While researching the Ultimate Museum PC it was hard to avoid the Tualatin, the final 0.13-micron incarnation of the Pentium III. With speeds up to 1.4GHz and 512KB on-chip L2 cache, a pair of PIII-S Tualatins should provide decent oomph. There’s just one problem—there doesn’t seem to be any Intel chipset which would support dual Tualatin processors.
Motherboards with dual Tualatin support do exist, but nearly all of them have either ServerWorks or VIA chipsets. While it is possible to run a Tualatin with an Intel 440BX/GX chipset, this requires an adapter or a modified processor. The 440BX/GX chipset is (in theory) fundamentally incompatible with Tualatin processors due to a difference in signaling voltage levels (AGTL vs. AGTL+).
Ironically, Intel itself sold server boards (SDS2, SAI2) with dual Tualatin support… boards built around chipsets from ServerWorks (the ServerSet III). This inevitably led to a few conspiracy theories.
The suspicion was that Intel did not want to promote Tualatin processors because they made the Pentium 4 look bad. A 1.4GHz Tualatin performed as well as a faster-clocked Pentium 4 and used plain PC133 SDRAM instead of expensive RDRAM.
Intel’s position at the time (2001) was clearly schizophrenic: There was a huge push to sell Pentium 4s, power-hungry and under-performing processors. For laptops, Intel initially had no choice but to sell Pentium IIIs. For servers, there were Pentium 4-based Xeons, but again power-hungry. For low-power 1U rack-mount servers, the Tualatin Pentium IIIs were an excellent choice.
For reference, the TDP of a 1.4GHz Pentium III-S was 32.2W (less than the 34.5W of a 600MHz Katmai Pentium III) while the TDP of an early desktop 1.7GHz Pentium 4 was 64W. The contemporary Foster Xeon processor had a TDP of 56W at 1.4GHz.
It is a question whether the Tualatin Pentium IIIs might have scaled past 1.4GHz or truly hit a wall, especially in light of the (much) later Pentium M. It’s certain that they didn’t get much love from Intel, with questionable “features” such as having multiprocessing support disabled in all but the server versions. Based on the performance of Tualatins relative to Pentium 4s, it’s no wonder there was suspicion Intel just didn’t want the Tualatins around.
Whatever Intel did back in 2001-2002, nowadays a Tualatin 1.4GHz Pentium III-S is quite easy to find. I myself ended up with a small pile of 1.26 and 1.4GHz Tualatins and a question: What’s a good board for these babies?
Dual Tualatin Boards
There is a handful of motherboards which support dual Tualatin processors. There are Intel’s own server boards, the SDS2, SCB2, (both extended ATX) and SAI2. All are based on the ServerWorks ServerSet III HE-SL chipset. This is a high-performance chipset with support for interleaved SDRAM memory; the boards support up to 6GB RAM. The boards come with relatively poor on-board graphics (ATI Rage XL) and no AGP slots; this is typical for server boards.
Supermicro built several P3TDE boards based on the ServerWorks HE, HE-SL, and LE chipsets; the P3TDE6-G with a 2x AGP Pro slot. Supermicro also offered P3TDD boards based on the VIA Apollo Pro 266T chipset. The P3TDDE board came with an AGP Pro 4x slot which should be capable of supporting quite fast graphics cards.
There were several other dual Tualatin boards based on the VIA Apollo 266 chipset: The AOpen DX37, MSI Pro266TD Master (MS-9105), and Iwill DVD266. The Iwill was rather unusual because it supported DDR SDRAM, although with questionable benefit.
Back in August 2001, Anandtech had a nice review of the then-current dual Socket 370 boards with an emphasis on desktop boards. The most exotic was the Acorp 6A815EP, the only one based on an Intel chipset (815EP), despite the fact that the 815EP did not officially support multiple processors. Many boards were based on VIA’s earlier Apollo Pro133 chipset, often with the dreaded 686B southbridge.
I must admit being tempted to get one of the ServerWorks-based boards, just to see what Tualatin Pentium IIIs can do with a high-performance chipset. It’s truly a shame that Intel did not build any dual Tualatin chipsets… but ultimately just a footnote in the annals of corporate stupidity.
Of course, that was also when AMD was entering the SMP market with the Athlon MP.
It certainly don’t help that Intel’s only chipsets to support dual 133Mhz P3 processors (820/840) required RDRAM, which turned out to be a disaster. In response, Intel altered the integrated graphics only 810E chipset to create the 815 with AGP support and support for PC133 RAM, but as you can see the single processor only and 512MB RAM limit remained.
I don’t know if RDRAM was a bigger disaster than the 820/840 chipsets themselves… but from what I can tell, both were effectively cancelled and motherboards based on the 820/840 chipsets appear to be more or less nonexistent today. The 840 chipset looked promising but then it went all bad. See eg. http://www.theregister.co.uk/2000/03/09/intel_840_chipset_problem_hits/
Interestingly, Supermicro still has the specs for their 840-based board on their website: http://www.supermicro.com/products/motherboard/archive/PentiumIII/840/PIIIDR3.cfm But for some reason, it’s not listed among their Pentium III boards: http://www.supermicro.com/products/motherboard/P3/
After reading the AnandTech roundup article I’d be most interested in playing with the iWill DVD266u-RN myself.
