Several days ago, one of our (Dutch) forum visitors asked about a Gigabyte motherboard with difference characteristics than the original model, and visible differences from the photo displayed by the shop where it was purchased. A bit of Googling and browsing through Gigabyte's website shows that the Taiwanese manufacturer makes new revisions of quite a few of its motherboards, mostly in the budget segment. For us reason to do some more research and buy and test the motherboards. The results are more than a little remarkable...
Gigabyte has been making new revisions of its motherboards, using the original product name, since time immemorial. This is nothing special, nor anything the manufacturer is less than open about. New motherboards are easily identified by the addition to their name: Rev. 2.0 or Rev 3.0 for example. However, lately these revisions appear to be primarily instigated by cost considerations. In the boards we tested, we observed very clear differences, such as a lowering of the number of CPU power supply phases and in one case even the removal of a secondary BIOS chip.
Two revisions of the B85M-HD3. Quite suitable for a game of "spot the differences"!
Although Gigabyte is fairly honest about the differing versions on its website (even though changes like a lower number of phases are not to be found in the official specification lists), this is not an approach we are enthusiastic about. The reason for that lies in the way Gigabyte brings the new products to market, using exactly the same manufacturer and UPC codes as for the original versions. Thanks to that, distributors and online shops (who work with automatic links based on these codes) cannot tell the difference. Hence, they can receive - and thus, sell - a different product, without being aware of the change in the product. Of course for you, the end user, there is no way at all to tell the difference, until you receive the product and open the box.
So, as a potential buyer you check specifications and photos of the product at the shops, but these are based on the first revision; once you order, you will probably receive a newer revision. Because price comparison engines such as Hardware.Info also use UPC and product codes to categorise information, these also don't spot new revisions and thus keep showing old specifications and photos. And thus this way too doesn't offer you as customer correct information about the product you are ordering.
New revisions of existing products without changing the codes unfortunately is hardly unique to Gigabyte. We have had our share of discussions on the topic with manufacturers of power supplies such as Nexus, Antec, Seasonic and XFX, who have the same bad habits. An important difference: in all cases we discussed with those brands, the new products were clearly superior to the predecessors. Although we remain of the opinion that significant product changes should come with new product codes, it is less of a problem to deliver a better product to people, than when it is a markedly worse one.
A more similar situation we encountered a while back with Kingston V300 SSDs, which suddenly turned out to use a different kind of flash memory than the type they had when we first tested them. This caused significantly lower performance in our tests. Kingston was of the opinion that there was no change to the official specifications, which also applies in most of the differences we spotted with Gigabyte's mainboards. However, just like with Kingston we are of the opinion that even though this may pass muster legally, it is a less than wholesome approach to doing business to supply a product to a customer that is clearly different from what they expect, when they have chosen it based on a positive review or viewing photos online.
When in doubt, test
In order to find out what's what we went to town with two entry level Gigabyte motherboards. Specifically, revision 2.0 of the Gigabyte B85M-D2V and B85M-HD3, obtained from Azerty.nl and Alternate.nl. Unfortuantely we didn't have a revision 1.0 of the first board, but we do have our original photos to base some analysis on. The second board still was present in our lab in all its 1.0 glory and hence we could do a number of tests to see if there were any differences in performance - and indeed there were...