I have 3dvagabond on ignore but that guy needs to consolidate his posts instead of tripleposting, wtf.
The inverse happens here. Ideas never come together at the same time, but i have to fuse 4 ideas or more in one post. Readers that don't read with the same caution all messages may miss my updates. My last post was updated five times.
The worse thing is that i can only talk to you all because we most use tech terms here(Brazilian with "video-game english level" here
).
TPU's power consumption metric is not very useful because they only measure power consumption in Crysis 2 (which happens to be a game that they do not even benchmark for performance!). The only proper way to gauge power efficiency would be to measure actual power consumed with each respective game, and then compare that to actual performance measured with each respective game using the same game sequence.
Power efficiency = perf. per watt. One cannot determine that without measuring both perf. AND power consumed on the same task, period.
When Anandtech measured both perf. and power consumed in Crysis 3, GTX 780 had ~ 15% better perf. per watt in comparison to R9 290X. Like it or not, GK110 is more power-efficient on average compared to Hawaii.
The best part of TPU reviews to me is the power consumption graphs. Their results question the disparity of power draw between GCN and Kelper cards that most review articles by tech sites shows so much.
I agree that TPU's choice of games ain't the best, more taking in mind that WoW is a 9 years old game that looks 15, Diablo 3 is almost in the 300 fps realm with these cards and Splinter Cell is a crappy 7.4 metacritic score game.
I want The Walking Dead by Telltales in the roundup. Haohaohao.
If Diablo 3 was not able to be run with a GTX 650Ti maxed out at 2500x1600, then i would not question its inclusion on the graphs. Any AMD/Intel APU can run D3 maxed out at 1080p...
TPU choses games based on popularity, but i think is not the better way to measure video cards power... Anyway is good to have so much games in a test suite.
My problem with their reviews is that nowdays TPU does more effort with the tests on Nvidia cards. Games that favor Kelper(AC IV, Batman, SC: BL) comes to TPU tests since their day 1 and BF4 only came to the test one month later. Have to mention too that they always use the newest driver for Nvidia and not for AMD cards....
PS: IMO, metacritic sucks...
Hate to say this; but, from what I see with AMD's position other then their OEM Air Thermal Metering Technology, it appears to indicate your statement is irrelevant (LOL): such that, if one intends to effectively OC Hawaii, you need to Water Cool and AIB Partners have no position but to design cooling systems based on water cooling. I've answered rhetorically.
In other words, AMD have pushed the Hawaii GPU and VRM to MAX Specs on Air with their so called Screaming Air Cooler but Water Cooling will bring about Cooler, Quieter and Better Performance.
In my opinion AIB Air Cooling Solutions are going to be Bitter for Hawaii, especially if you intend to OC. Not saying they will be quieter within a Nominal Range. Lets hope AMD Driver Development and Mantle will Over-Write my words and project AMD well ahead of nVidia, this time.
I know on forums many reference 290(x) owners overclocking the cards past 1100Mhz.
AMD's 7970 is designed by 250W tdp, but the card consume far less by TPU data. Its reference cooler is bad, but works good with the card.
7970Ghz reference(a card that you can't buy on market) have the same board aspects, but have the chip pushed so far and reference cooler speed up to maintain the card cool. Reference 7970Ghz reception was the same on tech sites.
290(x) shares the same problem: It's poorly cooled(in efficiency terms, not talking about noise here) no matter how much you speed up the fan, and because this suffer to maintain its max clock speed(But most of time runs at same speeds that GK110 cards run too...)
http://www.hardwarecanucks.com/foru...5172-sapphire-r9-290-4gb-tri-x-oc-review.html
OC results without clock graphs or any explanation. Sad that no one except PCPer worried about this.
1250 Mhz on reference PCB. I said you it's possible.
*Edited much later. I started to make this whole post 2 hours ago.*