ragenalien :
Meh, it's difficult to get excited about sata 3 SSD's anymore. Sure they're fast; I own a 840 pro myself, but they really can't get any faster until the next gen sata 4 makes it's debut.
That is so not true at all!!! There is tons of room for throughput improvement even on high end SSDs!
As awesome as SSD tech is, we need to remember that it is a tech that is still relatively in it's infancy. Just because it blows HDDs out of the water does not mean that it is anywhere near reaching it's potential. I run 2 SSDs in my main rig in RAID0. It is phenomenal getting a full 1GB/s of peak performance... but truth be told my average throughput is ~600MB/s, and my minimum benchmark throughput is somewhere around 375MB/s. Modern drives do not have the same max/min disparity as my 'ancient' Agility 3 SSDs, but they still have a long way to go to have drives that simply run flat out on any workload.
Other than raw performance there are power issues with running high end SSDs which need to be addressed as well. If you look at modern SSD power usage you will find it has been creeping up over the last few years, and now most mid to high end drives actually take the same or MORE power to run than their HDD counterparts. You still get the advantage of a much better 'race to sleep' as SSDs can better feed the processor so that it can be done and switch off faster, but those savings are beginning to be eaten into by the simple running of the drive itself. Moving to better chipsets, and better die shrinks will go a long way to getting power usage back down without sacrificing performance.
Hopefully they will get it down (and the price down) so that we can start seeing true SSD tech implemented into things like tablets before long, rather than having them rely on what is essentially a glorified SD card.
Another big issue with SSDs is their reliance on TRIM and other old garbage collection techniques. There is nothing particularly wrong with TRIM for home users where there are often long stretches of down time, but server system in much bigger enterprise environments (such as Netflix or Google) never really get much in the way of down time, and so they never get a real chance to run TRIM, which then degrades performance over time. This is something that Intel has worked a lot on, and which other SSD manufacturers are quickly trying to find ways around. Running GC in the background during use can help get more consistent performance all of the time, which will translate into massive business sales, which turns into larger volume production, which turns into lower prices for all of us.
And price is obviously the other big factor here. I picked up my first SSD for my wife's system almost 2 years ago when the old Solid 3 line hit $1/GB (that drive then died while under warranty and I got a free upgrade to a Vetrex 3 for her system), then I picked one up for my rig when the Agility 3 line hit $1/GB about a year ago, and then my 2nd Agility 3 on a crazy sale for $0.50/GB a few months later... but we are not seeing much in the way of SSDs for 50 cents per GB anymore, and higher performing drives are still sitting just under $1/GB. A good die shrink should go a long way to seeing prices come down a bit further which should help everyone on the consumer side while preserving some profits for manufacturers.
Yet another issue with today's SSDs is the 512GB limit imposed by modern controllers. Yes, you can purchase SSDs larger than 512GB, but the price is sky high because you have to pay for multiple controllers, plus some sort of on-board RAID chip to get them to work together. Newer chipsets coming out later this year should allow for 1TB of addressable space on each controller, which will cause the price of 1TB SSDs to fall in line with $/GB of normal drives, and also allow for 2TB SSDs to be made with an internal RAID scheme. Getting these bigger drives out the door is important for server applications where a minimum storage requirement is needed, and in things like laptops where you only have a single drive bay available.
Anywho, point being, there is lots of room for improvement in SSDs right now before we start complaining that SATA3 is holding drives back. A move to SATA 4 will cause the exact same problem that we have with SSDs in a RAID0 configuration. You will get very fast maximum throughput available... but the chipset will still hold 90% of your workload back to SATA2 levels like they do now. There is tons of room for improvement while we wait for SATA4 to be released in 2+ years.
Facebook
Twitter
Instagram