Staff Blog
12345
Rated: 100% (1 votes)
Loading ... Loading ...

Phenom in the Lab

James Gorbold

Posted in Staff on November 22, 2007 at 1:06 pm

As we didn’t want to publish any of AMD’s silly canned benchmarks from its ridiculous event in Warsaw last week we’ve managed to borrow a Phenom from some kind lads up North (i.e. not AMD, which hasn’t exactly been forthcoming with samples). The chip is currently benching and we should be able to get some results up later today.

Phenom in the Lab


 

10 Comments

Slightly off topic I know but……….
I’ve only been reading since issue 39 and I don’t know that much but I find out more and more thanks to you guys. If only I’d known about your mag before I built my pc. Anyway, as you do lots of labs tests and issue 52 did some motherboards I was wondering why cpc hasn’t reviewed the Intel Desktop Board DX38BT to see how their own ‘board stacks up against everyone else? is this test possible or do they not sell it direct to the public? Thanks for a great mag.
Andy

Comment by Andrew Allison - November 28, 2007 @ 11:12 pm

 

We have reviewed a few Intel motherboards before, but they are mainly designed for big system integrators so are usually quite hard to buy in retail. AFAIK the DX38BT isn’t available yet, although Intel’s PR may just have forgotten to send it to us.

Comment by jamesgorbold - November 29, 2007 @ 11:36 am

 

So, when are you going to bench a G0 stepping X3220 so we can see the overclocking difference against the Q6600? I mean, if you’re prepared to advocate lapping your IHS, surely you’d be jumping all over an extra 14°r;C of Thermal Spec?

Comment by Chris Cox - December 2, 2007 @ 9:38 pm

 

…and that’s what you get for trying to be smart and use the right character entity.

Comment by Chris Cox - December 2, 2007 @ 9:40 pm

 

We managed to get hold of a X3220 in time for this years CPU megatest (for Issue 53). Unfortunately we didn’t get a chance of overclock every CPU, but there is another much more pressing reason why the X3220 isn’t as interesting as you might at first think.

Comment by jamesgorbold - December 3, 2007 @ 11:10 am

 

Don’t keep us hanging on, man - what, for the love of God?

Comment by Chris Cox - December 3, 2007 @ 4:42 pm

 

Ok, so I assume you’re referring to the results of the prefetcher tuning, as seen in the latest megatest. To be honest, I hadn’t expected it to make such a difference (25% in games - wow) but the language used was a bit fuzzy and I’m still confused. I understand the server use-case scenario, but what exactly qualifies as a workstation application? If you were to ask me, I’d throw a guess in the direction of CAD/CAM and nonlinear video editing, but if so I’d imagine it would score higher in the Handbrake portion of your benchmarks. On the whole, I’d be prepared to put the other results down to margin of error of the tests - the X3210 outscored both the Q6600 and the X3220 in the multitasking test - but before I blow £10k of my boss’s money on building four systems to run Avid Xpress Pro, I’d love to be able to figure out what constitutes a workstation application in your wording (for example, an application that utilizes more than two cores?) so I can make the best purchase decision - as well as what to buy to put in my P5K-E and clock the nuts off.

Comment by Chris Cox - December 26, 2007 @ 11:10 am

 

Yes, it is pretty amazing how much of a difference the tuning of the pre-fetchers can have on game/application performance.

If the language used in the review was a bit fuzzy its because we used Intel’s own terminology (workstation/server). For workstation I’d also go with CAD, CAM, video, audio editing, 3D modelling, 3D rendering, 3D animation type applications - i.e. not strictly by the number of cores an application may be able to take advantage of, but by what the application is used for.

Comment by jamesgorbold - January 2, 2008 @ 12:04 pm

 

H’mmm. Without meaning to flog a dead horse, I’ve seen an ES X3220 that does 3.6 8×450 at 1.4v - if that’s representative of normal steppings - I defer to your expertise on this as I’ve no idea whether ES chips in general run better than production models - wouldn’t this mean that the odd few percent deficit in desktop applications (and possibly even the 25% in games) could be overcome with the right cooling setup?

Comment by Chris Cox - January 10, 2008 @ 2:58 pm

 

ES chips are a bit of a hit and miss affair - some overclock even better than retail chips, but some are so early they perform worse than retail chips.

Re the X3220 I don’t think its worth the risk over a cheaper Q6600 (especially if the latter is a G0 stepping). However, if you’re on a tight budget the X3210 is definitely worth buying, as by overclocking it you will be able to eliminate the performance drop caused by its pre-fetcher tuning.

Comment by jamesgorbold - January 10, 2008 @ 3:27 pm

 

Make a comment
  • * required
  • * required
Advertisement
Most commented posts
Highest Rated Blog Posts