I know that personas are effective during the design and development stage. They help take focus away from requirements and deliverables so we can focus on the user's goals. But if you don't know who you should test for, you can't actually test anything, right?
You don't need full-blown personas to test your product. You should have some idea of the characteristics and goals of your target user(s) for your product. This could be a persona, but could be much less well-defined and fleshed-out than that. Once you have defined those characteristics and goals, you use them to screen for participants in your user experience research.
Some user research is better than no user research. I wouldn't let a lack of personas block me from conducting user research. Do the best study that you can with the understanding that you currently have, and use what you learn in that study about both your product and your users to build a better product and a better understanding of who your users are and what they want to accomplish.
Fundamentally, you should have some idea of this before you've built the software/hardware. Whether you've got marketing personas or user personas, there should be some semblance of this information.
This will help you test some of the right things with the right people (users).
But, I'm not quite certain what you are looking for in your question. You can always test something, whether it's relevant depends on the goals and the users. But you can test.
So perhaps the question is: are the results of user testing conducted without personas useful?
I'd say yes, in fact, I'd say they can further inform your personas. Would personas help you better conduct testing? Probably. But this can be an iterative cycle.
If you are tasked with testing (software, hardware) but you have NO idea who will be using it, and aren't able to glean this information from business requirements, functional requirements, or even your own knowledge of the product, then you need more information, for sure. Is it personas? Maybe. Target audience? Yes.
In most cases its best to have personas but it also depends on what your testing and why. If you have a design that you just want to know if it makes sense in general then personas are not an absolute requirement. But if your design specifically addresses tasks meant for a specific persona then I would say they are required.
In my opinion, yes you do. Even if you don't have a working product to gather user feedback, you should perform some market research to determine the needs of potential users, and who those users would be.
That being said, in my experience, you should not be spending a ton of time on user personas. You just need to have the fundamental idea about who the user is, getting into the extreme details is unnecessary.
Like all things - it depends.
Persona are an artefact. They're the output of a period of user research. They're there to help you align the team, communicate research results, etc.
Are they always necessary? What happens if your team is talking to real users every week? Are the persona going to do a better job than regular contact with actual users? Possibly not.
What about the scenario where you're doing user testing as part of the user research that's helping you figure out your persona?
What happens when you're incrementally building and refining your persona during the product development process?
Persona are a particular practice that has certain benefits in certain contexts. They're not an automatic good.
Are they good in your context? I don't know. Think about what you want to do. Will persona help?
The answer is context-dependant. Look at Bloomberg terminal. So the key point is who your users are.
If your system is not shaped for specific users, my recommendation is to make some system exploration, which helps you to form some user model in your mind. Then write some scenarios of common use cases.
Then, perform heuristic evaluation (no users need at all!). Having the some user model in your mind, it's more preferable to invite more specific users for summative user test, while formative test could be fruitful even with average users.
So, what you absolutely need for user test are users, not person.