Subject: Re: [linux-audio-dev] Performance and Elegance? (Was: High Cost of
From: rob (rob_AT_kaybee.org)
Date: Wed May 16 2001 - 13:32:29 EEST
On Wed, 16 May 2001, Jay Ts wrote:
> Is the context switch "slowness" related directly to RAM "slowness"?
> Because if so, then there may be some light at the end of the tunnel.
>
> Either way, I'd like to point out a few things relative to the subject
> of RAM speed.
>
> First of all, it is possible to make fast RAM. Consider CPU registers
> or level 1 cache. Those are simply memory, and they run as fast as
> the CPU can handle the data. Basically, RAM is made of transistors,
> just like the CPU, and can be made to run just as fast. The technological
> limitation is mostly in the area of bus speeds, that is, connecting
> that fast memory to the CPU over a bunch of relatively long wires.
> The workaround is to put the fast RAM on chip, or very close to the CPU.
>
main memory will always be slower than registers. it's that whole memory
hierarchy thing working. on top of that you have different kinds of ram
with different tradeoffs: registers are quick usually using static ram,
main memory is slower and uses dynamic cells. the density achieved with
static ram is lower because of its complexity which in turn makes it more
expensive, hotter, etc.
even if you could put all your ram on chip, you eventually run into speed
of light problems. meaning you can only put so much ram close enough to
the cpu to be able to get bits out of it without stalling the execution
pipeline. (this means you want to have commonly used ram close to the
cpu...giving you a whole cache level of memory hierarchy again).
IPC is a bad way to go compared to in-process methods for a number of
reasons. if it turns out that tommorow some breakthough changes all this
then you can redesign it. no sin in that, nor would it be wasted effort.
please review the worse-is-better essay.
rob
----
Rob Melby
Georgia Institute of Technology, Atlanta Georgia, 30332
uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!gt4255a
Internet: rob_AT_kaybee.org, gt4255a_AT_prism.gatech.edu
This archive was generated by hypermail 2b28 : Wed May 16 2001 - 14:03:18 EEST