How I Got Full Windows XP Installed Under 2GB!
in [Laptops]
|
From: BillW50 on 4 Feb 2010 11:53 In news:hka924$8ui$1(a)reader2.panix.com, the wharf rat typed on Tue, 2 Feb 2010 22:30:28 +0000 (UTC): > In article <hk9f0f$rkc$1(a)news.eternal-september.org>, > BillW50 <BillW50(a)aol.kom> wrote: >> >> Secondly, these things are mass-produced. And being mass-produced, >> you only need a small sample. >> > > That's a false statement. Modern manufacturing uses statistical > quality control methods to reduce testing sample sizes by controlling > the process, but it doesn't eliminate quality testing. The samples > are much smaller than you might predict but a sample of 12 out of a > billion would never work. Well I never said I could speak for 12 billion flash drives. So get that idea right out of your head. Secondly, I've done my research and I am convinced they are reliable. In your mind, they are not. Which is fine by me. As there are more and more people learning the very same thing. And they are just becoming more and more popular than ever before. And before there was any study that I knew about the reliability of hard drives. I did my own study. I did a study on just 20 hard drives over 20 years and I came up with just under 7% early failure rate. I didn't know how close I really was until. Google later did their own study of hard drive reliability of 100,000 over 5 years. And their study showed I was only off by a small margin. Although if I only counted the drives just the last five years, my numbers would have been much lower. As my experience is that hard drives are generally far more reliable in recent years than they were 20 years ago. >> For example, when a cook makes a meal to feed an army. He/she tastes >> a small sample to make sure they made it with all of the right >> ingredients. Now the cook doesn't eat all of the food to test it, >> now do they? > > A cook performs many samples over time of a single item. Typical > mass production manufacturing performs a small number of samples over > time of a large population of items. Each sample size is dictated by > required tolerances (margins of error) and population size. > > (Anyway, cooking's an art not a science. Each dish is a little bit > different. One of the challenges a chef faces is developing > consistency.) I think many would disagree with you there. As that Taiwan electrolytic capacitor fiasco about 9 years ago is a good example. As *recipes* for making components are a well kept secret. Although a Taiwan company had stolen the *recipe* from a Japanese company and started to produce their own. But get this, the stolen *recipe* lacked one very important *ingredient*. And all of those Taiwan caps started failing left and right after about a year. I have one $800 Avatar computer that became worthless in 11 months because of this. The last time it was turned on, the video was in B&W and you could only run DOS. Anything running in 32 bit just wouldn't work. Manufacturing electronic components has their recipes, ingredients, batches, baking, etc. just like cooking does. As they use many of the same terminology. Thus I don't see them much different like you do. By the way, I was involved in component manufacturing when I worked for Magnavox. So were did you get your experience from again? >> I have all of these Asus 701/702 netbooks. Now are mine any different >> than other people's Asus 701/702 netbooks? Nope not really. > > That's another mistake. They are very different. Each part is > identical to the other parts in other laptops within certain > tolerances. > The machine is designed to work as long as each part is within those > tolerances even if it's slightly "different". You get very > interesting failures when several parts are just barely out of spec. My Asus notebooks work just like anybody else. You can talk all you want to but it doesn't change the truth. Of course components has to be within spec. Otherwise it is a failure. But there are millions of these things out there still working. Reports of failures are far and few in between. >> Your logic is far different than any I ever met, except from bozo >> the clown. > > If you actually remember Bozo you'd also remember that he > solved more mysteries than Scooby Doo. (The cartoon Bozo, that is.) Sorry, no I don't remember. Too involved with science and engineering I guess. >> For starters, take a 4GB SLC flash drive. To wear one out, you need >> to write 400TB worth of data to wear out each cell. > > No, you only need to have a failure in an important part of the > array. Because of file system structure a couple of dead cells in > the first meg or so would probably be fatal. > >> Second of