You could find it on the website of the German Magazine 'PC-Welt' late yesterday and then later on The Register as well. Intel is developing a chipset that will support (DDR?)-SDRAM. This chipset carries the code name 'Almador', not 'Armador', and is scheduled for the beginning of 2001. Our latest information shows however, that 'Almador' is not a chipset for Intel's next generation IA32 processor with the code name 'Willamette', but for the upcoming low-voltage 'Coppermine'. Those future 'Coppermine' processors are nothing mystical; they are not just notebook CPUs, but the next generation Pentium III processors manufactured in 0.13 micron process. Thus the GTL+ signals will carry 1.3 V instead of the 1.5 V that are used by 'Coppermine' Pentium III right now. You can imagine, that the lower GTL+-voltage requires a new chipset and that's what 'Almador' is going to be.

You have certainly read the Solano Preview yesterday and might remember my little quiz at the end of the article. Well, the answer is clear, the performance of Solano would obviously increase even more if DDR SDRAM was supported by it. Thus RDRAM would be declassed even more too. The 'Almador' info comes just at the right moment in time, because 'Almador' will obviously be the successor of Solano (aka 'i815'), and it comes with integrated 3D graphics as well. This is really good news and shows that Intel might have realized that they can score a huge amount of plus points with their worldwide customers if they offer alternatives to RDRAM. Now the only thing I want to hear is that there will indeed be a chipset for 'Willamette', which will support different memory than RDRAM as well. DDR SDRAM seems to be the most obvious choice right now, but I am aware that there are even faster alternatives to Rambus under development right now. It wouldn't surprise me if Intel was looking into those as well. This could mean that the Rambus star is slowly sinking.

Solano - A Few Possible Additions And Corrections

As I stated in the Solano Preview yesterday, we are lacking data sheets of i815, so that we cannot guarantee that all specs we published are completely correct. Our latest information suggests that we have to make a few corrections and additions.

Solano is probably able to clock the processor front side bus (FSB) asynchronously to the memory bus. A recent article at 'Electronic Buyer's News' suggests that. We can still not give you a definite statement however, since we are lacking hard data on this issue. Our test motherboard was definitely NOT able to do this and the PLL used on the board was an ICS-type and not an IMI.
Asynchronous FSB/memory clock would have the advantage that a 66 MHz FSB Celeron could still take advantage of SDRAM running at PC133 spec and you could still run a 133 MHz FSB Pentium III with your older PC100 SDRAM. If Intel wants to make i815 look like a real budget version, asynchronous FSB/memory timing would make a lot of sense. We also might have been off-mark with the amount of memory Solano is able to support. Good informed sources out of the motherboard business told us that i815 is only able to support up to 512 MB of main memory. I tried to verify this on our test board, but the pre-release board wouldn't even boot with one of my 256 MB registered DIMM modules. The price of the i815 chipset is supposed to be a hefty $55 , as opposed to $45 for BX. This price could be another trick Intel is using to make sure that i815-demand won't get too high. Two big Taiwanese motherboard makers told us that for them i815 is supposed to be available in decent amounts and that there wasn't any restriction to the OEM-market only. However, they do admit that not all motherboard makers might be in the same situation. Some people asked me if it was 'ethically correct' to review a pre-release chipset/motherboard. I am sorry guys, but didn't I praise i815 in the preview ? The results of beta-i815 were already promising and the final product can only be better than our test sample. Other people wondered why I didn't disclose the manufacturer of our test sample. Those must be very young readers. Maybe you don't realize that it could cost the head of the person who gave me this board without NDA if I would publish where it came from. Tom's Hardware has a long history of protecting its sources by all means. This won't change as long as I run this company.

You can see that information about the amount of available i815 chipsets as well as the OEM-restriction that Intel stated on its roadmaps so far is a bit inconsistent. I doubt that this inconsistency is due to false information. It seems more as if Intel is indeed considering turning away from Rambus, or at least diverting towards alternative solutions. I am sure that the Intel management is currently split in two halves, and so the official policy is not quite decided yet. The one half would probably prefer to forget that they've ever agreed to the Rambus deal, the other half still wants to hold on to Rambus. As much as I may have bullied Intel for the Rambus-situation so far, as much I would praise Intel for turning away from the stubborn Rambus aberration. A large production of i815 chipsets and the plan for the upcoming 'Almador' chipset with the possible support of DDR SDRAM could be read as a strong sign that Intel has finally got the message. Holding on to Rambus as the only solution is causing Intel a hefty image loss. Many OEMs turned away from Intel's chipset products, because RDRAM is too expensive, MTH was a bad solution even before the MTH-disaster and i810 is not interesting to their customers as well. Many OEMs are happily selling huge numbers of VIA platforms instead. Times could easily change again, and it's entirely up to Intel to let that happen. You could see in the Solano Preview that the upcoming i815 chipset is a good product. Once Intel realizes that good products are made for selling and not for mere politics they would get the first good press from me after a long history of criticism.

The Latest News From The MTH-disaster Front

The MTH-reboot/hang issue is already quite old news by now, but the computer users out there who are stuck with an MTH-motherboard have still not received any decent help. So far only distributors/OEMs/dealers, who had MTH-motherboards in stock were able to return those motherboards. This is valid for Intel as well as for other motherboard makers. The dealers were either receiving replacement boards and RDRAM memory or a refund.

What Happens With The Field Returns?

The real problem lies somewhere else though. Most of the boards and particularly the systems with MTH that were shipped to end-users have neither been replaced nor was there any refund. Most motherboard companies are trying to get around those really expensive replacements by denying that their motherboards are affected by the MTH-issue. One motherboard maker even came up with the fairy tale that a six-layer production would keep their boards from the reboot/hang problem. These statements are all incorrect and only more or less futile attempts to keep the customers from demanding the return/refund they are entitled to. Please be assured that every owner of an MTH-chip is indeed entitled to receive a replacement or refund, because Intel is not able to guarantee that the MTH-chip will work properly on any motherboard, regardless how many layers this board is using. Besides, the statement of an engineer from a well-known and well-respected motherboard maker goes "many layers are not any proof for quality, but for the inability of an engineer to do a reliable design with less layers".