LuaJIT is superior to anything else I've been able to find, it blows me away. Though there can always be something better!
That being said, the "better" things I'm thinking about could be added to LuaJIT just as well as they could be to other scripting languages. For example, the LuaJIT compiler could try its best to auto-vectorize (really really hard) or could leverage the developer with SIMD-like vector operations that would adapt to whatever SSE/NEON/...-like instructions the target platform has. GCC/Clang already dod this to some extent with their vector extensions. Although for many instructions you still need to use specific intrinsics which are not cross-architecture compatible of course. The ones that are:
- add/subtract
- mul/divide
- shuffle
Which are already pretty important. One could argue over implementing some "generic" intrinsics and possibly emulating them on platforms that don't have it. But then one could be left scratching one's head when suddenly everything gets slow.... So I can understand the gcc/clang devs.
I and a colleague have been looking a little bit at using Nimrod [1] as a game language.
Nimrod is pretty amazing — it's expressive and flexible (eg., type inference lets you omit types in most places, to the point that much of your code ends up looking like Python), and it's just amazingly fast, but not at the expense of ease of use.
What makes Nimrod particularly suited to game development is that it's designed to integrate well with C. Nimrod compiles to C, and the generated C code provides the necessary bindings so that you can easily call Nimrod functions from C without an intermediate translation layer.
In almost every way, Nimrod is the "speed of C/C++, ease of Ruby" replacement that we have been looking for. Go's performance has been too disappointing, while Rust is a wee bit too complex. Nimrod feels just right.
Very interesting. I'm currently still a bit enamored with Go as it allows me to be very productive and makes deployment, cross-compiling and concurrency such an absolute dream. The performance tradeoffs have been rather good to me I find, it's very speedy for the tasks I try to use it for (almost all of them web-based). But if Nimrod is better... well I can't be sad about that!
The foundation for SIMD vector operations has already been laid by Mike Pall, and his plan is to implement a pretty full API for supporting architectures. See:
I think he's looking for a sponsorship to accelerate this feature[1], so if you know of someone who might be in a position to sponsor the feature, it's a very worthy project!
That being said, the "better" things I'm thinking about could be added to LuaJIT just as well as they could be to other scripting languages. For example, the LuaJIT compiler could try its best to auto-vectorize (really really hard) or could leverage the developer with SIMD-like vector operations that would adapt to whatever SSE/NEON/...-like instructions the target platform has. GCC/Clang already dod this to some extent with their vector extensions. Although for many instructions you still need to use specific intrinsics which are not cross-architecture compatible of course. The ones that are:
- add/subtract - mul/divide - shuffle
Which are already pretty important. One could argue over implementing some "generic" intrinsics and possibly emulating them on platforms that don't have it. But then one could be left scratching one's head when suddenly everything gets slow.... So I can understand the gcc/clang devs.