[gentoo-user] AHCI/IDE-question

[gentoo-user] AHCI/IDE-question

[meino.cramer]

Hi,

I got a little confused about the sense or nonsense of AHCI vs. IDE.

I run a ASUS Crosshair IV Formula, which BIOS has a menu entry to 
configure the SATA ports either for IDE or AHCI or RAID. Forget RAID
for a momen -- I dont use it (nothing against RAID ! ;)

My box uses a linux 2.6.37 vanilla kernel.

The kernel config has been set to

    CONFIG_SATA_AHCI=y
    # CONFIG_SATA_AHCI_PLATFORM is not set

In the dmesg output I found this:

    pci 0000:00:11.0: set SATA to AHCI mode
    ahci 0000:00:11.0: AHCI 0001.0200 32 slots 4 ports 3 Gbps 0xf impl SATA mode
    ahci 0000:07:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3 impl SATA mode

despite the fact that AHCI is disabled in the BIOS settings (using
IDE).

I did an experiment an disabled AHCI in the kernel (to make the kernel
settings consistent with the BIOS.) 

Result: The kernel did not find the root partition.

In the meanwhile I do not understand all this never more.

Why does the kernel boots only, if the BIOS says "IDE!" and linux
insists on "AHCI!"...and waht ist the result?

Best regards,
mcc

Re: [gentoo-user] AHCI/IDE-question

[Volker Armin Hemmann]

On Friday 21 January 2011 19:45:07 meino.cramer@gmx.de wrote:
> Hi,
> 
> I got a little confused about the sense or nonsense of AHCI vs. IDE.
> 
> I run a ASUS Crosshair IV Formula, which BIOS has a menu entry to
> configure the SATA ports either for IDE or AHCI or RAID. Forget RAID
> for a momen -- I dont use it (nothing against RAID ! ;)
> 
> My box uses a linux 2.6.37 vanilla kernel.
> 
> The kernel config has been set to
> 
>     CONFIG_SATA_AHCI=y
>     # CONFIG_SATA_AHCI_PLATFORM is not set
> 
> In the dmesg output I found this:
> 
>     pci 0000:00:11.0: set SATA to AHCI mode
>     ahci 0000:00:11.0: AHCI 0001.0200 32 slots 4 ports 3 Gbps 0xf impl SATA
> mode ahci 0000:07:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3 impl
> SATA mode
> 
> despite the fact that AHCI is disabled in the BIOS settings (using
> IDE).
> 
> I did an experiment an disabled AHCI in the kernel (to make the kernel
> settings consistent with the BIOS.)
> 
> Result: The kernel did not find the root partition.
> 
> In the meanwhile I do not understand all this never more.
> 
> Why does the kernel boots only, if the BIOS says "IDE!" and linux
> insists on "AHCI!"...and waht ist the result?
> 
> Best regards,
> mcc

so, why are you doing soemthing incredible stupid in the first place?

Re: [gentoo-user] AHCI/IDE-question

[kashani]

On 1/21/2011 10:53 AM, Volker Armin Hemmann wrote:
>
> so, why are you doing soemthing incredible stupid in the first place?
>

How about you go have some coffee, maybe have a banana to even out the 
blood sugar, take a walk around the block, and try this email again 
without being a complete ass?

kashani

Re: [gentoo-user] AHCI/IDE-question

[Volker Armin Hemmann]

On Friday 21 January 2011 11:12:34 kashani wrote:
> On 1/21/2011 10:53 AM, Volker Armin Hemmann wrote:
> > so, why are you doing soemthing incredible stupid in the first place?
> 
> How about you go have some coffee, maybe have a banana to even out the
> blood sugar, take a walk around the block, and try this email again
> without being a complete ass?
> 
> kashani

I am sorry that over the years I lost my patience with none-existing problems.

Re: [gentoo-user] AHCI/IDE-question

[Mark Knecht]

On Fri, Jan 21, 2011 at 11:27 AM, Volker Armin Hemmann
<volkerarmin@googlemail.com> wrote:
> On Friday 21 January 2011 11:12:34 kashani wrote:
>> On 1/21/2011 10:53 AM, Volker Armin Hemmann wrote:
>> > so, why are you doing soemthing incredible stupid in the first place?
>>
>> How about you go have some coffee, maybe have a banana to even out the
>> blood sugar, take a walk around the block, and try this email again
>> without being a complete ass?
>>
>> kashani
>
> I am sorry that over the years I lost my patience with none-existing problems.
>
>

So why rip the guy a new one? You have years of experience? We don't
know meino's  experience level. He was clearly just doing experiments
and trying to learn something. You pop up and try to push him down
does him no good.

Mom said "If you don't have anything nice to say then say nothing at all".

- Mark

Re: [gentoo-user] AHCI/IDE-question

[Volker Armin Hemmann]

On Friday 21 January 2011 11:35:06 Mark Knecht wrote:
> On Fri, Jan 21, 2011 at 11:27 AM, Volker Armin Hemmann
> 
> <volkerarmin@googlemail.com> wrote:
> > On Friday 21 January 2011 11:12:34 kashani wrote:
> >> On 1/21/2011 10:53 AM, Volker Armin Hemmann wrote:
> >> > so, why are you doing soemthing incredible stupid in the first
> >> > place?
> >> 
> >> How about you go have some coffee, maybe have a banana to even out the
> >> blood sugar, take a walk around the block, and try this email again
> >> without being a complete ass?
> >> 
> >> kashani
> > 
> > I am sorry that over the years I lost my patience with none-existing
> > problems.
> 
> So why rip the guy a new one? You have years of experience? We don't
> know meino's  experience level. He was clearly just doing experiments
> and trying to learn something. You pop up and try to push him down
> does him no good.
> 
> Mom said "If you don't have anything nice to say then say nothing at all".
> 
> - Mark

but: If you always ask and never research for yourself you will never learn.

Re: [gentoo-user] AHCI/IDE-question

[Dale]

Volker Armin Hemmann wrote:
> On Friday 21 January 2011 11:35:06 Mark Knecht wrote:
>    
>> On Fri, Jan 21, 2011 at 11:27 AM, Volker Armin Hemmann
>>
>> <volkerarmin@googlemail.com>  wrote:
>>      
>>> On Friday 21 January 2011 11:12:34 kashani wrote:
>>>        
>>>> On 1/21/2011 10:53 AM, Volker Armin Hemmann wrote:
>>>>          
>>>>> so, why are you doing soemthing incredible stupid in the first
>>>>> place?
>>>>>            
>>>> How about you go have some coffee, maybe have a banana to even out the
>>>> blood sugar, take a walk around the block, and try this email again
>>>> without being a complete ass?
>>>>
>>>> kashani
>>>>          
>>> I am sorry that over the years I lost my patience with none-existing
>>> problems.
>>>        
>> So why rip the guy a new one? You have years of experience? We don't
>> know meino's  experience level. He was clearly just doing experiments
>> and trying to learn something. You pop up and try to push him down
>> does him no good.
>>
>> Mom said "If you don't have anything nice to say then say nothing at all".
>>
>> - Mark
>>      
> but: If you always ask and never research for yourself you will never learn.
>
>    

Research doesn't always make things better.  I don't even want to count 
the number of times I have researched something, found info that doesn't 
make sense, then ask here and find out it is not anything like that 
anymore.  All the info I found was outdated.

If I can research something and find info that is recent, then that may 
help, otherwise, I ask here and get some up to date info from some of 
the smartest folks there is.  Prime example, if I have a problem with my 
DVD drive, I know there is a person here that will answer my questions 
and to put it simply, if he doesn't know the answer, we all got 
problems.  Hi Jörg.

On the other hand, we have other people on here that are good with 
servers, networks and various other things.  We also have people that 
explain things in a simple way usually with long posts.  Me and Duncan 
come to mind here but there are others too.  Hi Duncan.

So, Research can be a good thing and a person can learn a lot but it 
doesn't always answer your questions or answer them the way a persons 
needs them to be answered.  I ran into this recently with my router 
setup.  I read, even followed a howto, didn't work and didn't learn 
anything either.  Folks on here explained it and now I understand it and 
it works VERY well for me.  I don't think any amount of research would 
have ever got me to where I am in weeks but the folks on here had me 
running in a day or so plus I understand networks better.  I had that 
light bulb moment.  ;-)

Dale

:-)  :-)

Re: [gentoo-user] AHCI/IDE-question

[kashani]

On 1/21/2011 11:27 AM, Volker Armin Hemmann wrote:
> On Friday 21 January 2011 11:12:34 kashani wrote:
>> On 1/21/2011 10:53 AM, Volker Armin Hemmann wrote:
>>> so, why are you doing soemthing incredible stupid in the first place?
>>
>> How about you go have some coffee, maybe have a banana to even out the
>> blood sugar, take a walk around the block, and try this email again
>> without being a complete ass?
>>
>> kashani
>
> I am sorry that over the years I lost my patience with none-existing problems.
>

	Don't be sorry, just stop doing it. This mailing list isn't anyone's 
job and if you're not enjoying the people and the questions anymore it 
might be time for a break. As with all volunteer work sometimes you 
*will* need to take a break. Hell I've gone months without responding to 
a single thread and most of the time I only read 20% of the posts. Those 
are usually the threads that interest me (ask more Postifx, Mysql, 
Apache, etc, server questions!) and I don't really have time for much 
more than that.

	Also with angry one liners you yourself are missing a chance to learn 
something. Maybe the answer to "what's the purpose of your setup? It 
sounds fairly strange to me." would have been interesting. We might have 
found out about x kernel bug or weird hardware y. Or it may have been a 
half baked idea based on some lame blog that we'd all know was false. At 
worse I've just tossed "use AHCI and trying to set IDE with modern 
hardware might have issues" into the back of my brain.

kashani

Re: [gentoo-user] AHCI/IDE-question

[Mark Knecht]

On Fri, Jan 21, 2011 at 10:45 AM,  <meino.cramer@gmx.de> wrote:
>
> Hi,
>
> I got a little confused about the sense or nonsense of AHCI vs. IDE.
>
> I run a ASUS Crosshair IV Formula, which BIOS has a menu entry to
> configure the SATA ports either for IDE or AHCI or RAID. Forget RAID
> for a momen -- I dont use it (nothing against RAID ! ;)
>
> My box uses a linux 2.6.37 vanilla kernel.
>
> The kernel config has been set to
>
>    CONFIG_SATA_AHCI=y
>    # CONFIG_SATA_AHCI_PLATFORM is not set
>
> In the dmesg output I found this:
>
>    pci 0000:00:11.0: set SATA to AHCI mode
>    ahci 0000:00:11.0: AHCI 0001.0200 32 slots 4 ports 3 Gbps 0xf impl SATA mode
>    ahci 0000:07:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3 impl SATA mode
>
> despite the fact that AHCI is disabled in the BIOS settings (using
> IDE).
>
> I did an experiment an disabled AHCI in the kernel (to make the kernel
> settings consistent with the BIOS.)
>
> Result: The kernel did not find the root partition.
>
> In the meanwhile I do not understand all this never more.
>
> Why does the kernel boots only, if the BIOS says "IDE!" and linux
> insists on "AHCI!"...and waht ist the result?
>
> Best regards,
> mcc

Hi meino,
   It's disappointing that Volker insists on sending these pissy
little responses which don't advance the conversation. Sorry for that.

