The weakest drives got around 3000-5000, the strongest (mainly SLC, industrial grade) may have over 100 000 but thats not the (bad) consumer grade. The drives which have been made around 1 year ago had 10 000 rated cycles. As good as anything because there was no NAND who was weaker than 10 000
rated cycles. So its kinda the new development to go even lower on NAND because they simply want to save up on cost and expensive NAND is usualy the one with higher nm. So, kinda obvious that the drives will be cheaper than 1 year ago. Not because of the speed but because of "lack on quality". Lack of quality usualy means: 25 and below is "weak", 32-50 is "good" and above 50 is industrial grade (usualy only used on SLC). In a percentage i think 50 nm can handle twice the overwrites of 32 nm and 32 nm can handle about 20% more than 25 nm. The difference might not be huge but its still a improvement, bigger is simply better on that technology. SLC compared to MLC is probably x5 for the SLC. Industry might be telling us its x10 or even higher but i think x5 is the true value.
Besides, on extreme tests (
xtremesystems.org) , the weaker 128/120 GB SSDs (such as OCZ octane) died after about 300 TB of overwrites, the Vertex 4 died after 400 TB of overwrites. Compared to theyr NAND size it means that they was able to handle about 3000 cycles. The old Vertex 3 however, could handle up to 1200 TB (10 000 cycles) in term the controller will survive. I still think, the 10 000 is probably over provisioned, although there was in fact many drives able to reach or even exceed 1000 TB, but that is kinda the top and not the endurance from a average drive.
However, the question is, does it matter? My answer would be: Not really because most likely in like 90% of the defective drives, the controller chip was going boom and the NAND wear could still be at 100% (means no NAND died due to wear).
However, take note, as far as i was able to read on the xtremesystem board, the difference between the drive can be very heavy. Some drive failed after around 80 TB while the exactly same drive on another user-test was able to keep up 800 TB (Samsung 830) so the quality difference of NAND and the SSD as a whole can be very very big and it seems like there is no real quality check...(!) There is no reason, because the powerless consumer will simply have to RMA and cry and what else, and all they can do is... not that much at all. At least it was cheap they would say... well... nowadays the drives fortunately are getting cheaper but that was not like that 1 year ago. Still, at least the overrated prices are slowly coming down the roof. However, a SSD is still about 15 times more expensive in price/size ratio than a media (high density) HDD, thats a huge difference, so the SSD clearly is here in order to provide performance, nothing else.
The manufacturer doesnt truly care because all they are looking for is that the drive is able to survive those 3+ year of warranty period, so they dont need to hand out a replacement. Of course its pretty hypocritical when they say so many good stuff about SSD reliability on theyr website, but does any of those manufacturer provide a 10 year warranty period? I guess not, and if so, they clearly say: you got 3000 cycles and when that value is exhausted no warranty (even if this condition is done in only a single month). So its not easy to deal with and in order to backup that limitation they start to even underrating theyr drives rewrite cycles and we have the situation we got today... its getting lower and lower. I find it critical if we go below 3000 because that means that there is not truly any effort into quality check and they want to increase theyr "wield" to almost 100%... thats how they may save up on production cost on the customers back.
Recently my M4 SSD died after only 3/4 year of use, the SMART data was close to perfect and still 100% NAND condition. In my experience HDDs are less vulnerable to failure. Thats not even a miracle because a HDD is a very old and well known technology and it got so much improvements over the past several decades. While the SSDs as we know today is still a pretty young technology with lot of unknown issues.
About the rated 1 million hour MTBF of a SSD (vs "only" 100 000 of a HDD), i find it a hilarious number which is in no way accurate and kinda "taken out of the air". Its pretty possible that a HDD can live up to 10 years of daily use while most SSDs may stop working already, there isnt any proof that a SSD is capable of doing that. Not a single test is able to reflect that number, its fully synthetical and based on totaly unknown or unreal values. Its kinda comparable to the several million contrast ratio of nowadays TV. Either rated like that, or not rated at all. The most wise crack of them is Samsung. They got in mind they can avoid being called a liar by telling us "Mega contrast" but its kinda the same just with lesser transparency. No matter what, all those stuff are simply fantasy numbers, and they are even allowed to spell it out.. because there is no law "against lying".
One important reason for lowering the NAND quality so much is the insane failure rate of those SSDs (in a timeline of 2 years). The main cost however is the NAND, and thats gonna be very expensive for the manufacturer for every SSD that have died. Because a piece with the value of 20$ usualy busted up the whole drive... so they surely had in mind "We rather use inferior NAND and work in improvements on the controller than to always having to throw away expensive NAND".
Note: Thats just my view as a consumer and when i watch the recent development. I have zero knowledge about the industrys background and it would be hard to know because its pretty hidden, intransparency is still a big issue.
