Date: Tue, 25 Nov 2014 08:31:29 +0000 From: Mark R V Murray <mark@grondar.org> To: Warner Losh <imp@bsdimp.com> Cc: =?utf-8?Q?Dag-Erling_Sm=C3=B8rgrav?= <des@des.no>, Ian Lepore <ian@FreeBSD.org>, arch@freebsd.org Subject: Re: svn commit: r274739 - head/sys/mips/conf Message-ID: <B5A7728B-87EC-4D5F-9756-1FECFBC88D2E@grondar.org> In-Reply-To: <4DC9E093-C877-40D4-A998-2C94318FECA6@bsdimp.com> References: <201411200552.sAK5qnXP063073@svn.freebsd.org> <20141120084832.GE24601@funkthat.com> <AE8F2D30-7F91-4C90-B79A-D99857D8AED8@grondar.org> <20141121092245.GI99957@funkthat.com> <1416582989.1147.250.camel@revolution.hippie.lan> <026FEB8A-CA8C-472F-A8E4-DA3D0AC44B34@grondar.org> <1416596266.1147.290.camel@revolution.hippie.lan> <F017033A-B761-4435-A7F8-264D2F4662A0@grondar.org> <1416598889.1147.297.camel@revolution.hippie.lan> <86egsvueqk.fsf@nine.des.no> <1416691274.1147.339.camel@revolution.hippie.lan> <398A380D-49AF-480C-8842-8835F81EF641@grondar.org> <1416806894.1147.362.camel@revolution.hippie.lan> <18B8A926-59C0-49B4-ADA3-A11688609852@grondar.org> <1416841268.1147.386.camel@revolution.hippie.lan> <CC6B67E1-55A2-4952-AB43-5F6C787F629B@grondar.org> <86wq6k9okk.fsf@nine.des.no> <4C2BD093-BEA2-47F9-B575-90342712E9B2@bsdimp.com> <B7727BFD-FBEC-4574-8382-7A00AF6D27AF@grondar.org> <4DC9E093-C877-40D4-A998-2C94318FECA6@bsdimp.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> On 25 Nov 2014, at 07:59, Warner Losh <imp@bsdimp.com> wrote: >=20 >=20 >> On Nov 25, 2014, at 12:57 AM, Mark R V Murray <mark@grondar.org> = wrote: >>=20 >>=20 >>> On 24 Nov 2014, at 23:22, Warner Losh <imp@bsdimp.com> wrote: >>>=20 >>> What=E2=80=99s the minimum specs required for get_cyclecount()? Not = all of the >>> boxes that Ian was posting about have high-resolution time-keeping = counters >>> in hardware=E2=80=A6 Maybe there=E2=80=99s some underlying = expectation for this function >>> that these systems either aren=E2=80=99t providing or can=E2=80=99t = provide. >>=20 >> get_cyclecount should be a high-resolution timing counter linear in = time, with a minimum tick no more accurate than a single instruction = execution. In practice we take what we can get. It is used to measure = hopefully chaotic events in order to obtain environmental entropy. >=20 > So right. armv6 and mips have something akin to that, but IIRC, not = all arm v4/5 boxes can do much better than microseconds=E2=80=A6 That=E2=80=99s my understanding. I still don=E2=80=99t get the massive = differences; they should be tiny. M --=20 Mark R V Murray
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?B5A7728B-87EC-4D5F-9756-1FECFBC88D2E>