e_dawg
Storage Freak
---Check out the terrible performance of other MLC drives with random writes. The Intel drive has 2700x shorter latency than the OCZ core drive. Jesus. Looks like controllers are very important for SSDs.
[...]
One thing is for sure though: there's no way]/i] I'm buying a non-Intel MLC drive until the other manufacturers sort out that random write performance --it's like going back a decade in terms of performance... And no-one noticed this before? Come on benchmarking (benchmarketing?) community.
This is why I decided not to RAID 0 a couple OCZ 64 MB SSD's as my OS/apps/temp/swap drive. I figured write-back caching would be the best solution to compensate for slow random writes, so i entertained the notion of getting a 3Ware RAID controller with 256 MB of battery-backed write-back cache. But I am not sure if this solution really works in the real world, and I've read enough reports of problems with SSD's that I don't want to be the one that sorts through all the early adoption issues.
Nevertheless, slow random write performance is not a big issue with the majority of computing usage, as random read performance is still critical. A lot of the random read performance issues with rotating magnetic disks have been addressed with intelligent caching and pre-fetching both on the drive and in the OS (Vista is actually pretty good at this if you have enough memory), but SSD's still shine in general usage despite the random write performance. There could still be a way to use the OS to mitigate random write performance issues that hasn't been explored yet. I'm not sure how or why, but I suspect that there is a potential solution in there.