Slashdot Mirror
AMD — "We're Not Entirely Honest" About Batteries
Slatterz writes "In an apparent attack of the bleeding-obvious, an AMD rep has come clean and admitted (on behalf of the industry) that notebook and phone battery life figures are completely unreliable. AMD's senior vice president Nigel Dessau says that 'we are not being entirely honest with users about what PC battery life they can expect to actually experience.' He says AMD will now use a combination of idle time (where the machine is left to sit idle, and timed to see how long it takes for the battery to go dead), and 3DMark06 to measure battery life. Great in theory but some of the industry already bases battery figures on a two-test measurement, and the results are still wildly inaccurate."
"We're Not Entirely Honest" = We've been lying
This happens in every industry. Every industry picks a low baseline to measure their product by and then shows how high their product scores over that. I really never assume that things like "battery life, sound clarity, brightness, etc" will be THAT accurate because the bottom line is the manufacturer wants to sell this to me and the better he makes his product sound the more likely I am to buy it. I don't hold any ill will towards people who do this. It's called marketing and it happens all the time.
AMD: "We're not being entirely honest about our processor speeds... We're decided to use an external scale against which to measure our /actual/ CPU performance. ...(AMD) will market our new processors as having a "Performance Rating", which are not equal to, nor based directly upon the /physical/ oscillations of the chip itself. Instead we intend to include such factors as idle time (cut - Ed.)"
This is a concoction. My story, and their plan. Why, AMD?
1) Fully load the machine
2) Time until battery death
3) Advertise "minimum" battery life
What is wrong with that? Then I can expect at least 40 minutes of battery life and anything more than that is nice. You will generally not be fully loading the machine so it will always be more than 40 minutes anyway..
The disappearing pencil trick. Let me show you it.
They say they let it site idle, then do an intensive test, and average the two times to get some inaccurate number. Why not just present both numbers, and let users decide what that translates to for their usage? As the article even refers to, cars report two mileage numbers, so the idea isn't new.
My webcomic
battery lifetime. I maintain about 200 laptops, and the damn batteries are usually completely useless after about a year. Oh, and of course, the laptop has a 3 year warranty, and the batteries have 1 year warranties. You can extend that to two years -- it'll only cost you about as much as a second battery would to do so.
I am not left-handed, either!
It's only really useful in comparison with other models. Your actual mileage (or battery time) can and will vary depending on usage and maintenance.
Good luck running 3DMark06 on anything else than a Windows-based computer. No Linux, no Mac, no phones, etc.
Is being entirely honest.
..at least they're being honest about not being entirely honest..
Once I can recharge my battery in 4.5 seconds will it matter?
Nope.
Exaggeration when concerning any companys Product Isnt anything revolutionary. It kind of comes with the whole "buisiness" Career.
That's great, a clearly hostile post on how crappy AMD's battery tests were and will continue to be. Only there is no alternative suggested, or even a hint that the poster thinks there is an effective alternative. Battery life depends on usage and there is no good test. Get over it.
Whale
I can see more variance in cellphones because those are devices that are on 24 hours on battery and usage patterns are reflective of how many minutes a person has. So someone with 1000 minutes and unlimited sms/data is going to use theirs a lot faster than I, with 550 shared minutes and no data.
On laptops, I think we can get a little more predictability. First of all, I'd venture to say that at least 80% of the time, if the laptop is on battery, it's being used. I don't know of too many people who fire up a laptop and walk off. However the variance is in the type of use. A photoshopper or developer is going to put a lot more stress on the battery than a Word/IE user. A teen is going to stress it more than a octogenarian. And a gamer is going to beat it down more than anyway. Well, maybe not as someone folding@home.
I think the solution for this is for someone with enough clout to develop a standard test that cycles through heavy/light load every 20 minutes. Let it run until it powers off. I think this should be a 'measurement company' such as futuremark. HP/Apple/Dell are never going to agree on a test, but if futuremark creats 'wattmark' and it becomes standard, they'll all use it.
At that point the consumer can say, "Ok, this machine gets 6 hours on wattmark, I'm a LIGHT user, and I usually get 20% more than wattmark" or "I'm a gamer, and I only get half what wattmark says"
But with the vendors publishing their own magic numbers, and consumer has NO idea what THEY can expect out of that machine/battery.