   Not sure I can lend any weight to the argument but it's my belief
that your installation of Gentoo Linux isn't using BIOS to access the
disk at all. Once the system boots and loads the kernel, then the
kernel loads drivers (or uses what you built into the kernel) and
takes over control of the hardware using the AHCI drivers. If the
kernel doesn't use BIOS disk calls (INT13?) then it doesn't care what
the BIOS thinks because the BIOS is not longer involved. It just talks
directly to the hardware.

   I'm happy to be corrected (by Volker I'm sure) but that's my guess
as to what you're seeing.

   Good luck!

Cheers,
Mark

Re: [gentoo-user] AHCI/IDE-question

[Volker Armin Hemmann]

On Friday 21 January 2011 11:08:39 Mark Knecht wrote:
> On Fri, Jan 21, 2011 at 10:45 AM,  <meino.cramer@gmx.de> wrote:
> > Hi,
> > 
> > I got a little confused about the sense or nonsense of AHCI vs. IDE.
> > 
> > I run a ASUS Crosshair IV Formula, which BIOS has a menu entry to
> > configure the SATA ports either for IDE or AHCI or RAID. Forget RAID
> > for a momen -- I dont use it (nothing against RAID ! ;)
> > 
> > My box uses a linux 2.6.37 vanilla kernel.
> > 
> > The kernel config has been set to
> > 
> >    CONFIG_SATA_AHCI=y
> >    # CONFIG_SATA_AHCI_PLATFORM is not set
> > 
> > In the dmesg output I found this:
> > 
> >    pci 0000:00:11.0: set SATA to AHCI mode
> >    ahci 0000:00:11.0: AHCI 0001.0200 32 slots 4 ports 3 Gbps 0xf impl
> > SATA mode ahci 0000:07:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3
> > impl SATA mode
> > 
> > despite the fact that AHCI is disabled in the BIOS settings (using
> > IDE).
> > 
> > I did an experiment an disabled AHCI in the kernel (to make the kernel
> > settings consistent with the BIOS.)
> > 
> > Result: The kernel did not find the root partition.
> > 
> > In the meanwhile I do not understand all this never more.
> > 
> > Why does the kernel boots only, if the BIOS says "IDE!" and linux
> > insists on "AHCI!"...and waht ist the result?
> > 
> > Best regards,
> > mcc
> 
> Hi meino,
>    It's disappointing that Volker insists on sending these pissy
> little responses which don't advance the conversation. Sorry for that.
> 
>    Not sure I can lend any weight to the argument but it's my belief
> that your installation of Gentoo Linux isn't using BIOS to access the
> disk at all. Once the system boots and loads the kernel, then the
> kernel loads drivers (or uses what you built into the kernel) and
> takes over control of the hardware using the AHCI drivers. If the
> kernel doesn't use BIOS disk calls (INT13?) then it doesn't care what
> the BIOS thinks because the BIOS is not longer involved. It just talks
> directly to the hardware.
> 
>    I'm happy to be corrected (by Volker I'm sure) but that's my guess
> as to what you're seeing.

you are confusing bios calls and bios programming chips as.... also - is there 
any good reason to use IDE mode? Any? At all?

Re: [gentoo-user] AHCI/IDE-question

[Mark Knecht]

On Fri, Jan 21, 2011 at 11:16 AM, Volker Armin Hemmann
<volkerarmin@googlemail.com> wrote:
<SNIP>
>>
>>    I'm happy to be corrected (by Volker I'm sure) but that's my guess
>> as to what you're seeing.
>
> you are confusing bios calls and bios programming chips as.... also - is there
> any good reason to use IDE mode? Any? At all?

I don't believe I'm 'confusing bios calls with bios programming'. The
BIOS can do whatever it wants to in programming the chips as long as
grub can still find the kernel. After grub finds the kernel the kernel
is free to override whatever chip programming the BIOS has done and
reprogram the chips as it sees best.

I think the issue meino possibly has is that he likely didn't include
an Int13 type driver in the kernel or most likely his system would
have booted like it did in the _very_ old days.

I agree that there isn't any good reason I know of to use IDE mode
unless the other modes the BIOS provides don't work.

I cannot get into my Asus BIOS at the moment, but as I remember it
Asus gave me something like

IDE
AHCI
AHCI + compatibility

IIRC I had to use the last one to get mine to boot but I may be wrong
about that. I only mention this as meino is also using Asus so he
might look for similar options.

- Mark

Re: [gentoo-user] AHCI/IDE-question

[meino.cramer]

Mark Knecht <markknecht@gmail.com> [11-01-21 20:36]:
> On Fri, Jan 21, 2011 at 11:16 AM, Volker Armin Hemmann
> <volkerarmin@googlemail.com> wrote:
> <SNIP>
> >>
> >>    I'm happy to be corrected (by Volker I'm sure) but that's my guess
> >> as to what you're seeing.
> >
> > you are confusing bios calls and bios programming chips as.... also - is there
> > any good reason to use IDE mode? Any? At all?
> 
> I don't believe I'm 'confusing bios calls with bios programming'. The
> BIOS can do whatever it wants to in programming the chips as long as
> grub can still find the kernel. After grub finds the kernel the kernel
> is free to override whatever chip programming the BIOS has done and
> reprogram the chips as it sees best.
> 
> I think the issue meino possibly has is that he likely didn't include
> an Int13 type driver in the kernel or most likely his system would
> have booted like it did in the _very_ old days.
> 
> I agree that there isn't any good reason I know of to use IDE mode
> unless the other modes the BIOS provides don't work.
> 
> I cannot get into my Asus BIOS at the moment, but as I remember it
> Asus gave me something like
> 
> IDE
> AHCI
> AHCI + compatibility
> 
> IIRC I had to use the last one to get mine to boot but I may be wrong
> about that. I only mention this as meino is also using Asus so he
> might look for similar options.
> 
> - Mark
> 

Hi Mark,

 ...I got some timing problems here, it seems: The answers are comeing
 faster than the related questions are posted.
 Are we disturbing the Einstein/Rosen-continuum here and should better stop
 mailing with lightspeed???

 ;)

 My ASUS board offers:
 RAID
 IDE
 AHCI

 The help to both kernel options mentioned above is saying (beside
 other things): "If unsure, say N".

 That would lead to a unbootable system (at least with my setup...).

 One point for clarification:
 Grub has no problem with either settings in the BIOS. Even the kernel
 boots til the point when it wants to access the root partition.

 I will try to reboot the system with kernel using AHCI _and_ the BIOS
 set to AHCI...I will post the result of the stupid experiment in a
 moment...wait... 

Re: [gentoo-user] AHCI/IDE-question

[Mike Edenfield]

On 1/21/2011 2:48 PM, meino.cramer@gmx.de wrote:

>  My ASUS board offers:
>  RAID
>  IDE
>  AHCI
> 
>  The help to both kernel options mentioned above is saying (beside
>  other things): "If unsure, say N".

Which kernel options are you specifically looking it?

There isn't a single option that I see that says "AHCI", or "IDE".
There are two separate AHCI options and a few dozen SFF ("IDE") options.

In your case, there is no good reason to use the old IDE interface
instead of the newer AHCI SATA interface.  So, go with that.

>  That would lead to a unbootable system (at least with my setup...).

Just for a future reference... the fact that you know this would mean
you are no longer "unsure" and should feel free to say "Y" to whichever
option works. :)

--Mike

Re: [gentoo-user] AHCI/IDE-question

[meino.cramer]

Mike Edenfield <kutulu@kutulu.org> [11-01-21 21:28]:
> On 1/21/2011 2:48 PM, meino.cramer@gmx.de wrote:
> 
> >  My ASUS board offers:
> >  RAID
> >  IDE
> >  AHCI
> > 
> >  The help to both kernel options mentioned above is saying (beside
> >  other things): "If unsure, say N".
> 
> Which kernel options are you specifically looking it?
> 
> There isn't a single option that I see that says "AHCI", or "IDE".
> There are two separate AHCI options and a few dozen SFF ("IDE") options.
> 
> In your case, there is no good reason to use the old IDE interface
> instead of the newer AHCI SATA interface.  So, go with that.
> 
> >  That would lead to a unbootable system (at least with my setup...).
> 
> Just for a future reference... the fact that you know this would mean
> you are no longer "unsure" and should feel free to say "Y" to whichever
> option works. :)
> 
> --Mike
> 

hi Mike,

I am talking about this two options:

CONFIG_SATA_AHCI=y
# CONFIG_SATA_AHCI_PLATFORM is not set

Best regards,
mcc

Re: [gentoo-user] AHCI/IDE-question

[Mark Knecht]

On Fri, Jan 21, 2011 at 12:30 PM,  <meino.cramer@gmx.de> wrote:
<SNIP>
>
> hi Mike,
>
> I am talking about this two options:
>
> CONFIG_SATA_AHCI=y
> # CONFIG_SATA_AHCI_PLATFORM is not set
>
> Best regards,
> mcc
>
>
>

From Niko a few months ago:

http://www.gossamer-threads.com/lists/gentoo/user/217204

Cheers,
Mark

Re: [gentoo-user] AHCI/IDE-question

[Mike Edenfield]

On 1/21/2011 3:30 PM, meino.cramer@gmx.de wrote:
> Mike Edenfield <kutulu@kutulu.org> [11-01-21 21:28]:
>> On 1/21/2011 2:48 PM, meino.cramer@gmx.de wrote:
>>
>>>  My ASUS board offers:
>>>  RAID
>>>  IDE
>>>  AHCI
>>>
>>>  The help to both kernel options mentioned above is saying (beside
>>>  other things): "If unsure, say N".
>>
>> Which kernel options are you specifically looking it?
>>
>> There isn't a single option that I see that says "AHCI", or "IDE".
>> There are two separate AHCI options and a few dozen SFF ("IDE") options.

> I am talking about this two options:
> 
> CONFIG_SATA_AHCI=y
> # CONFIG_SATA_AHCI_PLATFORM is not set

Yes, those are just two different version of the AHCI driver.  And you
don't need either of them to boot, since AHCI is only one option for
accessing IDE drives.  On my systems that predate AHCI/SATA, for
example, I just use the legacy interface:

CONFIG_ATA_SFF=y
CONFIG_ATA_BMDMA=y
CONFIG_SATA_NV=y

But if AHCI is available, then obviously I'm going to use that instead.

--Mike

Re: [gentoo-user] AHCI/IDE-question

[meino.cramer]

Mark Knecht <markknecht@gmail.com> [11-01-21 20:36]:
> On Fri, Jan 21, 2011 at 11:16 AM, Volker Armin Hemmann
> <volkerarmin@googlemail.com> wrote:
> <SNIP>
> >>
> >>    I'm happy to be corrected (by Volker I'm sure) but that's my guess
> >> as to what you're seeing.
> >
> > you are confusing bios calls and bios programming chips as.... also - is there
> > any good reason to use IDE mode? Any? At all?
> 
> I don't believe I'm 'confusing bios calls with bios programming'. The
> BIOS can do whatever it wants to in programming the chips as long as
> grub can still find the kernel. After grub finds the kernel the kernel
> is free to override whatever chip programming the BIOS has done and
> reprogram the chips as it sees best.
> 
> I think the issue meino possibly has is that he likely didn't include
> an Int13 type driver in the kernel or most likely his system would
> have booted like it did in the _very_ old days.
> 
> I agree that there isn't any good reason I know of to use IDE mode
> unless the other modes the BIOS provides don't work.
> 
> I cannot get into my Asus BIOS at the moment, but as I remember it
> Asus gave me something like
> 
> IDE
> AHCI
> AHCI + compatibility
> 
> IIRC I had to use the last one to get mine to boot but I may be wrong
> about that. I only mention this as meino is also using Asus so he
> might look for similar options.
> 
> - Mark
> 

Hi,

I switched the BIOS from IDE (kernel is using AHCI) to AHCI (kernel
uses AHCI). The dmesg says (I did a dmesg | grep -i ahci now, previous
check was done with dmesg | grep AHCI only):

    solfire:/root>dmesg | grep -i ahci
    ahci 0000:00:11.0: version 3.0
*0* ahci 0000:00:11.0: PCI INT A -> GSI 19 (level, low) -> IRQ 19
*1* ahci 0000:00:11.0: irq 78 for MSI/MSI-X
*2* ahci 0000:00:11.0: AHCI 0001.0200 32 slots 6 ports 3 Gbps 0x3f impl SATA mode
*3* ahci 0000:00:11.0: flags: 64bit ncq sntf ilck pm led clo pmp pio slum part
    scsi0 : ahci
    scsi1 : ahci
    scsi2 : ahci
    scsi3 : ahci
    scsi4 : ahci
    scsi5 : ahci
    ahci 0000:07:00.0: PCI INT A -> GSI 44 (level, low) -> IRQ 44
*4* ahci 0000:07:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3 impl SATA mode
*5* ahci 0000:07:00.0: flags: 64bit ncq pm led clo pmp pio slum part
*6* ahci 0000:07:00.0: setting latency timer to 64
    scsi6 : ahci
    scsi7 : ahci

For me bare eye this looks like the kernel ha switched all seven ports
to AHCI. Lines marked with "*n*" are still a riddle to me. May be
Volker will give us some enlightment?
Why is line *1* of the first block missing in the second block,
Volker? Why is line *2* talking about "0x3f" while line  *4* is using
"0x3", Volker? Why differ line *5* from line *3*, Volker? What does
all these flags mean?

I find this interesting:

http://www.linuxquestions.org/questions/linux-hardware-18/6-tips-for-improving-hard-drive-performance-835034/


Best regards,
mcc

Re: [gentoo-user] AHCI/IDE-question

[Mike Edenfield]

On 1/21/2011 3:05 PM, meino.cramer@gmx.de wrote:

>     solfire:/root>dmesg | grep -i ahci
>     ahci 0000:00:11.0: version 3.0
> *0* ahci 0000:00:11.0: PCI INT A -> GSI 19 (level, low) -> IRQ 19
> *1* ahci 0000:00:11.0: irq 78 for MSI/MSI-X
> *2* ahci 0000:00:11.0: AHCI 0001.0200 32 slots 6 ports 3 Gbps 0x3f impl SATA mode
> *3* ahci 0000:00:11.0: flags: 64bit ncq sntf ilck pm led clo pmp pio slum part

>     ahci 0000:07:00.0: PCI INT A -> GSI 44 (level, low) -> IRQ 44
> *4* ahci 0000:07:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3 impl SATA mode
> *5* ahci 0000:07:00.0: flags: 64bit ncq pm led clo pmp pio slum part
> *6* ahci 0000:07:00.0: setting latency timer to 64

> For me bare eye this looks like the kernel ha switched all seven ports
> to AHCI. 

To say that the kernel "switched" the ports is probably misleading.  The
kernel is just trying to hook up the devices it finds to the device
drivers it has available.

You told the kernel it was ok to use the AHCI driver.  The kernel
located those ports, detected that they supported the AHCI interface,
and thus, attached the AHCI SATA driver to them.  If you had told your
BIOS that those ports should be operated as IDE instead of AHCI, then
the kernel wouldn't have found any supported AHCI hardware and you
probably wouldn't be able to boot.

> Why is line *1* of the first block missing in the second block,
> Volker? Why is line *2* talking about "0x3f" while line  *4* is using
> "0x3", Volker? Why differ line *5* from line *3*, Volker? What does
> all these flags mean?

You could also dig into the internals of the libahci.c driver to figure
out what all of those display items mean.  In this case, your seeing two
different PCI busses with slightly different capabilities; just off the
top of my head, one is probably a 6-port PCI-X bus and the other a
2-port PCI bus, but I'd have to go look up the specs to your motherboard
to really find out.

--Mike

Re: [gentoo-user] AHCI/IDE-question

[meino.cramer]

Hi Mike,

thank you for your explanations! :)

But....


Mike Edenfield <kutulu@kutulu.org> [11-01-21 21:48]:
> On 1/21/2011 3:05 PM, meino.cramer@gmx.de wrote:
> 
> >     solfire:/root>dmesg | grep -i ahci
> >     ahci 0000:00:11.0: version 3.0
> > *0* ahci 0000:00:11.0: PCI INT A -> GSI 19 (level, low) -> IRQ 19
> > *1* ahci 0000:00:11.0: irq 78 for MSI/MSI-X
> > *2* ahci 0000:00:11.0: AHCI 0001.0200 32 slots 6 ports 3 Gbps 0x3f impl SATA mode
> > *3* ahci 0000:00:11.0: flags: 64bit ncq sntf ilck pm led clo pmp pio slum part
> 
> >     ahci 0000:07:00.0: PCI INT A -> GSI 44 (level, low) -> IRQ 44
> > *4* ahci 0000:07:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3 impl SATA mode
> > *5* ahci 0000:07:00.0: flags: 64bit ncq pm led clo pmp pio slum part
> > *6* ahci 0000:07:00.0: setting latency timer to 64
> 
> > For me bare eye this looks like the kernel ha switched all seven ports
> > to AHCI. 
> 
> To say that the kernel "switched" the ports is probably misleading.  The
> kernel is just trying to hook up the devices it finds to the device
> drivers it has available.
> 
> You told the kernel it was ok to use the AHCI driver.  The kernel
> located those ports, detected that they supported the AHCI interface,
> and thus, attached the AHCI SATA driver to them.  If you had told your
> BIOS that those ports should be operated as IDE instead of AHCI, then
> the kernel wouldn't have found any supported AHCI hardware and you
> probably wouldn't be able to boot.

.....before the BIOS was set to IDE the kernel used AHCI.
After switching the BIOS to AHCI the kernel acts identical to 
what it did before.

So it seems, that the BIOS has no influence. Mark mentioned this
before.

Only switching the kernel to IDE (only for experimenting) leads
to a non-bootable system.


> > Why is line *1* of the first block missing in the second block,
> > Volker? Why is line *2* talking about "0x3f" while line  *4* is using
> > "0x3", Volker? Why differ line *5* from line *3*, Volker? What does
> > all these flags mean?
> 
> You could also dig into the internals of the libahci.c driver to figure
> out what all of those display items mean.  In this case, your seeing two
> different PCI busses with slightly different capabilities; just off the
> top of my head, one is probably a 6-port PCI-X bus and the other a
> 2-port PCI bus, but I'd have to go look up the specs to your motherboard
> to really find out.

Ok, I save this for tommorrow. Here it is now 21r:59 in the evening
and I stood up 3:40 this night/morning today.
I think, there is enough AHCI and IDE left for tommorow... :)


> --Mike
>

Thanks to you all who have helped ! 

mcc

Re: [gentoo-user] AHCI/IDE-question

[Mark Knecht]

On Fri, Jan 21, 2011 at 12:40 PM, Mike Edenfield <kutulu@kutulu.org> wrote:
<SNIP>
>
> You could also dig into the internals of the libahci.c driver to figure
> out what all of those display items mean.  In this case, your seeing two
> different PCI busses with slightly different capabilities; just off the
> top of my head, one is probably a 6-port PCI-X bus and the other a
> 2-port PCI bus, but I'd have to go look up the specs to your motherboard
> to really find out.
>
> --Mike
>
>

Maybe

lspci -t

Would show something?

