Jump to content

Talk:BogoMips

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

If we're going to include this table

[edit]

If we're going to include this table, we should probably also include the info from the bogomips FAQ #3.5 -- i.e., that it can be double this, for some systems. (e.g., an Athlon 900 will be around 1795 bogomips, not 900.) I'd write it myself, but I don't claim to understand exactly what they're saying...

Good suggestion. I don't say that I know BogoMips, but I'm probably just more audacious. So I added the change. But then as the BogoMips mini-Howto maintainer, this isn't entirely unfounded, I guess :-). Wim van Dorst 22:51, 2005 Mar 11 (UTC) PS. Good suggestions should not be unsigned (neither should bad ones).

And perhaps mention a classic Unix pseudo-MIPS measure?

[edit]

On Unix-like systems, back in the days when a one MIP machine was considered fast, a rough and ready MIPS estimate was based on this simple command:

echo "99k2vpq" | /bin/time dc > /dev/null

This told the reverse polish calculator dc to calculate to 99 digits the square root of 2, print it and quit. A one MIP machine would take roughly one second to perform that CPU-bound calculation. As machines grew faster, adding a conversion to octal was added to approximately double the time, making the command:

echo "99k2vp8opq" | /bin/time dc > /dev/null

... which would execute in one second for a 2 MIP machine.

[1] Lukekendall (talk) 10:22, 22 June 2020 (UTC)[reply]

References

Core Duo

[edit]

1594 bogomips for Intel Core Duo T2250, 1.73GHz in HP nx7400 [EY508ES] notebook

That sounds low compared to other "modern" CPUs (given that it stops around the P3 / Xeon) in the list in the main article... unless it's running at low Speedstep. At full whack, that's 0.92 per clock. At low speed (800mhz?), 1.99 per clock. Hmmm. —Preceding unsigned comment added by 82.46.180.56 (talk) 03:21, 10 April 2008 (UTC)[reply]

New kernels

[edit]

With 2.6 series kernels the bogomips ratio scheme has changed and I think the article should be updated. <unsigned>

what is a jiffy switch?

[edit]

I'm an experienced compiler writer, x86 assembly programmer, and electronics designer but I understand little of the example code and documentation of the "How are BogoMIPS computed?" section.

Perhaps the article should explain some of the linux/unix/kernel jargon terms (my guess) before using them!?

83.255.42.141 (talk) 17:09, 11 April 2010 (UTC)[reply]

Location of delay.c

[edit]

"i386 architecture delay_loop is implemented in /usr/src/linux/arch/i386/lib/delay.c" - Was this moved to /usr/src/linux/arch/x86/lib/delay.c? Don't see the i386 version in the 3.4.1 kernel I've been poking around in. — Preceding unsigned comment added by Derekmceachern (talkcontribs) 20:34, 20 November 2012 (UTC)[reply]

Motorola 8081?

[edit]

The only Google hits I found were Wikipedia derived. The red link alone was suspicious, and the addtion in February 2007 with no references by a long-vanished IP is also not reassuring. I've deleted it. It's not on the current list anyway. --Wtshymanski (talk) 16:54, 31 May 2017 (UTC)[reply]

H3 ARMv7 rev 5

[edit]

BogoMIPS value of 48.00 to be added to table on https://en.wikipedia.org/wiki/BogoMips#Proper_BogoMips_ratings

user@pcname:# cat /proc/cpuinfo processor : 0 model name : ARMv7 Processor rev 5 (v7l) BogoMIPS : 48.00 Features : half thumb fastmult vfp edsp thumbee neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae evtstrm CPU implementer : 0x41 CPU architecture: 7 CPU variant : 0x0 CPU part : 0xc07 CPU revision : 5

user@pcname:# lscpu Architecture: armv7l Byte Order: Little Endian CPU(s): 4 On-line CPU(s) list: 0-3 Thread(s) per core: 1 Core(s) per socket: 4 Socket(s): 1 Vendor ID: ARM Model: 5 Model name: Cortex-A7 Stepping: r0p5 BogoMIPS: 48.00 Flags: half thumb fastmult vfp edsp thumbee neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae evtstrm

But why is a citation needed

[edit]

There is a citation marked as needed on the following quote:

With the 2.2.14 Linux kernel, a caching setting of the CPU state was moved from behind to before the BogoMips calculation. Although the BogoMips algorithm itself wasn't changed, from that kernel onward the BogoMips rating for then current Pentium CPUs was twice that of the rating before the change. The changed BogoMips outcome had no effect on real processor performance.

Why is there a citation needed on something that is a simple fact? Hovering over the link reads "performance claim with no citation", but that's not at all what it is. The BogoMips algorithm is a (crude) measurement of processor performance - changing the way it is calculated very clearly does not effect real processor performance. This seems like putting a citation needed on a "claim" that the sum of two twos is four.

81.129.109.84 (talk) 13:40, 10 May 2020 (UTC)[reply]

The table of ratings of CPUs needs a serious rethinking

[edit]

While I think that illustrating what a BogoMIPS is really helpful, most of the numbers in the list are irrelevant to today's reader. Besides, the "meaning" of BogoMIPS varies wildly across different processors and architectures. Nanotik (talk) 20:14, 22 February 2024 (UTC)[reply]