Do not meddle in the affairs of sysadmins, for they are subtle, and quick to anger.
I'm shocked! SHOCKED I tell you! You mean the phones/laptops don't run as long as advertised? I can't believe this! It's impossible! Next you'll tell me a 8GB pendrive/SSD holds less than 2^33 bytes.
Shouldn't this be posted in the 'idle' section?
This guy's the limit!
Most folks are not going to tap their machines to run 100% on battery, like the 3dmark tests do, but they sure as hell won't leave it sitting idle. So what is the answer? Simple, what DO most folks do while they are on their laptop? Well, from what I have seen that is web browsing, webmail, IM, and document creation/editing.
It really shouldn't be hard to simulate those uses. Since you can get an Open Source app to do each of these jobs you could just build a testing suite consisting of FF3,OO.o, and pidgin and run it, having those apps fed some simulated work(a document fo Writer,a few tabs for FF3, and some basic chat for Pidgin) and see how long the batteries last. I don't know about you but I would rather have a number based on "average Joe" usage than the crap numbers they pushed before or the even more pointless numbers they will be pushing now. Then I would have a real rough estimate of what to expect and could shop accordingly.
Certainly seems like a better way IMHO than some 50/50 split between 3dmark and idle, don't you think?
ACs don't waste your time replying, your posts are never seen by me.
>>This happens in every industry
This is a bit different from a breakfast cereal saying "now even tastier" or a soap promising "more suds!" The first is subjective (personal preference) but the second is objective -- it can be quantified and proven/disproven.
In this case with batteries, rather than taking an actual measurement of performance, the industry is building an estimate from a combination of measured behavior + a calculation based on a performance variable. It's no different than the automobile industry stating "EPA Estimated MPG city/highway" which is not based on a dynamometer test or actual performance measurement but instead is calculated based on the amount of CO2 which exits the exhaust pipe of the car! Is it any wonder, then, that hybrid cars which shut off their gasoline engine when stopped and at low speed/light acceleration, would give grossly inflated figures? Well, they did (and do), which explains why real-world MPG is often far less than this calculated (not even simulated) performance.
In short, they're both lying and it's obvious. Yet companies wonder why consumers are so cynical and therefore difficult to reach with advertising.
What is needed is real-world testing -- dynamometer ("rolling test track") testing for autos where the wind resistance, temperature, barometric pressure, etc. can all be carefully controlled. Similarly with computers, a pure performance-based measurement is needed which should account for idle time, network activity, etc. Just as an automobile is not tested at full-throttle for 3 hours, neither should a PC, but instead a variety of benchmarks (gaming, web browsing, spreadsheet, word processing, ???) could show performance figures for various activities.
In short, manufacturers, we want real numbers free of hype.
Um, stupid idea, I know, but: Why not just gather voluntary data from actual users, preferably thousands of them, and use that?
Are YOU using the TOOL, or is the TOOL using YOU? Think about it!
Portable CD players. The maximum battery life hours on the box never mentioned if the tests had anti-shock turned on, how often the device was turned on and off, or the volume settings used (equalizers, bass boost, & volume). The reason for the ignorance for these details is the portability of AA batteries. Now that we live in an age of more fuel cell technology than we know what what to do with, shouldn't we save ourselves a headache and atleast make some standardized benchmark tests?
Give us "real" numbers like how long the battery will last sitting in a drawer or under "full load" in a particular device, and how long the battery will last under a variety of scenarios.
An "emergency" phone user is more interested in how long they can leave their phone in their glove compartment before recharging.
"Light" users want to know standby time and how many minutes of standby time they lose for every minute they talk.
"Heavy" users are more interested in talk time and how much "talk time" they lose if they leave their phone on but not charging overnight.
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
Comment removed based on user account deletion
I wrote a battery driver for a Windows CE device once. Here's how we did that.
There is an A/D line on the AC97 codec that we use as a measurement probe to the battery. Used that to determine the actual voltage being seen. Charged the device 24 hours, and ran a program that dumped that output to a file until it died.
