Anand has performance covered

[DW_Hornet]

http://www.anandtech.com/video/showdoc.aspx?i=3127

[zeus]

Interesting article. Damnidge, you should read this for your future build.

Ugh.. I have a 6300 which even though it is a 6000 line card, only have 2MB of cache.

[KillaDEATH]

I was actually thinking about doing this myself so I'm glad they've done it for me. Anandtech is a mostly respectable site so I would believe what this article states. Like the article, I don't think the flyby is that great of a benchmark because nothing is going on during the flyby. One thing they didn't mention that I feel ads to this is network traffic. When even a Core2Duo reports %100 CPU usage the difference between recorded demo and actual online play will be a factor in my opinion. But we'll see maybe when the non-beta demo comes out(if it comes out). On the GPU side of things we once again notice that the UT3 engine seems to like the ATI card. At the lower rez like 1280X1024 you see the 2900XT spank a 8800 Ultra by 10FPS. The difference swings the other way the higher the resolution but remember a 19" 4:3 LCD is only 1280X1024 and the 2900XT is close to half the price of a 8800 Ultra. I was interested in what a multi-GPU set-up would do but we'll have to wait or do the tests myself. I also would like to know what amount makes a difference between 2,4 and 8gigs of memory and if a few Vista features like Readyboost does anything for lower memory configs. Again something I might have to do myself when I get time. A real recordable benchmark would make me want to get right on that.

Zeus we knew the smaller cache would catch up to us when we bought it. Luckily we're able to make up for it by the huge overclock the 6300s are capable of doing.

Too bad they didn't throw a Phenom in there to make the last important comparison.....

[DW_Hornet]

You can do fly bye demos with bots going on, and that will tax your system even more because of the supposedly super advanced AI. However since you cannot watch a pre-recorded game (since this is a beta demo) where everything is not 100% consistent you are going to get variations (sometimes large) in your performance. The Bots will do as they please, and there is to much randomness to what will actually happen. So statistically speaking having no bots is better then having any.

Numerically with 12 bots on just about every map, my average frame rate (with my settings) usually drops a whopping 8-12 FPS.

As for Multi-GPU, they had stated you will only really gain performance from running at higher resolutions. Also keep in mind there is quite a bit of CPU overhead that is added. So some of these numbers may actually do down.

[DW_Hornet]

Map AVG frame % over 60 Bots
DM-HeatRay 176.89 99.60 Yes
DM-HeatRay 191.56 99.89 No

Difference = 14.67 .29

[zeus]

Zeus we knew the smaller cache would catch up to us when we bought it. Luckily we're able to make up for it by the huge overclock the 6300s are capable of doing.

Yes and no... This is the first time that I have seen a significant difference when talking about cache on the C2D. Previous benchmarks have seem to make it near insignificant.

However, part of the reason the original 6300s and 6400s were so good at overclocking was just having the 2MB of cache vs. 4MB.

So, it is a wash either way. And, I have had my system since March, so I have gotten use out of it. At some point, you have to just pull the trigger. ;)

I will probably wait for penryn before upgrading at this point if cache is going to make that much of a difference because they are getting bumbed up to 6MB of cache if I recall correctly.

[KillaDEATH]

Well then that's why I bought the 6300 back in December I think it was. I was convinced that when things like games started to finally make use of Dual cores the cache size would also show it's benefit. I since then bought a 6600. That would have been alot of die realestate to waste on that much cache had it not come to show a benefit. I've been happy with it so thats all that matters to me...

[KillaDEATH]

Hornet I understand where you're coming from and haven't tried it yet myself. I would still rather have something repeatable with bots to compare. I have more than one system that needs to be up to spec because I have more than one player in the house. I know my system will have no problem. I know a second machine will have very little problem with reduced settings. I may still have another machine to build so I'm very interested in what I can get by with at both ends of the scale. To be honest I'm not exactly sure I like the game enough to buy the multiple copies it would take for this to happen. I spent many hours playing CTF in the original so I'm a little burnt out on that. Invasion is what I like right now so I'll probably wait to see what happens with that before I buy that many copies.

[Damnidge]

I just read through it. Seems my current part list should suffice quite well.

The thing that caught my eye was the comment that the "sweet spot" for processors was the E6550 or the E4500. The E6750 is only $20 US more than the 6550, so it's a no brainer to go up to the 6750.

I am waiting for the great deals for parts like the case, power supply, fans, etc. - things not dependent on hardware decisions. The rest I am waiting until after the new GPUs and CPU come out next month per many people's advice.

[WHIPperSNAPper]

I found the study on the value of Level 2 CPU cache enlightening.