it isn't... It's obvious SFF would be hit by this "upper limit TDP war"... But noone is forcing anybody to run this CPU full throttle (=at turbo frequencies)! If you want, you can still run these at lower TDP! You just have choice to to do it know. With proper cooling, you can have more performance. Otherwise, more efficient focused CPUs (or TDP) are available...
Careful... I know I will get some nodding heads from this, but hear me out. I appreciate their tireless efforts in journalistic integrity and brand neutrality. Commendable as they are, GamersNexus is not—bold statement, but repeat—not the ultimate authority in technical know-how and experimental accuracy. Take this example from recent memory. With one of Adobe’s major spring 2018 updates, Premiere brought Intel hardware video encoding acceleration support to the table. Here, GamersNexus made a big mistake in concluding this was a unilateral step forward. They benchmarked the performance of Intel 8th Gen with this enabled compared to AMD Ryzen 2000 using the default software encoder. Fair? Hardly so. That hardware acceleration happens to be Quick Sync Video (QSV) which is still highly debated for its fidelity in streaming and podcasting circles. If they were on top of things, they should have gone back and also used Voukoder in Premiere enabling NVENC for the GTX 10 series graphics card they use in their test setup to provide another option in their test results. Arguably, QSV, even in its most recent revision, still is a far cry in video quality from a software encoder like the videophile standby x264. This example also bears some relevance here. I wonder if QSV is being used here in the streaming test as well. If so, that would account for the major difference we are seeing in the encoding workload power draw between AMD and Intel in the GamersNexus result set.True, and more information on what games they tested at 95w is needed, however I don’t think it’s unlikely to see only a small percentile drop in performance when capping the CPU to 95w as it doesn’t use much more than that while gaming at stock settings with MCE disabled.
GamerNexus plotted the power usage while playing and streaming Fortnite for instance and if you haven’t seen it then the results may surprise you
That is literally the only thing I can surmise as well. Plus, the recent pending sell-off of IM Flash to Micron leads me to believe they are already hurting, that they are doing everything humanly possible to perform balance sheet origami. As they say, desperate times call for desperate measures and such is par for the course for Intel for the last two years. Intel already pulled some crazy stunts with X299 when they knew ThreadRipper was coming, and I imagine this is yet another desperate reaction and this time for Zen 2.
What CPU's Delta T are you referring to?
nonspecific in general? o_o like imagine Intel pulls all TDP info from ark.intel.com
Interesting. Compared to the 2700X, they already have a definite lead, no doubt about it, when at the base TDP. For some reason, though, they seem to be pushing out much further the performance than appears necessary. It could very well be they are working to narrow the possible gap or deficiency they may have with Zen 2’s launch in six months from now.This is starting to look really bad. Golem.de tested the 9900K with in-spec power limits and couldn't get the processor to maintain advertised all-core frequencies.
When they bumped up the limit to 200W, suddenly things changed:
Not an insignificant difference in performance:
Look at the power consumption:
There's a bunch more if you click through the image table.
About the closest thing to that would be FrostyTech’s or HardwareSecrets’ outdated charts, and our own @IntoxicatedPuma’s charts that he is even now amassing as a Herculean effort.
I'll wait for mid 2019 to decide on my future cpu platform (Zen 2 vs 10nm Intel?)...2019 will be so interesting in cpu market...