Then fit a third order polynomial to the data. We use that to predict where you're at percentage-wise on the draining curve. Then we made the mistake of looking at the metrics for other batteries we got from the manufacturer.
As it turns out, the characteristics from one battery to the next varied wildly. Even after you average a dozen or so batteries you'd still get better results throwing darts at a dartboard.
In short, that 3DMark06 test is probably reading battery capacity from something similar. That would be worth looking at for another source of possibly bogus readings.
Weaselmancer
rediculous.
I'm sure the 5% or so of laptop users that have an AMD chip are absolutely crushed.
(feel free to check my figures, as I pulled them from the nether -- my nethers.)
It is by my will alone my thoughts acquire motion; it is by the juice of the coffee bean that the thoughts acquire speed
And fudging number has been around since the beginning of the industrial revolution. The story goes that James Watt wanted to sell his steam engine based on the number of horses replace, or equivalent. So he measured the amount or work horses could do over a period of time. The story goes he did not make his horses work very hard, and came out with a very low power. I am sure his reasoning was 'sustained work' was what was important. In any case, he was forced to up this number, but is was still considered low. But this is number we have. The horsepower is the amount of a work a unmotivated tired weak horse can do. The battery life is the maximum one can expect when on is not using the device for anything. The rated miles per gallon on a car is valid if one is driving around a parking lot at a constant speed with no accesories on and no one, not even the driver, in the car.
Any number listed in advertising copy is solely for advertising purposes. That is the rule. If everyone uses the same basis, no matter how flawed, such as horsepower, it is a fair relative comparison. Though the exact number might not make any sense, it is useful for ordering. In a educated society, therefore, we would use a scaled number rather than a fixed number that implies some level or precision and accuracy. But, as stated, we are not even educated enough to understand there is nothing magical about an set of headphones. Thank the gods for that, otherwise many companies would be out business and we would be in greater trouble than we are.
"She's a scientist and a lesbian. She's not going to let it slide." Orphan Black
base your battery life figures on full consumption specification of all the components at operating temperatures.
simple math which instantly gives you the worst case scenario, which is probably more reliable than current methods.
also provide a scaled estimate based on battery age from an average number of charges.
They're using their grammar skills there.
1000s of busy bee lawyers are buzzing about their offices scheming as we speak...
... isn't some developer's idea of benchmarks, but estimates based on actual Real World Use.
For example - my laptop* will last a day or two unplugged in sleep mode. There's one metric for you - how long will it last if you stick it in your bag and forget about it?
My laptop will last maybe an hour - if that - running Photoshop. That stresses the disk, the ram, pegs the processor, etc.
It might last through a DVD if I'm not doing anything else. Optical drive is the big drain here - newer machines can handle the rest while barely ticking over.
Slurping a bigass (4+ gig) video file off of a thumb drive through a USB1 port completely drained the battery. That's sustained, heavy use of ports and disk.
Ultimately, I don't care how long the laptop battery lasts with "average use." I care about how well it holds up to the very extremes - extreme neglect (how long can it sleep on a full or moderate charge?), regular-for-me use (fullbore photoshop until the charge meter hits red), and Worst Case Scenario (I'm in the middle of the country trying to get all of my critical files copied onto my ipod or other laptop-powered USB drive).
Apple's battery monitoring software gives a fairly accurate, frequently updated guesstimate of battery time remaining and charge time remaining - and you're only going to get those figures through stressing the machine the way you'll be using it - not through skimming the interwebs for benchmarks that care deeply about ways you'll never use the hardware.
In my opinion, it may be easier to develop better battery monitors than it will be to develop and publish accurate test suites.
* An old 12" Powerbook G4 that's way past its prime. Newer lappies will get longer life for these scenarios but I think they stand as ballpark figures.
I'm shocked! SHOCKED I tell you! You mean the phones/laptops don't run as long as advertised? I can't believe this! It's impossible! Next you'll tell me a 8GB pendrive/SSD holds less than 2^33 bytes.
Yeah, well they did something a little sneaky there - which is they don't use the same definition for "Gigabyte" as computer scientists often do...
