So the most important speed when it comes to SSD is the random speed performance. Suppose the next couple of years, the SSD will use sports cars to pick up data or even helicopters.
The marketing departments have done a great job pushing max speed when random is what really matters. I agree with you. It's all about random speed.
I'll have to disagree. You guys are mistaken in believing that random read and write are the *only* factors that matter. Granted, I'll agree that random READ speed is easily the most significant benefit SSDs have over traditional harddrives, but like most things, it is about achieving balance.
The random write 'issue' started because older and budget SSDs based on the JMicron controllers without RAM cache had such poor random write performance that they could actually make your system stutter and lag during heavy multi-tasking. Current-generation SSDs that use controllers from Samsung, Indilinx, and Intel don't have this problem.
Intel's X25M SSDs took random write performance to an extreme, offering 3-4X faster performance than the closest competition using Indilinx's SSD controllers (like the OCZ Vertex, G.Skill Falcon, Patriot Torqx). However, based on real-world benchmarks and testing from the likes of Anandtech, PCperspective, benchmarkreviews.com, etc, the extreme random write performance offers no tangible benefits to non-server (e.g. consumer/workstation) workloads. The Intel X25 level of random write performance is really only useful for server applications like databases and transaction processing where you have thousands of database entries or log file entries being written to disk every second. It seems that once you get past a certain threshold, around the performance of fast 10K RPM HD (1-2MB/sec -- yes MB/sec), more random write performance doesn't provide much in the way of advantages for laptop/desktop situations.
And Intel's SSD architecture that creates this high random write performance directly limits their sequential write performance, which can be vital in certain applications, like video/multimedia editing, file transfer, file copy, transferring from fast external drives/RAID, writing large amounts of data to disk from RAM, etc.
Similarly, while random read performance speeds up OS startup, application launching, etc, sequential read performance is CRITICAL for high-bandwidth applications like multimedia work, or other situations where you have large volumes of data moving in and out of storage.
that is what i wanted to hear
larger capacity, fixing control problems, and price drops!
I really hope to see a 100-120GB around 200 dollars. . . . i can wish cant't I?
You can pick up a 120GB G.Skill Falcon SSD (identical to OCZ Vertex) for $299 with $40 rebate at newegg.com. Not quite $200, but this is an excellent drive. Do NOT skimp with a cheap SSD.. you'll end up with a JMicron controller based POS that you'll be very unhappy with.
Yet everybody was conveniently forgetting all of this in the 70-page 1.5Gbps SATA thread and complaining that their sequential read/write speeds were being throttled when their random reads/writes were hardly affected at all.
Ruahrc
Not only were sequential transfers throttled, but data moving in and out of cache. There are plenty of benchmarks out there now that show that despite not impacting random read and write speed, the SATA/1.5 was having real-world impacts on OS boot, application startup, etc.
As much as the Apple zealots on here want to defend Apple for whatever reason, the SATA/1.5 situation was a legitimate issue that could have greatly impacted performance in many situations, and hence was quickly taken care of.
