The market is already saturated with Sandforce 2281 drives. Why bring another into the fray? The specs aren't even at the top of the pack of other drives that have been out for a long time now.
I think Hynix is doing this to build distribution channels for their future LAMD based drives. It's easier to begin with a lower volume product and test several distributors and then choose the best one. LAMD based SSDs definitely have potential to sell like hot cakes if the performance figures are to believe, so you don't want to screw up supply because of distribution issues.
Well, it doesn't always work like that. While you get more dies per wafer (assuming each die is still 64Gb), your yields are lower at first. 25nm process is now mature and yields are better than ever. Demand can also be high compared to supply, which keeps prices high.
Smaller process node does not mean an immediate price drop. I would say NAND price drops are fairly linear.
Yeah, that's why I said that it should be cheaper. Another factor you didn't mention is that this is so far the only known SSD to use 20nm NAND, which means there would be no competition in this area. Hence, there would be little reason to not charge a price that is only marginally lower until someone else introduces their own SSD with 20nm NAND.
The reason you posted above though is also reasonable and I didn't consider it. I guess I assumed that Hynix already knows who they want to distribute stuff through.
>Smaller process does not mean cheaper prices. First of all, I said it should be cheaper, not will be. On the contrary, it does. If it weren't (eventually) cheaper to use smaller process nodes, there would be absolutely no point in using them. Sandy Bridge vs Ivy Bridge is a good example. IB CPUs are better in every way, yet they are basically sitting at the same prices as SB before IB's release. Not a cheaper price but definitely more bang for your buck.
>Waffer prices goes up with smaller process. I assume you mean wafer, as I have never heard of a waffer. At first, yes. Eventually, however, the cost of fabricating chips at any particular process node drops to lower than the one before.
>That is why Marjory of micro processors are produced in 65nm or larger. I assume you mean majority since marjory is also something I've never heard of. Otherwise, I find that hard to believe. Intel, AMD and Nvidia have been releasing all of their microprocessors at smaller nodes since 2009. There is also the fact that virtually every consumer SSD uses NAND smaller than 65nm.
>About 2-3% is 28nm or less. Ignoring the fact that every SSD I know uses 25nm NAND and has done so since at least 2011, it is only recently that nodes smaller than 29nm became mature enough to move microprocessor fabrication to it. Hence, it's quite obvious that very few processors in use today would be using a node less than 29nm, if your figures are accurate that is.
If you read the Hynix press announcement it is 20nm "class" not 20nm. Samsung used the same PR trick to indicate anything from 29-20nm awhile ago when selling 27nm. If this was really 20nm, I'd expect a datasheet. IMFT on the other hand is 20nm.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
12 Comments
Back to Article
RaistlinZ - Monday, June 25, 2012 - link
The market is already saturated with Sandforce 2281 drives. Why bring another into the fray? The specs aren't even at the top of the pack of other drives that have been out for a long time now.Kristian Vättö - Tuesday, June 26, 2012 - link
I think Hynix is doing this to build distribution channels for their future LAMD based drives. It's easier to begin with a lower volume product and test several distributors and then choose the best one. LAMD based SSDs definitely have potential to sell like hot cakes if the performance figures are to believe, so you don't want to screw up supply because of distribution issues.Or at least this is one way of looking at this.
mckirkus - Tuesday, June 26, 2012 - link
Large Array of Mechanical Donkeys (LAMD) in case anybody was wondering.surt - Tuesday, June 26, 2012 - link
It's microscopic, not mechanical. What kind of sense does mechanical donkeys make?artifex - Friday, July 13, 2012 - link
At least in the mSATA form factor there is not an overabundance of vendors.Speaking of, it'd be really nice to see a comparison among some of them.
XJDHDR - Tuesday, June 26, 2012 - link
"Another interesting aspect of Hynix' SSD series is the fact that it's the first consumer SSD to utilize 20nm MLC NAND."There is the reason they bothered. With 20nm NAND, it should be cheaper than any other SandForce SSD available right now, which still use 25nm NAND.
Kristian Vättö - Tuesday, June 26, 2012 - link
Well, it doesn't always work like that. While you get more dies per wafer (assuming each die is still 64Gb), your yields are lower at first. 25nm process is now mature and yields are better than ever. Demand can also be high compared to supply, which keeps prices high.Smaller process node does not mean an immediate price drop. I would say NAND price drops are fairly linear.
XJDHDR - Tuesday, June 26, 2012 - link
Yeah, that's why I said that it should be cheaper. Another factor you didn't mention is that this is so far the only known SSD to use 20nm NAND, which means there would be no competition in this area. Hence, there would be little reason to not charge a price that is only marginally lower until someone else introduces their own SSD with 20nm NAND.The reason you posted above though is also reasonable and I didn't consider it. I guess I assumed that Hynix already knows who they want to distribute stuff through.
shompa - Tuesday, June 26, 2012 - link
Smaller process does not mean cheaper prices. This myth need to die.Waffer prices goes up with smaller process.
That is why Marjory of micro processors are produced in 65nm or larger.
About 2-3% is 28nm or less.
(not counting Intel. Since intel owns its owns fabs, they are not impacted in the same way with waffer prices)
XJDHDR - Tuesday, June 26, 2012 - link
>Smaller process does not mean cheaper prices.First of all, I said it should be cheaper, not will be. On the contrary, it does. If it weren't (eventually) cheaper to use smaller process nodes, there would be absolutely no point in using them. Sandy Bridge vs Ivy Bridge is a good example. IB CPUs are better in every way, yet they are basically sitting at the same prices as SB before IB's release. Not a cheaper price but definitely more bang for your buck.
>Waffer prices goes up with smaller process.
I assume you mean wafer, as I have never heard of a waffer. At first, yes. Eventually, however, the cost of fabricating chips at any particular process node drops to lower than the one before.
>That is why Marjory of micro processors are produced in 65nm or larger.
I assume you mean majority since marjory is also something I've never heard of. Otherwise, I find that hard to believe. Intel, AMD and Nvidia have been releasing all of their microprocessors at smaller nodes since 2009. There is also the fact that virtually every consumer SSD uses NAND smaller than 65nm.
>About 2-3% is 28nm or less.
Ignoring the fact that every SSD I know uses 25nm NAND and has done so since at least 2011, it is only recently that nodes smaller than 29nm became mature enough to move microprocessor fabrication to it. Hence, it's quite obvious that very few processors in use today would be using a node less than 29nm, if your figures are accurate that is.
someguy199 - Tuesday, June 26, 2012 - link
If you read the Hynix press announcement it is 20nm "class" not 20nm. Samsung used the same PR trick to indicate anything from 29-20nm awhile ago when selling 27nm. If this was really 20nm, I'd expect a datasheet. IMFT on the other hand is 20nm.Kristian Vättö - Tuesday, June 26, 2012 - link
You are actually correct, the Hermitage Akihabara's review actually mentions 26nm as well. I've updated the article to reflect this.