Specifically, in this case, they made each byte only 7.45 bits instead of the full 8 bits. Hence, when you boot up your OS and check the drive capacity, it'll say 7.45 GB instead of 8GB.
Bow-ties are cool.
So What?... As long as they begin to publish true measured figures (Idle & Loaded) as opposed to some guess that they would make, then spin in the marketing department, no-one can complain if their own results vary. All the manufacturer can do is publish what they can measure. Any specific user who claims that their performance differs needs to understand that that there is no more reliable way to predict performance for their specific situation. Therefor, this can only be seen as an improvement.
Keep passing the open windows...
Thank you for reminding me to charge my cellphone.
Battery life figures are utterly useless. CPU load varies from user to user (consider the guy who is checking email, vs the guy who is gaming) to the point where any attempt to measure battery life, other than full tilt 100% cpu utilization, is doomed to failure.
It'd be better to produce the 100% CPU battery life figure since then everyone's expectations are managed and all will perform at or above the published battery life numbers in the field.
The industry is ridiculous and opens itself up to class action lawsuits.
-Viz
Don't kid yourself. It's the size of the regexp AND how you use it that counts.
Argue all you want about whether the EPA mileage tests accurately reflect reality... at least you can be somewhat confident that manufacturer to manufacturer, car to car, the test was conducted in a CONSISTENT manner. That way you can compare one rating to another and use the information to make a decision about the RELATIVE performance of each vehicle.
What is needed is an industry standard way of measuring battery performance. Preferably one that:
1. Provides numbers based on different usage scenarios (running cpuburn/gpuburn, watching a DVD/BluRay, surfing the web, responding to emails while beastly Outlook is running, and sitting completely idle).
2. At least attempts to provide some gauge of how the stated performance will drop over time.
3. Is run against a relatively large number of samples, since we know that unit-to-unit variance can be very high.
Will this be costly? YES. Will it be the ultimate panacea that guarantees we will know EXACTLY how a battery will do with our unique usage patterns? Hell no.
But this is would be a LOT better than what we have now, which is every manufacturer making up their own benchmarking standard and publishing figures that are essentially impossible to compare from manufacturer to manufacturer and from model to model.
Who do battery operated computer devices use just one battery, when there are at least three distinct types of electrical demand that go on? Seems like they should have two or three different types of batteries, and the device is smart enough to switch to the appropriate one quickly. You have high draw and demand, max CPU, your drive spinning, the fans kick on, your wireless is activated, then there is more casual use, then sleep/hibernation. Couldn't different types of batteries address this better than trying a one size fits nothing perfectly scenario?
Same with electric or hybrid cars, they need fast huge release type batteries or ultracaps (or hydraulic pressurized type storage) just to get up to speed from a stop and for braking regneration, then it shoud switch to long haul slow release type batts (possibly much less expensive batts then as well).
of being honest with your mistakes and PR lies ? well. thats something new alright. intel, nvidia, microsoft would have taken a different approach than what you have done though.
Read radical news here
put a DVD in the drive and see how many minutes you get before the playback stops...
Donald 'Duck' Dunn: We had a band powerful enough to turn goat piss into gasoline.
Batteries are for cars and machinery. Hand-cranks are for laptops...
Egads, did you step off a time machine, or something? Times Square has been Disney clean for well over a decade...
How many hours playing WoW in a Starbucks does it do? How many hours could it do after the first charge out of the box, and how many hours did it do after 6 months of doing it every day? I am not interested in any other algorithm for measuring a laptop's battery life, unless it's a game that also uses wifi and graphics continuously and is better than WoW.
We have effective laws against false advertising now. If battery claims cannot be verified under the same conditions, companies should be sued out of business. The same goes true for automobile manufacturers. I have yet to see a car that gets even close to published MPG figures under normal conditions.
How can these companies continue lying to us and get away with it? Even if there is some kind of special estimate that they do (I don't know, removing certain services that come with the laptop, running the test at 45 degrees Fahrenheit, driving the car down a mountain with a tailwind, etc), that shouldn't be allowed. The figures aren't "MPG under ideal conditions", they are "Estimated Highway MPG" and "Estimated Battery Life".\