Dear Manufacturers, if there's a newer tech allowing better battery efficiency, don't reduce battery capacity to maintain profit margins but instead please keep the same capacity and bump up the price slightly. People will still buy them if your product has 12-18 hours battery life while the competition has a mere 4-6 hours. We really don't mind paying extra for longer worktime.
I totally agree with this, but it's less about price, and more about weight and form factor - thinner and lighter are selling points.
I'd settle for longer battery life in the same form factors we have now, but you know someone will try to sell us phones that have junk battery life and are 0.8mm thinner...
I would pay any amount of money for a high res, bright sun readable, 15 hour battery, i7 with 16 gb of memory. If that costs $5k I would pay it.
Edit: thinking about it; 'any amount' here, for me, would be $7k max, but I would pay that for this configuration. That makes it such a shame you cannot really expand laptops besides some ruggedized ones. They would make a base version for $1k and allow extension up to $5-6k with all the trimmings.
If you get an "old style" laptop, like the Lenovo T-series, you get the same 60Wh battery as you always have, fully updated to the latest tech.
Laptop makers always have had old style laptops available. You're welcome to buy them any time.
----------
Another issue though, is that ULV chips kinda suck compared to a typical Laptop M chip. Your 35W laptop chip in the bulkier Lenovo T430s is significantly faster than say... a Surface Pro (using a ULV chip).
Either way, there are laptops out there in every configuration. Small&light battery with ULV processor (Surface Pro), Big battery with ULV processor (Samsung Series 7 Ultra Touch), Big Battery with Standard Mobile Processor (Lenovo T430s).
If you don't like small and light, then pick up a bigger model. They exist, and they've been selling. If you want 10+ hours of battery life on tiny batteries, Intel even sells Clovertrail Atoms. All of which are lighter than an iPad and offer over 10 hours of battery life.
i've been holding off my laptop purchase until i got more than 8 hours of battery life. currently the best option i can think of is the x230 series, but the screen is soso.
I have expectations for haswell, and hopefully i can buy a new laptop then. But I'm not really holding my breath considering the mixed messasges we've been getting. I currently have a 480gb ssd and 8gb ram in my ultraportable. the fact that a lot of the ultrabooks have 4gb onboard, and no means to upgrade the harddrive bothers me.
Well battery life, at least as I've seen it on Macs keeps going up year after year so I'm pretty confident that trend will continue. Not sure about other manufacturers but only because I haven't kept up with them not a lack of confidence.
The title is misleading, it should be 50% more battery life than did their predecessors. What this means is that if their predecessor's lower power usage offered an additional 10 minutes of power, Haswell will now offer an extra 15 minutes.
Why do you say it means that? It doesn't look to me as if it means any such thing. I think it means: if you got 4 hours of battery life before, you may be able to get 6 hours of battery life with Haswell processors. (But I bet this will be "up to 50% extra" and will be based on overoptimistic assumptions about the power consumption of other components.)
I can't find any source with a more direct quotation of what Rani Borkar said, unfortunately.
Android can expose "which subsystem is using power" information to the user and applications, which is worth looking at. On phones, it's mainly screen and WAN radio.
That would require the CPU to currently be using at least 33% of the laptop power on average, which sounds high.
My phone is currently reporting 30% on wifi and 24% on the display - and this is a small screen device that spends most of its time with the screen off.
The story is presented to put the focus on the Haswell processor itself, and there will be saving there for sure. But I would expect that a good part if not most of the savings will come from the platform surrounding the processor. Intel has a lot of control there too, they also provide the chipset and define the requirements for the whole platform that ODMs then follow.
The PC platform for now is very inefficient. It uses standards with low power support, but typically a lot of elements are left always on and just waste power just because it is easier and up to now was not a big deal. But with the constant reduction of the CPU part now it becomes important to dynamically turn off the rest of the platform as much as possible, and not only the CPU internals. It's a given in the ARM world, and my understanding is that Intel finally brings this mindset to the PC with the Haswell platform. Which is very good and about time!
An example: the USB protocol defines Link Power Management, which allows to very dynamically put a USB link in low power (switch time is ~10 us). Supporting it will become required with Haswell from an old IDF presentation (that I've lost, so no link).
Another example: optimizing the voltages generation to avoid waste. We know part of it has been pushed into the chip with Haswell so for sure Intel looked into the whole thing.
This is obviously useful for laptops and tablets, but I would love to see this done thoroughly in mini-servers and HTPC devices. An x86 server doing nothing can burn 10 to 20W easily. A whole ARM tablet use less! I would love to see this reduced (out of principle, but also less heat and noise).
It can touch most part of the system and could be extended in time. For example in an ARM SoC for mobile it's possible to cut the internal display link between AP and LCD display if the image displayed doesn't change (the controller on the display locally cache the image). That saves the power from transferring data from the AP to the display over a high bandwidth external interface. That kind of thing could in time come to external display links.
The goal is to go all over the system with a fine comb, and make sure that when nothing happens no energy is wasted. We'll see how big a step the Haswell platform make in this direction, but it'll likely take several iterations.
Looked up up
"Power Consumption Breakdown on a Modern Laptop" by Mahesri et al. and "Component-wise energy breakdown of laptop" by Kothuru et al. on Citeseer.
According to published research from 2005 and 2010, with vintage laptop processors (which were more energy hungry than ones today), you can't get 50% more battery life by even if you eliminated the CPU power usage entirely -unless you're running a CPU pegging benchmark all day.
