- Joined
- Dec 22, 2000
Summary:
AMD has tested the Asus A7M266 DDR Socket A board revision 1.03 and found it to be unreliable at 266 FSB operation. AMD will not be certifying the V1.03 board for 266 FSB. They are currently testing a V1.04 board from Asus.
AMD's certified list for 1.2 Athlons is located at:
More Details:
Back on December 4th, The Register posted a story that Gigabyte had discovered a stability problem with the AMD 760 chipset at 266 FSB (http://www.theregister.co.uk/content/archive/15211.html). Apparently, additional RAM buffering was needed for stability at 266.
Gigabyte had, by the time of the story breaking, distributed pre-production models of the GA-7DX 266 DDR board to reviewers. Due to the stability problem, that board was never produced as the GA-7DX -- instead it was renamed the GA-7DXC (watch that "C" suffix, it's important), and the specifications on Gigabyte's site were changed to 200 FSB. Gigabyte's site also lists the GA-7DX, which has not yet shipped, as 266 capable. So, got that? The original 7DX that some sites reviewed became the 200 FSB only 7DXC, and the fixed board will eventually be sold as the 7DX. Beware, a number of merchants on PriceWatch are selling the 7DXC as a 266 FSB board! Just refer to the specs on Gigabyte's site for confirmation.
AMD didn't hear about Gigabyte's rename of the 7DX, so if you look at AMD's certification list, the board which is now known as the 7DXC is listed as "GA-7DX 2.4 (See Note)". The note says, and I quote, "200 MHz Front Side Bus ONLY!!" Yes, the caps and exclamation points are AMD's. The GA-7DX 3.0, to be sold as the "real" GA-7DX, has no such disclaimer attached.
Gigabyte did the right thing in renaming their 200-only board, and changing the specifications on their web site. Meanwhile, Asus shipped the A7M266 (which has 266 in its very name) as a 266 capable board. Strangely enough, although I have an A7M266 in my hands, purchased from a US retailer, there is no A7M266 listed on Asus' web site except in their press release from October 30th, and their BIOS update page (where it restates in a post dated 1/4/2001 that it's a 266 FSB board). Asus is clearly guilty of insufficient testing and should recall the 1.03 revision of the A7M266.
Now, you can put PC2100 DDR RAM on either the GA-7DXC or the A7M266, and run at 266 FSB, but it will not be reliable. Obviously it's reliable enough for a couple of sites to get through a round of benchmarks, but it's not something you'd want to run every day. Without ECC, memory can simply become corrupted, and that's what AMD says will happen at 266. Of course, just like with any other overclocking experiment, I'm sure it's possible for someone, somewhere to get lucky at 266 FSB with these boards, so keep that in mind if someone posts "It works for me", because the chances of it working for everyone are slim to none.
This whole thing raises some tricky questions regarding the shipping KT133A boards, such as the Abit KT7A/KT7A-RAID. Note that none of the KT133A boards are shown on AMD's certification list. It's not because they're too new -- if that were the case, then the other 266 DDR boards wouldn't be listed by AMD either. No, I'd have to say they aren't listed because their first revision may not be reliable at 133 FSB either.
AMD says the mainboard vendors are shipping these new platforms too soon (especially since no 133 FSB Athlons are available), but AMD has no control over them. With this debacle going on, I'd strongly recommend waiting to buy any 133/266 FSB Athlon board until it shows up on AMD's certified list. Remember what happened when people tried to use non-certified power supplies?
Good luck fellow hardware fanatics,
KC
AMD has tested the Asus A7M266 DDR Socket A board revision 1.03 and found it to be unreliable at 266 FSB operation. AMD will not be certifying the V1.03 board for 266 FSB. They are currently testing a V1.04 board from Asus.
AMD's certified list for 1.2 Athlons is located at:
More Details:
Back on December 4th, The Register posted a story that Gigabyte had discovered a stability problem with the AMD 760 chipset at 266 FSB (http://www.theregister.co.uk/content/archive/15211.html). Apparently, additional RAM buffering was needed for stability at 266.
Gigabyte had, by the time of the story breaking, distributed pre-production models of the GA-7DX 266 DDR board to reviewers. Due to the stability problem, that board was never produced as the GA-7DX -- instead it was renamed the GA-7DXC (watch that "C" suffix, it's important), and the specifications on Gigabyte's site were changed to 200 FSB. Gigabyte's site also lists the GA-7DX, which has not yet shipped, as 266 capable. So, got that? The original 7DX that some sites reviewed became the 200 FSB only 7DXC, and the fixed board will eventually be sold as the 7DX. Beware, a number of merchants on PriceWatch are selling the 7DXC as a 266 FSB board! Just refer to the specs on Gigabyte's site for confirmation.
AMD didn't hear about Gigabyte's rename of the 7DX, so if you look at AMD's certification list, the board which is now known as the 7DXC is listed as "GA-7DX 2.4 (See Note)". The note says, and I quote, "200 MHz Front Side Bus ONLY!!" Yes, the caps and exclamation points are AMD's. The GA-7DX 3.0, to be sold as the "real" GA-7DX, has no such disclaimer attached.
Gigabyte did the right thing in renaming their 200-only board, and changing the specifications on their web site. Meanwhile, Asus shipped the A7M266 (which has 266 in its very name) as a 266 capable board. Strangely enough, although I have an A7M266 in my hands, purchased from a US retailer, there is no A7M266 listed on Asus' web site except in their press release from October 30th, and their BIOS update page (where it restates in a post dated 1/4/2001 that it's a 266 FSB board). Asus is clearly guilty of insufficient testing and should recall the 1.03 revision of the A7M266.
Now, you can put PC2100 DDR RAM on either the GA-7DXC or the A7M266, and run at 266 FSB, but it will not be reliable. Obviously it's reliable enough for a couple of sites to get through a round of benchmarks, but it's not something you'd want to run every day. Without ECC, memory can simply become corrupted, and that's what AMD says will happen at 266. Of course, just like with any other overclocking experiment, I'm sure it's possible for someone, somewhere to get lucky at 266 FSB with these boards, so keep that in mind if someone posts "It works for me", because the chances of it working for everyone are slim to none.
This whole thing raises some tricky questions regarding the shipping KT133A boards, such as the Abit KT7A/KT7A-RAID. Note that none of the KT133A boards are shown on AMD's certification list. It's not because they're too new -- if that were the case, then the other 266 DDR boards wouldn't be listed by AMD either. No, I'd have to say they aren't listed because their first revision may not be reliable at 133 FSB either.
AMD says the mainboard vendors are shipping these new platforms too soon (especially since no 133 FSB Athlons are available), but AMD has no control over them. With this debacle going on, I'd strongly recommend waiting to buy any 133/266 FSB Athlon board until it shows up on AMD's certified list. Remember what happened when people tried to use non-certified power supplies?
Good luck fellow hardware fanatics,
KC
