conjunction with socket/slot power draws...