sechs
Storage? I am Storage!
Dunno, but Newegg appears to have just put it on sale with a slightly lower price.hmmm... Maybe there will be a price war???
The prices on SSDs seem to be falling rather quickly these days.
Dunno, but Newegg appears to have just put it on sale with a slightly lower price.hmmm... Maybe there will be a price war???
The prices on SSDs seem to be falling rather quickly these days.
Maybe the Revo would be a good choice. Basically two SSDs and a RAID controller slapped together on a single PCI-E card.I'm not installing an OS and buying apps for an old computer. That would be only if I build a new system eventually. I was hoping for 6Gbps SSDs to use on the controller, and maybe that will still be possible.
I'm hoping that this is a server... and then wondering why you'd then need SSDs in RAID 0.
With an SSD, the data transfer proportion of the speed equation is very much larger than with a HD so it is much easier to get a speed increase using raid.
Lets look at 64k blocks transfering at 100MB/s so a transfer takes .64ms
With an SSD the seek speed will be .1ms so a single drive file copy is a repeated seek +read data + seek + write data.
SSD: .1ms + .64ms + .1ms +.64ms = 1.48ms to transfer 64KB
For an average HD with a 100MB/s transfer speed and a 15ms seek speed the same operation takes this amount of time.
HD: 15ms + .64ms + 15ms + .64ms = 31.28ms to transfer 64KB.
Now lets do a raid 0 scenario for both doubling the transfer speed to 200MB/s.
raid 0 SSD: .1ms + .32ms + .1ms + .32ms = .84ms or a 76% speed improvement.
Raid 0 HD: 15ms + .32ms + 15ms + .32ms = 30.64 or a 2.1% speed improvement.
Thats why people want to raid SSD's!
I'm hoping that this is a server... and then wondering why you'd then need SSDs in RAID 0.
With an SSD, the data transfer proportion of the speed equation is very much larger than with a HD so it is much easier to get a speed increase using raid.
Lets look at 64k blocks transfering at 100MB/s so a transfer takes .64ms
With an SSD the seek speed will be .1ms so a single drive file copy is a repeated seek +read data + seek + write data.
SSD: .1ms + .64ms + .1ms +.64ms = 1.48ms to transfer 64KB
For an average HD with a 100MB/s transfer speed and a 15ms seek speed the same operation takes this amount of time.
HD: 15ms + .64ms + 15ms + .64ms = 31.28ms to transfer 64KB.
Now lets do a raid 0 scenario for both doubling the transfer speed to 200MB/s.
raid 0 SSD: .1ms + .32ms + .1ms + .32ms = .84ms or a 76% speed improvement.
Raid 0 HD: 15ms + .32ms + 15ms + .32ms = 30.64 or a 2.1% speed improvement.
Thats why people want to raid SSD's!
The average file size is about 500MB with some a few GB. I expect the access times not to be affected much by RAID 0, or is that the case?
I have no easy place left for the SSDs. The PS cage is gounded so I would not think there would be any interference, but are there any problems with sticking them on the PS? :spiderman:
Performance is awful. :tdown: Some users have complained about the need to wipe it first. :cursin:
And that's a RAID 6 of what make of SSD?Sounds like a decent workstation to me. RAID0 SSDs for OS/Apps/Cache/Page/etc and Massive RAID6 for DATA.
All things considered, the Sandstorm drive is quite impressive and nearly as fast as the X25-E. Unsurprisingly, RAID 0 does not help that much. I'm not sure where the bottleneck is. orc:
I thought Native Command Queuing was supposed to allow OSs to call for more than one thread at a time. NCQ has been around for 5+ years. I would have thought newer OSs and service packs would have caught up by now.
That is how it has been forever and I was hoping it would be better by now. Apparently not.The bottleneck is generally in the software.
You should use it only as a pass-through though. RAID the disks in Windows software RAID and you'll see better performance.
I was reading the other day that there are problems using benchmarking software on SandForce-based SSDs because: (1) they have no cache and therefore react different than cached controllers; and (2) they compress the data before write, so compressibility affects performance.Yeah, I have a strong suspicion that none of the benchmarking tools are actually accurate for SSDs, particularly SSDs in RAID.
Sandstorm :mrgrn:Say it with me SandForce!
Sandstorm was awesome. But I use Auzentech stuff now.
Did he mean Soundstorm?:scratch:
Did he mean Soundstorm?
Replaceable NAND modules?Why is this such a big deal??