According to their now deleted Reddit account, it will use a custom Proton build on Intel and Zen 4 CPUs. For every other CPU you will need a kernel module.
Correction: usermode is only supported by zen4 (or newer) and intel Ivy Bridge (or newer).
The reason is pretty simple: only from those generations onwards a new instruction was added (cpuid faulting) which is executed in the hardware itself, old generations simply lacks the silicon for it.
cpu faulting is used to trap usermode cpuid calls and can be used to return spoofed values, without needing an hypervisor.
yeah, more on that: linux kernel has support for cpuid faulting for supported intel cpus since 2017, while amd since only last year. Not sure why amd was slower on this.
I wonder if the new 5800X3D will support it. I’d give it maybe a 1% chance. I doubt they changed all that much. Sure would be nice for people on AM4, though.
I was referring to AMD bringing the 5800X3D out of retirement. They said that they redesigned it. It’s probably just moving the cache under the CPU like on Zen5 3D, but I feel like there’s a very slight chance.
No it cannot be done with proton alone. Proton/wine is only a translation layer: if a program makes cpuid calls, proton/wine cannot intercept them because cpuid is executed directly on hardware silicon.
According to their now deleted Reddit account, it will use a custom Proton build on Intel and Zen 4 CPUs. For every other CPU you will need a kernel module.
Correction: usermode is only supported by zen4 (or newer) and intel Ivy Bridge (or newer).
The reason is pretty simple: only from those generations onwards a new instruction was added (cpuid faulting) which is executed in the hardware itself, old generations simply lacks the silicon for it.
cpu faulting is used to trap usermode cpuid calls and can be used to return spoofed values, without needing an hypervisor.
If ivy bridge then from 3 gen which means pretty much any Intel made from 2012.
But for AMD from 7000 series which means CPU from 2022.
It sucks to be AMD user.
yeah, more on that: linux kernel has support for cpuid faulting for supported intel cpus since 2017, while amd since only last year. Not sure why amd was slower on this.
I wonder if the new 5800X3D will support it. I’d give it maybe a 1% chance. I doubt they changed all that much. Sure would be nice for people on AM4, though.
No, 5800x3d is zen3 sadly.
But you can actually test it with a small python script:
# docs: https://man.archlinux.org/man/arch_prctl.2 import ctypes import errno import mmap import os import signal import subprocess import sys SYS_arch_prctl = 158 ARCH_SET_CPUID = 0x1012 libc = ctypes.CDLL(None, use_errno=True) libc.syscall.restype = ctypes.c_long def arch_set_cpuid(enabled: bool) -> None: rc = libc.syscall(SYS_arch_prctl, ARCH_SET_CPUID, 1 if enabled else 0) if rc != 0: e = ctypes.get_errno() raise OSError(e, os.strerror(e)) def make_cpuid_stub(): # push rbx # xor eax, eax # cpuid # pop rbx # ret code = b"\x53\x31\xc0\x0f\xa2\x5b\xc3" mm = mmap.mmap( -1, len(code), flags=mmap.MAP_PRIVATE | mmap.MAP_ANONYMOUS, prot=mmap.PROT_READ | mmap.PROT_WRITE | mmap.PROT_EXEC, ) mm.write(code) addr = ctypes.addressof(ctypes.c_char.from_buffer(mm)) func = ctypes.CFUNCTYPE(None)(addr) func._mm = mm return func def child(): cpuid = make_cpuid_stub() print("Trying normal CPUID...") cpuid() print("Normal CPUID worked.") try: arch_set_cpuid(False) except OSError as e: if e.errno == errno.ENODEV: print("CPU does not support CPUID faulting.") return 2 raise print("CPUID disabled for this thread. Calling CPUID again...") cpuid() # should SIGSEGV -11 if faulting is supported print("Unexpected: CPUID did not fault.") return 0 if __name__ == "__main__": if len(sys.argv) > 1 and sys.argv[1] == "child": raise SystemExit(child()) p = subprocess.run([sys.executable, __file__, "child"], text=True) print(f"child exit code: {p.returncode}") if p.returncode == -signal.SIGSEGV: print("Result: CPUID faulting is supported and worked.") elif p.returncode == 2: print("Result: CPUID faulting is not supported by the hardware.") else: print("Result: test did not complete cleanly.")I was referring to AMD bringing the 5800X3D out of retirement. They said that they redesigned it. It’s probably just moving the cache under the CPU like on Zen5 3D, but I feel like there’s a very slight chance.
Yeah, Nah. That’s a nope, especially given it can be done in proton. Also, updated Denuvo in 3.2.1…
Denuvo, just say no.
No it cannot be done with proton alone. Proton/wine is only a translation layer: if a program makes cpuid calls, proton/wine cannot intercept them because cpuid is executed directly on hardware silicon.
Kinda sucks for other architectures if only ryzen 7000 is supported
Or if building proton for different architectures is feasible.