Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

They're just another product on the Moore's Law trajectory. Treat them the same as you do computers and phones -- plan to buy a new one every 2-3 years, and plan on only having the latest and greatest for about 12 minutes after you get it.


Wrong. HD speed improvements have been pretty flat, and has been far outpaced by HD capacity which is closer to Moore's Law. SSD offers a nice boost in speed. Don't just throw around Moore's Law as if you know something.


I think you might have misread kbob’s comment. I think he was talking about upgrading from one SSD to another newer SSD — not from HDD to SSD.


It would be stunning for SSDs to take off with exponential speed improvements. Sure, I expect them to get faster and faster, but I have no doubt that there's no Moore's Law at work in this case. I stand by my original comment.


It has nothing to do with speed, it has to do with transistor counts, which should increase disk size or lower cost.


I totally agree. Moore's Law is completely inapplicable when it comes to disk performance.


Which is a strawman of your own invention that you are being an arse about, kbob didn't mention SSD performance.


kbob was replying to a post about a "performance boost" and then claimed without qualification that HDs were "on the Moore's Law trajectory"... I didn't invent anything here.


Moore's law is about transistor counts, not speed, and is directly applicable to SSDs.


It really depends. Taking a hard black&white stance on other side is just silly.

Say you have an SSD that consists of 8x16GB chips. Internally it's implemented in something like a RAID0 array, which is why they're so fast. If the capacity doubles / size shrinks by a factor of 2 and someone decides to stick 16x16GB chips in the next gen SSD then you will see a speed increase. On the other hand if they go for 8x32GB you won't.


Moore's Law, as originally stated, is that the number of transistors on a chip doubles every 18 months. Flash memory is definitely on that trajectory, and SSD controllers look like they can make good use of increased transistor counts for a few generations.

My comment was directed to Matt_Cutts' complaint that he didn't know when to buy an SSD. I don't think it matters exactly when you buy one. (Though I personally am waiting to see how Intel's 25nm SSDs compare to those based on the SandForce SF-2000 controller around Jan/Feb 2011. Maybe I should practice what I preach.)




Consider applying for YC's Winter 2027 batch! Applications are open till November 2.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: