AMD Shanghai/Deneb Review Thread

[Uncle Jimbo]

Do both Amd and Intel monitor their core chip temp in the same way? I find it hard to believe that a chip running at 4ghz full load will only read 33C (the IR probe is at best questionable)

I was using the IR measurement of the cooler base, which showed 39C. While IR measurement can be questionable for precise measurement, it is good enough for a ball park estimate - even if it is off by a degree either way, it's only 5% or so. The right way to do that kind of test is with a heatsink with a calibrated sensor, which has been charaterised by performance against a resistive load with known power input. I'll do some testing like that once I can get my hands on a 940 Phenom II.

The internal temps are not useful for heatsink power calculations, or really as anything other than a reference, unless they have been adjusted for offset and curve (through software). Based on the cooler temp, I would think the internal temp was probably closer to 50C, maybe more, but that's just a guess.

If they can have a 4ghz chip only dissipating 95W then why not release a chip >3ghz or am I looking at the wrong numbers??

If AMD follows past practice, they will add higher end parts once they see what the bin split is from volume production... My suspicion is that the 140W TDP for platform designers was to accommodate higher power parts as a straight swap with no other changes, since the 3G parts are obviously way below that dissipation. Based on what happened with Barcelona, I think AMD is wise to under-promise and over-deliver this time around.