Areca 1231ML+BBU - which 4 GB stick will fit ?

[Spacedust]

I just go a new Areca ARC-1231ML+BBU to replace my current ARC-1210.

My current card is having 1 GB DDR2 ECC PC2-4200 533 MHz 2Rx8 from Hynix. I would like to upgrade to 4 GB. Do you know which modules will work for 100% ?

I can get KVR667D2Q8P5/4G cheap. It seems to be 4Rx8. Will it work ?

Areca told me that only 8 ranked modules will work.

[nanofrog]

According to Areca, it's only upgradable to 2GB of DDR2-533 ECC or non-ECC, using x8 or x16 ranking only (here).

4GB sticks of DDR2-4200 (533MHz) aren't available in x8 or x16 ranked sticks (too many chips = why 4GB isn't supported on that model).

That said, an ARC-1231ML running 2GB is fast as hell (one of the card's I have in my machine, and I can get 1.39GB/s <RAID 5 via 8x members> out of it due to the cache ).

Hope this helps.

[Spacedust]

Originally Posted by nanofrog

According to Areca, it's only upgradable to 2GB of DDR2-533 ECC or non-ECC, using x8 or x16 ranking only (here).

4GB sticks of DDR2-4200 (533MHz) aren't available in x8 or x16 ranked sticks (too many chips = why 4GB isn't supported on that model).

That said, an ARC-1231ML running 2GB is fast as hell (one of the card's I have in my machine, and I can get 1.39GB/s <RAID 5 via 8x members> out of it due to the cache ).

Hope this helps.

I've read some forums and there are people who use 4 GB 533 MHz and 667 MHz sticks in their cards! Some of them were using Kingston and some were using Crucial. 667 MHz will be downclocked to 533 MHz.

There are a few such sticks on eBay and I will give a try

Now I have to make some room for the BBU and I will remove my flashed Radeon, because Crossfire is not working as it should paired with Apple card. Currently I was able to get 755 MB/s reads and writes with 3 Kingston SSD drives on degraded RAID5 (one is out for warranty) in PCIex 4x slot.

Example: http://www.ramexperts.com/ddr2/serve...5-15-1-8v.html

[nanofrog]

Originally Posted by Spacedust

I've read some forums and there are people who use 4 GB 533 MHz and 667 MHz sticks in their cards! Some of them were using Kingston and some were using Crucial. 667 MHz will be downclocked to 533 MHz.

There are a few such sticks on eBay and I will give a try

Now I have to make some room for the BBU and I will remove my flashed Radeon, because Crossfire is not working as it should paired with Apple card. Currently I was able to get 755 MB/s reads and writes with 3 Kingston SSD drives on degraded RAID5 (one is out for warranty) in PCIex 4x slot.

Example: http://www.ramexperts.com/ddr2/serve...5-15-1-8v.html

I realize that it would downclock, but I've not seen x8 or x16 ranked sticks @ 4GB (DDR2 533 is getting hard to find, and what I found in a search earlier today was x1, x2, or x4 ranked, which will not work).

Denser memory wasn't available at the introduction of DDR2-533MHz, so they used more chips. But once the denser chips became available, most memory shifted both to increase capacity, but also to lower costs (starts high, then gets cheaper - at/near the end of a product cycle, the prices drop, making higher part count products financially viable). So it seemed reasonable to me that any x8 or x16 sticks at 4GB were long gone, and I'm not a fan of used/refurbished memory unless I know the person/source it's coming from (too easy to get bad RAM otherwise when used, and refurbished could mean just about anything).

Assuming the description is truly accurate in that link, and you're willing to fund it, go for it.

Another thing to consider however, that the performance gain as you increase is not linear, so the increase between 1 - 2GB vs. 2 - 4GB isn't as great (aka Diminishing Returns). Search this if you want.

[Spacedust]

Example: http://www.ebay.pl/itm/Viking-Hynix-...item416309e333

This is just normal Hynix 2Rx8 which should perfectly work.

