You're free to disagree, but that is how it has been defined since more or less forever by the press.
Also, what in the world did you just try to link to?
I disagree.
Launch = when they say it will be available, not when they release specs.
On paper = reality does not match expectations, not according to plan:
http://www.respondus7.com/CLDB/cldb-download-qa.php?instid=876543216
"judging something by how it has been planned rather than how it really works in practice:
The design certainly looks good on paper. Several candidates seemed suitable on paper but failed the interview."
The cooler is less effective because the die shrank proportionately more than the TDP.
Today, most companies, whether in the automotive, computer or smartphone industry paper launch their products because they use the paper launch window to build-up hype. It also gives them additional time to build inventory and manage inventory/pre-orders and logistics more efficiently. There is nothing wrong with paper launching imo as long as once the product goes on sale, the company has sufficient inventory for consumers to purchase it. For example, once Sony/Nintendo paper launch PS4K Neo / NX, there is nothing wrong with those consoles launching 3-6 months post announcement. However, if they sell out for 1-2 months starting from the date of release in retail/online, then it's a problem.
You're free to disagree, but that is how it has been defined since more or less forever by the press.
Also, what in the world did you just try to link to?
Sorry, that should have been Cambridge Dictionaries Online:
http://dictionary.cambridge.org/us/dictionary/english/on-paper
yep, by this logic Apple routinely "paper launches" iPhones. It's usually a couple of week gap between product announcement and availability.
The phrase "on paper" doesn't really have anything to do with the term paper launch.
The "paper" part comes from the phrase "on paper" not spec sheets. Launch on paper => paper launch.
Phased Retirement Set to Launch, on Paper at Least
http://www.fedweek.com/fedweek/phased-retirement-set-to-launch-on-paper-at-least/
The whole point is that this is not the same cooler as the GTX 980 though, but instead is similar to the cooler on the GTX 980 Ti.
1080 cooler = vapor chamber design (similar to 980 Ti, Titan X, Titan, 780 Ti)
980 cooler = heat pipes design
As such it should in theory perform better than the 980 cooler, but it doesn't really.
Going for 2.3-2.5 GHZ will make that close to 1 W/mm^2 if not more.Of course it performs worse..
980 Ti and Titan X have 601 mm² to transfer the heat/energy, while 1080 only have 314 mm².
250 W / 601 mm² = ~0.415 W per mm² (stock)
350 W / 601 mm² = ~0.582 W per mm² (max overclock)
Or in the case of the heatpipe based 980
165 W / 398 mm² = ~0.414 W per mm² (stock)
215 W / 398 mm² = ~0.540 W per mm² (max overclock)
Founders 1080
180 W / 314 mm² = ~0.573 W per mm² (stock)
230 W / 314 mm² = ~0.732 W per mm² (max overclock)
Going for 2.3-2.5 GHZ will make that close to 1 W/mm^2 if not more.
Can that be done without going below ambient?
That was my concern. If the heat could be transferred fast enough with a small delta T.Of course. Really, the cooler efficiency won't change a whole lot based on density, as the heat is spread quite effectively once it's in the bulk of the copper. Higher power density has a large effect on the benefit of better TIM though. When I was running some of my bitcoin mining machines quite hard, I could pull 900W into the board and likely 750W into the chip itself. All that was pulled though a total die area of 324mm² into dual fan closed loop cooler. Switching from the stock paste that came with a Swiftech H-220 to Liquid Ultra would give a 20C difference in die temps.
Of course. Really, the cooler efficiency won't change a whole lot based on density, as the heat is spread quite effectively once it's in the bulk of the copper. Higher power density has a large effect on the benefit of better TIM though. When I was running some of my bitcoin mining machines quite hard, I could pull 900W into the board and likely 750W into the chip itself. All that was pulled though a total die area of 324mm² into dual fan closed loop cooler. Switching from the stock paste that came with a Swiftech H-220 to Liquid Ultra would give a 20C difference in die temps.
That was my concern. If the heat could be transferred fast enough with a small delta T.
You're not talking big numbers here, so even a doubling wouldn't cause massive swings. 1W of board power per mm² is something we're already seeing, look at review of the MSI 390X Gaming[/img]; it's already at that number and it does so with an air cooler and mid 80s temps.
What ?
Maybe i'm misunderstanding, but are you telling me your card pulled 900 watt, with 750 watt going thru for the gpu die itself ?
Surely there is some kind of misunderstanding here
344 W / 550 mm2 = 0.625 W per mm2
All my numbers have been average power draw.. 1080 can peak much higher then 230 watts. (ive seen reviews having the 1080 very close to 400 watt in peaks)
What ?
Maybe i'm misunderstanding, but are you telling me your card pulled 900 watt, with 750 watt going thru for the gpu die itself ?
Surely there is some kind of misunderstanding here as iam a novice on mining. (nvidia
Sorry for the OT here.
Well, it's not a GPU. It's four 9mm square bitcoin mining ASICs on a common substrate, each die with its own 6 phase 300A VRM.
/end OT
Just a point that power density in the die to cooler interface can go much higher without issue. Nothing on that board draws any power other than the ASIC and any VRM losses unlike a GPU, so an equivalent GPU power density would be about 3W of (DC) board power per mm² of die area.
I was using the Furmark numbers, but your size on Hawaii is also off quite a bit.
What about a 250W GPU, 200 amps. Not impossible I think.:\ that means at 1.23v (usual OC voltage) that chip was consuming 600 amps. By comparison a welder requires only about 80. No way is that possible.
My bad, google misinformed/tricked me
https://www.google.no/search?client=opera&q=390x+die+size&sourceid=opera&ie=UTF-8&oe=UTF-8
Correct number is:
275 W / 438 mm2 = ~0.627 W per mm2 (stock)
344 W / 438 mm2 = ~0.785 W per mm2 (MSI R9 390X Gaming 8 GB)
data from https://www.techpowerup.com/gpudb/2663/radeon-r9-390x
You have to deduct memory and fan. 75W +5W? So really 264W / 438 mm^2 = 0.6W/mm^2.My bad, google misinformed/tricked me
https://www.google.no/search?client=opera&q=390x+die+size&sourceid=opera&ie=UTF-8&oe=UTF-8
Correct number is:
275 W / 438 mm2 = ~0.627 W per mm2 (stock)
344 W / 438 mm2 = ~0.785 W per mm2 (MSI R9 390X Gaming 8 GB)
data from https://www.techpowerup.com/gpudb/2663/radeon-r9-390x
Mind boggling when you think about it really.What about a 250W GPU, 200 amps. Not impossible I think.
Yeah, so if you compare that to TPU's average number for stock FE, 166W / 318mm² gives 0.522. That's still lower than stock 390X.