IDE bus speeds: Are you sure you have yours set correctly?

[guero61]

I always got tired of that kernel message about defaulting to an IDE speed of 33MHz and setting it with idebus=xx. That worked until I just noticed today, and it's a little different now.

Now, if you say idebus=66, you'll get a nasty message from this little print statement:

VP_IDE: User given PCI clock speed impossible (66000), using 33 MHz instead

So now, my nice little idebus setting, which was gaining me "so much" speed, doesn't work anymore??? WTF?

Dissecting the via82xxx.c file, I find that you may now specify this in a different manner (or maybe always have been able to). The "new" manner is "ide0=ata66 ide1=ata133" as a kernel option. Or however you implement that. I don't think I'd recommend using different speeds, though... For those of you interested, this is in lines 454-466 of /usr/src/linux-2.4.20-gentoo-r2/drivers/ide/via82xxx.c It also shows how idebus=66 got broken.


So, do you have your ide clock speeds set right? I know I do now -- "ide0=ata100 ide1=ata100". w00t.

[AlterEgo]

Sorry to start a discussion in a documentation-forum,
but I have some doubts about this.

My boot message, and it's correct for my setup.

dmesg|grep ide
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
ide0: BM-DMA at 0xfc00-0xfc07, BIOS settings: hda:DMA, hdb:pio
ide1: BM-DMA at 0xfc08-0xfc0f, BIOS settings: hdc:pio, hdd:DMA

The chipset is correctly identified:

dmesg|grep IDE
VP_IDE: VIA vt8235 (rev 00) IDE UDMA133 controller on pci00:11.1

And my harddisk /dev/hda reports the correct settings back to me (ata-100=udma-5):

hdparm -i /dev/hda
UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5

From that I gather that setting the IDE bus speed has no effect on high-performance (UDMA) drives at all. By doing nothing the drive functions at UDMA-100 anyway.

Therefore, the 33 Mhz bus speed for PIO devices (like CD players) is just fine, and nothing to worry about, since it will not affect performance in any way.

I'd love to see someone reporting (hdparm) performance gains from changing the IDE-bus speed.

[aardvark]

Of you have a normal PCI motherboard, your bus speed should always be 33 Mhz (ide runs over PCI bus). This is something else than the numbers ATA 66/100/133. That is is the speed in Mb/s that the bus supports as in protocol. A PCI bus of 33 Mhz and 32 bit can theoratically support a data throughput of 33 * 10^6 * 4 bytes / sec = 133 MB/sec, which is enough for all ATA modes available today.
(one byte is 8 bits.)

Normal ATA 133 is therefore at the limit of what a NORMAL Pci bus can suport. Of course there is also 64 bit PCI and/or 66Mhz Pci bus, but you normally won't have those. Serial ATA is TNG and I don't know how that is done

Conclusion: the ide bus (or PCI) speed at 33 Mhz is just fine.
Look for transfer mode instead: hdparm -X??

[guero61]

That's what I said in my original post -- I don't think it gains you anything, it was just an unconfigured "error" message that I always saw.

So now, my nice little idebus setting, which was gaining me "so much" speed, doesn't work anymore??? WTF?

THERE IS NOTHING TO GAIN, just your own pride.
I discovered this about 1:30 am localtime, and you know what happens then...

[aardvark]

That's what I said in my original post -- I don't think it gains you anything, it was just an unconfigured "error" message that I always saw.

So now, my nice little idebus setting, which was gaining me "so much" speed, doesn't work anymore??? WTF?

THERE IS NOTHING TO GAIN, just your own pride.
I discovered this about 1:30 am localtime, and you know what happens then...

Ooops sorry, I didn't read well enough... I thought you were wondering how to set it up..

[shie]

although, hdparm says:

Code: Select all

/dev/hdf:
 Model=MAXTOR 6L040J2, FwRev=AR1.0500, SerialNo=662210249226
 Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs }
 RawCHS=16383/16/63, TrkSize=32256, SectSize=21298, ECCbytes=4
 BuffType=DualPortCache, BuffSize=1818kB, MaxMultSect=16, MultSect=16
 CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=78177792
 IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
 PIO modes:  pio0 pio1 pio2 pio3 pio4
 DMA modes:  mdma0 mdma1 mdma2
 UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5 udma6

i get slow speeds (190/40), and the following from dmesg

Code: Select all

ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
    ide0: BM-DMA at 0xd800-0xd807, BIOS settings: hda:DMA, hdb:DMA
    ide1: BM-DMA at 0xd808-0xd80f, BIOS settings: hdc:pio, hdd:pio
    ide2: BM-DMA at 0x7800-0x7807, BIOS settings: hde:pio, hdf:pio
    ide3: BM-DMA at 0x7808-0x780f, BIOS settings: hdg:pio, hdh:pio
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide2 at 0x9000-0x9007,0x8802 on irq 10
ide3 at 0x8400-0x8407,0x8002 on irq 10
 /dev/ide/host0/bus0/target1/lun0: [PTBL] [524/255/63] p1
 /dev/ide/host2/bus0/target1/lun0: [PTBL] [4866/255/63] p1 p2 < p5 p6 p7 p8 p9 p10 p11 >
ide2: Speed warnings UDMA 3/4/5 is not functional.
ide2: Speed warnings UDMA 3/4/5 is not functional.
ide2: Speed warnings UDMA 3/4/5 is not functional.

seems like the ide2 channel (which includes my ata100 disc), can't work with udma 3/4/5 ???

any ideas?