The AGTL vs. AGTL+ issue from my experience is no big deal and I’ve had 3 Tualatin rigs based on the 440BX chipset. So yeah “in theory” the 440BX is not compatible but in practice it is. The CPU and socket modification is simple and reversible. All that is required is a board that can adjust the voltage down to the Tualatin voltage levels and a good board that handles a 133 MHz FSB. Unfortunately, my 10 minute search for a dual socket 370 board with a 440BX chipset that meets those requirements came up empty.
I’m hoping that the P2B-DS v1.06 with the appropriate slotkets is a very good way to make use of the PIII-S CPUs. If all goes well for me and you pursue this project, we could benchmark the VIA Apollo Pro 266T against the 440BX.
Cheers.
Yeah, that would be an interesting comparison! The Apollo 266 Pro chipset is roughly the same vintage as Intel’s 820/840, only without the silly RDRAM requirement. I may soon have one of the ServerWorks-based boards and am very curious about their performance as well.
As far as I know, there were no 440BX-based boards with “out of the box” Tualatin support. I suspect that boards based on 440BX stopped being manufactured sometime in 2000, not long before Tualatins became available. So Coppermine yes, but Tualatin no.
Actually I think there were almost no Socket 370 440BX boards… nearly all were Slot 1. Especially the dual-processor ones. I’m not aware of any Socket 370 440BX (or 440GX) board that would support even Coppermine Pentium IIIs.
Well I have an ASUS CUBX which is a 440BX Socket 370 board with Coppermine support. Loved it – still do. It was my main system for quite some time and became the central piece of a custom water cooled rig. At one point I put in a Celeron 1100A (Tualatin-256) in there which ran at 1540 (11×140) and that’s how I left it until recently when I dismantled it. I found that I had added dip switches to the board to change the voltage id of the chip to whatever I wanted and that I had modified the BIOS to inject the Tualatin microcode.
After the CUBX, the TUSL2-C was popular but it used the 815EP chipset. So yeah I doubt very much that there was ever any boards made with the 440BX chipset and Tualatin support. My other Tualatin rig used the 815EP chipset.
After that I was into the AMD Athlon processors until the P4 era was over.
Dualatins sound like they would kick any P4’s ass. Good luck.
hey there guys
i got given today an Aopen AX3S-U motherboard, and ive checked on Aopen’s site
for bios updates and one clearly lists -S processor support being added in a bios update dated may 2002. theres a number of variants of this board made by Aopen, all based either on the 815E (like mine) or the 815PE — not sure what the PE version brings to the table.. but i might guess USB 2.0 controller ?
anyways im excited to try out a 1.4ghz “-S” cpu in this board..
ive geekbench2’d two cpus in it so far, the 866Mhz coppermine from march 2000 scored 444, and the 1.3ghz Tualatin Celeron scored 557, i’ve read the 1.4-s goes up to 1100 in geekbench2 score which is slightly higher then my powermac g5 1.8ghz from 2003… and my Northwood pentium 4 2.4ghz from 2002.
Those Pentium III-S processors are good stuff. Combined with reasonably fast memory they pack quite a punch, and without melting the system. No wonder they were super popular in those rack-mount servers in the early 2000s. They were called “poor man’s Xeons” but it was more like smart man’s Xeons, without paying through the nose for slightly improved performance 🙂
If you can get a direct comparison, I’d be interested in the numbers for a Tualatin Celeron vs. a PIII-S running at the same frequency.
http://www.tualatin.de/phpbb/portal.php
here you can find all the things….
Note: Its hard to manage 1Gb of Ram withe Dualatins. Overclocking is possible on this P2B-DS v1.06!
One note about the Intel SDS2 and SAI2 boards: Intel didn’t want to release them.
Intel was “betting the farm” on RAMBUS. The 820 chipset’s “Memory Translator Hub” was a project to appease some of the parts of Intel that didn’t believe in RAMBUS – but no large effort was put in to it, which is why the MTH was released with a major bug.
After the 820 and 840 chipsets had been out for a little while, Intel management finally saw the writing on the wall and abandoned RAMBUS. The problem was, all departments had been told to embrace RAMBUS whole-hog. Intel had multiple 820 and 840-based server and workstation motherboards in the works when the decision was made.
They were cancelled just as final release preparations were being made. A crash program was instituted that came up with the SDS2 and SAI2.
There were many people screaming to have SDRAM-based server systems earlier, but upper management stuck with RAMBUS far too long.
Thanks for the comment. This explains some of the unusual behavior on Intel’s part.
I’ve always been curious as to why Intel made such a decision in the first place?
I think it was to prepare for the P4, the problem was that RDRAM was a poor fit for the P3. Even DDR don’t help P3 much either.
Do anyone know why they did not increase the RAM limit beyond 512MB when they developed the i815 chipset? Even i815P with no IGP had it I think.