Areca 1880

[Computurd]

modify volume setting. i believe it is on by default though, but not sure.

[SteveRo]

wow see your busting out the q6600! wow gotta love those ol beasts, i see you have a penchant for old hardware lately, and seems to suit you nicely. Good Job!

EDIT: if they gave points for PCMV you would be crushing

thanks - also - did you see that futuremark is no longer going to support old benchmarks - looks like this might include all the old 3dmarks.

http://futuremark.yougamers.com/foru....php?p=1463773

[Computurd]

wow that is crazy, i cant believe they are retiring them...but i guess progress moves on. i am wondering if they will release a new PCMark with the new 3DMark. isnt that usually how it works? wouldnt we need a DX11 pcmark as well?\

there are gonna be some sad overclockers with these changes. looks like submission to the bot will have to be with screenshots only!

[Computurd]

areca on left, lsi on right

[[XC] gomeler]

The 1880ix has the LSI2208 ROC, correct? If so that's a massive improvement in the controller capabilities

[Computurd]

well, mixed bag there. as you see the areca is tuned for low QD performance with its storport drivers. they are just fantastic they do not scale as high as the 9260 though, which is tuned for high qd loads. of course the lower qd load is what is more beneficial to us, so excellent job from areca there
the arefcca handles differently with the scsiport drivers. l will do some comparisons and post those results to compare to these...

[Tiltevros]

ARECA is here
controller check
1GB cache check
4GB cache check
installing the card pending.....

[Tiltevros]

the latest firmware was @ 19/9/2010?? for the 1880?

[[XC] gomeler]

well, mixed bag there. as you see the areca is tuned for low QD performance with its storport drivers. they are just fantastic they do not scale as high as the 9260 though, which is tuned for high qd loads. of course the lower qd load is what is more beneficial to us, so excellent job from areca there
the arefcca handles differently with the scsiport drivers. l will do some comparisons and post those results to compare to these...

Yup, that's what I was commenting on. Only at work do I hit queue depths that the 9260 would shine in. That being said I haven't measured the queue depth of PCM05/Vantage yet. It's on my immensely long list of things to do

[tiro_uspsss]

I'm a n00b.. & thus confused.. so I have a bunch of questions:

which is faster: 1880 or 1680? I know the 1880 is newer.. however the 1880 has a single core (yes?) 800mhz XOR.. the 1680 has a dual core 1200mhz XOR.. is one better than the other outright? or has each an advantage over the other? i recall when the 1680 came out, the dual core XOR *significantly reduced RAID rebuild times.. if DC helps in this, then why is the newer 1880 series single core?

thanks for all the help

[Nizzen]

I have 1680ix-12 2gb and 1880ix-24 4gb. 1880ix is faster on everything. Raid-5 and 6 , build time rebuild time, lower latency! and sata3/sas2

But 1680ix still kicking ass

[tiro_uspsss]

I have 1680ix-12 2gb and 1880ix-24 4gb. 1880ix is faster on everything. Raid-5 and 6 , build time rebuild time, lower latency! and sata3/sas2

But 1680ix still kicking ass

thanks!

I wonder how it does that considering the discrepancy in cores & mhz..

[Computurd]

much more powerful IOP, better design for latency, and also, the fast that it is based upon a new spec might help it a bit. remember, a celeron with a dual core would probably be slower than a nehalem with a single core just a newer more powerful proc.

[tiro_uspsss]

much more powerful IOP, better design for latency, and also, the fast that it is based upon a new spec might help it a bit. remember, a celeron with a dual core would probably be slower than a nehalem with a single core just a newer more powerful proc.

ah-ha! nice! thanks!

