Any speed up on 64-bit or 2+ core systems ?
 

A question to those, who have 64-bit or 2,3.. core CPU - is there any speed up in common PC performance? How many "bogomips" at dmesg output ?

Probably not.
I don't use a lot of "common" software, but web browsing, video all looks the same on my i7.
I'm expecting kernel compiles to be a different matter. Might even go back to gentoo ...

I have single and dual core 64-bit systems. I cannot visually see any difference in the performance of either. I'm not an avid gamer and the most resource intensive application I use is the Flight Gear simulator. Can see no differences in performance. I think any improvements here would require a better graphics adapter.

In the area of stats, I really have no basis for comparison since I have different CPUs with differing clock speeds. The only things in common are the software, memory, and graphics adapter.

Cheers, Lou

Do you compile custom kernel for i7 ?
I'm testing Slackware13-32bit parallel with Slack12-32 on the same PC now (rather old PC).
Common performance of out-of-a-box Slack13 kernel is lower then my custom kernel for Slack12 with some system tuning. So, I'll create custom kernel for Slack13, by no means.
And what about 64-bit and multicore systems? Any performance boost? For 3D rendering in example (using Blender)?

It's me again.

I forgot to point out that on my dual core system I've built my own kernel tailored to my AMD Athlon 64 and my system hardware. Still see no differences.

Lou

Well, I dunno about you people, but I definitely DO notice a difference, especially with some programs like encoding, encrypting, games, etc.

App start times also are a bit faster. Test it yourselves, run 'time' to time commands as they run.

I do compile my own kernel.

As I said I don't do a lot of that stuff, but I am planning on a lot of (threaded) emulation code. For day-to-day, not a lot of difference to the core-2 duo it replaced (burnt out on-board nvidia).
It's a new toy for now, but I'll get to do some serious processing soon.

Sorry, I'm probably leading this discussion astray. It's been so long since I ran 32-bit systems for "real" work, I can't offer a decent comparison.
I'll just quietly exit stage left ...

Danzille,

Thanks for the thanks. Like you, I'm fairly new to LQ but I think the Thanks are supposed to be reserved for cases where someone has helped you solve a problem. In this case, I think responding Yes to the Did you find this post helpful? would be more appropriate. I'll know soon enough if a moderator jumps in.

Cheers, Lou

I compile my own kernels for use on my x64x2 machine. I also tend to compile a lot of programs using slackbuilds.org. The difference in compile times is tremendous when taking advantage of both cores. I can only imagine what the 8 core processors are like in this regard.

I have a Core2Duo quad core, and I notice a huge difference in compile times. On one of my single core machines, kernel builds would take forever, now they take about 15 min or so.

I have a core2duo e6600 system (dual core) and an i7 920 system (quad core with hyperthreading) both do/have dual booted with slack13 and slack64-current. They both have 6GB RAM and comparable Nvidia cards.

I didn't notice a huge difference when I built the i7 system, but after I used that exclusively for a month, then fired up the core2duo I was shocked at how SLOW it felt. For building kernels and packages there is NO COMPARISON the i7 does in 4-5 minutes what the dual core core2duo did in 20-30. Of course that's using Things like web browsing are less noticeable. Although dumb things like better flash support make the 32 bit OS work a little better on the web.

One thing that makes my results a little less meaningful though is that I overclocked the i7 to 3333MHz whereas I left the e6600 at the stock 2.4GHz.

Bah, you young whipper-snappers! always in a rush! My old p3-800 takes 3+ hours to build generic+modules! ;)

LOL, one of the few good points in coming to linux late is having the hardware it took to run vi$ta, I guess. I don't know that I'd be running kernel rc's if it took half my shift to build!

It depends a lot on the size of the kernel, so that's not a good benchmark, try firefox instead, the size is more constant.

Or try:
http://www.phoronix-test-suite.com/