SGI: Hardware

SATA in O2 with ARS-2160H - Page 1

rooprob
Who joined Sept. 1, 2010, 6:20 a.m.
and authored 37 notes

Wrote the following at Nov. 22, 2012, 1:34 a.m...

Hi,

I'm experimenting with ARS-2160H SCSI SCA to SATA interface in my O2.

I can *sometimes* get the PROM hinv to see the SCSI device, however IRIX hinv doesn't subsequently see it when it boots, and theres no entry for it in /dev.
On reboot most of the time the prom ceases to see the SCSI device again until I've pulled it and changed some jumpers on the ARS.

ARS-2160H
Has ID jumpers on the front for SCSI IDs 0 through 15, delayed init, SE to LVD, write protect.

Tried (with no real difference)
Set to SCSI ID 1, 3, 15.
Explicitly closed "delayed init", (not sure exactly what that does)

Doesn't matter which SCSI slot I put the disk into either or if the boot disk is present.

(already on the bus: internal disk (2), internal dvd rom (4)

Currently the machine boots just fine with my usual 10K SCSI disk.

I've also noticed (but not fully recorded) there is a sequence to when it will detect it and when it wont.
When I plug the power cable in it immediately starts (without my presseing the power button on the front). In this case I know it wont detect it, and sure enough it's not listed in PROM hinv.
When I plug the power in and it remains off, when it boots prom sometimes sees it.
Seems to depend on when it type "OFF" at the prom or just pull the power out.

diegel
Who joined Nov. 17, 2009, 2:08 a.m.
and authored 158 notes

Wrote the following at Nov. 22, 2012, 3:44 a.m...

rooprob wrote:

Hi,
I'm experimenting with ARS-2160H SCSI SCA to SATA interface in my O2.

The ACARD adapters works fine for me. It is important to set the force SE jumper to work correctly in an O2.

_________________
:Tezro:

jan-jaap
Who joined June 17, 2004, 10:35 a.m.
and authored 3659 notes

Wrote the following at Nov. 22, 2012, 4:30 a.m...

Delayed init delays startup of the disk, usually to some value * scsi ID.
The idea is to prevent a big power surge when all disks spin up at the same time.

_________________
Now this is a deep dark secret, so everybody keep it quiet

It turns out that when reset, the WD33C93 defaults to a SCSI ID of 0, and it was simpler to leave it that way... -- Dave Olson, in comp.sys.sgi
Currently in commercial service:

(2x)

In the museum : almost every MIPS/IRIX system.
Wanted : GM1 board for Professional Series GT graphics (030-0076-003, 030-0076-004)

rooprob
Who joined Sept. 1, 2010, 6:20 a.m.
and authored 37 notes

Wrote the following at Nov. 22, 2012, 2:54 p.m...

Thanks for the infos. Diegel was right. It needed the SE jumper closed.

Two hours, job done. Cloned 70GB SCSI 10K disk to 120GB SSD, rebooted and all is well.
£50 for the SSD, and (cough) £more for the ARS.

Incidentally, for those feeling daring, you _can_ get two disks in an R10K O2 without breaking anything. It's a little precarious but the SSD on a naked ARS card (sans chassis) will hang in the otherwise obscured SCA socket. You can remove the PCI cage so long as you put the PCI card back in and leave enough space. I didn't have any other means to clone the disk.

duck
Who joined Oct. 27, 2003, 5:22 p.m.
and authored 384 notes

Wrote the following at Nov. 23, 2012, 10:29 p.m...

Now these converters look really nice. What would be a good european/nordic source for them I wonder.

_________________
:Octane:

halo , oct ane
N.B.: I tend to talk out of my ass. Do not take it too seriously.

mia
Who joined Feb. 19, 2003, 1:54 p.m.
and authored 877 notes

Wrote the following at Nov. 23, 2012, 11:02 p.m...

Too bad Irix doesn't have trim support though. You should run diskperf btw.

_________________
:Onyx2:

rooprob
Who joined Sept. 1, 2010, 6:20 a.m.
and authored 37 notes

Wrote the following at Nov. 26, 2012, 3:18 a.m...

mia wrote:

Too bad Irix doesn't have trim support though. You should run diskperf btw.

That's fair concern about TRIM and getting a definitive story around support, especially when masquerading as a SCSI disk.

I picked the old line of OCZ Vertex2 128GB as a) SATA 2 IO rate still exceeds the SCSI2 bus, b) end of line is cheap and c) came out when no OS natively supported TRIM. Although I'm no expert, the controller onboard SandForce SF-1200 takes care of marking deleted blocks via garbage collection routine obviating the need for TRIM (apparently).

The ARS manual had screenshots of a Windows device tree clearly displaying an ARS, yet on my test rig the SCSI disk was reported OCZ Vertex clearly from the drive itself. With the right knowhow it may be possible to probe more information from the drive itself.

My O2 with a real SCSI disk clearly makes write ops every few seconds. I have never figured out which service this was. With an SSD only time will tell if it eventually kills it just by being on.

Thinking about it I may buy a second one given built-in garbage collection is something they'll eventually remove as a "cost saving" as more operating systems take care of it natively, in a kind of race-to-the-bottom to winmodems. Indeed newer Samsung 830 I added to my Macbook benefits from TRIM support (apparently) which isn't available in Snow Leopard (whoops!).

rooprob
Who joined Sept. 1, 2010, 6:20 a.m.
and authored 37 notes

Wrote the following at Nov. 26, 2012, 2:31 p.m...

This is interesting. The O2 has a Ultra Wide SCSI 40MB/sec bus, so should I expect to see read performance at that rate?

diskperf seems to max out at reads 26MB/sec

Code:

mapleleaf 46# diskperf -D -W -r4k -m16m /root/testfile

#---------------------------------------------------------

# Disk Performance Test Results Generated By Diskperf V1.2

#

# Test name     : Unspecified

# Test date     : Mon Nov 26 22:18:31 2012

# Test machine  : IRIX mapleleaf 6.5 07202013 IP32

# Test type     : XFS data subvolume

# Test path     : /root/testfile

# Request sizes : min=4096 max=16777216

# Parameters    : direct=1 time=10 scale=1.000 delay=0.000

# XFS file size : 536870912 bytes

#---------------------------------------------------------

# req_size  fwd_wt  fwd_rd  bwd_wt  bwd_rd  rnd_wt  rnd_rd

#  (bytes)  (MB/s)  (MB/s)  (MB/s)  (MB/s)  (MB/s)  (MB/s)

#---------------------------------------------------------

4096    9.31    7.82    9.27    7.77    9.06    6.90

8192   14.26   12.08   14.26   12.12   14.18   10.90

16384   19.30   16.23   19.29   16.38   19.37   15.37

32768   23.80   20.08   23.63   19.90   23.79   19.24

65536   26.59   23.49   26.68   23.48   26.54   22.90

131072   28.36   24.91   28.37   24.87   28.55   24.57

262144   29.20   25.69   29.33   25.70   28.90   25.52

524288   29.88   26.02   29.82   26.06   29.78   26.03

1048576   30.10   26.27   29.98   26.23   30.14   26.21

2097152   30.25   26.27   30.35   26.33   30.33   26.27

4194304   30.75   26.25   30.25   26.80   30.47   26.96

Adrenaline
Who joined Feb. 10, 2005, 12:37 p.m.
and authored 447 notes

Wrote the following at Nov. 26, 2012, 2:52 p.m...

rooprob wrote:

mia wrote:

Too bad Irix doesn't have trim support though. You should run diskperf btw.

