Home
| Calendar
| Mail Lists
| List Archives
| Desktop SIG
| Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings Linux Cafe | Meeting Notes | Linux Links | Bling | About BLU |
I'd love to see someone benchmark the various multi-core CPUs when the host machine is really just running VMware.. Or does it really matter most on what you're doing in the VMs? I.e., if I'm building a VMware host server for VMs to act as a few servers (web/mail/mysql) and various build systems, does it really matter if I go Intel v. AMD? I was under the impression that AMD's memory bus utilization was significantly better than Intel, and I would guess that VMware would be more memory intensive than CPU intensive.. But I dont know. Has anyone seen any VM benchmarks? -derek Daniel Feenberg <[hidden email]> writes: > On Mon, 1 Oct 2007, Mark J. Dulcey wrote: > >> Stephen Adler wrote: >>> The reason I ask is not so much if SMP is supported or not, but all >>> the crazy bridge bus chip sets, and interrupt controller chip sets >>> etc. etc. When I got my Intel based Core 2 Duo system, (i.e. intel >>> motherboard) there was an issue with the current version of the >>> kernel and lack of support due to some chip set issue... I can't >>> remember how I dealt with it, (some odd boot parameter option or >>> something like that.) Has anyone had any experience using a core 2 >>> quad? Is it worth the extra money? >> >> The motherboard chip set could be an issue, though it's a separate >> one from the processor. Using a new distribution will help. >> >> I haven't had any personal experience with the quad. What I've heard >> is that the extreme (big $$) version isn't worth it unless you >> absolutely must have the fastest computer for bragging rights, but >> the less expensive ones (like > > We have 6 of the various Intel Quad-cores, each was the fastest > available when purchased. We use them exclusively for floating point > (statistical) calculations, and performance at that task is linear in > the number of cores. So if you have the right workload, they are > worthwhile. > >> the Q6600 for about $280 or the Q6700 for $545, current MicroCenter >> pricing) are worthwhile if your application mix will take advantage >> of it (multitasking or multithreaded apps). Although the Intel quads >> only offer so-so scaling, 4 cores are still better than two if you >> can keep them all working. (On the other hand, the faster dual-core >> will probably be better for gaming; an E6750 for $190 or an E6850 >> for $280 look sweet.) AMD's new Barcelona will scale better, but the >> baseline single-core performance isn't as high except perhaps in >> scientific computing, because its floating-point performance is very >> high. >> > > Oor next purchase will probably be AMD, for this reason. > > Dan Feenberg > >> Intel will have true quad-core processors (four cores on one chip >> rather than two chips in an MCM) next year, and those will be faster >> and scale better. But isn't that always the way?
BLU is a member of BostonUserGroups | |
We also thank MIT for the use of their facilities. |