|
|
|
|
|
|
by rurban
2092 days ago
|
|
|
What are you taking about? Warp is only a bit faster in some cases, and about 10% slower in typical benchmark code.
The speed improvements don't come from optimizations, but from skipping many costly optimizations. The "value" of these skipped optimizations is that some type optimizations are just too costly for them, using also a lot of memory. It's a typical jit trade-off. |
|
|