You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The ARM processors are becoming more and more popular.
However, the Source engine is still not supporting them.
Many Linux distributions support ARM processors, so there are real ARM platforms to run the games on.
As far as I know, ARMv7 processor are required to support VFP, so floating-point calculations shouldn't be very slow. NEON is not required, but we can build two sets of binaries for it enabled and disabled.
I have already ported the 2007 mathlib ssemath.h (almost the same as the 2013 version) to NEON intrinsics (with heavy benchmarking - vector ceil function, for example, is 6 times faster than its C equivalent, which is also used on x86, and vector sqrt is over 2 times faster) and some other things like tier0 SetupFPUControlWord and number of cores detection, so I can help you with the port.
The text was updated successfully, but these errors were encountered:
The ARM processors are becoming more and more popular.
However, the Source engine is still not supporting them.
Many Linux distributions support ARM processors, so there are real ARM platforms to run the games on.
As far as I know, ARMv7 processor are required to support VFP, so floating-point calculations shouldn't be very slow. NEON is not required, but we can build two sets of binaries for it enabled and disabled.
I have already ported the 2007 mathlib ssemath.h (almost the same as the 2013 version) to NEON intrinsics (with heavy benchmarking - vector ceil function, for example, is 6 times faster than its C equivalent, which is also used on x86, and vector sqrt is over 2 times faster) and some other things like tier0 SetupFPUControlWord and number of cores detection, so I can help you with the port.
The text was updated successfully, but these errors were encountered: