Date: Wed, 27 Jun 2001 22:33:22 -0500 From: Mike Meyer <mwm@mired.org> To: Steve Ames <steve@virtual-voodoo.com>, Soren Kristensen <soren@soekris.com> Cc: hackers@FreeBSD.ORG Subject: Re: Status of encryption hardware support in FreeBSD Message-ID: <15162.42370.469003.143882@guru.mired.org> In-Reply-To: <20010627221945.B92559@virtual-voodoo.com> References: <107.1bc2228.2868aa7a@aol.com> <3B3A7823.337CA425@soekris.com> <15162.39875.353224.757884@guru.mired.org> <3B3A9EFA.E65C2DC8@soekris.com> <20010627221945.B92559@virtual-voodoo.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Steve Ames <steve@virtual-voodoo.com> types: > On Wed, Jun 27, 2001 at 09:51:47PM -0500, Mike Meyer wrote: > > The crucial bottleneck for this kind of thing is the doubling > > time. Unless your special purpose hardware doubles in speed as fast or > > faster than general purpose CPUs, then eventually it's going to be > > slow, then expensive, and finally dead. Given the two doubling times > > and current relative speed, you can easily predict when general > > purpose CPUs well be faster and then when they will be more cost > > effective. At that point, your special purpose hardware is dead, and > > just waiting for the rest of the world to realize it. > > > > Given the predicted lifetime, you can make a rational decision about > > whether it's worth the effort to support the hardware. > > Can't you also make the assumption that hardware vendors will > be upgrading their product with equal vigor? Supporting hardware > now (when its faster than CPU) will make it much easier to support > it later (when its still faster than CPU). Actually, I *did* assume that. The question isn't about vigor, it's about results. That's why you need to know how fast the hardware vendors can double the speed of their product, so you can predict the points where software overtakes it and then becomes more cost effective - or demonstrate that that isn't going to happen. Past experience - LISP and Forth Machines, the Amiga graphics hardware, etc. - indicate that special purpose hardware doesn't get faster as fast as general purpose cpus. There's apparently more money to be made in making general purpose CPUs faster than in making special purpose hardware faster, so they throw more resources at it. Soren Kristensen <soren@soekris.com> types: > That's not really the point here, I was talking about lowest end > hardware compared to high end CPU. If we compare with high end hardware, > then we're talking about factor >50 faster than software.... There are > chips out that can do >1Gbit 3-DES, given a 64bit/66Mhz PCI bus. > > I'm just starting with a low end chip to complement my 133 Mhz 486 based > net4501 board, with the goal of low cost and low power, not absolute > performance. Replies below the quote if you want to keep context. Low end, high end, and current relative speed are immaterial. If the doubling time for your special purpose hardware isn't as good as that for general purpose cpus, then it's a dead end. If you're planning a product around it - or are deciding whether or not to support it - then you need to take that into account. This doesn't mean the product is useless; it just means you need to plan on dealing with it being outpaced by general purpose CPUs at some point. <mike -- Mike Meyer <mwm@mired.org> http://www.mired.org/home/mwm/ Independent WWW/Perforce/FreeBSD/Unix consultant, email for more information. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?15162.42370.469003.143882>