Status of the sata_sil driver for the VT8237

I was wondering, what is the state of the sata_sil driver for the VIA 
VT8237 southbridge (ie the one used commonly on their K8T 
motherboards).  Is this still beta?  Any known problems with it?

-ryan

J. Ryan Earl wrote:

I was wondering, what is the state of the sata_sil driver for the VIA 
VT8237 southbridge (ie the one used commonly on their K8T 
motherboards).  Is this still beta?  Any known problems with it?

Well prior to the most recent version, calling it "beta" was putting it 
kindly.  I would call it broken :) which was why it was marked with 
CONFIG_BROKEN...

As of the most recent 2.6.5-rc2-bk snapshots, libata and sata_sil should 
be pretty happy with each other.

Some of the VIA problems were actually platform-related.  "noapic", 
"nomce", turning -on- local APIC support, and a few other workarounds 
have been found for these.

If you're on x86-64, make sure you have the latest BIOS version, too.

	Jeff

Jeff Garzik wrote:

J. Ryan Earl wrote:

quoted

I was wondering, what is the state of the sata_sil driver for the VIA 
VT8237 southbridge (ie the one used commonly on their K8T 
motherboards).  Is this still beta?  Any known problems with it?

Well prior to the most recent version, calling it "beta" was putting 
it kindly.  I would call it broken :) which was why it was marked with 
CONFIG_BROKEN...

As of the most recent 2.6.5-rc2-bk snapshots, libata and sata_sil 
should be pretty happy with each other.

I meant sata_via, I dunno why I said sata_sil; Freudian typo.  I was 
thinking of the current sii3112 I use, which has been stable and 
blazingly fast for awhile with the IDE driver--and a little little patch 
I keep applying.  I'm looking to upgrade a server with 2 Raptors in it 
to an A64 running 64-bit Linux of course.  I wanna get the HDs off the 
PCI bus and onto the VT8237 southbridge.  I was wondering if the driver 
for -that southbridge- was still considered beta.

-ryan

Hi.

I meant sata_via, I dunno why I said sata_sil; Freudian typo.  I was 
thinking of the current sii3112 I use, which has been stable and 
blazingly fast for awhile with the IDE driver--and a little little patch 
I keep applying.  I'm looking to upgrade a server with 2 Raptors in it 
to an A64 running 64-bit Linux of course.  I wanna get the HDs off the 
PCI bus and onto the VT8237 southbridge.  I was wondering if the driver 
for -that southbridge- was still considered beta.

Many A64 boards today come with a Promise SATA controller as well -
works like a charm for me on an ASUS K8V Deluxe board here.

The driver is sata_promise but I guess you either knew that or
would be able to figure it out yourself :)

// Stefan

Stefan Smietanowski wrote:

Many A64 boards today come with a Promise SATA controller as well -
works like a charm for me on an ASUS K8V Deluxe board here.

That's an option, as is just using the sii3112 that's been working so 
well, but that still puts the disks on the PCI bus.  The whole point is 
to get 300MB/sec of burstable I/O off the 133MB/s bus that's shared with 
the dual gigE controllers and a firewire camera.

-ryan

J. Ryan Earl wrote:

Stefan Smietanowski wrote:

quoted

Many A64 boards today come with a Promise SATA controller as well -
works like a charm for me on an ASUS K8V Deluxe board here.

That's an option, as is just using the sii3112 that's been working so 
well, but that still puts the disks on the PCI bus.  The whole point is 
to get 300MB/sec of burstable I/O off the 133MB/s bus that's shared with 
the dual gigE controllers and a firewire camera.

Do you really plan to get 300MB/sec from two disks?  :)

	Jeff

Jeff Garzik wrote:

Do you really plan to get 300MB/sec from two disks?  :)

Yes, for about 50-100ms at a time, and -to- the disks, not from them; 
well, the caches anyway.  On 133MB/s PCI, more like 150-300ms of 
saturation on bursts, and shared with the NICs.  I have very short and 
bursty I/O patterns as is typical in real-time applications.  It's 
easier to guarantee latency when your pipes are always at a small 
fraction of their potential.  100ms higher latency, even for part of a 
second, is quite noticeable in what I serve.

-ryan

PS 2.6 dropped my latencies under load by about 90% over 2.4!  We're 
talking 40-80ms application response time under high load to 8-10ms 
under high load.  <3