The RNA-dependent RNA polymerase of the Tobamoviruses exists as a heterodimer expressed by read-through of the RdRP portion of the genome (open reading frames 1 and 2). The products of these ORFs are two proteins of 183 kDa (large replicase subunit) and 126 kDa (small replicase subunit), with the 126 kDa being produced approximately 10x more than the 183 kDa form. The functions and sizes of the various proteins are described well in Watanabe et al., (1999):
The genome of tobacco mosaic virus (TMV) consists of a single-stranded RNA molecule of about 6,400 nucleotides in length with positive polarity, which encodes at least four polypeptides: 126- and 183-kDa proteins required for transcription and replication (hereafter referred to as the 126K and 183K proteins, respectively), a 30-kDa (30K) protein for cell-to-cell virus movement in infected plants, and an 18-kDa protein for virus coat formation. The sequence of the 126K protein is encoded by the 5′-proximal region of the viral genome and includes the methyltransferase and RNA helicase motifs, while the 183K protein is a read-through protein of the 126K open reading frame (ORF) and contains, in addition to the above two motifs, the RNA-dependent RNA polymerase motif. The RNA polymerase is considered to be involved in both transcription and replication (8). From sequence analysis, it is believed that the viral RNA polymerase contains the 183K protein as a catalytic subunit, but the precise molecular compositions of transcriptase and replicase have not yet been determined.
A common feature of RdRPs in RNA viruses is that they exist as heteromers. A very well known example of this being the RdRP of influenza virus (-ssRNA), which exists as a heterotrimer consisting of the PB1, PB2 and PA subunits produced from the eponymous genome fragments. However, (many) other examples exist in the virus world, including in another plant virus genus Potyvirus, which, in common with Tobamoviruses and 90% of plant viruses, is a positive sense single-stranded RNA virus. The features of the Potyvirus RdRP have just been published, and to quote the linked article (and references therein):
The RdRp–RdRp self-interaction seems to be a common feature for positive-sense, single-stranded RNA viruses, including insect-, animal- and human-, and plant-infecting viruses [38,49,50,51,52,53]. The dimerization or oligomerization of RdRps may increase the stability of these enzymes and protect against degradation.
As you can see from the information provided in the top quote, the functions of the different subunits of the protein are different but complementary. The 126 kDa contains a helicase and methyltransferase motifs, while the 183 kDa functions as the polymerase as well as containing the same helicase and methyltransferase motifs. Lewandowski and Dawson (2000) found that the 183 kDa subunit was capable of performing all the functions listed above, acting as the full RdRP, but with the 126 kDa subunit those functions were performed ~10 times faster. They also found that mutating a base in the helicase domain of the 126 kDa (183 kDa supplied by a helper virus that only expresses the 183 kDa form) resulted in no replication of the RNA with the mutant 126 kDa indicating that this protein is essential to RNA replication. It should also be noted here as per comments, that the evidence for heterodimerism of the two forms is not in the form of a crystal structure, but in a 1:1 stoichiometry in immunoprecipitation as discussed in the Watanabe paper linked above. To my knowledge no-one has produced a full crystal structure of the RdRP of any Tobamovirus.
However, as you might have noticed when you looked at the information on the link you provided, that there seemed to be only about 4 ORFs, and a small number of proteins produced, and you might be thinking something along the lines of
"how does a virus manage to work when it only produces so few
The answer to which is viral proteins perform many functions (this feature isn't specific to viral proteins). In particular, the small subunit (126 kDa) seems to act as a suppressor of the host silencing RNA system (siRNA). siRNA systems in plants function a bit like an immune system response - they signal within the cell and extra-cellularly to induce a viral RNA degrading response, so that the virus can not easily spread or establish itself in the host. I would speculate that what better place to suppress siRNA functions than at the site of viral RNA replication itself.