INC -----> New hardware for 6 boxing. (OC progress update)

[moosejaw]

Just curious what stepping 920 you got from Tiger. On the CPU box there is a white sticker with a bunch of scan codes, prduct code, UPC, etc.

One number will be "S-spec: SLBXX

If it's SLBCH , it's a CO stepping
If it's SLBEJ, it's DO stepping


mine is a CO ;(

I'll check my box tonight. Crossing my fingers for the DO. I have been working on setting up the HTPC to get my tv recordings back up before I miss any favorite shows (read: Eliza Dushku).

I'm also upgrading to a new i7 soon, for those that haven't bought a 920 yet thought I'd reccomend trying to get a chip with D0 stepping as they supposably overclock to 4.0+ speeds even better and easier

http://www.anandtech.com/weblog/showpost.aspx?i=584

I'm currently able to run 6 with ok performance on a q9450 (oc to 3.2) w/8gigs ram, a GTX295, 30" dell and vista x64. I experimented alot and what I am doing to increase performance (using Innerspace) is to set all windows to 1024x768 resolution and I set the 5 slaves to render 512x384 windows with 100% zoom on rollover (this way they expand to 1024x768). After I launch I resize the main window to be much larger. I give up mouse repeating this way because the resolutions are different but I can play the main in a much larger window this way and I don't use window swapping but rather just mouse over slave windows (which expand to full size) when I need to.

I believe EQ2 uses alot of software rendering techniques (or just very bad code) and a move to a i7 cpu should provide a big boost. For those that have moved to i7, have you tried it with and without hyperthreading to see which gives better performance in this case?

Also I just saw this Asus board that can run 2x socket 1366 xeons http://www.newegg.com/Product/Produc...sus%20Z8NA-D6C which might be even better with 8 real cores vs 4 on the 920.

As I said above the Quad 6600 was a real dog with 6. I am hoping that I use the full features of windowsnapper and repeater with the I7.

I agree with the "bad code" analysis. I will do some testing with different configs to see what works best. I am thinking of locking instances to each core thread (except the main) for the full hyperthreading and assigning all cores to all instances with hyperthreading off. I know that core/thread priority was a little bit funky on the Q6600 and I am not sure if that was M$oft thread scheduler, IS or EQ acting up. Most likely it was thread scheduler due to the heavy load saturation. Either way I should have some testing numbers by this weekend.

And stop showing me gear I can't afford!! :P