Messages in this thread | | | Date | Tue, 24 Feb 1998 12:57:09 -0800 | From | "Nate Tuck" <> | Subject | Re: GGI, EGCS/PGCC, Kernel source |
| |
At 08:15 PM 2/24/98 +0000, Alan Cox wrote: >Gigabit lan DMA versus main memory - latency yes bandwidth is ceasing to be >an issue
How long until gigabit ethernet has all the kinks worked out of it? Let's say second half of '99. You can probably buy a PC system with 1.6 Gigabytes/s of memory bandwidth and a gigabyte/sec going to graphics in that timeframe. I still see an 8X performance hit on the network. ;)
>Existing hardware on all Linux supported platforms barring SGI. Nobody >else has hardware video contexts.
How long until 3D Labs and E&S catch up? Assuming they haven't already.
>So the PC case is always going to be >2 (not 3) times as bad - I can compute in my mitshm object thanks
Not without making sourcing the data from shared memory an API requirement. Or dynamically checking on every API entry whether the data come from mitshm. Bleh.
nate
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |