TheR

Promise SuperTrak EX8350 PCIe x4

Recommended Posts

Anyone using v2.2.0.23 of the webpam for windows software?

I'm trying to locate the ncq settings for my seagate 7200.9 drives ... but don't see it in any of the menu/drive setting tabs.

It seems that the promise card has problems with some seagate drives. Have a look at the compatablilty list. The old webpam & bios uses to have NCQ but the new version removes it.

I have also contacted promise support about the slow transfer speed. They will only accept IOMETER benchamrks. 100% read and then 100% write. Whe I did the IOMETER benchmark for them, my write was over 250MB/s but the read was peaking at 110MB. Something it not right.

I would suggest that anyone looking for a decent Raid 6 card look elsewhere.

So basically, they will only accept benches from a program that gives results completely out of line with what every other program does? That's awsome! It's like a Car dealership being like;

"well, meinke, Goodrich, autozone, etc all said you have bad brakes, but our iobrake test showed that they were the best breaks in history, so nothing's wrong."

Share this post


Link to post
Share on other sites

It sounds like this card will work fine as long as you have NCQ disabled and write cache enabled, is that about right? Or does the harddrive you use have an impact because I was considering using it some of the new Seagate 750GB drives?

If there is no one that can describe an experience using the EX8350, should I just get the Areca ARC-1220 instead?

Share this post


Link to post
Share on other sites
It sounds like this card will work fine as long as you have NCQ disabled and write cache enabled, is that about right? Or does the harddrive you use have an impact because I was considering using it some of the new Seagate 750GB drives?

If there is no one that can describe an experience using the EX8350, should I just get the Areca ARC-1220 instead?

I've got a EX8350, and so far I'm pretty happy with it. I've only had it a few weeks, and have just started to copy everything over from my old, slow but stable linux software raid 5 setup.

Here is some bonnie++ test results, first with NCQ enabled, and then disabled. I've included a software raid 1 with 2 74GB raptors for comparison. I also left my copy process running in the background, doing around 1MB/s random writes. That messes the result up a tiny bit I imagine, but shouldn't be too bad.

As you can see, having NCQ enabled improved my results by around 10%.

Promise EX8350 with 8 Maxtor Maxline III, 250GB (one hot spare, raid 5. That is, 7 active drives with the total storage capacity of 6 drives)

Filesystem: ext3 with dir_index
Computer: Opteron 165 with 4GB RAM on a Asus A8N32-SLI deluxe (the promise card didn't fit into the 4x slot because of the heatpipe on the mobo. But it fit great into the bottom 16x slot. And i'm not gonna have 2 graphics cards anyways.)

NCQ Enabled:

Version  1.03	   ------Sequential Output------ --Sequential Input- --Random-
				-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine		Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
XXXX			 6G 40852  97 179055  80 80614  25 40485  91 221867  27 281.8   0
				------Sequential Create------ --------Random Create--------
				-Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
		  files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
			 16 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++
XXXX,6G,40852,97,179055,80,80614,25,40485,91,221867,27,281.8,0,16,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++




NCQ Disabled:

Version  1.03	   ------Sequential Output------ --Sequential Input- --Random-
				-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine		Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
XXXX			 6G 39732  95 126287  58 68988  21 37088  85 205217  26 226.3   0
				------Sequential Create------ --------Random Create--------
				-Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
		  files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
			 16 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++
XXXX,6G,39732,95,126287,58,68988,21,37088,85,205217,26,226.3,0,16,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++

Same computer, 2 WD Raptor 74GB, Linux software raid 1

Version  1.03	   ------Sequential Output------ --Sequential Input- --Random-
				-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine		Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
XXXX			 6G 39557  94 65722  28 26792  10 32853  72 52296   6 394.4   1
				------Sequential Create------ --------Random Create--------
				-Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
		  files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
			 16 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++
XXXX,6G,39557,94,65722,28,26792,10,32853,72,52296,6,394.4,1,16,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++

I've yet to see how it does in the real world, with lots of random reads and writes. Perhaps i'll start 10 or so parallell bonnie++'s, and add everything up and see how much it differs from the above results.

Share this post


Link to post
Share on other sites

Nah, I didn't do the file tests (at least it didn't do them for me. Maybe i should try again?) , only the bulk test. That comes to 6 files, 1GB each, being written, read and rewritten using different methods (getc()/putc() and fprintf perhaps?)

Performing a test with thousands of files will probably depend more on the filesystem chosen. ext3, resiser3 and xfs perform around the same with the bulk test though. I'm using ext3 because it's served me well for a long time and with dir_index its fast even for large directories.