Probably Intel marketing managed to fool the reporter into translating CPU power efficiency changes into battery life increases.
Every time I see one of these things it is fully exciting. Then the actual product gets bogged down in patent bullshit and we never see it.
What happened to the CPU heatsink where the whole damn thing rotated? If the fan in my laptop was as efficient as that thing promised to be I'd get an extra hour or so already.
[+] [-] sigkill|13 years ago|reply
[+] [-] zdw|13 years ago|reply
I'd settle for longer battery life in the same form factors we have now, but you know someone will try to sell us phones that have junk battery life and are 0.8mm thinner...
[+] [-] tluyben2|13 years ago|reply
Edit: thinking about it; 'any amount' here, for me, would be $7k max, but I would pay that for this configuration. That makes it such a shame you cannot really expand laptops besides some ruggedized ones. They would make a base version for $1k and allow extension up to $5-6k with all the trimmings.
[+] [-] dragontamer|13 years ago|reply
Laptop makers always have had old style laptops available. You're welcome to buy them any time.
----------
Another issue though, is that ULV chips kinda suck compared to a typical Laptop M chip. Your 35W laptop chip in the bulkier Lenovo T430s is significantly faster than say... a Surface Pro (using a ULV chip).
Either way, there are laptops out there in every configuration. Small&light battery with ULV processor (Surface Pro), Big battery with ULV processor (Samsung Series 7 Ultra Touch), Big Battery with Standard Mobile Processor (Lenovo T430s).
If you don't like small and light, then pick up a bigger model. They exist, and they've been selling. If you want 10+ hours of battery life on tiny batteries, Intel even sells Clovertrail Atoms. All of which are lighter than an iPad and offer over 10 hours of battery life.
[+] [-] rjzzleep|13 years ago|reply
I have expectations for haswell, and hopefully i can buy a new laptop then. But I'm not really holding my breath considering the mixed messasges we've been getting. I currently have a 480gb ssd and 8gb ram in my ultraportable. the fact that a lot of the ultrabooks have 4gb onboard, and no means to upgrade the harddrive bothers me.
[+] [-] achughes|13 years ago|reply
[+] [-] Zikes|13 years ago|reply
[+] [-] gjm11|13 years ago|reply
I can't find any source with a more direct quotation of what Rani Borkar said, unfortunately.
[+] [-] AndrewDucker|13 years ago|reply
[+] [-] zdw|13 years ago|reply
Android can expose "which subsystem is using power" information to the user and applications, which is worth looking at. On phones, it's mainly screen and WAN radio.
[+] [-] keeperofdakeys|13 years ago|reply
[+] [-] tveita|13 years ago|reply
My phone is currently reporting 30% on wifi and 24% on the display - and this is a small screen device that spends most of its time with the screen off.
[+] [-] yaantc|13 years ago|reply
The PC platform for now is very inefficient. It uses standards with low power support, but typically a lot of elements are left always on and just waste power just because it is easier and up to now was not a big deal. But with the constant reduction of the CPU part now it becomes important to dynamically turn off the rest of the platform as much as possible, and not only the CPU internals. It's a given in the ARM world, and my understanding is that Intel finally brings this mindset to the PC with the Haswell platform. Which is very good and about time!
An example: the USB protocol defines Link Power Management, which allows to very dynamically put a USB link in low power (switch time is ~10 us). Supporting it will become required with Haswell from an old IDF presentation (that I've lost, so no link).
Another example: optimizing the voltages generation to avoid waste. We know part of it has been pushed into the chip with Haswell so for sure Intel looked into the whole thing.
This is obviously useful for laptops and tablets, but I would love to see this done thoroughly in mini-servers and HTPC devices. An x86 server doing nothing can burn 10 to 20W easily. A whole ARM tablet use less! I would love to see this reduced (out of principle, but also less heat and noise).
It can touch most part of the system and could be extended in time. For example in an ARM SoC for mobile it's possible to cut the internal display link between AP and LCD display if the image displayed doesn't change (the controller on the display locally cache the image). That saves the power from transferring data from the AP to the display over a high bandwidth external interface. That kind of thing could in time come to external display links.
The goal is to go all over the system with a fine comb, and make sure that when nothing happens no energy is wasted. We'll see how big a step the Haswell platform make in this direction, but it'll likely take several iterations.
[+] [-] bornhuetter|13 years ago|reply
[+] [-] zurn|13 years ago|reply
According to published research from 2005 and 2010, with vintage laptop processors (which were more energy hungry than ones today), you can't get 50% more battery life by even if you eliminated the CPU power usage entirely -unless you're running a CPU pegging benchmark all day.
Probably Intel marketing managed to fool the reporter into translating CPU power efficiency changes into battery life increases.
[+] [-] pi18n|13 years ago|reply
What happened to the CPU heatsink where the whole damn thing rotated? If the fan in my laptop was as efficient as that thing promised to be I'd get an extra hour or so already.
[+] [-] seunosewa|13 years ago|reply
[+] [-] whichdan|13 years ago|reply
[+] [-] macavity23|13 years ago|reply
[+] [-] bornhuetter|13 years ago|reply
[+] [-] dragontamer|13 years ago|reply
50% more efficiency entails a 50% heat reduction. This battery issue is a very very good thing for Intel to tackle.
[+] [-] zokier|13 years ago|reply
[+] [-] AndreasFrom|13 years ago|reply
[+] [-] Create|13 years ago|reply
https://en.wikipedia.org/wiki/Transactional_Synchronization_...
[+] [-] pekk|13 years ago|reply