- Mark

firefly ~ # lspci -t
-[0000:00]-+-00.0
           +-02.0
           +-16.0
           +-16.2
           +-16.3
           +-19.0
           +-1a.0
           +-1b.0
           +-1c.0-[01]--
           +-1c.4-[02]--
           +-1d.0
           +-1e.0-[03]--+-00.0
           |            +-01.0
           |            \-02.0
           +-1f.0
           +-1f.2
           \-1f.3
firefly ~ #

Re: [gentoo-user] AHCI/IDE-question

[meino.cramer]

Mark Knecht <markknecht@gmail.com> [11-01-22 03:04]:
> On Fri, Jan 21, 2011 at 12:40 PM, Mike Edenfield <kutulu@kutulu.org> wrote:
> <SNIP>
> >
> > You could also dig into the internals of the libahci.c driver to figure
> > out what all of those display items mean.  In this case, your seeing two
> > different PCI busses with slightly different capabilities; just off the
> > top of my head, one is probably a 6-port PCI-X bus and the other a
> > 2-port PCI bus, but I'd have to go look up the specs to your motherboard
> > to really find out.
> >
> > --Mike
> >
> >
> 
> Maybe
> 
> lspci -t
> 
> Would show something?
> 
> - Mark
> 
> firefly ~ # lspci -t
> -[0000:00]-+-00.0
>            +-02.0
>            +-16.0
>            +-16.2
>            +-16.3
>            +-19.0
>            +-1a.0
>            +-1b.0
>            +-1c.0-[01]--
>            +-1c.4-[02]--
>            +-1d.0
>            +-1e.0-[03]--+-00.0
>            |            +-01.0
>            |            \-02.0
>            +-1f.0
>            +-1f.2
>            \-1f.3
> firefly ~ #
> 

Hi Mark,

I got this (lspci -tv):

-[0000:00]-+-00.0  ATI Technologies Inc RD890 Northbridge only single slot PCI-e GFX Hydra part
           +-00.2  ATI Technologies Inc Device 5a23
           +-02.0-[08]--+-00.0  nVidia Corporation Device 0de1
           |            \-00.1  nVidia Corporation Device 0bea
           +-04.0-[07]--+-00.0  JMicron Technology Corp. JMB362/JMB363 Serial ATA Controller
           |            \-00.1  JMicron Technology Corp. JMB362/JMB363 Serial ATA Controller
           +-05.0-[06]----00.0  VIA Technologies, Inc. VT6315 Series Firewire Controller
           +-06.0-[05]----00.0  Marvell Technology Group Ltd. 88E8059 PCI-E Gigabit Ethernet Controller
           +-07.0-[04]----00.0  NEC Corporation uPD720200 USB 3.0 Host Controller
           +-0d.0-[02-03]----00.0-[03]--+-00.0  Oxford Semiconductor Ltd OX16PCI954 (Quad 16950 UART) function 0 (Uart)
           |                            \-00.1  Oxford Semiconductor Ltd OX16PCI954 (Quad 16950 UART) function 1 (8bit bus)
           +-11.0  ATI Technologies Inc SB700/SB800 SATA Controller [AHCI mode]
           +-12.0  ATI Technologies Inc SB700/SB800 USB OHCI0 Controller
           +-12.2  ATI Technologies Inc SB700/SB800 USB EHCI Controller
           +-13.0  ATI Technologies Inc SB700/SB800 USB OHCI0 Controller
           +-13.2  ATI Technologies Inc SB700/SB800 USB EHCI Controller
           +-14.0  ATI Technologies Inc SBx00 SMBus Controller
           +-14.2  ATI Technologies Inc SBx00 Azalia (Intel HDA)
           +-14.3  ATI Technologies Inc SB700/SB800 LPC host controller
           +-14.4-[01]--+-06.0  Brooktree Corporation Bt878 Video Capture
           |            \-06.1  Brooktree Corporation Bt878 Audio Capture
           +-14.5  ATI Technologies Inc SB700/SB800 USB OHCI2 Controller
           +-16.0  ATI Technologies Inc SB700/SB800 USB OHCI0 Controller
           +-16.2  ATI Technologies Inc SB700/SB800 USB EHCI Controller
           +-18.0  Advanced Micro Devices [AMD] Family 10h Processor HyperTransport Configuration
           +-18.1  Advanced Micro Devices [AMD] Family 10h Processor Address Map
           +-18.2  Advanced Micro Devices [AMD] Family 10h Processor DRAM Controller
           +-18.3  Advanced Micro Devices [AMD] Family 10h Processor Miscellaneous Control
           \-18.4  Advanced Micro Devices [AMD] Family 10h Processor Link Control

Re: [gentoo-user] AHCI/IDE-question

[Mark Knecht]

On Fri, Jan 21, 2011 at 6:11 PM,  <meino.cramer@gmx.de> wrote:
<SNIP>
>>
>
> Hi Mark,
>
> I got this (lspci -tv):
>
> -[0000:00]-+-00.0  ATI Technologies Inc RD890 Northbridge only single slot PCI-e GFX Hydra part
>           +-00.2  ATI Technologies Inc Device 5a23
>           +-02.0-[08]--+-00.0  nVidia Corporation Device 0de1
>           |            \-00.1  nVidia Corporation Device 0bea
>           +-04.0-[07]--+-00.0  JMicron Technology Corp. JMB362/JMB363 Serial ATA Controller
>           |            \-00.1  JMicron Technology Corp. JMB362/JMB363 Serial ATA Controller
>           +-05.0-[06]----00.0  VIA Technologies, Inc. VT6315 Series Firewire Controller
>           +-06.0-[05]----00.0  Marvell Technology Group Ltd. 88E8059 PCI-E Gigabit Ethernet Controller
>           +-07.0-[04]----00.0  NEC Corporation uPD720200 USB 3.0 Host Controller
>           +-0d.0-[02-03]----00.0-[03]--+-00.0  Oxford Semiconductor Ltd OX16PCI954 (Quad 16950 UART) function 0 (Uart)
>           |                            \-00.1  Oxford Semiconductor Ltd OX16PCI954 (Quad 16950 UART) function 1 (8bit bus)
>           +-11.0  ATI Technologies Inc SB700/SB800 SATA Controller [AHCI mode]
>           +-12.0  ATI Technologies Inc SB700/SB800 USB OHCI0 Controller
>           +-12.2  ATI Technologies Inc SB700/SB800 USB EHCI Controller
>           +-13.0  ATI Technologies Inc SB700/SB800 USB OHCI0 Controller
>           +-13.2  ATI Technologies Inc SB700/SB800 USB EHCI Controller
>           +-14.0  ATI Technologies Inc SBx00 SMBus Controller
>           +-14.2  ATI Technologies Inc SBx00 Azalia (Intel HDA)
>           +-14.3  ATI Technologies Inc SB700/SB800 LPC host controller
>           +-14.4-[01]--+-06.0  Brooktree Corporation Bt878 Video Capture
>           |            \-06.1  Brooktree Corporation Bt878 Audio Capture
>           +-14.5  ATI Technologies Inc SB700/SB800 USB OHCI2 Controller
>           +-16.0  ATI Technologies Inc SB700/SB800 USB OHCI0 Controller
>           +-16.2  ATI Technologies Inc SB700/SB800 USB EHCI Controller
>           +-18.0  Advanced Micro Devices [AMD] Family 10h Processor HyperTransport Configuration
>           +-18.1  Advanced Micro Devices [AMD] Family 10h Processor Address Map
>           +-18.2  Advanced Micro Devices [AMD] Family 10h Processor DRAM Controller
>           +-18.3  Advanced Micro Devices [AMD] Family 10h Processor Miscellaneous Control
>           \-18.4  Advanced Micro Devices [AMD] Family 10h Processor Link Control
>
>
>
>

Interesting. This is (apparently) the Asus/AMD version of the
motherboard I'm using which is Asus/Intel. (Both ROG) It's almost
feature for feature identical

http://www.newegg.com/Product/Product.aspx?Item=N82E16813131352&cm_re=Rampage_II_extreme-_-13-131-352-_-Product

I'm using the Intel Core i7-980x Extreme in mine, so it's 6 cores/12
threads. Here's mine again:

c2stable ~ # lspci -tv
-[0000:00]-+-00.0  Intel Corporation X58 I/O Hub to ESI Port
           +-01.0-[01]--
           +-03.0-[02]--
           +-07.0-[03]--+-00.0  ATI Technologies Inc Device 68b8
           |            \-00.1  ATI Technologies Inc Device aa58
           +-14.0  Intel Corporation 5520/5500/X58 I/O Hub System
Management Registers
           +-14.1  Intel Corporation 5520/5500/X58 I/O Hub GPIO and
Scratch Pad Registers
           +-14.2  Intel Corporation 5520/5500/X58 I/O Hub Control
Status and RAS Registers
           +-14.3  Intel Corporation 5520/5500/X58 I/O Hub Throttle Registers
           +-1a.0  Intel Corporation 82801JI (ICH10 Family) USB UHCI
Controller #4
           +-1a.1  Intel Corporation 82801JI (ICH10 Family) USB UHCI
Controller #5
           +-1a.2  Intel Corporation 82801JI (ICH10 Family) USB UHCI
Controller #6
           +-1a.7  Intel Corporation 82801JI (ICH10 Family) USB2 EHCI
Controller #2
           +-1b.0  Intel Corporation 82801JI (ICH10 Family) HD Audio Controller
           +-1c.0-[07]--
           +-1c.2-[06]----00.0  Marvell Technology Group Ltd. 88E8056
PCI-E Gigabit Ethernet Controller
           +-1c.4-[05]--+-00.0  JMicron Technology Corp. 20360/20363
Serial ATA Controller
           |            \-00.1  JMicron Technology Corp. 20360/20363
Serial ATA Controller
           +-1c.5-[04]----00.0  Marvell Technology Group Ltd. 88E8056
PCI-E Gigabit Ethernet Controller
           +-1d.0  Intel Corporation 82801JI (ICH10 Family) USB UHCI
Controller #1
           +-1d.1  Intel Corporation 82801JI (ICH10 Family) USB UHCI
Controller #2
           +-1d.2  Intel Corporation 82801JI (ICH10 Family) USB UHCI
Controller #3
           +-1d.7  Intel Corporation 82801JI (ICH10 Family) USB2 EHCI
Controller #1
           +-1e.0-[08]----02.0  VIA Technologies, Inc. VT6306 Fire II
IEEE 1394 OHCI Link Layer Controller
           +-1f.0  Intel Corporation 82801JIR (ICH10R) LPC Interface Controller
           +-1f.2  Intel Corporation 82801JI (ICH10 Family) 4 port
SATA IDE Controller
           +-1f.3  Intel Corporation 82801JI (ICH10 Family) SMBus Controller
           \-1f.5  Intel Corporation 82801JI (ICH10 Family) 2 port
SATA IDE Controller
c2stable ~ #

So you actually have two SATA controllers (just like mine)

>           +-04.0-[07]--+-00.0  JMicron Technology Corp. JMB362/JMB363 Serial ATA Controller
>           |            \-00.1  JMicron Technology Corp. JMB362/JMB363 Serial ATA Controller

>           +-11.0  ATI Technologies Inc SB700/SB800 SATA Controller [AHCI mode]

Depending on your need for eSATA (I use it for external backups) you
might look at using JMicron controller for your CD if it will find it.

