All I saw in the nvidia RAID BIOS was stripe size. Oh well, seems to be running great anyway.Quote:
Originally Posted by thinkingbear
Printable View
All I saw in the nvidia RAID BIOS was stripe size. Oh well, seems to be running great anyway.Quote:
Originally Posted by thinkingbear
If you read this thread it is already answered also, yes stripe size is in bios, but clusters is in os, several methods for selecting cluster size are mentioned. Mabe if I get some spare time today, or if someone else feels generous we could list them together.
Someone mentioned using win xp x64
Someone mentioned partition magic
Someone mentioned using an already partitioned and formatted disk made by existing os
that was from memory, did I miss any?
My bad, got stripe and cluster size mixed up, sorry.Quote:
Originally Posted by Kilyin
Decided to try some new drives... Western Digital Caviar SE16 250GB SATA II HDD's - RAID 0 32k stripe.
Really bad burst speeds. Latency is okay i guess...just hoping I can improve it somehow. This is no way is better than Hitachi's which i had previously...
These drives aren't so quiet...
There's no tools to change audible seek modes.
Has 16 MB's of cache, but no real difference in performance. Kinda disappointed...
Anyone using these also?
Have you tried IBM/Hitachi feature toolQuote:
Originally Posted by Reinvented
http://www.techspot.com/downloads/53...ture-tool.html
or Maxtor's AMSET.
I have used the feature tool on drives of manufacturers other than IBM/Hitachi
I'm not gonna risk if it's especially not warranted or even mentioned as doable.
have you tried doing actual file transfers?Quote:
Originally Posted by Reinvented
Yeah...still pretty crappy.Quote:
Originally Posted by safan80
Did u run benchmarks one your other drives before u installed and setup the WD's? Mind showing us?
32k Stripe size, but what cluster size?
How many drives in the array?
Are they all useing the same firmware?
What controller are you useing?
What drivers are u useing with the controller?
Is the OS installed on the array or seperate drive?
and lastly HDtach is the last program i would use to get relieable benchmarks with.
Quote:
Originally Posted by Reinvented
sound, you didn't install the nvidia ide sw driver did you?Quote:
Originally Posted by soundx98
Yes, I did
Disabled read caching, however burst speed went up, just not as high as yours even with 2 drives. Still got better on Hitachi's...except for latency.Quote:
Originally Posted by soundx98
On the WD drives, latency increased 3ms, and read went down to 99. :(
As Delerious suggested. try some other banchmarks as well.
Also remember that with a larger array (500G in your case, 750G) in mind that size will also affect Average read. Run your HD Tach again (long run)./
I'm using a 4 partitions on my array.
1st - 50G XP32 OS
2nd - 50G x63 Edition OS
3rd - 150G Games
4th - 500G general storage
The ist part of the array (or drive is always) the fastest.
Your average read on the 1st party of your array looks like it's averaging 120-125. Easily twice as fast as any single drive.
Again, these ARE synthetic benchmarks.
And, like a futuremark score, they are seldom repeatable.
Try the latest ATTO (need to register) as well.
My HD tach scores are dead on. I can run them one day save them and run them another and re-load saved and not be able to distinguish them. They are exactly the same.Quote:
Originally Posted by soundx98
That's amazing.
I just ran them three times in a row and something always changes slightly.
While the burst graph looks almost identical, there is always a slight difference in random access, cpu utilzation, or average read.
I thought it was caused by the random nature of the reads.
I stand corrected, but I did say "seldom" :)
How about ATTO - same identical readings?
my burst and cpu util will vary slightly, but the ra time and graph are identical every time
that was my point, all the readings will seldom be identical. :)
Areca 1230 256MB
4 x 16MB 74 Raptor
TCQ/NCQ + read/write cache enabled
32K/32K
http://www.xtremesystems.org/forums/...1&d=1149628351
6 x 16MB 74 Raptor
32K/32K
http://www.xtremesystems.org/forums/...1&d=1149628418
:slap: i expected better scores @ 6 rappies more like 450+MB/s.. ATTO scores are far from pretty.. slightly better than 4 rappies.. was gonna go for 1GB cache but am waiting on 2GB modules if ever available..
anyone know where i can download winbench99? i just cant find a link..
-
Where do you live? :clap:
lol.. seattle region
NapalmV5
I'm sure 8 raptors would do much better..
http://tweakers.net/reviews/557/24
the areca cards seem to really pick up speed with 8 drives.
the areca cards only support a max of 1GB cache.
try this program
http://nodesoft.com/DiskBench/Default.aspx
copy file with stopwatch, write benchs, batch file creations, actually measures speeds, not synthetic... great way to see differences between RAID10 and RAID5 and also different stripe sizes
tnx! for the link.. sure will read it..Quote:
Originally Posted by safan80
actually it does support more but only ~1GB module is available.. it should support 2GB, i will ask Tony from support on this..
cool.. I'll do that after I finish converting my array to GPTQuote:
Originally Posted by Hassan
http://www.microsoft.com/whdc/device...e/GPT_FAQ.mspx
funny quote from the faq
Quote:
53. Can a disk be converted from GPT to MBR, and vice versa?
Yes, but only if the disk contains no partitions or volumes. Any data on the disk will be destroyed.