Is there are way, short of hacking the kernel, to force it to use
p5_mmx? If the machine is a file server then this might be useful.
Andy
>From [EMAIL PROTECTED] Wed Apr 18 22:55 PDT 2001
>Subject: Re: raid5 checksumming speed
>From: Gregory Leblanc <[EMAIL PROTECTED]>
>To: [EMAIL PROTECTED]
>Mime-Version: 1.0
>X-Mailing-List: [EMAIL PROTECTED]
>
>On 18 Apr 2001 18:31:02 -0700, Andy Arvai wrote:
>>
>> I was looking at my system log file at the output from the raid drivers
>> and I noticed the following:
>>
>> raid0 personality registered
>> raid1 personality registered
>> raid5 personality registered
>> raid5: measuring checksumming speed
>> 8regs : 1342.400 MB/sec
>> 32regs : 906.400 MB/sec
>> pIII_sse : 1990.800 MB/sec
>> pII_mmx : 2214.000 MB/sec
>> p5_mmx : 2354.000 MB/sec
>> raid5: using function: pIII_sse (1990.800 MB/sec)
>> md driver 0.90.0 MAX_MD_DEVS=256, MD_SB_DISKS=27
>> md.c: sizeof(mdp_super_t) = 4096
>>
>> Why does the driver choose pIII_sse vs. the faster p5_mmx?
>
>I'm beginning to think about putting this in the RAID FAQ (which I've
>been lax about lately, sorry). The reason it uses the PIII stuff
>instead of others is that it takes advantage of the SSE instruction sets
>in the CPU, and as such, uses significantly less CPU cycles to get the
>same work done. So while the absolute bandwidth that it can handle may
>be less, it's probably a better choice, as it leaves more of the CPU to
>do real work.
> Greg
>
>-
>To unsubscribe from this list: send the line "unsubscribe linux-raid" in
>the body of a message to [EMAIL PROTECTED]
>
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to [EMAIL PROTECTED]