Share this post


Link to post
Share on other sites
It sounds like this card will work fine as long as you have NCQ disabled and write cache enabled, is that about right? Or does the harddrive you use have an impact because I was considering using it some of the new Seagate 750GB drives?

If there is no one that can describe an experience using the EX8350, should I just get the Areca ARC-1220 instead?

That's the way it is for me, when I disabled NCQ and enabled write cache performance sky rocketed. I haven't tested any really though database workloads but for a file server it works great.

Share this post


Link to post
Share on other sites

I just wanted to add that in my two tests above write cache was enabled for both. I've never tried without it, but I imagine that performance would suffer as it would with any storage solutions. If you're paranoid about loosing data, consider getting a UPS to allow a clean shutdown. A decent 1000VA unit doesn't cost a lot these days, at least a lot less than the promise board itself ;).

I've had problems with random hangs (during little, or no, disk load, so not promise boards fault), but it's probably the motherboards fault. Replacement being shipped in tomorrow, and then I'll test everything again, with write cache disabled maybe? If somebody wants it I can make a 2x2 matrix with the results. That is, turing on and off write cache and NCQ?

Share this post


Link to post
Share on other sites
Yeah, that would be great if you would create a matrix like that.

All right, here it is:

Writing only:

  |W. Cache  |
  | ON | OFF |
------+----+-----+
N ON  | 186| 43  |
C ----+----+-----+	
Q OFF | 146| 40  |
------+----+-----+

Reading only:

  |W. Cache  |
  | ON | OFF |
------+----+-----+
N ON  | 200| 136 |
C ----+----+-----+	
Q OFF | 199| 131 |
------+----+-----+

As you can see, NCQ makes a marginal difference only, compared to write caching. You'd want to have that enabled if you want good performance. It is possible to get around 200MB/s read and write, and probably more with faster drives. This is the same setup as before, but with another motherboard.

Share this post


Link to post
Share on other sites
Hello All - 1st post :)

After installing an EX8350 on a MSI K8N Neo4 MB w/4x Seagate 120G SATA drives

When I boot the system the card identifies itself SuperTrak EX8350/8300 BIOS version 2.8.0.25

Scanning disks...... about 250 dots - takes about 3 minutes

Warning - Something wrong with your hardware!.

The system hangs here.

I have rebooted several times with and without the drives connected to the card - all 4, 2 drives, 1 drive - the result is the same every time - it takes 2-3 minutes in the Scanning Disks mode then fails

Is there something wrong with my card or the drives

what can I try next.

Please Help!!!

Thank you,

Will :)

Will, did you ever get a solution to this?

I have an EX8350. I put the latest Firmware & BIOS on it, and I'm getting this problem.

On a different EX8350, I'm seeing a different problem. There, the card finds the logical drives, but fails to successfully start loading my OS (which is Linux). I get through all the Grub levels, then the system stops.

Does anyone have any ideas?

Thanks.

Share this post


Link to post
Share on other sites
This is the same setup as before, but with another motherboard.

What motherboards did you try? I'm trying to find which motherboards are known to work well with this card.

Share this post


Link to post
Share on other sites

Right now i'm using a MSI K8N Diamond Plus, for about a week, and so far no crash or anything else weird. I'm going to try the asus board in another computer, and hopefully it will crash a few times within a month and confirm its diagnosis ;)

If you want to use ECC i think you need a socket 940 motherboard and an s940-opteron to go in it. I'm not aware of any socket 939 boards that support ECC, but I haven't looked that hard I guess.

Share this post


Link to post
Share on other sites

All in all, i've been using this controller for a few months now, with no major issues. Compared to other people in this thread, either i've been lucky, or skilled enough to do everything right ;)

We're serving around 3k domains with the cluster that this controller acts as the file store for (including some high-profile torrent sites with millions of hits per day), and so far so good. During the testing phase i've tried to happy fellow things up by messing with the settings and even unplugging drives while running, and everything kept working when i expected it to.

Performance is also significantly higher than our old software based raid 5 system (right now we've got lots to spare). I understand that this i Promise's attempt to enter the premium market, and although it's not flawless (personally i'd like a CLI interface, and a little higher performance sometimes) it's good enough. When we run out of diskspace in around 3 years i'd be happy to buy whatever Promise has on the market, unless something drastic makes me change my mind ;)

I've posted some test results before, in this thread. This however concludes my participation here, as i'm going on an extended vacation :)

Share this post


Link to post
Share on other sites

Im having trouble getting both my nF4 NVRaid 0 Array and Promise EX8350 working together.

Basically if I have no devices on the card, the pc boots fine from the NVRaid, but as soon as I put a hard drive on the EX8350, the NVRaid fails to boot.

