Need hardware upgrade advice - should I go dual core or quad core?

[Xpleet]

meh i would say get 45NM quad. People say you cant oveclock Q9300 much because of 7.5 multiplier. Well i think its bullshit, to get good overclock from 65NM you need some HARD ASS cooling to go over 3.2ghz. Q9300 can  get 3.2GHZ with default cooling.

Q6700 too.

Easily. Cooler still running on 2/3 fanspeed on full load lol.

[Xpleet]

What this benchmark shows is disappointing.

Q9300 OC reaches 5100 points at 3.5 gigaherz! 5100 is what I get with my Q670@3,2!!!

[buchvecny]

well... thats pretty strange. Must be the cache. However in overall tests q9300 is superior by 4-5 percent. Its less effective when playing games (cache). However its more effective when encoding video (sse4.1).

[FrankB]

...not sure if the 3DMark performance would tell you anything about render performance, especially when multi-threaded.

[Xpleet]

...not sure if the 3DMark performance would tell you anything about render performance, especially when multi-threaded.

Yes you can.

If you do compare quad to quad. Definatly. Compared mine with a dualcore I had earlier and it fits perfectly. Also compared my non-OC with now OC'ed q67.

3DMark06 runs a test that pushes the CPU to it's max. It's known to greatly advantage Quadcores over Dualcores (like it's supposed to be) in opposition to games, which put Quad and Dual in the same box because it's hard putting physics, computing, sound and other minor things in 4 equal threads. Because Physics and Computing is still the two most relevant and heavy things, therefore Dualcores are still being produced today!!! You simply don't need more there.


Hey this is the most official and professional 'BENCHMARK' and who is to say that you can not really compare performance in it?

[FrankB]

I don't know, I'm not an expert, but I was under the impression that those 3D benchmarks don't use the CPU cores in the way a renderer would, and that much of the 3DMark performance would be related to the GPU power....

[Xpleet]

I don't know, I'm not an expert, but I was under the impression that those 3D benchmarks don't use the CPU cores in the way a renderer would, and that much of the 3DMark performance would be related to the GPU power....

These are CPU-only tests. Sort of extra scores just for the CPU. They run a test that runs just 3fps in 640x480. These results from the quadcores are real I can confirm, I tested myself.

[BlueRose]

  Im a bit lost, what are you guys saying here?

Im still not certain whether to go dual or quad - I like the idea of dedicating 3 cores to TG and leaving one to do other things, that is tempting.  Pricewise its much the same so for me at this point its about performance.

[nikita]

Quad of course.

[rcallicotte]

What he said.

[PG]

Go for a Quad only if it has high spec, i.e 2.66Ghz or more, 8MB cache ideally, the 45nm yorkfields are very good and can be overclocked a lot more than 60nm's without heating up too much because you don't need to increase the voltage unless you go for 1.3GHz overclocking rates or higher.

[neon22]

So in NZ something like this is pretty good (the US based people can feel sorry for us now OK)
http://www.pbtech.co.nz/index.php?item=WKSPB2191&PHPSESSID=99d7c0777405f70da7ca30b326eefec7
from: http://www.pbtech.co.nz/index.php?page=2&s=PB%20SYSTEM

[Xpleet]

Go for a Quad only if it has high spec, i.e 2.66Ghz or more, 8MB cache ideally, the 45nm yorkfields are very good and can be overclocked a lot more than 60nm's without heating up too much because you don't need to increase the voltage unless you go for 1.3GHz overclocking rates or higher.

the tests i showed above speak for themselves.

[PG]

Not really, benchmark programs are notorious for being one track minded. When it comes to real world testing they just aren't a true representation. The Geforce 9600 for example, didn't do brilliantly inder 3d mark, yet it runs call of duty 4 better than an 8800GTX. The intel core 2 duo E8500 had a massive improvement over the 8400 under 3d mark, yet in real world testing it makes no significant difference and can;t be overclocked nearly as far. I wouldn't advise taking benchmark results too seriously. Find tests from game and other program benchmarks. There needs to be a comparison between programs.

[rcallicotte]

Tom's Hardware is a good place to look for these sorts of tests.  And Sharky's.