-----------------
Shie Erlich
Krusader Krew

[xunil]

Here's what the idebus kernel command-line option actually means (taken from the kernel IDE documentation, /usr/src/linux/Documentation/ide.txt):

Code: Select all

 "idebus=xx"            : inform IDE driver of VESA/PCI bus speed in MHz,
                                where "xx" is between 20 and 66 inclusive,
                                used when tuning chipset PIO modes.
                                For PCI bus, 25 is correct for a P75 system,
                                30 is correct for P90,P120,P180 systems,
                                and 33 is used for P100,P133,P166 systems.
                                If in doubt, use idebus=33 for PCI.
                                As for VLB, it is safest to not specify it.
                                Bigger values are safer than smaller ones.

[shie]

i know what idebus is, i was more worried about the line

ide2: Speed warnings UDMA 3/4/5 is not functional.

[xunil]

Right, I was replying to the thread, not you. However, your problem may have to do w/ your IDE configuration in the adapter's BIOS (whether that be your motherboard's BIOS or the BIOS on the PCI card if its an expansion adapter). Most modern IDE adapters' BIOSes allow the user to configure which "speeds" the channel can use. Your BIOS may have it configured to do PIO modes only. OTOH, if you're getting 190 for hdparm -T and 40 for hdparm -t, that's actually quite good for a single IDE hard drive. In fact, *no* consumer-grade hard drive (IDE or SCSI) can actually do more than 50-60 for hdparm -t. The ATA66/100/133 stuff is meant for doing RAID, not boosting the performance of individual drives.

[taskara]

so assuming I have a mainboard with 66 mhz pci bus (like dual amd machines) and that the ide controller is supported by the linux kernel, I make sure idebus=66 and I should have a peak theoretical bandwidth of 266 MB/sec for my HDDs.

Impossible to achieve sure, but what if I had a 66 mhz 3Ware serial ata hardware raid controller configured with 4X WD Raptop drives in raid 0, I may be able to surpase 133mb/sec ?

hmmm I doubt they will be that fast, but still that's a good experiment I reakon..

[Marctraider]

How can it be that on my Satellite Pro 480CDT idebus=66 DOES work?
And, it affects the HDparm results.
For the buffers only though...

It's a 4GB hd running at Pio mode 4,
The speed is 6mb/s sais hdparm, and for the other onces is was first 33mb/s
now it's 60 - 66mb/s (thanks to idebus=66)

Anyways my laptop died, could this happened because of the idebus=66?


(Edit, it didnt work at all, i just configured my bios wrong or something, i believe this option even killed my laptop)

[xunil]

so assuming I have a mainboard with 66 mhz pci bus (like dual amd machines) and that the ide controller is supported by the linux kernel, I make sure idebus=66 and I should have a peak theoretical bandwidth of 266 MB/sec for my HDDs.

Impossible to achieve sure, but what if I had a 66 mhz 3Ware serial ata hardware raid controller configured with 4X WD Raptop drives in raid 0, I may be able to surpase 133mb/sec ?

hmmm I doubt they will be that fast, but still that's a good experiment I reakon..

idebus only affects PIO modes which are CPU-intensive. ATA modes will use the highest PCI frequency available, so there's no benefit to setting idebus unless you're using PIO modes, and there's no benefit to using PIO modes if ATA modes are available.

[taskara]

we are talking bus speed aren't we? not transfer modes.

I dunno.. too tired..

[xunil]

idebus was added for IDE drivers that could not automatically detect the PCI or VLB bus speed, such as the generic IDE driver which often cannot, and some dumb drivers that don't try to detect the bus speed and just assume it to be one value if not otherwise specified. PIO transfer modes won't work w/o the proper bus speed, so idebus doesn't make your IDE devices any faster -- it just makes them work.

[Marctraider]

Of you have a normal PCI motherboard, your bus speed should always be 33 Mhz (ide runs over PCI bus). This is something else than the numbers ATA 66/100/133. That is is the speed in Mb/s that the bus supports as in protocol. A PCI bus of 33 Mhz and 32 bit can theoratically support a data throughput of 33 * 10^6 * 4 bytes / sec = 133 MB/sec, which is enough for all ATA modes available today.
(one byte is 8 bits.)

Normal ATA 133 is therefore at the limit of what a NORMAL Pci bus can suport. Of course there is also 64 bit PCI and/or 66Mhz Pci bus, but you normally won't have those. Serial ATA is TNG and I don't know how that is done

Conclusion: the ide bus (or PCI) speed at 33 Mhz is just fine.
Look for transfer mode instead: hdparm -X??

That the IDE bus runs over PCI is not entirely true, my IDE bus runs over ISA, but still i have a PCI bus in my laptop, which my cardbus and video and stuff uses :)

for everyone else, i can set idebus=66 without problems, but it doesnt affect performance at all (first i thought it did, but i was wrong)

Even worse: I think my laptop was destroyed because of this, hardware doesnt necessarily wants to like this option.
O well, replaced the mainboard, and it runs fine again :)