BTW: Areca is testing firmware 1.50 for their cards. Currently only 1882 users can get it.

********* Change Log For V1.50 Firmware

2010-12-22
1 BOOT_WITH_PHY_DISABLED for 1880
(A) DHCP must be disabled
(B) IP should be leave unchanged (192.168.1.100)
(C) EtherNet port must be connected directly
(D) One controller is set to bootup with PHY disabled (spBootPhyDisabled=1)
(E) When spBootPhyDisabled is changed from 1 to 0, the controller will
send a command to inform other controller to stop its action and
wait for 10 seconds to receive the reply, then enable the PHY of the
controller.
(F) Limitation:
CACHE may be lost if enabled
MIGRATION point may be lost if just in MIGRATING
REBUILDING point may be lost if just rebuilding, will restart from
previous point
2010-12-29
1 FIX SNMP with MAX_ENCLOSURES not 8
2011-1-12
1 Add SUPPORT_ZIPPY_POWER for ARC1201
2011-1-20
1 Fix ARC1880 dual controller control
2011-1-24
1 Fix Raid5060 write through volume cannot be deleted after write
2011-1-27
1 Fix OfflineRiadSet:if raidset already spun down, no start/stop unit is required
2 Add SUPPORT_RAID1_BG_INIT (background initialization for raid 1)
2011-2-8
1 Fix ARC1880 EPCIE activity/status LED
2011-2-18
1 ARC1880:set HM_FAN_DIVISOR to 8 to support low RPM fans
2 ARC1880:RAID1 background init supported
2011-3-7
1 TCP.c __tcp_write:add timeout check (NESSUS security test ether net hang)
2011-3-7
1 Fix IOP3XX/MRVL5182 model pci HBA raid card with spin down in FreeBSD
incorrect sense data is sent to OS, OS need the correct sense data for
error recovery
2011-3-15
1 VolumeInit is called after device initialization, if HDD insert is signaled
between HDD scan and HDD checkin may cause hang
2011-3-18
1 Fix ARC1680/1880+supermicro expander under voltage error reported.
supermicro report volatge code 1/5/6 dynamically. patch the code to treat
1/5/6 to no error
2011-3-23
1 Fix NESSUS cause CheckPassWord hang,
2011-3-24
1 Add cacheLock in rebuild.c to fix heavy loading may cause race
condition for cache allocation.
2 Add BSemaTakeH for rebuild.c to improve init/rebuilding speed
2011-3-29
1 Add COMPATIBLE_RAIDSET: to create old style raid set on new SAS controller
new style raidset:max 128 volumes
old style raidset:max 16 volumes
2011-4-7
1 Fix JBOD empty slot LED error in StaggerDelay routine use HM_LED_LOCATE1
2 CLEANUP_TASK_MGNT:fix HDD timeout and task management request from pl lib
need to clean up task management after AbortTask
3 PL library phase 9 is used (MPT and PL)
2011-4-8
1 ARC1880:add model name extension display
2 ARC1880:Add -4i/-8i support (no expander support)
2011-4-14
1 PCI SETTABLE_DATA_XFER_LEN
2 FIX_SAS_SATA_BRIDGE Fix SAStoSata bridge problem:LSI6G
3 Fix HDD temperature display problem
2011-4-22
1 Add ARC1880... SES2 code 5 (not installed for power and fan)
2011-5-10
1 Add log for Lost migration point if migration is aborted or stopped due to
volume failed
2011-5-27
1 AJA_TUNING, remove OPTIMIZE_HD_EDIT
2 ARC1880:add DELL-MD1000 support
3 ARC1680:Add AJA support
2011-6-2
1 Fix All Model:CompleteRebuild, when aborting initializing and rebuilding
need to reset volInitLba to zero
2011-6-10
1 Add update EXPANDER firmware feature for ARC1880/8366... 6G models
2011-6-17
1 Add SUPPORT_PHY_CONTROL
(A) LSI6G model
(B) Marvell SATA (6081 model)
(C) TBD MRVL6145/LSI3G/IOP348
2 Show EXPANDER information on browser with clear PHY error log
(A) LSI6G model completed
(B) TBD LSI3G/IOP348
3 Add update expander firmware
(A) Supported for LSI 6G model controller and 6G expander
(B) 3G expander, there is probelm identifying data file ID ....
2011-6-23
1 LSI 6G model and pure SATA model:add SUPPORT_PHY_CONTROL
2011-6-24
1 SUPPORT_LARGE_BLOCK_IO
2011-6-30
1 LSISAS2208:fix L2 invalidate bug
2011-7-13
1 Fix ARC1880:model name error
2 Fix ARC1880:battery checking, need to set batFlag to 0 before read chip ID
3 Add SUPPORT_PM2_SETTING for ARC1880/1882
(A) The setting will take effect after next power cycle
(B) ARC1680:due to intel transport reject the PM2 enabled HDD, this feature
cannot work with ARC1680
2011-7-15
1 Disable >= 256K stripe size support if HDDs do not support LBA48
2 Disable migration to >= 256K stripe if HDDs do not support LBA48
2011-7-22
1 Fix NEW_SMART_HDD_TEMP smart HDD temp value error in smart array
2011-7-26
1 Add SSD Support
2 Support RaidLevel 100 and RaidLevel 00
(http/vt100/gui/panel, for GUI, R00/R100 already supported)
3 Add RAID0 no init (rescue) feature for SSD guard feature
4 Add hot spare type
Original global hot spare is modified to global for separate disk type
Add global for All type
// 1. Local hot spare:do not check disk type
// 2. Local to ENC:check disk type
// 3. Global For Disk Type:check disk type
// 4. Global For All:do not check disk type
Hot plug auto rebuild-->SSD only rebuild SSD, HDD only rebuild HDD
2011-8-4
1 ARC1882:Add/fix SDRAM ECC handling and scrubbing
2 Fix ARC1882 with 1M stripe size and write through stack underflow
2011-8-5
1 Fix PM2 support for ARC1882/1880
2 Fix R5060 background init race condition with host access
2011-9-9
1 Hitachi HDD xfer rate programming
2 WD HDD Xfer Rate Control
2011-9-15
1 VOLUME_CHECK_ERROR_TYPE:to show volume check errors
parity_error_count/media_error_count
2011-9-19
1 Fix SCSI write same program flow bug for more than 2TB
2011-9-22
1 Fix ARC1882:BIOS IO hang
2 Add support for SAS model with SATA Write Same
ARC1680/1880/1882
2011-9-26
1 Fix ARC1680/1880:RWPassThrough problem with large read ahead
2011-9-28
1 Fix ARC1880/1680+LSI+Intel SSD
2011-10-17
1 Fix foreground init display percentage error
2011-11-3
1 Fix LSI PL library problem
2011-11-7
1 Fix LSI SAS problem of timer
2011-11-10
1 Fix PL Phase11 + LSI SAS_SATA bridge problem
2011-11-21
1 Cache structure redesign
2011-12-6
1 Fix ArcHttp update firmware protocol error
2011-12-9
1 IMPROVE_DISK_WT_REBUILD
3 SUPPORT_WRITE_PROTECT
4 PATCH_CAPACITY_FOR_NEW_STRIPE
5 Add display for volume info
7 PATCH_INTEL510_SSD
2012-1-4
1 Build V1.50 for some models with fixed bugs

