That is another part of the reason I ALWAYS disable Defender on any mining machine.
 I WISH I could disable that bloody Cortana wasteage as well, but I don't have many Windows-based mining machines and they're all getting moved to LINUX anyway.

Gaming doesn't present a CONSISTANT high-level load to a GPU.
Comparing based on GAMING stats means nothing.
It has been shown a LOT of times that mining puts a LOT more stress and that most algorithms soak a LOT more power than any gaming does.

Why do you keep talking about MEANINGLESS Teraflops figure, that has absolutely ZERO to do with mining performance.

 And, your links do NOT say anything about undervolting being available, the standard RX series cards have the SAME "voltage control" optimization logic that your links are talking about - that logic goes back to AT LEAST the R9 3xx series, and has been shown to NOT optimize well in a mining environment.
The only difference is how they TUNED the logic for the APU involved - which is also the case in every other APU and recent GPU that AMD has built, they ALL have different "tuning" to that logic depending on the model.

 Do try READING the links you post before you point to them as demonstrating information that IS NOT THERE.

 Have you ever actually USED an AMD APU before?

There are VERY few places that have no competition at all - virtually NONE in the US.

 You are NOT limited to cable or DSL - you also have Sat service (Exceed worked fine for me for 2 YEARS of mining, abet at a cost of perhaps half a percent higher on stales when mining Bitcoin, NOTICEABLY LESS on anything else), Cellular service usually from A FEW providers, some areas still have DIAL UP available.

I'd be willing to BET you actually have at least 4 and more likely 6-8 actual ISP options, if you don't insist on "I gotta have HUGE BANDWIDTH" when mining doesn't need it.

 As oppposed to Bitmain that would have let them burn, based on PAST lack of effort at fixing issues in a timely way on the S9.

 1050 ti aren't old tech - SAME Pascal generation as the 1060 through 1080ti.

 I won't bother with them due to the excessively low rig density, but they look to be a tolerable entry-level mining card for folks on a LOW budget, and they come in "low profile" options for folks with small cases that want to upgrade from older stuff.



 On a different subject, I got a ".01" payment from Nicehash today - and noticed they changed their "payment policy" to "irregular" for small payments.
 Perhaps they are now going to cluster together the small payments to keep the fees under control?

 2 reasons.

 1) Better efficiency. In the long run, the A4 A4+ and A6 will still be profitable when the L3+ turns into a money-loser.
 2) RELIABILITY. Innosilicon has a TON BETTER reputation for reliability than Bitmain recent generations, and an INFINITELY better reputation for FIXING issues when they do crop up.

 Overall though I'd prefer the A4+ *IF THE MOQ WAN'T SO INSANELY HIGH* due to the better flexability from it's much lower power draw.

 WHY would you need to update a BIOS that is NOT IN USE AT ALL by the time these bugs CAN be invoked?

 I can't speak to 16.04, but XUbuntu 14.04 was no problem at all - just "dd cloned" one of my existing Nvidia-based systems and away it went (once I got the hardware installation "oops, what did I forget" issue worked out).

 You DO get close to twice or sometimes a hair MORE than twice if you don't insist on pointing a 1080 at an algorithm (ethhash) that is IS NOT WELL SUITED FOR.

 You can also get a TON better efficiency on some algorithms - ZEC in particular a 1060 (per ALL of the postings I've seen) struggles to hit more than 3 sol/watt while a 1080 easily hits 4.

 For ETH, the 1060 is a better option - but that's NOT the only option out there, and judging a card JUST on it's ETH performance is STUPID.

EBWF seems to return "instantaneous" power readings, like nvidia-smi but not always on the same timing.

 You learn to compensate for that with a bit of practice.

I'm not saying it's crap over JUST the false positives - though it is FAR WORSE on those in general than any of the other major anti-virus venders.
It also offers the LEAST PROTECTION by a wide margin in EVERY COMPARATIVE TEST it has had done against ANY of the other major anti-virus products, even the FREE ones.

 PDUs are useful when you CAN'T change your wiring and outlets (most apartment situations), and in some cases might be a lower-cost option to calling in an electrician to rewire your place when that IS an option.

 Nope.
 Unless you are going to push the card for max hashrate, in the 200+ TDP range, you won't see any significant difference other than the SC might run a degree or 2 warmer.

Greater demand than supply, like MOST "mining-related" hardware since March or so of last year.

 7nm hasn't left the lab yet - the only "chips" that MIGHT have been produced to date have been internal testing stuff, NOT FOR SALE TO ANYONE.
 There is no "very very expensive to be used" at all, much less "practically".

 10nm is only JUST hitting real production out of Samsung and Intel.
 TSMC and GF have announced and have been working on "7nm" but they're way behind the curve as their "7nm" processes are widely reported to be looking more like "10nm" out of Intel or Samsung than "actual 7nm feature size".

Just got a .01 payout today.

 Given my current issues with both Bittrex and Poloniex, I'm seriously considering moving ALL my hash back to Nicehash - they have LESS of my money held up at this point and at least an ESTIMATED timeframe for payout on it.

 Better cooling that doesn't throttle the boost clock as much, and you're pushing high TDP (probably 220-250 watts?).

 I saw the same thing out of my AORUS card vs my other 1080 ti models though not as much (EVGA SC, ASUS Turbo, Gigabyte Windforce) when I was pushing all of them during testing.

 Bitmain has been pumping out thousands of L3/L3+ units over the last year - but difficulty has not gone up all that much.

 Keep in mind that UNLIKE the situation with X11, there already was a quite large "installed base" of Scrypt ASIC miners in place prior to introduction of the L3 - and that the L3 wasn't "night and day" more efficient than some of them so most or all of the predating miners are STILL MINING.

 The Scrypt situation is closer to the SHA256 situation, except that efficiency gains have been even HARDER to come by.


 Price NEVER follows difficulty.
 Price is always a factor that leads TO difficulty increased over time.
 There is NO SUCH THING as "the price should increase to compensate for that", never has been, never will be.

 You would be better off replacing that junk Windows Defender thing with a REAL anti-virus program.
 It does NOT keep you safe, and it is NOTORIOUS for how often it "false flags" legitimate programs for no reason.

The limit is that the iGPU won't have enough power available to run at full speed consistantly - keep in mind that the actual RX 580 can soak 175 watts or more ITSELF when mining ETH, the iGPU is NOT going to have that full power available - there are other things in the box to run, and there is a lot of question as to IF you can undervolt an iGPU in a console at all to allow it to run at lower power draw.

 Vapor chamber cooling does not address the power limits - that addresses COOLING issues, and given the usage of vapor chamber cooling on the GTX 1080 ti FE cards AND HOW POORLY IT PERFORMS COMPARED TO ANY 3'D PARTY SOLUTION FOR COOLING I wouldn't be bragging on the stuff.

 Quoting TeraFlop figures is MEANINGLESS. Mining is 100% Integer operations, FP operations don't mine.
 

 Console APUs are based QUITE CLOSELY on the same-generation "general purpose" APUs. It's going to be interesting to see how the "Bristol Ridge" series compares to the APU in the Xbox One X.
 "SEMI-custom"  after all.