Specification and Test Normalisation

I think that it is impressive that an fundamentally limited product can meet or beat a significant superior spec'ed product.

I would like to see product normalisation introduced into testing.

This would be where the performance of two products are normalised based on the specifications of the inferior product, i.e. the superior product has part of it's specification disabled to the point where it matches the inferior product. For example Caching is restricted, external memory speed and simm count is limited.

In this case the Intel Core i7 Chip would loose part of it's cache and be limited to pairs of memory simms.

If this cannot be applied to the system for testing then results of tests should be extrapolated.

Equally the results of test can be extrapolated to indicate what would happen if the AMD product had extra cache and support for 3 memory simms with a higher speed.

By j_woolliscroft on 13 Aug 2009

Big White Space

Could you please put 'related' links here, underneath the advertisers and to the right of the comments. Links such as 'Motherboards, processors, memory, hard disks and all the other components' would come in handy for many people -I'm sure. Thanks

By nicomo on 13 Aug 2009

What I mean is that people will not have to scroll back up to the top menu and filter - instead on the right could be a filtered menu related to the article - in this case AMD Mobos would be shown top-most along with other AMD processors.

By nicomo on 13 Aug 2009

missing details

What about power consumption?
For many organisations this can be a critical factor but is almost always omitted from PCPRO reviews.

By darkhairedlord on 13 Aug 2009

missing details

ah, ha, I see the new website has a specs tab!
appologies for my oversight.

By darkhairedlord on 13 Aug 2009

nice price but a bit of a goldfish?

Darrien, when the AM3 platform was introduced there was a memory controller issue whereby you could only use 1 dimm slot per channel, unless you throttled back the RAM to 1066 MHz. AMD said at the time they were working on a fix but having scoured loads of forums I can't find anything, but have you heard anything? As it stands it's a serious limitation on the platform as there's no point in having ddr3 Ram at those speeds, which effectively puts a 4GB limit on Ram versus Intel's 12GB maximum which isn't especially futureproof.

By felefant on 13 Aug 2009

darkhairedlord: Good point. Sorry for hiding that information away... especially since AMD only announced after the launch that the TDP is not actually 125W after all, but a whopping 140W! Apologies - we've updated the review now.

felefant: Happily, it seems motherboard manufacturers have solved that problem: the MSI 790FX-GD70 board we used in our tests officially supports 16GB across 4 DIMMS at speeds up to a remarkable DDR3-2133. If you like, I can stick some fast DIMMs in our 965 system and take a capture in CPU-Z so you can see for yourself...? (It'll have to be next week though as I'm out of the office today!)

By DarienGS on 14 Aug 2009

Nice one, that's good to know, but it's ok I can take your word for it if you like! I thought I'd wasted my money but from what you say it looks like flashing the bios will fix it as I have the same 790 chipset. Have to admit I was a bit worried after reading a feb. article with 200 odd comments from fanbois flaming each other within 48 hours...

By felefant on 14 Aug 2009

Specification and Test Normalisation

I think that it is impressive that an fundamentally limited product can meet or beat a significant superior spec'ed product.

I would like to see product normalisation introduced into testing.

This would be where the performance of two products are normalised based on the specifications of the inferior product, i.e. the superior product has part of it's specification disabled to the point where it matches the inferior product. For example Caching is restricted, external memory speed and simm count is limited.

In this case the Intel Core i7 Chip would loose part of it's cache and be limited to pairs of memory simms.

If this cannot be applied to the system for testing then results of tests should be extrapolated.

Equally the results of test can be extrapolated to indicate what would happen if the AMD product had extra cache and support for 3 memory simms with a higher speed.

By j_woolliscroft on 16 Aug 2009

RAM support confirmed

To felefant and any others who may be concerned about AM3's memory support: I've added a screenshot to this review's picture gallery that shows CPU-Z confirming 8GB of DDR3-1600 in dual channel mode. Hope this clears up any doubts!

By DarienGS on 17 Aug 2009

wow, I didn't realise they even made RAM that fast! thx

By felefant on 20 Aug 2009