Date: Sat, 21 Oct 2006 02:23:24 -0700 From: soralx@cydem.org To: freebsd-hardware@freebsd.org Subject: Re: Quiet computer Message-ID: <200610210223.24178.soralx@cydem.org> In-Reply-To: <44u0241hf4.fsf@be-well.ilk.org> References: <44wt738057.fsf@be-well.ilk.org> <200610150041.59870.soralx@cydem.org> <44u0241hf4.fsf@be-well.ilk.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> Fair enough. Its speed on your own system may matter less to you than > you think, as well. ;-) yeah, but it's that feeling of disappointment: you expect a marvel of hardware crypto engineering, and get something that's totally unexciting :) Still, hardware encryption speed is importand in many cases, and especially for a server. > > just curious whether the speed VIA claims, 25Gbps(!) peak is achievable > Peak rate is a useless number. If it were sustainable, it wouldn't be > quoted as a "peak" rate; and if it's not sustainable, it doesn't > affect the perceived performance. quoting a peak rate is indeed useless, I hoped VIA were nice enough not to employ this type of marketing trick. Supposedly, it it said to be peak because it's only achievable on their fastest 2GHz C7? But the magnitude of numbers from benchmarks suggest otherwise. > No, not at all; I didn't mean to imply I was. I have the slowest (and > lowest powered) Via chips I could get. The ubench numbers are > slightly lower than the Pentium II system it replaced, at a similar > clock rate. Note that I didn't mean the numbers were bad, either... okay, by now we have enough performance data to be able to determine how fast (or slow) these VIA systems are. Merci to all who did the testing. The conclusion is that these VIA chips are by no means champions performance-wise; this fact makes them run cool as well. If performance matters at all, it'd be better to wait until AMD or Intel release low-power versions of their chips (they do have some now, and had them awile ago, but these don't come close to the low power dissipation of VIA's slower CPUs), or pay for performance (electricity and noise). (all this is obvious, of course, but now we have numbers to help determine where that line between fast-and-power-hungry and slow-and-cool is). [SorAlx] ridin' VN1500-B2
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200610210223.24178.soralx>