As I said earlier, I have my BIOS set to 'Enhanced' and then 'IDE'.
Works fine for me and I get quite good speed on my mdadm-based RAID6.
I'm sure it's doing SATA no matter what the MB manual implies.

- Mark

Re: [gentoo-user] AHCI/IDE-question

[Volker Armin Hemmann]

On Friday 21 January 2011 21:05:30 meino.cramer@gmx.de wrote:
> Mark Knecht <markknecht@gmail.com> [11-01-21 20:36]:
> > On Fri, Jan 21, 2011 at 11:16 AM, Volker Armin Hemmann
> > <volkerarmin@googlemail.com> wrote:
> > <SNIP>
> > 
> > >>    I'm happy to be corrected (by Volker I'm sure) but that's my
> > >> guess
> > >> as to what you're seeing.
> > > 
> > > you are confusing bios calls and bios programming chips as.... also
> > > - is there any good reason to use IDE mode? Any? At all?
> > 
> > I don't believe I'm 'confusing bios calls with bios programming'. The
> > BIOS can do whatever it wants to in programming the chips as long as
> > grub can still find the kernel. After grub finds the kernel the kernel
> > is free to override whatever chip programming the BIOS has done and
> > reprogram the chips as it sees best.
> > 
> > I think the issue meino possibly has is that he likely didn't include
> > an Int13 type driver in the kernel or most likely his system would
> > have booted like it did in the _very_ old days.
> > 
> > I agree that there isn't any good reason I know of to use IDE mode
> > unless the other modes the BIOS provides don't work.
> > 
> > I cannot get into my Asus BIOS at the moment, but as I remember it
> > Asus gave me something like
> > 
> > IDE
> > AHCI
> > AHCI + compatibility
> > 
> > IIRC I had to use the last one to get mine to boot but I may be wrong
> > about that. I only mention this as meino is also using Asus so he
> > might look for similar options.
> > 
> > - Mark
> 
> Hi,
> 
> I switched the BIOS from IDE (kernel is using AHCI) to AHCI (kernel
> uses AHCI). The dmesg says (I did a dmesg | grep -i ahci now, previous
> check was done with dmesg | grep AHCI only):
> 
>     solfire:/root>dmesg | grep -i ahci
>     ahci 0000:00:11.0: version 3.0
> *0* ahci 0000:00:11.0: PCI INT A -> GSI 19 (level, low) -> IRQ 19
> *1* ahci 0000:00:11.0: irq 78 for MSI/MSI-X
> *2* ahci 0000:00:11.0: AHCI 0001.0200 32 slots 6 ports 3 Gbps 0x3f impl SATA
> mode *3* ahci 0000:00:11.0: flags: 64bit ncq sntf ilck pm led clo pmp pio
> slum part scsi0 : ahci
>     scsi1 : ahci
>     scsi2 : ahci
>     scsi3 : ahci
>     scsi4 : ahci
>     scsi5 : ahci
>     ahci 0000:07:00.0: PCI INT A -> GSI 44 (level, low) -> IRQ 44
> *4* ahci 0000:07:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3 impl SATA
> mode *5* ahci 0000:07:00.0: flags: 64bit ncq pm led clo pmp pio slum part
> *6* ahci 0000:07:00.0: setting latency timer to 64
>     scsi6 : ahci
>     scsi7 : ahci
> 
> For me bare eye this looks like the kernel ha switched all seven ports
> to AHCI. Lines marked with "*n*" are still a riddle to me. May be
> Volker will give us some enlightment?
> Why is line *1* of the first block missing in the second block,
> Volker? Why is line *2* talking about "0x3f" while line  *4* is using
> "0x3", Volker? Why differ line *5* from line *3*, Volker? What does
> all these flags mean?
> 

you know - there are websites for that. Google is your friend. But even a 
glance would reveal to you:
two different chips.
One using MSI for interrupts the second not.

> I find this interesting:
> 
> http://www.linuxquestions.org/questions/linux-hardware-18/6-tips-for-improvi
> ng-hard-drive-performance-835034/

it is a start. But the first link there... just saying.. there is no magically 
correct value for stride or chunk.

Oh and if you are using AFT drives make sure the partitions are set up 
correctly.

Also:
https://ata.wiki.kernel.org/index.php/Main_Page

Re: [gentoo-user] AHCI/IDE-question

[meino.cramer]

Volker Armin Hemmann <volkerarmin@googlemail.com> [11-01-22 03:04]:
> On Friday 21 January 2011 21:05:30 meino.cramer@gmx.de wrote:
> > Mark Knecht <markknecht@gmail.com> [11-01-21 20:36]:
> > > On Fri, Jan 21, 2011 at 11:16 AM, Volker Armin Hemmann
> > > <volkerarmin@googlemail.com> wrote:
> > > <SNIP>
> > > 
> > > >>    I'm happy to be corrected (by Volker I'm sure) but that's my
> > > >> guess
> > > >> as to what you're seeing.
> > > > 
> > > > you are confusing bios calls and bios programming chips as.... also
> > > > - is there any good reason to use IDE mode? Any? At all?
> > > 
> > > I don't believe I'm 'confusing bios calls with bios programming'. The
> > > BIOS can do whatever it wants to in programming the chips as long as
> > > grub can still find the kernel. After grub finds the kernel the kernel
> > > is free to override whatever chip programming the BIOS has done and
> > > reprogram the chips as it sees best.
> > > 
> > > I think the issue meino possibly has is that he likely didn't include
> > > an Int13 type driver in the kernel or most likely his system would
> > > have booted like it did in the _very_ old days.
> > > 
> > > I agree that there isn't any good reason I know of to use IDE mode
> > > unless the other modes the BIOS provides don't work.
> > > 
> > > I cannot get into my Asus BIOS at the moment, but as I remember it
> > > Asus gave me something like
> > > 
> > > IDE
> > > AHCI
> > > AHCI + compatibility
> > > 
> > > IIRC I had to use the last one to get mine to boot but I may be wrong
> > > about that. I only mention this as meino is also using Asus so he
> > > might look for similar options.
> > > 
> > > - Mark
> > 
> > Hi,
> > 
> > I switched the BIOS from IDE (kernel is using AHCI) to AHCI (kernel
> > uses AHCI). The dmesg says (I did a dmesg | grep -i ahci now, previous
> > check was done with dmesg | grep AHCI only):
> > 
> >     solfire:/root>dmesg | grep -i ahci
> >     ahci 0000:00:11.0: version 3.0
> > *0* ahci 0000:00:11.0: PCI INT A -> GSI 19 (level, low) -> IRQ 19
> > *1* ahci 0000:00:11.0: irq 78 for MSI/MSI-X
> > *2* ahci 0000:00:11.0: AHCI 0001.0200 32 slots 6 ports 3 Gbps 0x3f impl SATA
> > mode *3* ahci 0000:00:11.0: flags: 64bit ncq sntf ilck pm led clo pmp pio
> > slum part scsi0 : ahci
> >     scsi1 : ahci
> >     scsi2 : ahci
> >     scsi3 : ahci
> >     scsi4 : ahci
> >     scsi5 : ahci
> >     ahci 0000:07:00.0: PCI INT A -> GSI 44 (level, low) -> IRQ 44
> > *4* ahci 0000:07:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3 impl SATA
> > mode *5* ahci 0000:07:00.0: flags: 64bit ncq pm led clo pmp pio slum part
> > *6* ahci 0000:07:00.0: setting latency timer to 64
> >     scsi6 : ahci
> >     scsi7 : ahci
> > 
> > For me bare eye this looks like the kernel ha switched all seven ports
> > to AHCI. Lines marked with "*n*" are still a riddle to me. May be
> > Volker will give us some enlightment?
> > Why is line *1* of the first block missing in the second block,
> > Volker? Why is line *2* talking about "0x3f" while line  *4* is using
> > "0x3", Volker? Why differ line *5* from line *3*, Volker? What does
> > all these flags mean?
> > 
> 
> you know - there are websites for that. Google is your friend. But even a 
> glance would reveal to you:
> two different chips.
> One using MSI for interrupts the second not.
> 
> > I find this interesting:
> > 
> > http://www.linuxquestions.org/questions/linux-hardware-18/6-tips-for-improvi
> > ng-hard-drive-performance-835034/
> 
> it is a start. But the first link there... just saying.. there is no magically 
> correct value for stride or chunk.
> 
> Oh and if you are using AFT drives make sure the partitions are set up 
> correctly.
> 
> Also:
> https://ata.wiki.kernel.org/index.php/Main_Page
> 

Hi Volker,

I have done the partition alignment thingy when I installed my disk a
year ago (or so). Thanks for the hint anyway... :)

If you know a webite, which explains all that low level stuff like the
flags I mentioned I would be happy, if would be so kind to post the 
link here.

Thanks a lot for your help in advance!
Best regards,
mcc

Re: [gentoo-user] AHCI/IDE-question

[meino.cramer]

Mark Knecht <markknecht@gmail.com> [11-01-21 20:16]:
> On Fri, Jan 21, 2011 at 10:45 AM,  <meino.cramer@gmx.de> wrote:
> >
> > Hi,
> >
> > I got a little confused about the sense or nonsense of AHCI vs. IDE.
> >
> > I run a ASUS Crosshair IV Formula, which BIOS has a menu entry to
> > configure the SATA ports either for IDE or AHCI or RAID. Forget RAID
> > for a momen -- I dont use it (nothing against RAID ! ;)
> >
> > My box uses a linux 2.6.37 vanilla kernel.
> >
> > The kernel config has been set to
> >
> >    CONFIG_SATA_AHCI=y
> >    # CONFIG_SATA_AHCI_PLATFORM is not set
> >
> > In the dmesg output I found this:
> >
> >    pci 0000:00:11.0: set SATA to AHCI mode
> >    ahci 0000:00:11.0: AHCI 0001.0200 32 slots 4 ports 3 Gbps 0xf impl SATA mode
> >    ahci 0000:07:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3 impl SATA mode
> >
> > despite the fact that AHCI is disabled in the BIOS settings (using
> > IDE).
> >
> > I did an experiment an disabled AHCI in the kernel (to make the kernel
> > settings consistent with the BIOS.)
> >
> > Result: The kernel did not find the root partition.
> >
> > In the meanwhile I do not understand all this never more.
> >
> > Why does the kernel boots only, if the BIOS says "IDE!" and linux
> > insists on "AHCI!"...and waht ist the result?
> >
> > Best regards,
> > mcc
> 
> Hi meino,
>    It's disappointing that Volker insists on sending these pissy
> little responses which don't advance the conversation. Sorry for that.
> 
>    Not sure I can lend any weight to the argument but it's my belief
> that your installation of Gentoo Linux isn't using BIOS to access the
> disk at all. Once the system boots and loads the kernel, then the
> kernel loads drivers (or uses what you built into the kernel) and
> takes over control of the hardware using the AHCI drivers. If the
> kernel doesn't use BIOS disk calls (INT13?) then it doesn't care what
> the BIOS thinks because the BIOS is not longer involved. It just talks
> directly to the hardware.
> 
>    I'm happy to be corrected (by Volker I'm sure) but that's my guess
> as to what you're seeing.
> 
>    Good luck!
> 
> Cheers,
> Mark
> 

Hi Mark,

thank you for your kind words. There is no need to feel sorry for
others. The behaviour of those are definelty neither your fault nor
your problem ;)

For me I have learned that it doesn't matter, whether the result
of an experiment is positive or negative as long as one is prepared
to learn from it. If one insisits on doing only so called "non stupid
things" one will miss a lot of results sooner or later... ;) ;)

I thought (which implies "I dont know for sure"), that the BIOS do
enable/disable certain features, the kernels reads that settings and
act accordingly -- but definitely this is not true for all settings.

