It seems that AMDTech got out there and make a whooping review of different motherboard makers based on the i965 chipset. Well, all I can say is that they are learning when judging what motherboard to get. I have been saying this in the forums for such a long time: it's not all about performance; warranty and support must also be top notch.For extreme hardcore enthusiasts, it's not much because most of the time, their warranty is void anyway. But for 99.999999999% of the population out there (yes, I am exaggerating) performace difference of 0.00000001+/- between manufacturers isn't much of a big deal. Oftentimes, users will end up having a nightmare of an experience RMAing their broken board. Anyway, if you woud like to read on their thoughts, expand this posting...
Final Words
After twenty five pages of information we can safely say that an Intel P965 is still an Intel P965 no matter which motherboard you place it in. This should come as no big surprise as the days of one motherboard manufacturer truly outperforming another one at stock clock speeds with the same chipset is over. The basis of competition is now on features, price, warranty, appearance, overclocking capability, accessories, availability, and reliability. If that sounds a lot like the motherboard being a commodity item, it is, as the consumer expects or demands a certain level of performance and support from all motherboards. They will pay extra for the features that are deemed important based upon their needs. One only has to look at the fallout and consolidation of the motherboard suppliers over the past couple of years to realize this simple fact.
While we believe performance is still extremely important, it is no longer the only reason to consider a particular motherboard supplier. In our opinion, the features, price, warranty, support, and reliability of the motherboard should be first on your list as base performance will generally be equal among all motherboards based on the same chipset. To a lesser degree even the performance amongst various chipsets for the same CPU family performs almost identically now. It comes back to features and support for making that final decision. We would love to make a final decision on the boards we tested today but we have another eight or so left to present to you. We can draw a few conclusions from the results and information we presented after testing these boards for past few weeks. Our first conclusion is that the P965 was released too early and the first motherboards while being very solid from a features and quality viewpoint had very immature BIOS releases.
The inability to boot many boards with most performance oriented memory modules was inexcusable. Do not get us wrong as it took two to tango this sorry dance. We blame the memory manufacturers also as they were just as guilty by having modules in the market with SPD settings that assumed the board would boot at 2.0V or higher. While the motherboard manufacturers will state they followed Intel's 1.8V requirement they are still guilty for not having the BIOS cycle properly to recognize the memory speed, voltage, or timings after the POST issue. The majority of these issues have been solved with the latest BIOS or SPD releases. We still cringe when installing a new memory module but now concentrate on how well it performs instead of crossing our fingers and calling the psychic hotline.
Other issues included incompatibilities with the new SATA/IDE controller chips resulting in PIO mode operation or failure to see the drive at all. This goes back to quality control and pushing an early release. It also brings up the point that Intel in their infinite wisdom decided to pull PATA support from this chipset when over 98% of optical drives are still based on PATA technology. They could have waited for the Bearlake chipset next year to do this. This change also increased the cost and complexity of the motherboards as you now introduce another chipset on each and every motherboard sold.
Overall, the BIOS releases have greatly matured and the amount of issues have declined sharply over the last two months. In this case we have to give credit to the motherboard and memory suppliers for acting quickly... but only after upsetting an untold number of customers. While there are still a few nagging issues here and there, like getting AHCI to work on the Intel ICH8R equipped boards without an engineering degree and a day off, we are seeing most of the issues being reduced now to what we have come to live with in an open PC hardware world. Those issues would be where a combination of parts that you would never expect to be used together creates an issue that is extremely difficult to fix or even diagnose. We still get mail on a fair share of those and will report our findings in the final article. We can safely say that with out list of components used in our test bed that our motherboards were virtually error free of issues that would cause instability or create a no POST situation.
Overall, the Gigabyte GA-965P-DS3 performed the best in our benchmarks when not overclocked. This is a very good accomplishment and shows a level of consistency and fine tuning that was not matched by the other suppliers. However, the total margin of victory over the other P965 motherboards is less than one percent. It took several hundred hours of benchmarking to come to this conclusion and without those benchmarks it would have been impossible to tell the difference between any of the P965 motherboards tested today.
The overclocking capability of ASUS motherboards continues to be impressive in their mid-range series. If you are looking to get the highest possible overclocking results with an E6300 or E6400 processor then we would recommend the ASUS P5B-E 1.02G at this time. Of course it is not available yet (it will be shortly) and unless you have expensive PC2-8000 RAM then chances of going over 500FSB are not that great with the ASUS P5B-E 1.01G board. However, it is hard to complain about the ASUS 1.01G board with 7x490FSB results using mid-range PC2-6400 memory. Gigabyte has not solved their Micron D9 1GB module issue yet so they are stuck in the 450FSB range with our test components. Biostar has a 500FSB level BIOS limit but more importantly you cannot find the board. That brings us to Abit and the AB9-Pro which has improved its overclocking capabilities a great deal since we first reviewed the board but apparently has reached its limits at this time. If you have an E6400 or E6600 then this board still has great overclocking potential for the majority of users.
When it comes to features we really liked the ASUS and Abit boards as they provided just about every possible option on a midrange/performance motherboard that one could want. While the Gigabyte and Biostar boards are also feature rich, they both lack Firewire support which should be a given on boards in this price range. We have to have to give Biostar a gold star for overall layout design although the location of the 24-pin and 4-pin ATX power connectors are a detraction. The ASUS and Gigabyte motherboards have fairly standard layouts that we could live with on a daily basis but we have to wonder what the layout design group at Abit was thinking when they placed the IDE and two SATA connectors between the PCI Express x1 and PCI slots. We will call it a creative design inspiration at this time to be nice as we still like the board's overall capabilities.
The Analog Devices AD1988A HD Audio Codec really put the screws to the Realtek ALC-88x series of HD Audio Codecs in our EAX 2 gaming tests. The EAX sounds were clear and concise unlike some of the warbling and muddy sound generated by the Realtek codecs in our Battlefield 2 test. While the audio quality of both codecs was almost equal in our standard game, DVD video, and audio tests we still think the ADI solution had the superior overall audio quality.
Wrapping up part one, there was not a real loser in this group. Each board has its strengths and weaknesses so it comes down to what the individual user wants in a board that will meet their needs. We received varying support from each supplier and not in a way you would think. Our direct support was excellent but we wanted to find out how well the manufacturers supported a retail customer so we acted like one. We logged on to their support forums if available and asked questions about our issues or those of other users. We purchased retail boards (when possible) and called technical support. We emailed, faxed, or otherwise bugged the hell out of some customer support personnel for the last six weeks. We will provide our results in the final article and these results will help determine our Editor's Choice awards. After all, it's not just about performance anymore.
Source:Intel P965: Mid-Range Performance Sector Roundup
No comments:
Post a Comment