all, solid state drives have proved reliable and are used >> for a main operating system. As many use them all of the time > > Many people buy lottery tickets, don't back up their data, and use > their birthday as their Paypal password. It's not correct because > everyone does it. That's called the "appeal to common practice", and > is shown to be bad logic by everyone's mother: > > "If Johnny jumped off the roof would YOU jump off the roof too??" Apples and oranges. I have no problems running Windows or Linux on flash drives and I do so everyday. You can believe they are not reliable as much as you wish. It doesn't bother me one bit. And I have found flash drives many times more reliable than hard drives so far. >> And in the next >> year or two, half of all laptops are said will be using the solid >> state type instead of conventional hard drives. >> > Oh, and I suppose they'll all be using Rambus memory and we'll be > driving to Best Buy in our flying cars? Nope, I haven't heard that one yet. >> And no, they are not too expensive. You can buy laptops with 1TB or >> more with a solid state drive right now. And some of the cheapest >> netbooks are using solid state drives already. So they are very >> affordable. > > Terabyte SSD drives retail for about $3,800. That's hardly price > competitive with traditional magenetic storage. That isn't a lot. The first GUI personal computer was designed back in '71 with a list price of $10,000. My first laptop back in '84 cost over $2,000. People who buys powerful game machines and servers, this is reasonable. I've heard that some servers are now using flash drives instead of hard drives. >> You haven't done any research at all. > > At least I looked up the price of the things on newegg before > talking about how affordable they are. $3800 isn't affordable unless > you're talking about a car. When you are talking about powerful laptops or desktops, that is a common price range. And it doesn't sound out of line to me. After all back in the late 80's I bought $1000 worth of GEOS OS and software applications. Sounds a bit expensive I know. But it was still cheaper than a Mac. <grin> And I know you like to argue and are often full of it. But I know that the prices will come down very quickly in the next few years at any rate. I know you said you don't believe it, while history shows you are wrong in either case. -- Bill Gateway M465e ('06 era) - Windows XP SP3
From: the wharf rat on 5 Feb 2010 02:31 In article <hkeu26$9kg$1(a)news.eternal-september.org>, BillW50 <BillW50(a)aol.kom> wrote: > >Well I never said I could speak for 12 billion flash drives. Sure you did. You said that you'd sampled 12 of them and so you could conclude that all 12 billion will never fail because none of yours had. Which IS kinda silly, don't you think? >drives. I did my own study. I did a study on just 20 hard drives over 20 >years and I came up with just under 7% early failure rate. I didn't know >how close I really was until. Ok, I have to ask: Which 40% of that second drive failed? The top half or the bottom half? > >Manufacturing electronic components has their recipes, ingredients, >batches, baking, etc. just like cooking does. As they use many of the >same terminology. Thus I don't see them much different like you do. By Oh, great Ghu, no. Modern mass manufacturing carefully designs processes to eliminate as much variability as possible. In fact, you try to eliminate the actual *variables*. What you can't eliminate you design elaborate process controls for. The end result really *IS* a sort of recipe that you can more or less duplicate at will (but not entirely: look at the scrap rates at different chip fabs for instance). But cooking CAN'T be reduced to recipes because there's no way to eliminate the myriad variables, and because it's a manual process you can't really build in good controls. That's why any old person can't just follow a recipe and turn out a dish worthy of a four star chef. Take a simple thing like baking a loaf of bread: how alive is the yeast? How much moisture is in the flour? How warm is the room? It's a much more fluid and dynamic process than anything you see in a manufacturing plant. >My Asus notebooks work just like anybody else. They may WORK like all the other samples but they're NOT identical. They don't clone them, you know. They build them out of parts, each of which varies just slightly from all the rest. Sometimes you get one that varies TOO much. Those are the ones people complain about on Usenet. >> Terabyte SSD drives retail for about $3,800. > >That isn't a lot. That's an awful lot for just a terabyte drive. I can build an entire SAN array for 4 grand. Fibre channel.
