Intel x25-V

[jalyst]

I'm trying to decide between 2x X25-V, 1x X25-M 80GB, or 1x OWC Mercury 50GB (unlikely)
I'm almost settled on 2x X25-V's, but I'm having a hard time justifying their slightly higher cost.

It's hard to explain my workload in detail at this stage, all I know is it'll house....
*Win7 & some multimedia related apps
*Stripped-down Ubuntu + MythTV + XBMC/Boxee + maybe LXDE & NAS related s'ware.

It'll primarily be a media playback, media capture/transcode, & storage/bu device. i.e. HTPC/PVR & to a lesser extent, NAS.
There'll be at least one 1TB 7.2k drive for storage, & this is prolly also where captured DVB will be dumped.
LT I have plans for a dedicated NAS device with a more sophisticated array/config of HDD's.

I wonder how 2x X25-V performance compares "across the board" to these two drives & the Corsair F100? (100GB SF-1200)
Do the Anandtech bench tools allow one to see all this, or are there supplementary resources?*
And I wonder how their performance differences compare to their price differences??

I'm concerned my usage pattern will be random enough to accelerate degradation of 2x X25-V's in RAID-0.
Hence requiring me to set aside more space than I can spare to mitigate it...
To be safe, I'm pretty sure I'll need 60GB for both OS's, their apps, page file etc, but I'd be surprised if I need more than that.
Will I notice enough of an advantage in my workload to justify their higher cost & potentially higher rate of degradation/wear?

*Anand's X25-V RAID review seems to include Mercury 50GB & the X25-M in all the benches, but no Corsair F100.
*I'll need to analyse this data more carefully soon to help me towards a decision....

**update**
So GullLars has recommended 1x X25-V, 2x Kingston V+ G2 64GB, + my storage disks (which initially will only be 1 or 2 1TB HDD)
The idea being that the X25-V is the OS-SSD, & the RAID-0 of Kingston's is my scratch-disk with high sustained r/w.
Does everyone else agree with this basic concept (seems pretty sound to me), perhaps you agree with the topology but not the disks picked?

[SteveRo]

Makes sense to me.

[GullLars]

Maybe it's a bit off-topic, but i heard you lost the PCmark Vantage WR to Corsair Labs, SteveRo.
They apparently needed a new 8-core Xenon @ 5Ghz and 11 Corsair F200 (sandforce 200GB) SSDs from an Areca 1680. I'm a bit confused why they didn't use 12...
Anyways, funny to know you are competitive with the storage labs of major companies :P

Do you plan to get an Areca of the new 6Gbps series when it comes out and reclaim the throne with a bit of L2N (or L2H)?
Maybe you could contact Fusion-IO and ask if they would lend you 2x ioDriveDuo 320GB (highest performance point), or an ioDrive Octal as a PR stunt. Btw, i forgot if it's possible to run PCmark on non-bootable storage... If not, that won't be an option

[GullLars]

Earlier i had just skimmed page 4 and 5, i've now read this thread in it's entirity and have downloaded SteveRo's IOmeter result files "stevero_x25v_1231ML4G" and "(stevero_)12x_R0_acard_9010_1231ML4G"
BTW, i'm fairly impressed with 12x your Acard numbers Stevero, 12K IOPS @ 4KB random read QD 1, and over 40K IOPS @ 4KB random read QD 4. The party stopper is your IOPS roof at 50-55K IOPS. As far as i can tell, this clearly comes from the 1231ML4G.
Here is a bit of quick math to support the statement:
You have 12x Acard connections, wich comes from 6 devices. Even if each device don't support NCQ, you should at least get 6x the IOPS @ QD1 performance, and hopefully 12x IOPS @ QD1.
4KB IOPS @ QD1 = 12K
4KB IOPS @ QD4 = 42-44K, so almost perfect scaling, as is to be expected from statistical distribution and controller overhead.
4KB IOPS @ QD8 = 50K, clearly hitting some kind of either diminishing returns or bottleneck.
4KB IOPS max is at QD 16 = 55K, and stays there up to QD 128 (wich is max meassured). This indicates you saturated the RAID controllers (none-cached) IO performance.
To support this, max IOPS meassured at 512B, 1KB, and 2KB were also ~55K.

I think i will make the new thread Anvil mentioned in february, and that i never got around to, either later tonight or tomorrow. I will post a link here when i have done so.
For starters, it will be about gathering IOmeter data from as many setups as possible for 0,5KB-64KB random read (and write for those who want to), and I will filter and analyze it, and then post comparative graphs for interresting parts and more general 3D diagrams of (f.ex.) x=Block size, y=QD, z=IOPS (or bandwidth). I already have data for about 10-15 such setups, and will likely start off the thread with 2 posts, the second showing some graphs and diagrams, and possibly tables.
I hope many of you extreme users will contribute with data, as it will not only help comparing SSDs, and scaling of RAID, but also comparing controllers (RAID and HBA), and possibly drivers.

