Recent content by algu01

Seems like the latency could be well below 4 mS even running over the virtualized NIC: http://www.netlib.org/lapack/lawnspdf/lawn185.pdf Look at page 8, figure 3, around 200 microseconds in end-to-end latency running MPI in the PS3 cluster. Well I guess that is the final nail in the coffin...

I have never understood how ping can report figures below 10 mS on operating systems where any given thread has a maximum timeslice of 10 mS. This at least was the case with Windows NT and I assume the later derivatives. Anyway, I guess my nice, fat PS4 speculation is debunked then.

I guess we're posting almost simultaneously :) . Ok, I was under the impression that 4 mS was really below the latency of an ethernet-connection (especially after reading a paper about the high latency of the virtualized NIC in the PS3), but I really pulled that one from the top of my head...

I don't really care about the multiple viewports, it's the rendering at 240 Hz that is the interesting fact. 240 Hz would roughly be 1 full frame rendered and synchronized over gigabit ethernet in 4 mS. Perhaps someone could tell us if this is really possible?

Hi all, first post for me, but I've been reading the forums for quite some time. Can one really do linear scaling at 240 Hz over virtualized NICs, with the latency involved and all? I think 3 of the boxes are not doing anything and the fourth contains a new Cell configuration. Could this new...