Comments on: SanDisk cheat to increase random SSD write speeds by 100x http://www.crunchgear.com/2008/11/05/sandisk-cheat-to-increase-random-ssd-write-speeds-by-100x/ Gadgets, gear and computer hardware. Fri, 04 Dec 2009 16:13:26 -0500 hourly 1 By: Myron http://www.crunchgear.com/2008/11/05/sandisk-cheat-to-increase-random-ssd-write-speeds-by-100x/comment-page-1/#comment-976488 Myron Fri, 30 Jan 2009 18:27:21 +0000 http://www.crunchgear.com/?p=52249#comment-976488 SanDisk is making a big marketing deal of this but has not yet delivered a product. Other manufacturers have quietly delivered this promised performance level for months already using similar and possibly better algorithms. For example, see the latest Samsung and Intel drives. SanDisk is making a big marketing deal of this but has not yet delivered a product. Other manufacturers have quietly delivered this promised performance level for months already using similar and possibly better algorithms. For example, see the latest Samsung and Intel drives.

]]> By: Howard http://www.crunchgear.com/2008/11/05/sandisk-cheat-to-increase-random-ssd-write-speeds-by-100x/comment-page-1/#comment-945821 Howard Thu, 06 Nov 2008 15:05:36 +0000 http://www.crunchgear.com/?p=52249#comment-945821 "... it would be pretty taxing on the RAM" Fortunately the RAM (DRAM, PSRAM, SRAM) does not have the same wear out mechanism as the Flash. A RAM cache, and an intelligent controller, are needed to minimize the number erase cycles to the flash. Without them, as in some low cost SSD, the useful life of the SSD could be significantly shorter. At this point in time there are not any JEDEC standards on the RQA items that are required to evaluate the claims of the SSD vendors. If there is data to back up the claims, it's methodology is unique to the vendor. SanDisk may be 'tooting their own horn' but should be get a 'thanks' for working to create a meaningful industry standard. “… it would be pretty taxing on the RAM”

Fortunately the RAM (DRAM, PSRAM, SRAM) does not have the same wear out mechanism as the Flash. A RAM cache, and an intelligent controller, are needed to minimize the number erase cycles to the flash. Without them, as in some low cost SSD, the useful life of the SSD could be significantly shorter.

At this point in time there are not any JEDEC standards on the RQA items that are required to evaluate the claims of the SSD vendors. If there is data to back up the claims, it’s methodology is unique to the vendor. SanDisk may be ‘tooting their own horn’ but should be get a ‘thanks’ for working to create a meaningful industry standard.

]]> By: Doug http://www.crunchgear.com/2008/11/05/sandisk-cheat-to-increase-random-ssd-write-speeds-by-100x/comment-page-1/#comment-945792 Doug Thu, 06 Nov 2008 11:57:32 +0000 http://www.crunchgear.com/?p=52249#comment-945792 I don't believe you are reading it correctly. The section on the Extreme FFS talks about using a page-based system which separates the physical location of the data from its logical location. I.e. two blocks of a data which might be logically sequential are not necessarily placed next to each other physically. Instead, the blocks are written to a physical page which is most convenient and efficient, and logically linked into place. There doesn't seem to be any mention of RAM cache in the article that I can see. I don’t believe you are reading it correctly. The section on the Extreme FFS talks about using a page-based system which separates the physical location of the data from its logical location. I.e. two blocks of a data which might be logically sequential are not necessarily placed next to each other physically. Instead, the blocks are written to a physical page which is most convenient and efficient, and logically linked into place.

There doesn’t seem to be any mention of RAM cache in the article that I can see.

]]>