There are an enormous number of partial i-stems in the third declension and many of them have variable case endings depending on the period or whim of the author, so compiling a full list would be a difficult task. Even listing the exceptions would be tricky since often a word may have, for example, an i-stem genitive plural and occasionally not. Another thing to remember is that a 3rd declension adjective (most of which are pure i-stem) used substantively as a noun is an i-stem noun.
a 3rd declension noun has i-stem if:
1) parisyllabic nouns in -is or -es, e.g. ciuis, -is ; nubes, -is. exceptions: canis, panis, iuuenis.
It is worth noting that all masc/fem
pure i-stems fall into this category, besides the four ending in -er:
imber, linter, uter, and
uenter.
2) imparisyllabic nouns, with genitive's -is preceded by more than one consonant, e.g. ars, artis ; nox, noctis.
cliens [
client(i)um] is occasionally an exception.
3) neuter nouns ending in -e, -al, -ar; e.g. mare, animal, exemplar. exceptions: sal.
sal is no exception because it is not a neuter noun.
4) some others:
anas, anatis - duck
dos, dotis - dowry
fraus, fraudis - fraud
fur, furis - thief
glis, gliris - dormouse
lis, litis - lawsuit
mas, maris - male
mus, muris - mouse
nix, niuis - snow
strix, strigis - grove (doesn't seem Classical)
uis, uiris - strength
A lot of these would so rarely be used in the genitive plural (or even in the plural at all) that it is hard to say.
uis should be obvious since it has the ablative singular -i.
Something important to understand about i-stems that is rarely ever mentioned is why they are called "i-stem". It is not because of a set of rules that these nouns utilize special endings (though those rules are useful in practice); i-stem nouns use the same 3rd declension endings as all other 3rd declension nouns, it is just that their stem ends in a vowel rather than a consonant, causing certain anomalies.
For example, a "normal" 3rd declension noun such as
rex, called a consonant stem (or more specifically a mute stem) for the final letter of its stem, reg-. To the stem you add the regular endings -s, -is, -i, -em, -e, -es, -um, -ibus, -es, -ibus.
Typical reasoning says that with an i-stem like
ignis, the stem is ign-. To that stem you add different endings than you would to a non i-stem:
-is, -is,
-i, -em,
-i, -es,
-ium, -ibus, -es (
-is), -ibus. In reality, the stem is igni-, hence its name as an i-stem; when you combine igni- and -um you get the genitive plural
ignium, no different from combining reg- and -um to get
regum. You can add -s to igni- or reg- to get the nominative singular, so once again i-stems do not differ from consonant stems. In the same way, you can think of
maria as being mari + a instead of the less accurate mar + ia.
I believe that thinking of i-stems in this way, as having the same endings as consonant stems, is a much better approach and more appropriate than regarding them as irregular.
There are a few complications though, when the case ending begins in a vowel other than u there is contraction or omission of a letter. igni+is = ignis, igni+i = igni, igni + ibus = ignibus. Admitting to contraction occurring regularly in Latin declension might just tarnish its reputation of being a neat and orderly language however.

flebile nescio quid queritur lyra, flebile lingua murmurat exanimis, respondent flebile ripae