Areca 1880

well this sucks, gonna have to pull up lame for a bit with regards to my quest for the #4 spot...my hardware isnt working as well with only 6 drives, the thing about these gen 1 drives is that i have to have tons of them to get any real justice out of them...i will see what this weeks IDF has to say about the G3 specs this week before i pull the trigger on something...not gonna spend 300 bucks on 3 new vertex gen 1 though, i guarantee that :)

this week will be busy digesting IDF info for sure though, and maybe some downtime to do something i haven't done in a while...game! i will do some gaming comparison graphs though..

Quote:

---

Originally Posted by SteveRo

@ Mr Anvil - just noticed you uped your pcmv score by about 2500 pnts!
Very nice work - over 30K!

---

I've got a dozen scores over 30K, all very consistent.
The top score was using 7 Vertex 2 60GB/100GB drives.
I've practically got the same score using 4 Vertex 2 100GB (mix of LE/V2)

Meaning, the number of drives does not seem to matter, it's controller bound. (that is using the SF drives, the C300s might prove otherwise)
I need the 4GB stick and start doing some OC to get more points :)

@CT
Are you sure you need that many drives, 7 should do OK.

well maybe...i guess it depends upon my configuration. however, these drives dont have the brute power of the SF, so it takes tons to get any good numbers. will make it cheaper to upgrade in the future though...

after some talking with the main office of ARC they send me this thing

with datatransfer-guessed/advised of 2016MB with RAID 0 and 1720 MT with RAID 5 (reading, 512K, 16 outstanding IOs)

so the 1880 has cap limit at 2GB/s Cool :)

Got the tracking # for my 4GB stick today :)

gah fun time for comp @ #5 position is almost over!
nice pic tilt...amazing that they still havent widely released the specifications yet, maybe they arent entirely sure yet of what it can do :)

Hello :D

awesome :) is it the kingston stick?

It's supposed to be the Unigen stick.

I'm almost ready to start the sqlio thread, I just need to create some references.

Here's the 1880 1G running 4R0 Vertex 2, testing cache only. (testfile 1GB)

edit:
Currently running the same tests on the 9260 with the FP key :), finally got it.

those vertex are looking great. do you think that 4 vertex2 would saturate the controller? surely more than 5 wouldnt be needed...
have a bbu inbound due today btw :)

awesome, you will love the FP!

Keep in mind that the above test is *cache only*, I'll do the same test using a 2-4GB file.

iops wise (@4K) they would saturate the controller, as for throughput, well, you know that answer :)

Current firmware gives about 114K iops, 4R0 Vertex 2 should reach ~120K iops. (if the controller scaled to 120K iops, that is)

As for Vertex 2 Vs C300, let's see how they turn out in the SQLIO thread.
There will be a lot more than just 4K iops in that thread.

edit:

I've done a few tests with the FP key and that little thing surely boosts iops to a new level :)
143K iops (4K) using 1GB testfile and 4R0 Intel G1 80GB SSDs.

BTW, price per kilogram for the FP key would be outrageous :)

Quote:

---

Originally Posted by Computurd

awesome :) is it the kingston stick?

---

It is the "areca" memory Unigen :)

I found some 4k IOPS hidden in the memory :ROTF:

@ Mr Nizzen - very nice numbers! :yepp::up:

LOL worth more than gold per ounce Anvil :)

A bit misleading as the Areca is all in-cache and the 9260FP works with a testfile thats 2x the size of the cache.
+
Areca = 4R0 Vertex 2 100/120GB
9260FP = 4R0 Intel G1 80GB

strange that the 9260 is outperforming it at higher QD even without using cache... can you give us your impressions of the fastpath performance, is there a notable difference between the 1880 and 9260 w/FP in normal usage to you?
I think they are both stellar, but in real world the 9260 might pull out ahead, with its more mature drivers. the 1880 of course benches like hell though! :)

anvil try using 8 drives ;) with FP LSI is still in the lead with high queue depths

Quote:

---

Originally Posted by Nizzen

Hello :D

---

hello to cache :)

@CT
I don't know yet, this is the first thing running off the FP key :)
I kind of expected this result though, the Areca is awesome at low QD and the LSI is awesome at high QD.

@Tilt
I'll do that :)
Maybe tomorrow, gotta try to finish the SQLIO thread and then you can bring on your own results where we can compare.

nizzen i have a question...is the 1880IX identifying the device as a sata 3 ? also would a single device run at 6gb/s speeds?


god i can taste number 4, im @ 32364.....GAH!!!!!:brick::cord::stick::cheer2:

Ha ha, finally got a tracking number from Areca! I should be seeing an 1880ix-12 soon. The tracking says it was shipped at 5pm yesterday in Taiwan but should be delivered... ... ...today before 10:30am? Well that's clearly incorrect. Changing the timezone view to by destination shows that it was picked up today at 6pm my time (?), in transit as of 9:30pm (?), and will be delivered tomorrow morning. That's still bizarre.

Anyway. Here's a fun bit about this particular card I will be receiving: it's a pre-final R&D card featuring a wire jump on the board itself and with no BBU support. Normally that might not be something to get excited about, but comparing the pre-final R&D copy with a final one might give us some insights about the process the card went through and how much consideration the engineers put into it (eg. was the wire soldered properly and tight to the board). I am told that I may/may not get to see a production sample as well once their spare bin is replenished, and until then comparisons will probably have to be made between my photos and someone else's.


Now I really need to give a hard think about what I'm going to use for drives in this thing. Do I pull the trigger on some C300's or wait for the new products due out this fall? With the lifetime this controller will probably see in the market I don't think waiting would necessarily be the worst idea, but sooner means I can have fun sooner. Hmmm. I really need to get in with someone who makes SSD's now.

Looking at the pcmv scores that CT has been pulling down with ancient vertex gen 1s and 920 (no 6 core nor improved instruction set) -
It looks like the 1880 with 4GB cache just screams with whatever you attach it to.
It might be fun to try 5xR0 of inexpensive SSDs, maybe 5x small SF drives or even the ancient x25-v 40s.

a

Quote:

---

Originally Posted by Serra

Ha ha, finally got a tracking number from Areca! I should be seeing an 1880ix-12 soon. The tracking says it was shipped at 5pm yesterday in Taiwan but should be delivered... ... ...today before 10:30am? Well that's clearly incorrect. Changing the timezone view to by destination shows that it was picked up today at 6pm my time (?), in transit as of 9:30pm (?), and will be delivered tomorrow morning. That's still bizarre.

Anyway. Here's a fun bit about this particular card I will be receiving: it's a pre-final R&D card featuring a wire jump on the board itself and with no BBU support. Normally that might not be something to get excited about, but comparing the pre-final R&D copy with a final one might give us some insights about the process the card went through and how much consideration the engineers put into it (eg. was the wire soldered properly and tight to the board). I am told that I may/may not get to see a production sample as well once their spare bin is replenished, and until then comparisons will probably have to be made between my photos and someone else's.


Now I really need to give a hard think about what I'm going to use for drives in this thing. Do I pull the trigger on some C300's or wait for the new products due out this fall? With the lifetime this controller will probably see in the market I don't think waiting would necessarily be the worst idea, but sooner means I can have fun sooner. Hmmm. I really need to get in with someone who makes SSD's now.

---

@ Mr Serra - looks like you could get 5x C300 64's from the egg for about $730.
I would imagine the resale on those should be pretty good if you don't keep them too long.