[SteveRo]

GullLars - much thanks for the data assessment - yes, I realized when I generated/reviewed the data myself that I was controller bound.
I will replace the 1231ML at some point - will it be an areca 1880 or something else - hard to say yet

[GullLars]

If you are willing to sacrafice the partition being bootable, you could go for pass-through on an LSI 9211-8i, although that would only give you 8 ports and have some CPU overhead, you will be free to scale to 300K IOPS, and no bandwidth limitations using SATA 3Gbps units.
I don't remember if i still have your benchmark data for a single Acard with single and dualport, i looked now but couldn't find it. If you could run a single Acard in both single and dual port mode from ICH10R with the same config, it would be really helpfull for estimating scaling, and would make a good addition to data and graphs i already have.

BTW, the only columns i use is; Block Size, Queue Depth, IOPS, MB/s, average accesstime, and max accesstime. It takes me about 2-3 minutes to filter each csv file, and a bit more when you have 8 (logical) processors since i also have to filter the rows.

[jalyst]

@GullLars
Have you ever considered offering to do some review work for Anand?
You could really contribute to "beefing them up" a bit more!

[GullLars]

I have been offered to write for 3 tech-news sites, and been offered to write for a couple of independent sites, but i'm a university student, so there's a time issue, and I live in Norway, so there's a geographical issue as well for doing reviews of units.
However, I have contributed to major tech sites and exchanged mails with editors and reviewers. The review of 2R0 x25-V Anand recently did i had been nagging him to do for a couple of months now in every SSD article he wrote, and i've also been campaigning for a QD 32 part in random read/writes, since he only tests at QD 3, and while it may be representative of some real life usage, it is highly misrepresentative of the avalible performance from the SSDs. Not to mention SATA NCQ spec is 32 (or in practice 31 + one ongoing).
I was also responsible for getting Tweaktown to switch from CDM 2.2 to 3.0 TP, and getting them to use AS SSD. I am now nagging Chris Ramseyer (head of testing) to start listing PCmark Vantage total HDD score, and AS SSD total score. And I've been exchanging mails with him regarding a larger test of 1-4R0 low capasity SSDs soon, wich i hope gets made.
I was also in contact with Toms Hardware about a year ago, but i don't remember regarding what... (EDIT: I think the editor wanted me to do reviews for the european branch after a tester forwarded a mail i sent them...)
I've also gotten 4-5 sites to start using AHCI/RAID mode instead of IDE when testing SSDs, it was mostly in early 2009.

Oh, yeah, hardware.no posted an article in january 2009 linking to the SSD thread on the norwegian forum diskusjon.no that i'd started dec 2008 giving me cred (diskusjon.no is where I know , Anvil, Nizzen, and Ourasi from)
But that's enough about me, just thought i'd list a few things to show i've contributed

EDIT2: BTW, it just hit me what could be really interresting to test for us "enthusiasts"/"powerusers": What setup for $1000 gets you the most performance for *General usage (OS+apps), *Poweruser (OS+apps), *Scratch-disk (workspace and highly sequential-bound things).

[jalyst]

Oh I never doubted your levels of contribution towards the enthusiast community.
I just figured it could be better leveraged via a big tech site like Anandtech etc.
But you've cited some good reasons why that might be difficult...
Great to have chaps like you & other helpful/inquisitive users around anyway!

[SteveRo]

btw im in the mood of buying this

The SC199 Cheetah Y-Series has R/W speeds up to 290/280 MB/s in 2.5" and 3.5" SATA form factors - which approaches the theoretical speed limit of the SATA-II protocol. It also delivers impressive R/W IOPS of up to 50,000/45,000 respectively.

but i cant find a store lol?????

50k/40k 4KB iops?

[Tiltevros]

1k=1.000 so 50k=50.000 read / 40k=40.000 write IO/s

http://www.storagesearch.com/foremay.html

[GullLars]

Yes, 50K read 40K write 4KB random IOPS. I've seen these numbers elsewhere also.
Here's a page with a bit more info on these drives, and other SSDs from foremay:
http://www.storagesearch.com/foremay.html
I'm not surprised you can't find them in e-tail, as they are enterprise drives.

BTW, OCZ has released Z-drive R2 now, and it's bootable
http://www.ocztechnology.com/product...press-ssd.html
512GB with 1,3GB/s read, 550MB/s write.
1-2TB with 1,4GB/s read, 950MB/s write.
4KB random IOPS is listed as 29.000 read and 7.200 write in the product sheet PDF.
The editor at storagesearch.com comments:

R/W IOPS are 29,000 and 7,200 respectively - an order of magnitude slower than the fastest PCIe SSDs today - but nevertheless useful for many applications - unlike the original Z-Drive (March 2009) which so slow that it couldn't be regarded as a serious contender.

If i remember correctly, this is based on an LSI 92xx-8i. I am not sure wich SSD controllers are used in the RAID array, or the stripe size. Cache is listed as 512MB, but if that's the RAID controller cache, or the sum of SSD caches i don't know either.