Does the contents of a harddisk differ when written with AHCI
compared to a disk which is written with IDE?

If NO _AND_ only the kernel sets the AHCI- odr IDE-protocol, then
the harddisk should be readable in either case.

If the BIOS _and_ the kernel settings are defining, how to talk 
to the disk, then it may happen, that there is only the "sound of silence"
between kernel and hardware if before the BIOS set up the SATA-chips
differently to what the kernel wants to talk.

But again, these are only thougts drifting in the dark.

I tried to shed some more light (for getting greater shadows ;) )
by posting my question here... ;) 8)

May be I should do some more stupid things??? ;)

Thanks again for your help and your words, Mark!
Have a nice weekend!
Best regards,
mcc

Re: [gentoo-user] AHCI/IDE-question

[Mark Knecht]

On Fri, Jan 21, 2011 at 11:40 AM,  <meino.cramer@gmx.de> wrote:
<SNIP>
> Hi Mark,
>
<SNIP>
> I thought (which implies "I dont know for sure"), that the BIOS do
> enable/disable certain features, the kernels reads that settings and
> act accordingly -- but definitely this is not true for all settings.
>

Certainly true for some hardware, like clocks, etc.

For disk controllers AFAIK the goal is to give the boot loader a
chance to boot. After that it doesn't, in general, matter what the
BIOS did.

For instance, modern SATA controllers use DMA. BIOS and older
operating systems like DOS didn't know much, if anything, about DMA,
so BIOS leaves that turned off. The kernel turns that on.

> Does the contents of a harddisk differ when written with AHCI
> compared to a disk which is written with IDE?
>

TTBOMK no. Other things like file system type, etc., change what's on
the disk, but the disk store so many bytes/sector and that's just the
way it works.


> If NO _AND_ only the kernel sets the AHCI- odr IDE-protocol, then
> the harddisk should be readable in either case.
>

Certainly, which is why you could build this system using AHCI and
then move it to some other system and read the disk using DOS.
(Assuming DOS could understand the file system like FAT, etc.)

> If the BIOS _and_ the kernel settings are defining, how to talk
> to the disk, then it may happen, that there is only the "sound of silence"
> between kernel and hardware if before the BIOS set up the SATA-chips
> differently to what the kernel wants to talk.
>

BIOS sets up the system hardware so the boot loader can get the kernel
image off the disk. The kernel is read into memory using these
settings. At that point there aren't any more disk reads for a while.
The kernel executes and starts resetting the hardware through driver
loads, etc. This is why one controller could be set to use a SATA
Drive by itself or RAID.

> But again, these are only thougts drifting in the dark.
>
> I tried to shed some more light (for getting greater shadows ;) )
> by posting my question here... ;) 8)
>
> May be I should do some more stupid things??? ;)
>

Ain't no such thing a stupid question. Only thing to do when
experimenting is ensure you aren't risking data you care about. I
would do these experiments on a new clean system. I would not do them
on a system that has stuff I care about unless I had known good
backups.

> Thanks again for your help and your words, Mark!
> Have a nice weekend!
> Best regards,
> mcc
>

You too sir!

Cheers,
Mark

Re: [gentoo-user] AHCI/IDE-question

[meino.cramer]

Mark Knecht <markknecht@gmail.com> [11-01-21 21:04]:
> On Fri, Jan 21, 2011 at 11:40 AM,  <meino.cramer@gmx.de> wrote:
> <SNIP>
> > Hi Mark,
> >
> <SNIP>
> > I thought (which implies "I dont know for sure"), that the BIOS do
> > enable/disable certain features, the kernels reads that settings and
> > act accordingly -- but definitely this is not true for all settings.
> >
> 
> Certainly true for some hardware, like clocks, etc.
> 
> For disk controllers AFAIK the goal is to give the boot loader a
> chance to boot. After that it doesn't, in general, matter what the
> BIOS did.
> 
> For instance, modern SATA controllers use DMA. BIOS and older
> operating systems like DOS didn't know much, if anything, about DMA,
> so BIOS leaves that turned off. The kernel turns that on.
> 
> > Does the contents of a harddisk differ when written with AHCI
> > compared to a disk which is written with IDE?
> >
> 
> TTBOMK no. Other things like file system type, etc., change what's on
> the disk, but the disk store so many bytes/sector and that's just the
> way it works.
> 
> 
> > If NO _AND_ only the kernel sets the AHCI- odr IDE-protocol, then
> > the harddisk should be readable in either case.
> >
> 
> Certainly, which is why you could build this system using AHCI and
> then move it to some other system and read the disk using DOS.
> (Assuming DOS could understand the file system like FAT, etc.)
> 
> > If the BIOS _and_ the kernel settings are defining, how to talk
> > to the disk, then it may happen, that there is only the "sound of silence"
> > between kernel and hardware if before the BIOS set up the SATA-chips
> > differently to what the kernel wants to talk.
> >
> 
> BIOS sets up the system hardware so the boot loader can get the kernel
> image off the disk. The kernel is read into memory using these
> settings. At that point there aren't any more disk reads for a while.
> The kernel executes and starts resetting the hardware through driver
> loads, etc. This is why one controller could be set to use a SATA
> Drive by itself or RAID.
> 
> > But again, these are only thougts drifting in the dark.
> >
> > I tried to shed some more light (for getting greater shadows ;) )
> > by posting my question here... ;) 8)
> >
> > May be I should do some more stupid things??? ;)
> >
> 
> Ain't no such thing a stupid question. Only thing to do when
> experimenting is ensure you aren't risking data you care about. I
> would do these experiments on a new clean system. I would not do them
> on a system that has stuff I care about unless I had known good
> backups.
> 
> > Thanks again for your help and your words, Mark!
> > Have a nice weekend!
> > Best regards,
> > mcc
> >
> 
> You too sir!
> 
> Cheers,
> Mark
> 

:) Thanks for your explanations, Mark!

I have a complete mirror of the harddisk in question on another
identical harddisk...
Despite "others" may think...I am not /that/ stupid, hahahahahahahaha!
:) 8) X)

Best regards,
mcc

Re: [gentoo-user] AHCI/IDE-question

[meino.cramer]

Mark Knecht <markknecht@gmail.com> [11-01-21 21:04]:
> On Fri, Jan 21, 2011 at 11:40 AM,  <meino.cramer@gmx.de> wrote:
> <SNIP>
> > Hi Mark,
> >
> <SNIP>
> > I thought (which implies "I dont know for sure"), that the BIOS do
> > enable/disable certain features, the kernels reads that settings and
> > act accordingly -- but definitely this is not true for all settings.
> >
> 
> Certainly true for some hardware, like clocks, etc.
> 
> For disk controllers AFAIK the goal is to give the boot loader a
> chance to boot. After that it doesn't, in general, matter what the
> BIOS did.
> 
> For instance, modern SATA controllers use DMA. BIOS and older
> operating systems like DOS didn't know much, if anything, about DMA,
> so BIOS leaves that turned off. The kernel turns that on.
> 
> > Does the contents of a harddisk differ when written with AHCI
> > compared to a disk which is written with IDE?
> >
> 
> TTBOMK no. Other things like file system type, etc., change what's on
> the disk, but the disk store so many bytes/sector and that's just the
> way it works.
> 
> 
> > If NO _AND_ only the kernel sets the AHCI- odr IDE-protocol, then
> > the harddisk should be readable in either case.
> >
> 
> Certainly, which is why you could build this system using AHCI and
> then move it to some other system and read the disk using DOS.
> (Assuming DOS could understand the file system like FAT, etc.)
> 
> > If the BIOS _and_ the kernel settings are defining, how to talk
> > to the disk, then it may happen, that there is only the "sound of silence"
> > between kernel and hardware if before the BIOS set up the SATA-chips
> > differently to what the kernel wants to talk.
> >
> 
> BIOS sets up the system hardware so the boot loader can get the kernel
> image off the disk. The kernel is read into memory using these
> settings. At that point there aren't any more disk reads for a while.
> The kernel executes and starts resetting the hardware through driver
> loads, etc. This is why one controller could be set to use a SATA
> Drive by itself or RAID.
> 
> > But again, these are only thougts drifting in the dark.
> >
> > I tried to shed some more light (for getting greater shadows ;) )
> > by posting my question here... ;) 8)
> >
> > May be I should do some more stupid things??? ;)
> >
> 
> Ain't no such thing a stupid question. Only thing to do when
> experimenting is ensure you aren't risking data you care about. I
> would do these experiments on a new clean system. I would not do them
> on a system that has stuff I care about unless I had known good
> backups.
> 
> > Thanks again for your help and your words, Mark!
> > Have a nice weekend!
> > Best regards,
> > mcc
> >
> 
> You too sir!
> 
> Cheers,
> Mark
> 


Last thing which remains is: Why does the help of the kernel says
to both AHCI-settings: "If unsure, say N"... ?

Re: [gentoo-user] AHCI/IDE-question

[Mark Knecht]

On Fri, Jan 21, 2011 at 12:11 PM,  <meino.cramer@gmx.de> wrote:
<SNIP>
>
>
> Last thing which remains is: Why does the help of the kernel says
> to both AHCI-settings: "If unsure, say N"... ?
>

Because if you're not sure you have SATA then you don't need the driver? :-)

It could just as easily say "Y" or "M" but they mostly seem to opt for
a smaller kernel in those defaults instead of more functionality.

Just my opinion.