[nanofrog]

I'm usually a bit too nervous to buy memory off of eBay (much harder to do returns in my experience).

As per the new firmware, it looks like those LSI chips they went with are giving them a real headache.

[Spacedust]

Originally Posted by nanofrog

I'm usually a bit too nervous to buy memory off of eBay (much harder to do returns in my experience).

As per the new firmware, it looks like those LSI chips they went with are giving them a real headache.

There is some specialized company in Poland offering 4GB KINGSTON DDR2 667MHz PC2-5300 ECC Registered DIMM KVR667D2Q8P5/4G in a good price, but will it work ? It's a Quad Rank x8. I've asked Areca, maybe they will know.

[nanofrog]

Originally Posted by Spacedust

...It's a Quad Rank x8...

Unfortunately, No.

Since that particular Kingston stick is Quad Ranked (x4, sometimes written as Rx4; R = rank, followed by xn, n being the rank value) will not work.

BTW, when I checked for memory on your behalf (quick search, not every memory supplier I usually do when actually buying myself), I went to Kingston to search, as their ValueRAM is usually quite good, competitively priced, and the search results are accurate (also the fact I look for new rather than used, unless there is absolutely no choice).

This is part of the reason I was surprised you located a 4GB*1 x8 ranked stick (linked in a previous post).

[Spacedust]

Originally Posted by nanofrog

Unfortunately, No.

Since that particular Kingston stick is Quad Ranked (x4, sometimes written as Rx4; R = rank, followed by xn, n being the rank value) will not work.

BTW, when I checked for memory on your behalf (quick search, not every memory supplier I usually do when actually buying myself), I went to Kingston to search, as their ValueRAM is usually quite good, competitively priced, and the search results are accurate (also the fact I look for new rather than used, unless there is absolutely no choice).

This is part of the reason I was surprised you located a 4GB*1 x8 ranked stick (linked in a previous post).

From my experience the worst is Micron (3 failures in 1,5 year time). Now my Mac Pro is using 5x Micron, 1x Kingston, 1x Nanya, 1x Hynix

[nanofrog]

Originally Posted by Spacedust

From my experience the worst is Micron (3 failures in 1,5 year time). Now my Mac Pro is using 5x Micron, 1x Kingston, 1x Nanya, 1x Hynix

Recently the worse I've dealt with is OCZ memory (non-ECC), and they source the chips from major vendors. But it's standard memory, not ECC (don't usually see an issue there <ECC>, and when it does happen, ESD is usually the cause, as the installer didn't know about it/take precautions while handling it).

[Spacedust]

Just got mail from Areca:

> ARC1231ML do not support 4GB cache and it support 2GB cache since PCB
> revision 1.0.
> so you can not upgrade the cache on your card based on the card's
revision.

No upgrade for me :/ I got rev. B.

[Spacedust]

2GB is supported but not 4GB.
and please make sure the module you choosed do met the controller requirement. due to the processor limitation, controller do not works with every 2GB modules.

So 2 GB will fit They're so cheap in Poland. Only about $30.

[nanofrog]

Found out I'm not as crazy as I might seem on the surface after all, didn't you...

[Spacedust]

Originally Posted by nanofrog

Found out I'm not as crazy as I might seem on the surface after all, didn't you...

Just got the 2 GB stick and all working. Now getting 1,4 GB/s speed read/writes !!!

[nanofrog]

Originally Posted by Spacedust

Just got the 2 GB stick and all working. Now getting 1,4 GB/s speed read/writes !!!

Now you know what that kind of speed feels like. Just don't drool in the keyboard...

[Spacedust]

Originally Posted by nanofrog

Now you know what that kind of speed feels like. Just don't drool in the keyboard...

Sometimes my card is being detected as PCI-E 2x instead of 8x. Card is installed in a second slot 16x 2.0.

How to fix it ?

[nanofrog]

Originally Posted by Spacedust

Sometimes my card is being detected as PCI-E 2x instead of 8x. Card is installed in a second slot 16x 2.0.

How to fix it ?

Detection is in the hardware.

You might try getting your hands on as close to 100% Isopropyl Alcohol as possible, and clean the contacts (card and slot if you can get in there; this is where ESD swabs and brushes can help, and they can be found on eBay). 100% Ethyl Alcohol will work as well. Just make sure you're in a well ventilated area, as 100% alcohol is flammable.

