Rudimental
Expert Member
- Joined
- Jan 6, 2009
- Messages
- 1,457
- Reaction score
- 2
It is basic comprehension: Huge discrepencies (2x-6x) lead to relatively small effects. The 10-20% difference between throughput can be called 'pretty major', but when compared to the difference in pure random r/w speads, it is small. I never once said that the Vertex 3 wasn't better in that regard. I always said that the relative differences show that huge small block random r/w speeds do not translate directly into huge gains. 100% increase in read speeds and up to a 500% increase in write speeds lead to an overall increase of about 10-20% throughput in real-world testing, and this increase must also factor in a 5-10% increase in simultaneous r/w speed.
That such a large increase in pure isolated benchmark performance leads to a result that is an order of magnitude smaller in real-world testing indicates that the returns are extremely marginal, and thus that performance in this area above a certain amount in synthetic benchmarks is not very important.
Also, no, it would not be a failure, because the overall performance in real world application is what matters, and I am going to stick with my view that the Intel is good enough in this regard. I would say that you just cannot see the forest for the trees. You can't seem to understand that people don't just constantly do small block random r/w, that not everyone who buys an SSD is a power user whose only concern is pure price/performance, and that Intel's apparent reliability (real or only apparent) in the SSD market is a big draw for many people.
come on Rudimental where are you....
lunchtime, lol I needed to eat. I cannot subsist argument alone
