Quote:
Originally Posted by Atmos
From memory I get about 4,500-5000 as a total performance, would that mean that you’d be getting ~4x the speed increase?
Curious as to whether it works that way
|
It depends on whether your workload is similar to the benchmark, Colin
My old system is an i7-960 that also does around 5,000. On my tests so far the new system is 4 to 10 times faster on time-consuming processes like calibration, registration, drizzle integration and linear normalization.
Quote:
Originally Posted by lazjen
If I'm reading those benchmarks correctly, there's no real advantage using a RAM disk? Assuming I have not made a mistake there, I guess this would be because of how much memory you've got and Linux's excellent page/buffer support. It's excellent news as you don't have "fear" a power outage and lose your swap cache.
|
Yes, I think that's right, Chris. The improvement from the RAM disk is fairly small. The comparison is with a NVMe SSD. If I was using a SATA SSD or, heaven forbid, a spinning disk then the difference would likely be significant.