I'd watch out for that Vertex 2 or any SF based SSD. Had several SF-XXXX series SSDs die on me (Agility 2s/3s, Vertex 1s/2s/3s, Corsair Force 3s, SanDisk Ultra Extreme) over the years. The only drives that hadn't failed were my Samsung 830s and my new 840 Pros. I should note I do still have 1 30gb Vertex Plus as a caching drive in a ZFS Pool that's been running for 2 years now without errors. So your mileage may vary.

_________________
:Indigo:

33mhz R3k/48mb/XS24 :Indy:

150mhz R4400/256mb/XL24 :Fuel:

600mhz R14kA/2gb/V10

8x1.4ghz Itanium 2/8GB :O3x08R:

32x600mhz R14kA/24GB :Tezro:

4x700mhz R16k/8GB/V12/DCD/SAS/FC/DM5 :O3x0:

2x700mhz R16k/4GB

jan-jaap
Who joined June 17, 2004, 10:35 a.m.
and authored 3659 notes

Wrote the following at Nov. 27, 2012, 12:22 a.m...

rooprob wrote:

diskperf seems to max out at reads 26MB/sec

What's the CPU spec of the machine? Maybe it can't handle (much) more than that. I've got a QL2342 in my 400MHz R12000 O2, that's a dual 2Gb/s FC card. I don't remember the exact numbers, but it was rather underwhelming

Adrenaline wrote:

Had several SF-XXXX series SSDs die on me (Agility 2s/3s, Vertex 1s/2s/3s, Corsair Force 3s, SanDisk Ultra Extreme) over the years. The only drives that hadn't failed were my Samsung 830s and my new 840 Pros.

Lost a Corsair M4 myself last week. I've got an SF based SSD in my 2011 MBP (an OWC Mercury Extreme 6G ), seems to work fine (after Apple replaced the SATA cable of the MBP, that is).

_________________
Now this is a deep dark secret, so everybody keep it quiet

It turns out that when reset, the WD33C93 defaults to a SCSI ID of 0, and it was simpler to leave it that way... -- Dave Olson, in comp.sys.sgi
Currently in commercial service:

(2x)

In the museum : almost every MIPS/IRIX system.
Wanted : GM1 board for Professional Series GT graphics (030-0076-003, 030-0076-004)

rooprob
Who joined Sept. 1, 2010, 6:20 a.m.
and authored 37 notes

Wrote the following at Nov. 27, 2012, 4:09 a.m...

Interesting info's, thanks.

My O2 is R12K 400, so as good as it gets. During diskperf osview was telling me 50% idle, with the rest in %sys time and some %iowait. So it's not all CPU, but it is a lot of kernel time which all makes sense.

I can do some experiments with NetBSD to get another baseline. Or I could do some work.

Anecdotally when I performed the xfsdump/xfsrestore from the 10K SCSI disk to the SSD it was all %sys time to almost 100% for the whole 2 hours or so. gr_osview was a solid red bar

Back of an envelop tells me that was about 11-15MB/second, which seems a little low for two disks.

Ah well. With the filesystem IO cache it tells me I get 50MB/sec and it's all totally fine for what I use it for. If I squeeze any more out of it I'll let you know. I seem to remember from way back that fx lets you set lowlevel caching options on the disk. I wont disable fsync.

Quote:

Now these converters look really nice. What would be a good european/nordic source for them I wonder.

Nice guys at this outfit Active Network Systems Ltd | Tel: +44 (0)1480 437997 | Fax: +44 (0)1480 436031 | [email protected] | www.ans-ltd.co.uk

Pricey at £150 inc VAT and delivery for one but they did say they can get a price on a bulk order. If we can get the price down to £100 with others here I'll pitch in for one more.

duck
Who joined Oct. 27, 2003, 5:22 p.m.
and authored 384 notes

Wrote the following at Nov. 28, 2012, 3:26 a.m...

rooprob wrote:

Quote:

Now these converters look really nice. What would be a good european/nordic source for them I wonder.

Nice guys at this outfit Active Network Systems Ltd | Tel: +44 (0)1480 437997 | Fax: +44 (0)1480 436031 | [email protected] | http://www.ans-ltd.co.uk

Pricey at £150 inc VAT and delivery for one but they did say they can get a price on a bulk order. If we can get the price down to £100 with others here I'll pitch in for one more.

Waargh! That's way out of my toy budget I'm afraid.

_________________
:Octane:

halo , oct ane
N.B.: I tend to talk out of my ass. Do not take it too seriously.

canavan
Who joined July 21, 2003, 5:19 a.m.
and authored 687 notes

Wrote the following at Dec. 13, 2012, 1:06 p.m...

Concerning the lifetime of an SSD under IRIX, has anyone thought about using fx and the various options for mkfs_xfs (such as sunit, swidth , lazy-count) to reduce write amplification?

Here's a forum thread of some NeXT users http://www.nextcomputers.org/forums/viewtopic.php?t=3057 , where someone recommends using hdat2 to reserve part of the SSD as a Host Protected Area, such that a larger fraction can be used for wear leveling.

Has anyone run any benchmarks on an ACARD-connected SSD in something more modern (IP35)?

Rev.Bubba
Who joined Oct. 19, 2003, 12:21 p.m.
and authored 260 notes

Wrote the following at April 11, 2013, 7:33 p.m...

Well, I guess I will jump on this thread. I just received my ACARD 2160H today and I have had SGINERD's O2+ waiting for something special. I am going to waste an Intel 710 300gb SSD on mine. The Intel 710 was designed to be used behind raid arrays without TRIM. Mostly found in Sun and Cisco UCS servers. It has an extra 180gb NAND on the drive and the advanced firmware for garbage collection should do the trick.

I will be that douche driving down the street in a $500 car sporting $5k rims.

rooprob wrote:

Incidentally, for those feeling daring, you _can_ get two disks in an R10K O2 without breaking anything. It's a little precarious but the SSD on a naked ARS card (sans chassis) will hang in the otherwise obscured SCA socket. You can remove the PCI cage so long as you put the PCI card back in and leave enough space. I didn't have any other means to clone the disk.

That seems like a job for a 3D printer, rooprob. I am sure someone on here could design the right bracket for it. :roll:

_________________
Always act incompetent, so they expect less of you and your job will be much easier.

mia
Who joined Feb. 19, 2003, 1:54 p.m.
and authored 877 notes

Wrote the following at April 11, 2013, 8:47 p.m...

Make sure to run diskperf this could be interesting.

_________________
:Onyx2:

mapesdhs
Who joined Nov. 10, 2003, 4:17 p.m.
and authored 2183 notes

Wrote the following at April 12, 2013, 3:59 a.m...

rooprob writes:
> That's fair concern about TRIM and getting a definitive story around support, especially when masquerading as a SCSI disk.

It doesn't matter. Vertex2Es were optimised to run well on OSs that don't support TRIM, because lots of people used them with XP.
My V2E/120 is faster now than it was when I first installed it in my Fuel several years ago.

> I picked the old line of OCZ Vertex2 128GB ...

Actually 120GB, but yes, perfect choice. The Vertex3 and Vertex4 are less effective under IRIX.

The Corsair F60 is similar to the V2E (same controller), but not quite as quick (less optimised fw).

> controller onboard SandForce SF-1200 takes care of marking deleted blocks via garbage collection routine obviating the need for TRIM (apparently).

It does this very well under IRIX. See my old post about SSD performance with Fuel.

> ... With an SSD only time will tell if it eventually kills it just by being on.

Unless O2 behaves very differently to Fuel, it will be fine.

Ian.

_________________
SGI Systems/Parts/Spares/Upgrades For Sale: http://www.sgidepot.co.uk/sgidepot/
[email protected] , [email protected] , +44 (0)131 476 0796, check my auctions on eBid!

mila
Who joined Dec. 31, 2009, 4:20 a.m.
and authored 284 notes

Wrote the following at April 12, 2013, 9:27 a.m...