From: BillW50 on 7 Feb 2010 09:36 the wharf rat wrote: > In article <hkeu26$9kg$1(a)news.eternal-september.org>, > BillW50 <BillW50(a)aol.kom> wrote: >> Well I never said I could speak for 12 billion flash drives. > > Sure you did. You said that you'd sampled 12 of them and so you > could conclude that all 12 billion will never fail because none of yours > had. Which IS kinda silly, don't you think? I said no such thing! I also said I have a friend living in France that burns up his cheesy flash drives in two months. Which I find is odd. As I have never seen any that bad yet over on this side of the big pond. >> drives. I did my own study. I did a study on just 20 hard drives over 20 >> years and I came up with just under 7% early failure rate. I didn't know >> how close I really was until. > > Ok, I have to ask: Which 40% of that second drive failed? The > top half or the bottom half? What? No, I had 3 out of 20 drives fail within 20 years. That is just under 7% failure rate. Google tested 100,000 drives and came up with a very similar failure rate. They had a smaller percentage, but Google's was only tested for 5 years and not 20 years like my study. So the slight difference makes since. >> Manufacturing electronic components has their recipes, ingredients, >> batches, baking, etc. just like cooking does. As they use many of the >> same terminology. Thus I don't see them much different like you do. By > > Oh, great Ghu, no. Modern mass manufacturing carefully designs > processes to eliminate as much variability as possible. In fact, you try > to eliminate the actual *variables*. What you can't eliminate you design > elaborate process controls for. The end result really *IS* a sort of > recipe that you can more or less duplicate at will (but not entirely: > look at the scrap rates at different chip fabs for instance). Same holds true for cooking as well. I had worked for a time for a company who manufactured food ingredients for other processed food companies. And the same thing applies. And they threw out tons of ingredients from time to time because a given batch didn't come out to their high standards. > But cooking CAN'T be reduced to recipes because there's no way > to eliminate the myriad variables, and because it's a manual process you > can't really build in good controls. That's why any old person can't just > follow a recipe and turn out a dish worthy of a four star chef. Take a simple > thing like baking a loaf of bread: how alive is the yeast? How much > moisture is in the flour? How warm is the room? It's a much more fluid > and dynamic process than anything you see in a manufacturing plant. Same is true in component manufacturing. I worked in both fields. Where did you get yours again? >> My Asus notebooks work just like anybody else. > > They may WORK like all the other samples but they're NOT identical. > They don't clone them, you know. They build them out of parts, each of which > varies just slightly from all the rest. Sometimes you get one that varies > TOO much. Those are the ones people complain about on Usenet. Of course. Very few of them though. As the vast majority of them work as they are suppose to. And if you are one of the unlucky ones, you just take it back and get one that actually works. >>> Terabyte SSD drives retail for about $3,800. >> That isn't a lot. > > That's an awful lot for just a terabyte drive. I can build an > entire SAN array for 4 grand. Fibre channel. My first laptop which costs $2000 back in '84. And that was a lot of money for just 128kb of storage too. And I would have gladly paid 4 grand more for an extra terabyte drive. Or even just a gigabyte drive. ;-) -- Bill Asus EEE PC 702G4 ~ 2GB RAM ~ 16GB-SDHC Ubuntu 9.10 Netbook Remix
From: the wharf rat on 7 Feb 2010 15:34
In article <hkmj5j$5mt$1(a)news.eternal-september.org>, BillW50 <BillW50(a)aol.kom> wrote: > >What? No, I had 3 out of 20 drives fail within 20 years. That is just >under 7% failure rate. Google tested 100,000 drives and came up with a >very similar failure rate. They had a smaller percentage, but Google's >was only tested for 5 years and not 20 years like my study. So the >slight difference makes since. > Oh, for crying out loud... So you're telling me that you ran a sample of 20 drives for 20 years under controlled conditions, and during that time you had three failures. You're therefore calculating a failure rate of 3 years out of 400 "drive years" and improperly rounding down to 7%. That's complete and utter bullshit. You didn't run an array under controlled conditions for 20 years. And if you did you'd have a 100% failure rate because NO drives will run for 20 years under ANY conditions. We'll overlook your mathematical errors since your data is imaginary. Or maybe you're telling me that during a 20 year period you owned a total of 20 drives, three of which failed in some way, and that you don't think the fact that they had 20 different hardware architectures and 20 different working environments is material. If I could stop laughing I'd try to explain why asserting that this is a 7% failure rate is ridiculous but I can't so I'll simply point out that 3/20 =.15 not .07. Your constant argument by authority would be more convincing if you sounded like an authority rather than a lame attempt to pass the Turing test. >Same holds true for cooking as well. I had worked for a time for a >company who manufactured food ingredients for other processed food >companies. And the same thing applies. And they threw out tons of >ingredients from time to time because a given batch didn't come out to >their high standards. That's not cooking. THAT is manufacturing. |