Posts: 1370 from 2003/2/15
From: Central Europe
@ Velcro_SP
Quote:
That CNET report could be a misquote. It would be silly for Max Seybold to claim the maximum power is 2 watts, because the claim could be easily refuted. On the other hand I do grant you that he's put the 2 watts figure out there repeatedly in a non-specific and potentially misleading way. I just don't think he's lied about it. Consumers are expected to be able to see through such puffery, as I and you and most others did. Do you attack the vacuum cleaner salesman for saying "this vacuum cleaner will run forever," or the cafe that says "we have the best coffee in the world?"
Sigh.
If a car dealer advertized that a specific car goes as fast as 120mph even though its maximum speed is 60mph and you sued him after buying this car from him, you would have a slam-dunk case in court.
If, on other hand, you sued a cafe for saying they sell the best coffee in the world which you happen to disagree with, your chances at winning would be rather slim.
The distinction should be quite obvious to any semi-intelligent adult.
Quote:
Beyond that I would hope that for example an EeeBox consuming 36, 40 or 65 watts (see the Wikipedia net-top table) probably having more memory and probably costing a lot more can outperform the CherryPal consuming less than 7. By even bringing in performance, I think you've conceded that CherryPal has better wattage than most X86 net-tops.
Zylesea has explained to you here on MorphZone that his EeePC laptop consumes roughly 10 watt in battery mode based on his real-life experiences. That number is
for an entire computer, which includes a display, touchpad, keyboard, etc.
My argument has been in the past that the Cherrypal's real life power consumption numbers do not look very impressive in comparison.
Of course, you choose to rather talk about how some ominous "average x86 net-top" uses more power, which is completely besides the point. As Zylesea also explained, his EeePC consumes considerably more power when he plugs it into a wall socket. The reason is quite simply that netbook manufacturers think that extended battery time is more important than performance when you are on the go. When you are at home, the device switches from power-saving to full-power mode in order to provide as much speed as possible, hence the much higher power consumption.
In conclusion, it is obviously possible to configure an Atom platform to consume less than 10 watt of power if you do not mind a lower performance. Unlike the Cherrypal box, however, an Atom-based platform can also provide much higher performance (and consume substantially more power) in situations when you might need it. It is clearly a far more flexible solution in the context of desktop and mobile computers.
Quote:
And I think if the special chips in CherryPal were supported in software it would get close to Fit-PC 2 performance.
I think you are entirely wrong about this.
The MPC5200B was advertized by Freescale to achieve 760 MIPS. They say the same about the MPC5121e. The Intel N270 (1.6Ghz) that is commonly found in netbooks is said to achieve 3300 MIPS. The FitPC2 uses Z series Atom processors that are more energy efficient but about 25% slower at the same clockspeed (random source: http://www.netbux.de/wp-content/myfotos/09-april/atom-n270.vs.n280.vs.z530.jpg). So, it is not unreasonable to make the educated guess that the Z530 at 1.6Ghz should achieve about 2500 MIPS.
And, yes, MIPS is not a perfect benchmark but it is useful enough to make rough comparisons, especially when performance differences are as huge as in this case. Based on common sense, it is difficult to believe somebody could write a "magic driver" that suddenly makes the LimePC come even close to matching the Fit PC 2.
The same is true about the PowerVR unit that is in fact widely used. This not "alien" hardware that nobody knows how to write a driver for. It is comparably slow because it was designed with cellphones in mind and it is certainly a great choice for mobile appliances, which explains why it is so successful in this market space. Comparing its data sheet with desktop graphic chips from a few years back should make it quite obvious how limited the offered performance really is. Again, there is no "magic driver" that is going to overcome the fundamental limitations of this hardware.
