while im waiting to see if crossflashing and or an obnixous 80$ LSI cable will let me do NVMe raid1 on this card ("Tri-mode") , ie the saga over at this thread:
https://forums.servethehome.com/ind...3408-got-myself-a-530-8i-on-ebay.21588/page-4.
i decided to run some normal ssd/sas raid benchmarks for my notes, and post them here too.
some notes:
im running the latest LENOVO FW (jan 2019 release), not LSI fw, but i dont think that should matter much for these tests.
+ This card (9440-8i) has NO cache, so keep that i mind for RAID 5/6 performance
+ system is baremetal, x10 SRL-F , e5 2620 v3 , 64g d4 ecc
+ SSDs ARE CAPABLE OF SAS3 12g, (and card showed link at sas3 12g, but bc of the TQ backplane, i had to lock the cards link rate at sas2 6g, via LSA gui. I tried running at sas3 but had nonstop issues, as expected)
+so far i have NOT been able to pull SMART data from the cards behind this adaptor, ive only tried with windows smartmon tools , and w latest version of HardDisk Sentinel (both of which normally pull smart data from drives behind LSI cards, maybe it bc of the lenovo OEM fw?) (not good!)
lenovo spec PDF: https://lenovopress.com/lp0651.pdf
lsi spec pdf: https://cdn.competec.ch/documents/5/8/587385/DE_Datenblatt_SAS-9460-8i.pdf
drives:
i have/using 6x of this SSD (nice!): 200g -HUSMM812CLAR200 (PDF specs - https://documents.westerndigital.co...-sas-series/data-sheet-ultrastar-ssd800mm.pdf)

and using 1x of this SSD (nicer!): 200g - -HUSMM1620ASS204 (PDF specs -
https://documents.westerndigital.co...sas-series/data-sheet-ultrastar-ssd1600mm.pdf

a question im curious about and hope others can answer- Seeing all the specs above, and my test results below, do you think the results would have been different if i had SAS3 Links / DA backplanes ?
I normally do FIO benches too, but didnt have time this time (and i usually prefer to run FIO on linux not win).
---
Not so good that raid 50 performance is worse than raid5. may show that card/card cpu is being maxed (which it really shouldnt be at this late generation LSI card)
also , please correct me if im wrong, but i would have expected to see a bit more performance out of all these benchmarks, considering the drives used (esp the RAID 5, even with no cache onboard). But maybe im wrong, as i think with no cache onboard, these are essentially SYNC / Write Through writes? (ie not async nor write back), so maybe then they are good?
im about to swap it with a Adaptec 71645 to compare speeds (even though that adpatec is one gen behind this LSI, so i should be using a adapatec 8805, but that card is being used currently)





https://forums.servethehome.com/ind...3408-got-myself-a-530-8i-on-ebay.21588/page-4.
i decided to run some normal ssd/sas raid benchmarks for my notes, and post them here too.
some notes:
im running the latest LENOVO FW (jan 2019 release), not LSI fw, but i dont think that should matter much for these tests.
+ This card (9440-8i) has NO cache, so keep that i mind for RAID 5/6 performance
+ system is baremetal, x10 SRL-F , e5 2620 v3 , 64g d4 ecc
+ SSDs ARE CAPABLE OF SAS3 12g, (and card showed link at sas3 12g, but bc of the TQ backplane, i had to lock the cards link rate at sas2 6g, via LSA gui. I tried running at sas3 but had nonstop issues, as expected)
+so far i have NOT been able to pull SMART data from the cards behind this adaptor, ive only tried with windows smartmon tools , and w latest version of HardDisk Sentinel (both of which normally pull smart data from drives behind LSI cards, maybe it bc of the lenovo OEM fw?) (not good!)
lenovo spec PDF: https://lenovopress.com/lp0651.pdf
lsi spec pdf: https://cdn.competec.ch/documents/5/8/587385/DE_Datenblatt_SAS-9460-8i.pdf
drives:
i have/using 6x of this SSD (nice!): 200g -HUSMM812CLAR200 (PDF specs - https://documents.westerndigital.co...-sas-series/data-sheet-ultrastar-ssd800mm.pdf)

and using 1x of this SSD (nicer!): 200g - -HUSMM1620ASS204 (PDF specs -
https://documents.westerndigital.co...sas-series/data-sheet-ultrastar-ssd1600mm.pdf

a question im curious about and hope others can answer- Seeing all the specs above, and my test results below, do you think the results would have been different if i had SAS3 Links / DA backplanes ?
I normally do FIO benches too, but didnt have time this time (and i usually prefer to run FIO on linux not win).
---
Not so good that raid 50 performance is worse than raid5. may show that card/card cpu is being maxed (which it really shouldnt be at this late generation LSI card)
also , please correct me if im wrong, but i would have expected to see a bit more performance out of all these benchmarks, considering the drives used (esp the RAID 5, even with no cache onboard). But maybe im wrong, as i think with no cache onboard, these are essentially SYNC / Write Through writes? (ie not async nor write back), so maybe then they are good?
im about to swap it with a Adaptec 71645 to compare speeds (even though that adpatec is one gen behind this LSI, so i should be using a adapatec 8805, but that card is being used currently)







Last edited: