3 ms·

"get passed over because companies decide to fund some new(er) shiny thing instead" is not a fair or reasonable assessment. Why would a startup that's racing ag

by jph00 3y ago

"get passed over because companies decide to fund some new(er) shiny thing instead" is not a fair or reasonable assessment. Why would a startup that's racing against time and money decide to create a new language when something already exists that can meet their needs with just a little tweaking? My post embeds my recent keynote at JuliaCon in which I lay out some of the shortcomings of the language, as I see it. MLIR didn't even exist when Julia was created. A language that can really take advantage of MLIR, and is designed from the ground up for modern accelerators, is not something that can be readily created by tweaking Julia AFAICT.

buggy6257 3y ago
> Why would a startup that's racing against time and money decide to create a new language when something already exists that can meet their needs with just a little tweaking? Have you MET programmers? The idea that we/they would do this is not anywhere close to surprising.
- UncleOxidant 3y ago
  Exactly. Lattner himself has been involved in some of this - Swift for Tensorflow comes to mind. I'm old enough to remember when Swift for Tensorflow was going to be the new ML language hotness: https://www.youtube.com/watch?v=drSpCwDFwnM https://www.youtube.com/watch?v=drSpCwDFwnM and https://www.youtube.com/watch?v=s65BigoMV_I https://www.youtube.com/watch?v=s65BigoMV_I
UncleOxidant 3y ago
> MLIR didn't even exist when Julia was created. A language that can really take advantage of MLIR, and is designed from the ground up for modern accelerators, is not something that can be readily created by tweaking Julia AFAICT. MLIR is another level of intermediate representation that sits at a higher level than LLVM-IR. Should the Julia folks decide that MLIR would offer advantages for optimization it wouldn't be difficult for them to add in an MLIR intermediate stage - that's kind of the whole point of an IR. But again, that would take some development effort - if the kind of investment that Mojo was getting were available for Julia it would get done. Nothing about the language itself would need to be impacted, it's the mid and backend where MLIR can offer advantages.

by jph00 3y ago

buggy6257 3y ago

> Why would a startup that's racing against time and money decide to create a new language when something already exists that can meet their needs with just a little tweaking? Have you MET programmers? The idea that we/they would do this is not anywhere close to surprising.

UncleOxidant 3y ago
Exactly. Lattner himself has been involved in some of this - Swift for Tensorflow comes to mind. I'm old enough to remember when Swift for Tensorflow was going to be the new ML language hotness: https://www.youtube.com/watch?v=drSpCwDFwnM https://www.youtube.com/watch?v=drSpCwDFwnM and https://www.youtube.com/watch?v=s65BigoMV_I https://www.youtube.com/watch?v=s65BigoMV_I

UncleOxidant 3y ago

> MLIR didn't even exist when Julia was created. A language that can really take advantage of MLIR, and is designed from the ground up for modern accelerators, is not something that can be readily created by tweaking Julia AFAICT. MLIR is another level of intermediate representation that sits at a higher level than LLVM-IR. Should the Julia folks decide that MLIR would offer advantages for optimization it wouldn't be difficult for them to add in an MLIR intermediate stage - that's kind of the whole point of an IR. But again, that would take some development effort - if the kind of investment that Mojo was getting were available for Julia it would get done. Nothing about the language itself would need to be impacted, it's the mid and backend where MLIR can offer advantages.