Doom Marine Posted November 30, 2012 While we share a different technological paradigm than cyberdemon butt, he has brought up important points concerning the sacrifice of endurance in the reduction of costs. 1. MLC is acceptable, TLC is approaching toilet paper disposability, no thank you. 2. Which is sooner to happen given similar start-times and regular usage: an SSD running out of write/erase cycles, or a spinning drive fucking up? Another factor affecting write/erase endurance is the ever-shrinking lithography process. When SSDs were 34nm, their write/erase endurance were rated at 5,000. With current 25nm SSDs, it's reduced to 3,000. TLC makes it even worse, down to 1,000. The question is, at what point will the end-user find it unacceptable? NAND flash doesn't seem to have a very bright future, given phase-change memory and memristors are on the horizon. 0 Share this post Link to post
Super Jamie Posted November 30, 2012 Doom Marine said:Which is sooner to happen given similar start-times and regular usage: an SSD running out of write/erase cycles, or a spinning drive fucking up? Does it matter? Existing electronic component failure rates tend to indicate the drive controller will die before either the platter or flash, in most cases. 0 Share this post Link to post
printz Posted November 30, 2012 I'd rather the software become obsolete before any hardware fails. 0 Share this post Link to post
Maes Posted November 30, 2012 Super Jamie said:Does it matter? Existing electronic component failure rates tend to indicate the drive controller will die before either the platter or flash, in most cases. Get a drive controller chip from a HDD. Throw in several like it on a SSD ????? PROFIT!!! (If you sell or service SSDs, that is...) 0 Share this post Link to post
Doom Marine Posted November 30, 2012 He's saying that HDD's controllers are more reliable, and that throwing such reliable controllers (in multiple redundancies) onto SSDs would solve the hardware controller reliability issue. ...I was able to extract useful information out of trolling. I'm just that good. 0 Share this post Link to post
exp(x) Posted November 30, 2012 Doom Marine said:given phase-change memory and memristors are on the horizon Not to mention STT-MRAM which has effectively infinite endurance. EDIT: I suppose that falls under memristors 0 Share this post Link to post
Ladna Posted November 30, 2012 SSDs are amazing. I have a 60GB OCZ Vertex2 and it changed my life. A simple look at even the worst SSD benchmarks show them blowing HDDs away, even fast HDDs like 15k Cheetah drives or whatever. I also have a 1.5TB RAID 10 array (4 750GB drives), and I back all this up onto a 2TB disk over eSATA. Any technology can fail and you'll find horror stories everywhere. Buy an SSD, put all the stuff you use every day on it, and back your stuff up once if not twice. Don't listen to haters. Also Fraggle won the thread like 500 years ago. 0 Share this post Link to post
Doom Marine Posted December 1, 2012 Maes said:A man's troll is another man's gadfly. What shall be the appropriate punishment for you? Corrupter of Youth? 0 Share this post Link to post
Doom Marine Posted December 2, 2012 Lookie here, another potential paradigm change for the lifespan of flash memory: Taiwan engineers defeat limits of flash memory 0 Share this post Link to post
printz Posted December 2, 2012 Ladna said:Any technology can fail and you'll find horror stories everywhere. I hate when this phrase is used to excuse bad hardware. And not any technology can fail, some survive long enough until their service is useless, others die too soon and cause losses. THERE'S A DIFFERENCE. 0 Share this post Link to post
exp(x) Posted December 2, 2012 printz said:I hate when this phrase is used to excuse bad hardware. And not any technology can fail, some survive long enough until their service is useless, others die too soon and cause losses. THERE'S A DIFFERENCE. Do you have any idea how hard it is to get billions of deep submicron transistors and interconnects to work perfectly and detect any that aren't working or will fail before they should? 0 Share this post Link to post
Super Jamie Posted December 2, 2012 printz, are you making the argument that we've had ~60 years since the transistor went into most retail electronics, and we should be able to get reliability much better than 97% ±2% by now? If so, I'd never considered that point before, but I agree. 0 Share this post Link to post
Maes Posted December 2, 2012 When you multiply an apparently low error rate (1 in 10^12 is common today, for conventional hard disks, ) with a faster increase in media size or read/write operations volume, interesting things start to happen, to say the least ;-) 0 Share this post Link to post
Super Jamie Posted December 2, 2012 Completely forget just hard drives, all electronics have a failure rate. TVs, dishwashers, car alarms, defibrillators, space shuttles. Haven't we been doing electronics long enough to get it right? Evidently not. 0 Share this post Link to post
Doom Marine Posted December 3, 2012 Super Jamie said:Completely forget just hard drives, all electronics have a failure rate. TVs, dishwashers, car alarms, defibrillators, space shuttles. Haven't we been doing electronics long enough to get it right? Evidently not. Of course we can get it right, but it will cost more than the optimal price in a competitive market, thus won't sell in volumes that would maximize profitability. It's not just about the quality of the technology, but cost-effectiveness as well. For SSDs, take Intel vs OCZ for example: where Intel is one of the best in terms of reliability, OCZ just makes cheap products that are more error-prone, but sells in much higher volume. 0 Share this post Link to post
Bucket Posted December 3, 2012 Apparently printz has the inside track on when different kinds of hardware are meant to fail, and which are designed to fail before they're meant to fail. 0 Share this post Link to post