I have the latest Firmware/BIOS/Drivers/Webpam

Product Name SuperTrak EX8350

BIOS Version 2.9.0.15

Firmware Version 1.2.0.17

Driver Version 2.9.0.6

I have disabled the Sil3114 controller also.

My motherboard is a DFI LanParty uT nF4 SLi-DR 'Venus'.

Share this post


Link to post
Share on other sites

I am running EX8350 with more or less no problems on SuSE 10.0. More problems is of course when kernel gets updated ;). Speeds are around 50MB/s on 8 disk RAID 5 array.

What I would like to know is which Linux distros do you run EX8350 on. Promise officialy supports only SuSE and RedHat with drivers so far and I am really interested changing to Ubuntu 606.

by

TheR

Edited by TheR

Share this post


Link to post
Share on other sites

I have also been running this nice piece of hardware on my gentoo box for some time now. After I found a kernel patch everything worked like a charm.

I have a RAID 1 setup with two 200GB disks.

System info:

05:0e.0 RAID bus controller: Promise Technology, Inc. Unknown device 8350

Subsystem: Promise Technology, Inc. Unknown device 0374

Flags: bus master, stepping, 66MHz, medium devsel, latency 96, IRQ 17

Memory at d4000000 (32-bit, non-prefetchable)

Expansion ROM at d3000000 [disabled]

Capabilities: [c0] Power Management version 2

Capabilities: [d0] Message Signalled Interrupts: 64bit+ Queue=0/1 Enable-

Capabilities: [e0] PCI-X non-bridge device

Mainbord is: nVidia Corporation CK804

Patches that I used:

http://forums.storagereview.net/index.php?...99entry228899

http://kernel.org/pub/linux/kernel/people/...supertrak.patch

http://kernel.org/pub/linux/kernel/people/...5-support.patch

Share this post


Link to post
Share on other sites

I've been using it for a few months now with no major problems. I'm using Debian GNU/Linux so there was no pre-compiled driver, but the one I downloaded from www.promise.com and compiled myself worked good. I ended up having to upgrade the firmware on my harddrives, and I had some motherboard issues, but thats a different story. Performance has been good, and so far no reliability issues. I have 8 maxtor maxline III's, 250GB, and typical performance is around 200MB/s with 7 of them in a RAID-5 array.

Share this post


Link to post
Share on other sites

Hi

i compiled a new Kernel with the mm patch wich works fine. So I made a Raid 5 with 4 300GB Samsung harddiscs and tried if there is

any notification if i disconnect a harddisc. But there is no beeping and no alert. Even when I reboot the system I only see, that the Raid is listed as "critical" but no beeping. The webpam software doesn't run with the mm patch so i can't configure anything.

How did you solved the problem? Can you run the webpam config tool? Is there an other way to activate at minimum the buzzer on the raid controller? This would be enough for me.

best regards

smjor

Share this post


Link to post
Share on other sites

Hi, I ran into the same problem you did last night when I was setting up a new linux file server (Fedora Core 5, Promise EX8350). I was able to get around the problem - not in the most pretty way, but it works. The problem is in the Makefile - the line that tries to determine if it is a 32-bit or 64-bit environment isn't working correctly, so really I just removed the check and set it to always be 64-bit.

The problem occurs around the following lines:

CONFIG_AMD64=$(shell cat $(KERNEL_BLD_DIR)/.config | grep CONFIG_X86_64 | awk -F= '{print $$2}')

ifeq ($(strip $(CONFIG_AMD64)),y)

ARCH_TYPE=x86_64

else

ARCH_TYPE=i386

endif

The string stored in the CONFIG_AMD64 variable ends up returning a few extra lines, that aren't properly stripped off in the if statement. But since we know we're on a 64-bit platform, we can simply remove this check altogether, and simply set the ARCH_TYPE variable to x86_64:

CONFIG_AMD64=$(shell cat $(KERNEL_BLD_DIR)/.config | grep CONFIG_X86_64 | awk -F= '{print $$2}')

ARCH_TYPE=x86_64

You can remove the CONFIG_AMD64 bit as well, I just left it in to clearly illustrate the before and after. A more elegant solution would be to fix the check so that it handles both environments correctly, but as I'm only building this for this platform not worth the time. Anyway, this worked for me, hopefully it will for you too. Good luck!