[Spacedust]

Originally Posted by nanofrog

Detection is in the hardware.

You might try getting your hands on as close to 100% Isopropyl Alcohol as possible, and clean the contacts (card and slot if you can get in there; this is where ESD swabs and brushes can help, and they can be found on eBay). 100% Ethyl Alcohol will work as well. Just make sure you're in a well ventilated area, as 100% alcohol is flammable.

I need to test it out, it happens only I start my Mac Pro the first time in current day when it's cold boot. Other boots are working fine at 8x

This is how it looks:

[nanofrog]

Check the Areca's Logs (not the Change Log in the latest firmware revision), and see if it's showing an Under-volt Condition, particularly on the +3.3V rail (bad +3.3V rail indicates a failing PSU, as that is what the PCIe slot provides to the card; all other voltages are derived from that by the card's own voltage regs).

[Spacedust]

Originally Posted by nanofrog

Check the Areca's Logs (not the Change Log in the latest firmware revision), and see if it's showing an Under-volt Condition, particularly on the +3.3V rail (bad +3.3V rail indicates a failing PSU, as that is what the PCIe slot provides to the card; all other voltages are derived from that by the card's own voltage regs).

It looks like this:

Fan Speed N.A.
Battery Status Charged(100%)
CPU Temperature 56 ºC
Ctrl Temperature 49 ºC
Power +12V 12.038 V
Power +5V 4.999 V
Power +3.3V 3.296 V
SATA PHY +2.5V 2.496 V
DDR-II +1.8V 1.840 V
PCI-E +1.8V 1.840 V
CPU +1.8V 1.840 V
CPU +1.2V 1.216 V
DDR-II +0.9V 0.912 V
Hdd#1 Temperature 30 ºC
Hdd#2 Temperature 30 ºC
Hdd#3 Temperature 40 ºC
Hdd#4 Temperature 40 ºC
Hdd#5 Temperature --
Hdd#6 Temperature --
Hdd#7 Temperature --
Hdd#8 Temperature --
Hdd#9 Temperature --
Hdd#10 Temperature --
Hdd#11 Temperature --
Hdd#12 Temperature --

Seems to be ok, I've just set stagger controll to 0.3 and my card is booting much faster. Instead of 46 seconds just 14

[nanofrog]

Originally Posted by Spacedust

Seems to be ok, I've just set stagger control to 0.3 and my card is booting much faster. Instead of 46 seconds just 14

Hopefully, it will be OK from this point on.

But you see what happens when the boot process isn't being completed in the available time (insufficient current can cause the entire system to hang if the OS/Boot disk is attached to the RAID card).

[Spacedust]

Originally Posted by nanofrog

Hopefully, it will be OK from this point on.

But you see what happens when the boot process isn't being completed in the available time (insufficient current can cause the entire system to hang if the OS/Boot disk is attached to the RAID card).

Cleaned all contacts and still the same. When on cold boot I got 2x, reboot and I got 8x. It seems to be a bug inside firmware. All other reboots and boot within several hours are 8x too.

[Spacedust]

Got reply from Areca:

It is not a firmware bug but a known issue between the chipset in this machine and the processor in 1231.
there have one known link speed issue between 5 series chipset and IOP34x, the problem is the link speed may changed from x1 to x8. this issue have no fix solution available. if you got incorrect link speed, the only possible solution is reboot system once until you got the properly x8 link speed.

another solution only available in older macpro machines. as i known, older macpro machines have an utility can configure the link speed of each slot.
you can use this utility to force configure the specific slot with x8 link speed if it is available in your machine.

[nanofrog]

Originally Posted by Spacedust

Got reply from Areca:

Really odd situation. You'll have to deal with it (reboot until it's linked up at x8 lanes).

[Spacedust]

Originally Posted by nanofrog

Really odd situation. You'll have to deal with it (reboot until it's linked up at x8 lanes).

It always happens on first cold boot (after night in the morning)