Cheers,
Mark

Re: [gentoo-user] AHCI/IDE-question

[Dale]

meino.cramer@gmx.de wrote:
>
> Last thing which remains is: Why does the help of the kernel says
> to both AHCI-settings: "If unsure, say N"... ?
>
>
>    

I just built a rig with a Gigabyte mobo.  Mine has setting like yours.  
I asked on here and was told that AHCI is the "new way" to do things.  
So, I set mine to that and it has worked fine.

The only issue I did have is not being able to boot from a CD/DVD with 
it set to AHCI.  No idea why that matters.  My DVD drive is SATA too.  I 
need to play with that more later on.  See if it was that or something 
else that I missed.  Thought I would mention that just in case you try 
to boot a CD or something and get a nasty error message or something.  
May want to file that in the back of your brain for future reference.

Dale

:-)  :-)

Re: [gentoo-user] AHCI/IDE-question

[meino.cramer]

Dale <rdalek1967@gmail.com> [11-01-21 21:44]:
> meino.cramer@gmx.de wrote:
> >
> >Last thing which remains is: Why does the help of the kernel says
> >to both AHCI-settings: "If unsure, say N"... ?
> >
> >
> >   
> 
> I just built a rig with a Gigabyte mobo.  Mine has setting like yours.  
> I asked on here and was told that AHCI is the "new way" to do things.  
> So, I set mine to that and it has worked fine.
> 
> The only issue I did have is not being able to boot from a CD/DVD with 
> it set to AHCI.  No idea why that matters.  My DVD drive is SATA too.  
> I need to play with that more later on.  See if it was that or 
> something else that I missed.  Thought I would mention that just in 
> case you try to boot a CD or something and get a nasty error message or 
> something.  May want to file that in the back of your brain for future 
> reference.
> 
> Dale
> 
> :-)  :-)
> 

Hi Dale,

*IDEA* 

Mark said, that the kernel alone is defining, whether to talk AHCI or
IDE to the harddisk and there is not a single result here from my
experiments, that makes me believe, that Mark is wrong with that...

Why not to switch back to IDE in the BIOS, which again makes it
possible to boot from USB/DVD since it is used far before the
kernel image takes over.

When the kernel boots, the chips are "brainwashed" and after that they are
"thinking AHCI" instead of IDE....

? may be an option ?

Best regards,
mcc

Re: [gentoo-user] AHCI/IDE-question

[Mark Knecht]

On Fri, Jan 21, 2011 at 12:53 PM,  <meino.cramer@gmx.de> wrote:
<SNIP>
> Hi Dale,
>
> *IDEA*
>
> Mark said, that the kernel alone is defining, whether to talk AHCI or
> IDE to the harddisk and there is not a single result here from my
> experiments, that makes me believe, that Mark is wrong with that...
>

Mark also said 'Don't trust Mark' so be careful about anything Mark
says here! ;-))


> Why not to switch back to IDE in the BIOS, which again makes it
> possible to boot from USB/DVD since it is used far before the
> kernel image takes over.
>

If Mike's assessment is correct then the reason not to do that is that
once the kernel boots it wouldn't know that the hardware is AHCI
capable and you almost certainly would get lower performance.


> When the kernel boots, the chips are "brainwashed" and after that they are
> "thinking AHCI" instead of IDE....
>

I think it's more what Mike suggested. The chips have a control bit in
them that changes the way they work. If the bit is set to IDE then
then chips never tell the kernel that they can do SATA so the kernel
never tries.

I booted into BIOS here. I have an option called "SATA Configuration"
which is set to Enhanced. It offers Disabled and Compatible also.

BIOS then gives me another choice "Configure SATA as" which I have set
to IDE. It offers RAID and AHCI also.

This allows the system to boot CDs and still allows the kernel to run
SATA at full speed.

- Mark

[gentoo-user] Re: AHCI/IDE-question

[walt]

On 01/21/2011 01:46 PM, Mark Knecht wrote:

> I booted into BIOS here. I have an option called "SATA Configuration"
>
> BIOS then gives me another choice "Configure SATA as"

What was that programmer smoking?  Some of the equipment I use at work
offers choices that are that dumb, and the error messages are completely
useless without the 300-page manual :(

Some programmers are alive only because it's (still) illegal to kill them.

Re: [gentoo-user] Re: AHCI/IDE-question

[Mark Knecht]

On Fri, Jan 21, 2011 at 4:59 PM, walt <w41ter@gmail.com> wrote:
> On 01/21/2011 01:46 PM, Mark Knecht wrote:
>
>> I booted into BIOS here. I have an option called "SATA Configuration"
>>
>> BIOS then gives me another choice "Configure SATA as"
>
> What was that programmer smoking?  Some of the equipment I use at work
> offers choices that are that dumb, and the error messages are completely
> useless without the 300-page manual :(
>
> Some programmers are alive only because it's (still) illegal to kill them.
>

Yeah, it was a strange one. IIRC it's an AMI BIOS on that machine.

There's no consistency between vendors and even the same vendors seem
to change what things are called from board to board...

- Mark

Re: [gentoo-user] AHCI/IDE-question

[Alan McKinnon]

Apparently, though unproven, at 22:37 on Friday 21 January 2011, Dale did 
opine thusly:

> meino.cramer@gmx.de wrote:
> > Last thing which remains is: Why does the help of the kernel says
> > to both AHCI-settings: "If unsure, say N"... ?
> 
> I just built a rig with a Gigabyte mobo.  Mine has setting like yours.
> I asked on here and was told that AHCI is the "new way" to do things.
> So, I set mine to that and it has worked fine.
> 
> The only issue I did have is not being able to boot from a CD/DVD with
> it set to AHCI.  No idea why that matters.  My DVD drive is SATA too.  I
> need to play with that more later on.  See if it was that or something
> else that I missed.  Thought I would mention that just in case you try
> to boot a CD or something and get a nasty error message or something.
> May want to file that in the back of your brain for future reference.


My notebook works like that too. 

Hard disk works fine when everything is set to AHCI, but then the system won't 
boot from CD. So I enabled the IDE driver and the IDE driver for CD-ROMs.

My take on this is that Dell had a vast stock of cheap-skate CD-ROM hardware 
and used them up. The engineering logic would have been "it doesn't matter 
that we use the slow interface for that device, it's still faster than we can 
get the data off the media."

-- 
alan dot mckinnon at gmail dot com

Re: [gentoo-user] AHCI/IDE-question

[Dale]

Alan McKinnon wrote:
> My notebook works like that too.
>
> Hard disk works fine when everything is set to AHCI, but then the system won't
> boot from CD. So I enabled the IDE driver and the IDE driver for CD-ROMs.
>
> My take on this is that Dell had a vast stock of cheap-skate CD-ROM hardware
> and used them up. The engineering logic would have been "it doesn't matter
> that we use the slow interface for that device, it's still faster than we can
> get the data off the media."
>
>    

And I thought there was something weird with me on this one.  o_O  I did 
switch it back to AHCI after I got done booting the CD thingy.  I really 
can't tell any difference in speed between the two and neither could 
hdparm -tT either.

root@fireball / # hdparm -Tt /dev/sda

/dev/sda:
  Timing cached reads:   6408 MB in  2.00 seconds = 3205.06 MB/sec
  Timing buffered disk reads:  328 MB in  3.00 seconds = 109.22 MB/sec
root@fireball / #

I get about the same either way.  Could that mean that when the kernel 
boots that it switched over to AHCI regardless of the BIOS setting?  
This is a little info too:

root@fireball / # dmesg | grep -i ahci
[    0.827837] ahci 0000:00:11.0: version 3.0
[    0.827855] ahci 0000:00:11.0: PCI INT A -> GSI 22 (level, low) -> IRQ 22
[    0.828285] ahci 0000:00:11.0: AHCI 0001.0100 32 slots 6 ports 3 Gbps 
0x3f impl SATA mode
[    0.828840] ahci 0000:00:11.0: flags: 64bit ncq sntf ilck pm led clo 
pmp pio slum part ccc
[    0.830342] scsi0 : ahci
[    0.830734] scsi1 : ahci
[    0.831103] scsi2 : ahci
[    0.831474] scsi3 : ahci
[    0.831843] scsi4 : ahci
[    0.832204] scsi5 : ahci
root@fireball / #

Someone may can talk be into rebooting and switching AHCI off and 
testing again.  Big may there.  ;-)

Dale

:-)  :-)

P. S.  Seriously off topic.  I used hugin the other day to stitch 
together about a dozen pics from a 10Mpxl camera.  It was awesome to 
watch all four cores crunch on that thing.  It was fast too.  Going from 
a single 2.5Ghz CPU to a four core 3.2Ghz CPU is a huge difference.

[gentoo-user] Re: AHCI/IDE-question

[Nikos Chantziaras]

On 01/22/2011 12:31 AM, Dale wrote:
> Alan McKinnon wrote:
>> My notebook works like that too.
>>
>> Hard disk works fine when everything is set to AHCI, but then the
>> system won't
>> boot from CD. So I enabled the IDE driver and the IDE driver for CD-ROMs.
>>
>> My take on this is that Dell had a vast stock of cheap-skate CD-ROM
>> hardware
>> and used them up. The engineering logic would have been "it doesn't
>> matter
>> that we use the slow interface for that device, it's still faster than
>> we can
>> get the data off the media."
>>
>
> And I thought there was something weird with me on this one. o_O I did
> switch it back to AHCI after I got done booting the CD thingy. I really
> can't tell any difference in speed between the two and neither could
> hdparm -tT either.

hdparm measures raw throughput when reading continuously from one 
position to another.  AHCI improves performance when the disk needs to 
read from several different places, which is the case in every day use. 
  It does this by providing a feature similar to what SCSI provides: 
native command queuing (NCQ).  You can read about what this is and why 
we want it here:

http://en.wikipedia.org/wiki/Native_Command_Queuing

Re: [gentoo-user] Re: AHCI/IDE-question

[Dale]

Nikos Chantziaras wrote:
> On 01/22/2011 12:31 AM, Dale wrote:
>> Alan McKinnon wrote:
>>> My notebook works like that too.
>>>
>>> Hard disk works fine when everything is set to AHCI, but then the
>>> system won't
>>> boot from CD. So I enabled the IDE driver and the IDE driver for 
>>> CD-ROMs.
>>>
>>> My take on this is that Dell had a vast stock of cheap-skate CD-ROM
>>> hardware
>>> and used them up. The engineering logic would have been "it doesn't
>>> matter
>>> that we use the slow interface for that device, it's still faster than
>>> we can
>>> get the data off the media."
>>>
>>
>> And I thought there was something weird with me on this one. o_O I did
>> switch it back to AHCI after I got done booting the CD thingy. I really
>> can't tell any difference in speed between the two and neither could
>> hdparm -tT either.
>
> hdparm measures raw throughput when reading continuously from one 
> position to another.  AHCI improves performance when the disk needs to 
> read from several different places, which is the case in every day 
> use.  It does this by providing a feature similar to what SCSI 
> provides: native command queuing (NCQ).  You can read about what this 
> is and why we want it here:
>
> http://en.wikipedia.org/wiki/Native_Command_Queuing
>

Other than copying a file and using time to measure how long it takes, 
what is the best test of a hard drive's speed?