Hi,

Just a question. If you not have trim support in the driver, how can the drive itself make "garbage collection" without knowing the filesystem?

This flash must be erased prior to write which means that if I write a file foo.c and then rm foo.c, the file system only removes the "pointer" to the file, so how can the SSD firmware "erase" it?

Or does the "latest" SSD without "need" for trim just are pretty quick in erasing then non 0xff sectors in the flash?

Michael

_________________
--
No Microsoft product was used in any way to write or send this text.
If you use a Microsoft product to read it, you're doing so at your own
risk.

Rev.Bubba
Who joined Oct. 19, 2003, 12:21 p.m.
and authored 260 notes

Wrote the following at April 12, 2013, 11:16 a.m...

mia wrote:

Make sure to run diskperf this could be interesting.

I will be happy to do just that. I am sure I will need to call on a few of you to help me with running these tests.

mila wrote:

Just a question. If you not have trim support in the driver, how can the drive itself make "garbage collection" without knowing the filesystem?

This flash must be erased prior to write which means that if I write a file foo.c and then rm foo.c, the file system only removes the "pointer" to the file, so how can the SSD firmware "erase" it?

Or does the "latest" SSD without "need" for trim just are pretty quick in erasing then non 0xff sectors in the flash?

It has been a while since my various reviews of these and other enterprise drives and I maybe misusing the term "garbage collection" but it more boiled down to the modifications to the firmware found in on Intel 320 to make it more aware of the additional NAND left for redundancy and such. Either way it made me focus on acquiring these at the best price possible for a small CloudStack setup I am working on with some older Altix XE servers.

Lower power consumption and noise is important but not as important as data integrity and knowing what I want to end up using the O2+ for, I want to feel rather good about what is to be stored on it. I will then build me a primary box out of a Fuel with an Intel 710 for boot and a couple striped Intel DC S3700 drives. I assume there are no pci based SAS controllers supported under IRIX that are 6Gbps?

Rev.

_________________
Always act incompetent, so they expect less of you and your job will be much easier.

robespierre
Who joined Sept. 12, 2011, 1:28 p.m.
and authored 318 notes

Wrote the following at April 12, 2013, 11:52 a.m...

hi, all SSDs have garbage collection. they know which blocks have ever been written, and which have been erased to zero. this information is necessary to provide wear leveling.

the problem with hosts that do not provide "trim" or "drop" signals is that blocks that the filesystem considers free cannot be freed on the SSD side. there is nothing magic about a trim command, it is just shorter and faster than overwriting the block with zeros. so as writes accumulate over time, the wear leveler has a smaller set of free blocks to use.

the other major SSD problem, write amplification, has more damaging effects when the wear leveler has a smaller set of free blocks to use.

_________________
:PI:

Winnili
Who joined Sept. 15, 2012, 5:37 a.m.
and authored 178 notes

Wrote the following at April 12, 2013, 12:41 p.m...

Interesting thread, but won't it ultimately be limited to the O2's slower SCSI bus speed? (Not to spoil things by pointing out something ‘obvious’, I still really like this idea. Anything that involves the prolonged use of SGI gets my approval!)

How easy and practical, or even possible, would it be to add an Ultra-3/Ultra160 HBA inside the O2 and to which degree would the O2's overall bandwidth deal with it acceptably? And which of which, if any, work inside the O2? Routing a cable to the disk riser/backplane would be a bit tricky, I'd imagine...

I should note that I've never owned an O2 (although it's never too late for that, bust just saying), so I'm just curious and wondering.

_________________
:Tezro:

• Offering various remaining systems and parts, several interestingly compatible with both IRIX and OpenVMS ( AXP and I64 );
• Looking for an SGI O3000 IP59 1-GHz MIPS R16000 quad-processor node board (for a Tezro).