| ttoine | hello | 13:15 |
|---|---|---|
| scott-wokr | good morning everyone | 14:06 |
| len-dt | scott-wokr, good morning... | 14:08 |
| scott-wokr | oops | 14:10 |
| === scott-wokr is now known as scott-work | ||
| === drupin is now known as zz_drupin | ||
| len-dt | holstein, just a note.. I have been doing a little reading and it appears there is a trade off between throughput and latency. Higher latency/larger chunks of data works better for throughput... and most of the version changes for PCIe have been made to increase throughput, | 21:28 |
| len-dt | but... to make bigger data chunks requires bigger headers. | 21:28 |
| len-dt | For low latency the data chunks need to be smaller. | 21:29 |
| len-dt | on a high throughput system that means the data headers may be as big as the data. | 21:29 |
| len-dt | For most uses, bigger throughput seems faster... program loads faster off the disk for example | 21:30 |
| len-dt | Audio has different needs. | 21:31 |
| len-dt | All that to say, the low latency kernel should not slow the system down unless something is using the lowlatency parts, in which case the desktop may seem slower | 22:18 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!