Also, does or can the kernel override the BIOS setting?  I think it uses 
AHCI no matter what is in the BIOS.  It seems it would be at least some 
difference in speed.

Nice link.  The picture explained it best.  I'm sort of simple.  lol

Dale

:-)  :-)

[gentoo-user] Re: AHCI/IDE-question

[Nikos Chantziaras]

On 01/22/2011 09:50 AM, Dale wrote:
> Nikos Chantziaras wrote:
>> On 01/22/2011 12:31 AM, Dale wrote:
>>> [...]
>>> And I thought there was something weird with me on this one. o_O I did
>>> switch it back to AHCI after I got done booting the CD thingy. I really
>>> can't tell any difference in speed between the two and neither could
>>> hdparm -tT either.
>>
>> hdparm measures raw throughput when reading continuously from one
>> position to another. AHCI improves performance when the disk needs to
>> read from several different places, which is the case in every day
>> use. It does this by providing a feature similar to what SCSI
>> provides: native command queuing (NCQ). You can read about what this
>> is and why we want it here:
>>
>> http://en.wikipedia.org/wiki/Native_Command_Queuing
>
> Other than copying a file and using time to measure how long it takes,
> what is the best test of a hard drive's speed?

By running a benchmark tool that does exactly this.  IOzone is a nice one:

   http://www.iozone.org

It's in portage: "app-benchmarks/iozone".


> Also, does or can the kernel override the BIOS setting? I think it uses
> AHCI no matter what is in the BIOS. It seems it would be at least some
> difference in speed.

The kernel can't change this setting and has no access to it whatsoever.

Re: [gentoo-user] Re: AHCI/IDE-question

[Dale]

Nikos Chantziaras wrote:
> On 01/22/2011 09:50 AM, Dale wrote:
>> Nikos Chantziaras wrote:
>>> On 01/22/2011 12:31 AM, Dale wrote:
>>>> [...]
>>>> And I thought there was something weird with me on this one. o_O I did
>>>> switch it back to AHCI after I got done booting the CD thingy. I 
>>>> really
>>>> can't tell any difference in speed between the two and neither could
>>>> hdparm -tT either.
>>>
>>> hdparm measures raw throughput when reading continuously from one
>>> position to another. AHCI improves performance when the disk needs to
>>> read from several different places, which is the case in every day
>>> use. It does this by providing a feature similar to what SCSI
>>> provides: native command queuing (NCQ). You can read about what this
>>> is and why we want it here:
>>>
>>> http://en.wikipedia.org/wiki/Native_Command_Queuing
>>
>> Other than copying a file and using time to measure how long it takes,
>> what is the best test of a hard drive's speed?
>
> By running a benchmark tool that does exactly this.  IOzone is a nice 
> one:
>
>   http://www.iozone.org
>
> It's in portage: "app-benchmarks/iozone".

I installed it but trying to figure out how to use it.  Jeez, what a man 
page.  O_O

>
>
>> Also, does or can the kernel override the BIOS setting? I think it uses
>> AHCI no matter what is in the BIOS. It seems it would be at least some
>> difference in speed.
>
> The kernel can't change this setting and has no access to it whatsoever.
>

I didn't think so but thought it worth asking.  Seems to be some 
confusion on this.

Dale

:-)  :-)

Re: [gentoo-user] Re: AHCI/IDE-question

[Mark Knecht]

On Sat, Jan 22, 2011 at 1:15 AM, Dale <rdalek1967@gmail.com> wrote:
<SNIP>
>>> Other than copying a file and using time to measure how long it takes,
>>> what is the best test of a hard drive's speed?
>>
>> By running a benchmark tool that does exactly this.  IOzone is a nice one:
>>
>>  http://www.iozone.org
>>
>> It's in portage: "app-benchmarks/iozone".
>
> I installed it but trying to figure out how to use it.  Jeez, what a man
> page.  O_O
>

Yeah, it's a tough one. Also, very slow, assuming I understand it
correctly. You have to use file sizes larger than the memory of the
system, so on a 24GB system it takes (literally) a day or two to run.
(Assuming I actually understood the man page!) ;-)

- Mark

Re: [gentoo-user] Re: AHCI/IDE-question

[Dale]

Mark Knecht wrote:
> On Sat, Jan 22, 2011 at 1:15 AM, Dale<rdalek1967@gmail.com>  wrote:
> <SNIP>
>    
>>>> Other than copying a file and using time to measure how long it takes,
>>>> what is the best test of a hard drive's speed?
>>>>          
>>> By running a benchmark tool that does exactly this.  IOzone is a nice one:
>>>
>>>   http://www.iozone.org
>>>
>>> It's in portage: "app-benchmarks/iozone".
>>>        
>> I installed it but trying to figure out how to use it.  Jeez, what a man
>> page.  O_O
>>
>>      
> Yeah, it's a tough one. Also, very slow, assuming I understand it
> correctly. You have to use file sizes larger than the memory of the
> system, so on a 24GB system it takes (literally) a day or two to run.
> (Assuming I actually understood the man page!) ;-)
>
> - Mark
>
>
>    

I found this command and it worked pretty well.  It does take a good 
while to run tho.  It wasn't to bad on my new rig but the old rig did 
take a little while.

iozone -R -l 5 -u 5 -r 4k -s 100m -F /home/f1 /home/f2 /home/f3 /home/f4 
/home/f5 | tee -a /tmp/iozone_results.txt &

The results were much different than what hdparm shows.  It shows the 
3Gbs/sec like they advertise they can do.  I wonder which is more 
accurate?  :/

Dale

:-)  :-)

Re: [gentoo-user] Re: AHCI/IDE-question

[Mark Knecht]

On Sun, Jan 23, 2011 at 4:02 AM, Dale <rdalek1967@gmail.com> wrote:
> Mark Knecht wrote:
>>
>> On Sat, Jan 22, 2011 at 1:15 AM, Dale<rdalek1967@gmail.com>  wrote:
>> <SNIP>
>>
>>>>>
>>>>> Other than copying a file and using time to measure how long it takes,
>>>>> what is the best test of a hard drive's speed?
>>>>>
>>>>
>>>> By running a benchmark tool that does exactly this.  IOzone is a nice
>>>> one:
>>>>
>>>>  http://www.iozone.org
>>>>
>>>> It's in portage: "app-benchmarks/iozone".
>>>>
>>>
>>> I installed it but trying to figure out how to use it.  Jeez, what a man
>>> page.  O_O
>>>
>>>
>>
>> Yeah, it's a tough one. Also, very slow, assuming I understand it
>> correctly. You have to use file sizes larger than the memory of the
>> system, so on a 24GB system it takes (literally) a day or two to run.
>> (Assuming I actually understood the man page!) ;-)
>>
>> - Mark
>>
>>
>>
>
> I found this command and it worked pretty well.  It does take a good while
> to run tho.  It wasn't to bad on my new rig but the old rig did take a
> little while.
>
> iozone -R -l 5 -u 5 -r 4k -s 100m -F /home/f1 /home/f2 /home/f3 /home/f4
> /home/f5 | tee -a /tmp/iozone_results.txt &
>
> The results were much different than what hdparm shows.  It shows the
> 3Gbs/sec like they advertise they can do.  I wonder which is more accurate?
>  :/
>
> Dale

Hi Dale,
   Good command, but it's missing one parameter that will likely make
your results more meaningful. Try it again with the -e parameter
added. I'll limit the tests also to just the initial write and read
tests to give you something else to thiink about:

iozone -R -l 5 -u 5 -r 4k -s 100m -i 0 -i 1 -e -F /home/f1 /home/f2
/home/f3 /home/f4 /home/f5 | tee -a /tmp/iozone_results.txt &

   I'll be more than happy to explain later why I suggested the
changes, but in the spirit of someone's earlier comments about people
doing research, give some thought to the results you get and see if
you can explain to yourself why -e makes a difference. Do you believe
the results you get are meaningful?

   If you want another good experiment, remove -e and make the files
much larger, like 2g if you have an 8GB or DRAM. Again, explain to
yourself why this makes a difference. If you do this experiment you
_really_ should just do the -i 0 test only your first time through...
;-) (HINT: You might want to open another terminal, run top and when
in top hit iz and maybe m if you're not showing memory usage...)

Cheers,
Mark

Re: [gentoo-user] AHCI/IDE-question

[Neil Bothwick]

On Fri, 21 Jan 2011 19:45:07 +0100, meino.cramer@gmx.de wrote:

> I did an experiment an disabled AHCI in the kernel (to make the kernel
> settings consistent with the BIOS.) 
> 
> Result: The kernel did not find the root partition.
> 

You disabled the driver for your hard disk controller and you wonder why
the kernel cannot find the hard disk? Did you remember to build in
support for the specific chipset drivers?


-- 
Neil Bothwick

... Never say anything more predictive than "Watch this!"

Re: [gentoo-user] AHCI/IDE-question

[Paul Hartman]

On Fri, Jan 21, 2011 at 12:45 PM,  <meino.cramer@gmx.de> wrote:
>
> Hi,
>
> I got a little confused about the sense or nonsense of AHCI vs. IDE.
>
> I run a ASUS Crosshair IV Formula, which BIOS has a menu entry to
> configure the SATA ports either for IDE or AHCI or RAID. Forget RAID
> for a momen -- I dont use it (nothing against RAID ! ;)

Consider "IDE" to mean "Compatibility mode".

I just RTFM for your manual, it tells that if you want to use NCQ or
hot-plugging you must enable AHCI in BIOS.

I have a Gigabyte motherboard and it's the same way. If I use "IDE"
mode, kernel must use a different driver (it's still not really IDE.
When I use AHCI, it uses the ahci driver in kernel and all features
and full speed are available.

I think it's a bad name, just like "Keyboard: DOS/USB" setting, when
"DOS" really means "PS/2 Emulation from BIOS" or whatever.

So, use AHCI, be happy. ;)