[Tiltevros]

can anyone make me a profile for IOmeter to run it on ARC-1880 with 4 GB's i think i found something unique (120.000 IO's on 4k random )

[Nizzen]

can anyone make me a profile for IOmeter to run it on ARC-1880 with 4 GB's i think i found something unique (120.000 IO's on 4k random )

I got 140672 IOPS with QD 16 in Iometer :p

Rename to .rar

This is the testfile I used

[Tiltevros]

I got 140672 IOPS with QD 16 in Iometer :p

Rename to .rar

This is the testfile I used

why do i have that kind of latency in AS SSD???
in all the others i have 0.11ms

[Tiltevros]

ok here we go Nizzen i did ur test file as u have it I'm hitting 130.000 IO's 4k random file


just delete the .txt extension to view it
if someone can make it propper xls file it will be wonderfull to make a graf

[Computurd]

hmm you are getting bad latency cause you need vertex!! hiding yesterday?
tell us of your PCMV results...i know your busy you greek (you-know-what)

[odditory]

so what is the cap limit of this controller??

If you're talking read/write, then according to LSI at least peaks for the SAS2108 RoC are 2875MB/s read and 1800MB/s write.

The 1880ix has the LSI2208 ROC, correct? If so that's a massive improvement in the controller capabilities

Nope, SAS2108. Wish I'd seen page 1 of this thread before I removed my heatsink a few weeks ago!

SAS2208 based cards in 2011 will very well be monsters though especially for parity calc on account of dual core 800Mhz PPC and a claimed 400,000 max IOPS for raid 5/6/10. Whether they'll be capable of higher peak in JBOD/Raid0 remains to be seen. PCIe 3.0 as well.

@Computurd: nice thread, btw. it'll be interesting to see if Areca cooks something up for SSD's ala FastPath but I have my doubts. They've hinted at it in emails back and forth so we'll see.

[Computurd]

SAS2208 based cards in 2011 will very well be monsters though especially for parity calc on account of dual core 800Mhz PPC and a claimed 400,000 max IOPS for raid 5/6/10. Whether they'll be capable of higher peak in JBOD/Raid0 remains to be seen. PCIe 3.0 as well.

yup. will be a very exciting year, finally some controllers that will handle the load
interesting to see the low QD and how it scales...

[Tiltevros]

inside info the SAS2208 will be an HBA with 16-32MB cache so dont be so happy about it

[Computurd]

well yea but there will be many different flavors. the 2108 is on HBA and raid cards...

[Tiltevros]

http://service.futuremark.com/compare?pcmv=379778

HDD Test Suite 131381

HDD Test Suite
HDD 1
HDD - Windows Defender 680.09 MB/s
HDD 2
HDD - gaming 492.41 MB/s
HDD 3
HDD - importing pictures to Windows Photo Gallery 707.92 MB/s
HDD 4
HDD - Windows Vista startup 513.55 MB/s
HDD 5
HDD - video editing using Windows Movie Maker 595.41 MB/s
HDD 6
HDD - Windows Media Center 1230.82 MB/s
HDD 7
HDD - adding music to Windows Media Player 377.17 MB/s
HDD 8
HDD - application loading 585.37 MB/s

[odditory]

well CT i think you've convinced me to dump one of my 1880i's and grab an 1880ix-12 w/ 4GB module in its place. main concern about the ix's was the integrated expander, given that the one employed on the 1680ix's was fairly problematic and a lot of drives didn't work with it, but this new one is solid and implementation is solid (based on some research and Q&A w/ areca for an upcoming review), its multipathed to the RoC so all 8 lanes are communicating at 6Gbps per. OT, i think I finally found the first flaw with my beloved HP SAS Expanders and that's that they only negotiates @ 3Gbps to SATA-III drives, whereas the integrated LSI expander on the 1880ix *does* negotiate a 6Gbps link to SATA-III drives. i'll have to confirm once I get it, its only a theory based on various benches of C300's I've seen.

anyway I guess there's still no consensus on best bet for 4GB module - based on everything I've read (here and elsewhere) it seems the parity chip(s) are pivotal in whether it works with the 1880ix or not. when you look at similarly spec'd modules, the presence or absence of the parity chip also seems the dividing line between whether its a $125 or $450 part. i know a guy at kingston, i might ask him what it would take for a small production run on a discontinued SKU. i'm 99.9% sure the news won't be good though.