• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

What needs to happen for next quantum leap in speed?

T

TemjinGold

Diamond Member
So I've been curious: We know that SATA3 SSDs bench 2x or more versus SATA2, yet everyone says it's not worth your money to upgrade because you won't see the speed difference. It used to be that the spindle was holding back speed. What needs to change for SSDs to have noticeable improvements again? Is something else the bottleneck now? Something wrong with SATA3 itself?
 
When you look at the specs on the SSDs, you'll see one big number trumpeted for the speed. But that's only for a particular type of access. Other random accesses/writes are much slower, and typical computer usage may need those random accesses/writes to feel faster. There is much room for improvement for those specs, with lots of room to grow while still being far below the SATAIII threshold.
 
There's nothing wrong with SATA3 itself. The reason why some people say it's not worth upgrading from SATA2 to SATA3 is because the 4K read/writes have not yet bottlenecked SATA2, so it won't be any different in SATA3. These small 4K read/writes are what make up most of the OS background operations, such as booting, writing to logs, loading menu options, etc. So I'd say these 4K read/writes are the current "bottleneck".

I suppose for SSDs to gain noticeable improvements again, its latencies and 4K speeds would have to be improved and approach the levels of RAM. If you've ever ran a virtual machine's virtual HDD on RAM, or searched a bunch of txt files for a particular string in RAM, you'll see just how ridiculously fast and responsive RAM is, even compared to SSDs.
 
Last edited:
Its pretty obvious to me that its going to be the transition to PCI-express.

The early pci-e SSD models are already moving at 1,500 MB/s, aren't they?
 
For the biggest revolution in computing since the first transistor and microprocessor combined, weneed a breakthrough in material physics that will allow us to make a cheap practical high capacity non volatile universal main memory that completely eliminates the need for or concept of a third tier storage device.

A memory as fast as SRAM that is so fast the CPU can have 64 cores instead of wasting 2/3 the die on caches, infinite read/write cycles like DRAM, and high density and powerless refreshless long term data retention of Flash. Only then will we see the biggest revolution in computer performance since the abacus and start realizing some of the technology we can only dream of in modern science fi movies.
 
quantum leap
Click to expand...
In physics, a quantum (plural: quanta) is the minimum amount of any physical entity involved in an interaction
A quantum leap is the smallest possible leap.
 
Last edited:
TemjinGold said:
So I've been curious: We know that SATA3 SSDs bench 2x or more versus SATA2, yet everyone says it's not worth your money to upgrade because you won't see the speed difference. It used to be that the spindle was holding back speed. What needs to change for SSDs to have noticeable improvements again? Is something else the bottleneck now? Something wrong with SATA3 itself?
Click to expand...

The limitation is the Flash RAM itself. We'll keep getting minor improvements with each new smaller process but its unlikely to bring much extra speed and may bring with it a lot of problems (the number of erase cycles is dropping).

A big jump will probably be ReRam or the so called MemRister. Basically we need a better technology for storing bits without constant power that can switch at considerably higher speeds than that of todays flash RAM. Otherwise extra speed is linked with extra cost by having ever bigger and wider channelled controllers.
 
Well, I suppose the SSD manufacturers could always start, sort of, RAIDing multiple "drives" together and just call that thing a single drive. So you buy a 256GB drive, but really it's a RAID of two 128GBs so it runs really really fast. They could do this with existing technology, just when production/supply increases enough to drop prices where it's affordable to do it, maybe even just do the safer version of RAID where you take two 256GB drives and RAID them for 256GB total, but twice the performance and redundancy/safety too? I dunno, I just think that when prices fall, you can really play around with existing tech to get good performance.
 
KingFatty said:
Well, I suppose the SSD manufacturers could always start, sort of, RAIDing multiple "drives" together and just call that thing a single drive. So you buy a 256GB drive, but really it's a RAID of two 128GBs so it runs really really fast. They could do this with existing technology, just when production/supply increases enough to drop prices where it's affordable to do it, maybe even just do the safer version of RAID where you take two 256GB drives and RAID them for 256GB total, but twice the performance and redundancy/safety too? I dunno, I just think that when prices fall, you can really play around with existing tech to get good performance.
Click to expand...

Start? There have been such drives around for years.
 
You must log in or register to reply here.
Back
Top