Visually, I am not aware of an obvious difference. The Paxville motherboards have a different part number when ordering -- JG454 as opposed to HG593, if ordering through Spare Parts.
It's good you got back in touch; there are three other items that you would want. The VRM for the second cpu is different as well, with a part number of KG229. The heatsink is bigger, with a part number of KD513. Finally, the fan and shroud assembly is different, with a part number of PD690.
So to sum up, if you were ordering this through Spare Parts, you would want JG454, KG229, and two each of KD513 and PD690. Otherwise, you would want Paxville-capable components; avoid the explicitly-labeled Irwindale and Nocona Xeon motherboards as they won't support the Dual-Core.
I read this thread with interest, and I just want to add some information. I got two 670, one of them with possibilities to run dual core, and the other one not.
You don´t need to look at the product code or number, you just check out the processor information in the bios setup. If it is possible to run dual core, it has an option under the processor info that give the information if the processor is multiple core or not. That line is not present in the bios setup on the other 670.
Hi! I´ve been checking around for more info on this subject, and on Intel´s site Processorfinder( http://processorfinder.intel.com/List.aspx?ParentRadio=All&ProcFam=528&SearchKey= )they state that the SL8MA is single core, and the only dual core 2.8 ghz is the SL8UB. I also have been locking at the SL8UD, which is 3.0 ghz and very similar to SL8UB, the only thing that differs is the 4 mb cache and the multiplier(15 instead of 14). Maube that´ll work as well?
I'm also intrested in an answer if the SL8UD works in the Precision 670 with the lastest MB.
I have also found that it is the SL8UB that is dual core according to Intel, which is confusing, since that is has been suggested that the SL8MA, which is just a single core, and that the passmark tests has been giving results with a 33% increase in performance against, in reality, a faster CPU configuration.
Is this the human factor or can the placebo-effect work on computers as well? Did the 670 is question also believe that it was equipped with dual xeon dual core and performed accordingly? Has AI already come such a long way?
SL8UD is is just a 3GHz version of the SL8MA. Power is at 165W vs. 135W. Both are dual core. The single core Xeons are 110W. I'd worry about sufficient cooling on the SL8UD. I just upgraded my PE 2800 to the SL8MA, had to swap the mobo from V2 to V4 and get the upgraded heatsinks for it. I installed Server 2008 64bit and here are my PassMark CPUMarks. 2x 3GHz single Core Xeons = 1008, 2x 2.8GHz SL8MA = 1841. The HD config is 9x 15K HDs in a RAID 5. Overall PassMark, 627 ->808. In 2010 $, I did the upgrade for ~$200 from eBay resellers. It makes a nice home server :-).
I hope you're still in the forum because I'd like to know if you ever went further. Several of us are looking at upgrading to the Tulsa, Processor Number 7140M aka SL9HA, or 7150N aka SL9YR using the XC837 board. Please let us know how you're faring.