I also have problems compiling the open-source driver from promise (open source version 2.9.0.10 on an AMD64 gentoo linux with 64bit kernel, latest BIOS for the EX8350)

this is the error I receive:

tresor shasta # make KERNEL_SOURCE_DIR=/usr/src/linux
make V=1  CC=cc LD=ld ARCH=i386 -C /usr/src/linux SUBDIRS=`pwd` modules
make[1]: Entering directory `/usr/src/linux-2.6.15-gentoo-r5'
mkdir -p /root/shasta/.tmp_versions
make -f scripts/Makefile.build obj=/root/shasta
 cc -Wp,-MD,/root/shasta/.shasta.o.d  -nostdinc -isystem /usr/lib/gcc/x86_64-pc-linux-gnu/3.4.5/include -D__KERNEL__ -Iinclude  -include include/linux/autoconf.h -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -ffreestanding -Os	 -fomit-frame-pointer -pipe -msoft-float  -fno-unit-at-a-time -march=k8 -Iinclude/asm-i386/mach-default -Wdeclaration-after-statement  -D_LINUXDRIVER -D_MMMAP_ -DMOONFANG -D_X8632B -D_32BPLATFORM -I/usr/src/linux/include -I/usr/src/linux/drivers/scsi -I/usr/src/linux/include -I/usr/src/linux/drivers/scsi	  -DMODULE -DKBUILD_BASENAME=shasta -DKBUILD_MODNAME=shasta -c -o /root/shasta/shasta.o /root/shasta/shasta.c  Building modules, stage 2.
make -rR -f /usr/src/linux-2.6.15-gentoo-r5/scripts/Makefile.modpost
 scripts/mod/modpost   -i /usr/src/linux-2.6.15-gentoo-r5/Module.symvers vmlinux /root/shasta/shasta.o
 cc -Wp,-MD,/root/shasta/.shasta.mod.o.d  -nostdinc -isystem /usr/lib/gcc/x86_64-pc-linux-gnu/3.4.5/include -D__KERNEL__ -Iinclude  -include include/linux/autoconf.h -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -ffreestanding -Os	 -fomit-frame-pointer -pipe -msoft-float  -fno-unit-at-a-time -march=k8 -Iinclude/asm-i386/mach-default -Wdeclaration-after-statement	 -DKBUILD_BASENAME=shasta -DKBUILD_MODNAME=shasta -DMODULE -c -o /root/shasta/shasta.mod.o /root/shasta/shasta.mod.c
 ld -m elf_i386 -r -o /root/shasta/shasta.ko /root/shasta/shasta.o /root/shasta/shasta.mod.o
ld: Relocatable linking with relocations from format elf64-x86-64 (/root/shasta/shasta.o) to format elf32-i386 (/root/shasta/shasta.ko) is not supported
make[2]: *** [/root/shasta/shasta.ko] Fehler 1
make[1]: *** [modules] Fehler 2
make[1]: Leaving directory `/usr/src/linux-2.6.15-gentoo-r5'
make: *** [default] Fehler 2

Seems like it's not recognizing the 64 bit environment? Any ideas anyone?

Share this post


Link to post
Share on other sites

My card runs a little hot here, but thats not really suprising with 32 degress roomtemperature and around 42-45 degrees case temperature, I will mount a little 40mm fand on it which I can control.

But thats not the reason I am posting here, I have some problems with the Webpam:

I am running Gentoo on a 64bit environment and somehow I can only get Webpam to work right after installation :blink: If I reboot and try to activate it again I have no luck and just get some error messages.

Has anyone figured out how to run Webpam in 64bit linux?

Share this post


Link to post
Share on other sites

Could a few more of you guys please post some write benchie for this card? (raid 5?)

i have been looking into it and for $200 less than the areca, it will do just fine as a home file storage space.

Thanks!

Share this post


Link to post
Share on other sites

I haven't run any benches lately but I also use the 8350 for home file serving and I haven't been able to stress it in any way. Copying from a raid0 to a raid5 seems limited by the read-speed of the raid0 so I'm pretty happy :). Even small operations are very fast, stuff like finding 3000 files in a 70000 file archive and then deleting them is incredibly fast.

Note:

Make sure to check your settings, some slight changes between write-back cache or not and NCQ or not makes huge impact on performance.

Share this post


Link to post
Share on other sites

Enhance Technology at www.enhance-tech.com supports RAID 6. They also have a proven solution with a PCIe card via SATA cards that enables you to get 2GB Fibrechannel speeds at eSATA prices. Much more cost effective to Promise or any other that I know of.

It is official. Promise's PCIe RAID controller. http://www.promise.com/product/product_det...&product_id=156

What do you think? Will it work on SLI board (2x PCIe x8) althow it is PCIe x4 card.

I'am thinking AMD X2 4400+, decent SLI board (ASUS), 64bit Linux, 2TB of disk space.

by

TheR

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now