System Builder Marathon: Overclocking

[muk]

Over the week we build and test low-, mid- and higher-cost systems at standard and overclocked rates. Then we compare them and look at price-performance. Today we compare overclocking capabilities.

http://www.tomshardware.com/2008/03/27/system_builder_marathon/index.html

 

[retro77]

With .34 overclock, it sure made a difference on the benchmarks. So maybe .34 is a lot when spread across 4 cores.

 

[cleeve]

Probably more to do with the fact the Phenom was only 2.2 GHz in the first place.

 

[Crashman]

2.2? Damn, I was thinking 2.4. Oops.

 

[sojrner]

curiosity question rather than a critique: With the radeon not oc'ing well on the oc'd proc, did you guys do a side compare between the proc oc vs. oc'ing the graphics on that low-cost system? [:mousemonkey:2] I am merely curious if with the games the oc'd radeon might prove the more worthy component to clock up? (depending on how much it ramps up that is) I know your decision was to keep it a "multimedia" comp so I understand sticking w/ the proc, but for curiosity's sake did you check the other direction a bit too?


interesting thought though is with the gains versus clock increase... you guys got healthy oc's on the top two systems and yet a much smaller bench gain as a result on average. I am thinking that as # of cores becomes higher and apps become more thread optimized we will start seeing that disparity increase. Where before the raw speed was king now the # of threads are becoming the new race... and I don't think we can "overclock" that number.

 

[randomizer]

I want that high end rig... alot.

 

[JeanLuc]

$837 seems a lot of money for a low end 'cheap' computer.

 

[cleeve]

With the radeon not oc'ing well on the oc'd proc, did you guys do a side compare between the proc oc vs. oc'ing the graphics on that low-cost system?

Good question, in fact it did cross our minds. Unfortunately we were VERY pressed for time this round.

What I really would have liked to do was compare our previous low-end system (based on an E6750) with our current low-end system. On the one or two tests I did for curoiosity's sake, the stock Phenom 9500 beat the E6750: having this in the article might have calmed alot of people down. Usually it's not a big deal because we already have the benchmarks, but since we revamped our whole benchmarking suite - and even changed the OS to Vista - it wasn't possible without a complete re-do, and like I said, we were really pressed for time.

 

[Arbie]

I certainly appreciate the work that Tom's staff puts into these builds and comparisons, whether or not you happen to get every detail I'm interested in. I check the site every day.

But here's one complaint: the comparison charts constantly change the order of items. One chart has cards A,B,C, and the next presents as A,C,B, and the next as C,B,A etc. This seems pointless and makes it much harder to scan the results - which is the opposite of what you're working for.

In the future, please choose a sequence and stick to it throughout the review.

Thanks,

Arbie

Arbie

 

[sojrner]

@cleeve: I understand.

@arbie: I agree, consistency is king

 

[Crashman]

Tom's Hardware charts are always organized by "best performance", hence the lack of "consistent order".

 

[randomizer]

Tom's Hardware charts are always organized by "best performance", hence the lack of "consistent order".

Maybe you should ask what the majority of people prefer. I don't mind either as I always read the labels anyway.