[DragoonXX]

ok, I have a few questions:
-I've got a Windows XP pro system (don't have money right now to get 7 x64) and Just got 2 X-25V drives, and have them in a raid 0 array (128k stripe, diskpart align=1024 offset- like one of about...a half billion different configs)
latest intel RST reads one drive as internal, the other as external (asus P5Q pro board) my numbers seem to be similar to others (I'll have to do a bench run after my current render is done..so..about 2 more days, lol)

but what can people reccomend for things for me to try to change (other than upgrading to win7)?
-I had a huge pain getting an nlited install to work properly, once im done school for the summer I'll be able to have some time to actually tweak out the pre-install
what I've done so far:
-I've disabled defrag (actually removed it from the install environment), moved pagefile off to my HDD, disabled indexing, prefetch, etc
-updated firmware on both drives (newest firmware- did it first thing)
-BIOS is set to RAID, although, my BIOS revision is relatively old now, I'll upgrade it later as well
-set it to a ~56GB partition in the 128k stripe RAID0

any and all help would be appreciated.
and again, I'll post benchmarks (as the drives stand now, once the current render is complete- I'm doing a test to see if the SSDs do make any real difference on render times in REVIT)

[mbreslin]

I just got a x25v to have something else with an os for testing and such, even the -v is a fast little sucker!

[DragoonXX]

Yeah, I agree, they are speedy buggers, I got mine on sale for 100 bucks each

[SteveRo]

Playing around with a Frankenstien Raid (4x R0 on areca 1231ML4g - 2 drives x25-v and 2 drives Kingston 40's cross flashed to intel's f/w) - 128k stripe -

[Biker]

<snip>

Other things you could do:

Relocate your browser(s) cache to HDD.

Relocate My documents do HDD.

Thats all really. You have done everything else

Playing around with a Frankenstien Raid (4x R0 on areca 1231ML4g - 2 drives x25-v and 2 drives Kingston 40's cross flashed to intel's f/w)

[DragoonXX]

forgot to mention that I did put the FF cache on HDD...my docs to HDD...I don't often save stuff to it, thats what the storage drive is for (2 SSDs, 2 HDDs, 2 ODDs, lol)

[GullLars]

SteveRo, what is your PCmark Vantage HDD total score with those 4? 56.000 like the ORB window in the back says?
And have you tried them on ICH10R? You may get higher scores there.
And once again, your 1231ML seems to top out around 60K IOPS. 4x x25-V should be able to pass 100K, maybe even 120K.

I think Anvil is tinkering with stripe sizes on ICH10R from 4K-128K and PCMV HDD test, maybe we'll se some tables later.

[jalyst]

I'll have Win7 X64 installed with just one X25-V for now...
So no need to bother with this sort of stuff as it has TRIM right?

Thank-you.

what can people reccomend for things for me to try to change (other than upgrading to win7)?
what I've done so far:
-I've disabled defrag (actually removed it from the install environment),
-moved pagefile off to my HDD,
-disabled indexing, prefetch, etc

Relocate your browser(s) cache to HDD.
Relocate My documents do HDD.

[SteveRo]

SteveRo, what is your PCmark Vantage HDD total score with those 4? 56.000 like the ORB window in the back says?
And have you tried them on ICH10R? You may get higher scores there.
And once again, your 1231ML seems to top out around 60K IOPS. 4x x25-V should be able to pass 100K, maybe even 120K.

I think Anvil is tinkering with stripe sizes on ICH10R from 4K-128K and PCMV HDD test, maybe we'll se some tables later.

Good morning,
Yes - when I did the 1231-9211-9260 compare back in December we saw that the ich10 was better than all three in the pcmv HDD test.
Back then, it was interesting that the pcmv suite test shuffled the deck again (vs. the HDD test) making the best from first to last - 1231, 9211, ich10 and the 9260.
But that was with the 1231ml-2g not 4g (2 vs 4GB cache) and that was with early LSI f/w.
The pecking order might be much different if I reran that test today.
I should be able to get an indication of that - relative to the LSI anyway - I have a 9211 on the way, much thanks to Paul.

Question for someone with the LSI 9211 and new f/w - Does the new 9211 f/w allow stripe size changes?
I think the acards usually like smaller stripes for higher performance.

[DragoonXX]

this is my 2x r0 X-25V 128k stripe



oddly, intel's RST shows one drive as external, one as internal

[SteveRo]

Dragoon - very nice! I assume intel ich? ich10?
Try 16k stripe. Is write back cache off?

[jalyst]

Tempting me to buy a 2nd...
But 1st I want to get 2 of the cheapest SSD's I can find that have much better seq. r/w.
To be used in a RAID-0 as a scratch-disk

[DragoonXX]

Dragoon - very nice! I assume intel ich? ich10?
Try 16k stripe. Is write back cache off?

ICH 10 (P5Q pro), write back is on, never noticed any differences between it on and off (edit, noticing lower performance with it off, benches with write-back off below)
and I will try 16k stripe after exams and the week of surveying we've gotta do (and when I update the bios)