link do download
http://www.hdtune.com/download.html
my results
http://img354.imageshack.us/img354/3...lrailj6.th.png
two of these http://www.scan.co.uk/Products/Produ...oductID=469115
is this how it should be?
Printable View
link do download
http://www.hdtune.com/download.html
my results
http://img354.imageshack.us/img354/3...lrailj6.th.png
two of these http://www.scan.co.uk/Products/Produ...oductID=469115
is this how it should be?
Just having a defrag and Ill try it, although Ive only got a single raptor and not raid :)
mines about the same on a raid 0 sata2 setup, get the drop off around 70% as well..
3Ware 9500S-8 4x300G ST7200.9 RAID5
http://www.tsubasaria.com/3Waew-9500S8-RAID5.PNG
Western Digital 3200JD
http://www.tsubasaria.com/WD3200JD.PNG
Maxtor Diamondmax 10 300GB IDE
http://www.tsubasaria.com/6B300R0.PNG
Seagate 7200.8 300GB SATA1
http://www.tsubasaria.com/ST3300831AS.PNG
Seagate 7200.9 500GB SATA2
http://www.tsubasaria.com/ST3500841AS.PNG
(No, you don't see it wrong, IT IS slowerthan the old generation)
Clunk, defrag does not affect the result at all. Its not like ATTO and Sandra
HD Tune: SAMSUNG HD501LJ Benchmark
Transfer Rate Minimum : 21.9 MB/sec
Transfer Rate Maximum : 78.9 MB/sec
Transfer Rate Average : 56.4 MB/sec
Access Time : 14.9 ms
Burst Rate : 147.1 MB/sec
CPU Usage : 4.6%
2 x WD 74GB in RAID0 on the onboard Intel i965 controller.
http://www.thisisgav.co.uk/tmp/hdtune_hex.png
I have 2 WD 3200KS drives (not RAID. The D: Drive is OK but the C: Drive Windows boot disk is VERY spikey...... hmmmm need a fast replacement boot disk.. (both new last January....)
Seagate Barracuda 7200.7 160GB
(Serial ATA-100, 7,200 RPM, 8MB - MPN: ST3160827AS
http://img182.imageshack.us/img182/5990/seagatebb0.png
Is this buggered ?
Maxtor DiamondMax Plus 9 160Gb
Model no. 6Y160M0 - SATA 8 MBytes
http://img213.imageshack.us/img213/8034/maxtorqb3.png
It probably just indicates that there was some other activity on the disk at the time.
I'm new here so I can't post URLs yet, but I got:
Transfer Rate
Minimum: 37.4 MB/sec
Maximum: 105.4 MB/sec
Average: 72.3 MB/sec
Access Time: 14.2 ms
Burst Rate: 963.7 MB/sec
CPU Usage: 19.5%
Graph started with rather big changes in spikes getting less so towards the end, with an average decline starting around 85%.
2x 120gb Maxtor Diamond+9 8mb cache or something like that, I forget :D. They're in RAID 2+0 Stripe on a Promise Fasttrack controller =).
It's a cracker:lol:
4x80GB Hitachi sata2
http://homepages.nildram.co.uk/~chsm...b-64stripe.png
also worth a look:-
HDTune
has some nice comparison fetures in it:-
Here compares dfast outer edge matrix raid based on ich8r(commando) v ich9r(p5k)
http://homepages.nildram.co.uk/~chsm...taid_moved.JPG
LOL, that burst rate is sooo wrong :P
I wouldn't worry DataMatrix, transfer rate only has advantages in very specific applications. General Windows work and gaming can't really justify it. Awesome for editing media though. :D
Yep but intel have done a freat job of making it consistently wromg comparisons between :-Quote:
LOL, that burst rate is sooo wrong
a) different drivers
b) Ichr8 v ichr9
The real reason behind the burst rate is the impact of write back cache
see Write Back Cache Impact
I've heard people say similar things about a sports car. :lol: Oh you can only do 70 on the motor ways. There are times when it's just fun to do.Quote:
I wouldn't worry DataMatrix, transfer rate only has advantages in very specific applications. General Windows work and gaming can't really justify it.
2 x Seagate 320GB 7200.10 on NVRAID 0, P5N32-E SLi PLUS (This is my main drive. OS/Games/Downloads/etc...)
http://craigcrawford.no-ip.org/images/hdtune1.png
2 x Western Digital Caviar 80GB on ITE IT8212 RAID 0, P5N32-E SLi PLUS (This is mostly storage and pagefile)
http://craigcrawford.no-ip.org/images/hdtune2.png
Both of the above arrays done at the same time
http://craigcrawford.no-ip.org/images/hdtune3.pnghttp://craigcrawford.no-ip.org/images/hdtune4.png
DataMatrix, there is definitely something wrong with your Nvidia driver / configuration, it is way too slow for 7200.10. The IDE RAID speed is sub-par, you would expect at least 55MB/s average even with the crap Nvidia PCI bus.
that's a very small stripe , intel default to 128k stripe most use 64kb which is whhat the raids in my examples are based on. as others have said there is definately something wrong with those results, nvidia raid is not noted for great performance but is genarally way better than that. Try My Raid guide, and nitty gritty of strip sizes for more info or check out nvidia's web site they usually have a half decent guide., you can download.
luck:)
i think your raid perfoirmance is suffering due to your stripe size. Also the lower your stripe size the more load your cpu is taking compare your cpu usage with mine mine 4~6% yours is 8~15.7% I'd suggest experimenting with say 64kb & see what your results are.
Areca Raid5
http://www.shaithis.me.uk/files/ShaiRAID.png
Yes, but while a sportscar can go faster if required, hard-drives can only delivery performance where it's requested. Your game loading times aren't 4 times faster - most won't even improve at all. Where transfer rate comes in it working with large continuous file reading or writing - uncompressed audio, video or large images, and some sorts of databases. For everyday use the advantages of a striped RAID setup are very limited. 99% of other applications rely heavily on access rate, and transfer rate is limited to the speed of the slowest drive in that case.
Well I'll be buying a P5K and an extra 2 7200.10 320GBs at the end of the month so I may get a RAID card like shaithis. been looking at them, and they look mighty tasty (although the P5K only supports 1 16x & 4x PCI-E when the Areca needs 8x so it may be slightly bottlenecked).
Or I may just use the Intel raid of the P5K which also looks mighty tasty.
Sorry, so if that's the case why have you gone to the expese of buying 2x wd raptor'sd spending approx £156 74GB Western Digital Raptor SATA 16MB £78 If what you claim is correct you've poured money away ?? £1.054/GBQuote:
Your game loading times aren't 4 times faster - most won't even improve at all. Where transfer rate comes in it working with large continuous file reading or writing - uncompressed audio, video or large images, and some sorts of databases. For everyday use the advantages of a striped RAID setup are very limited. 99% of other applications rely heavily on access rate, and transfer rate is limited to the speed of the slowest drive in that case.
My drives ny comparison are mucho cheapo 80 Gb HGST (IBM/Hitachi) HDS728080PLAT20 Deskstar 7K80, IDE (PATA), 7200 rpm, 2MB Cache, 9.0 ms Total £106, £0.331/GB From what i've heard i've also got an advantage in sound, having quiter drives. I had a pair of therse drives on my old amd rig & added a further 2 for fun , as they are cheap.
But it seems that with the exception of access speed the benchmark favours the 4xhitachi
Looking at your benches i'd be particularly concerned about the 10.2% CPU utilisation, compared with my 6%, there's 4% that could be being utilised by something else.
Now if the real world performance is how you claim i've wasted £102 compared to £156. I've got 320Gb of storage v 148Gb.
Because I do do video editing, and I got the drives cheap :P
I've also got 2 (36GB) at work, and it's good for some of the audio work I'm sometimes laboured with. Not really large enough for the video work I do, so I do that at home. These were old drives I used to have at home, and I wouldn't have bought them specifically for work.
Even if you do do basic editing, I'd still argue it's not particularly good value for money unless you're a professional or do so much editing that you're losing time due to using only a single hard drive, and if you're a pro you'd like a hell of a lot more than a few Raptors striped.
Raptors started off black, then they had a clear top, and now striped, what next? :D
I'm sorry you lost me on that argument seems to contradict itself??:)Quote:
Even if you do do basic editing, I'd still argue it's not particularly good value for money unless you're a professional or do so much editing that you're losing time due to using only a single hard drive, and if you're a pro you'd like a hell of a lot more than a few Raptors striped.
@DataMatrix
Don't get carried away the ichr9 shows about the same erformance as ichr8 (commando 965 chipset). What i find superior is matrix raid which imo beats nvidia raid hands down.Quote:
I may just use the Intel raid of the P5K which also looks mighty tasty
Yes you can get some great add on cards, shaithis's Areca Raid5 look superb but i'm not too familiar with raid add on cards. the cost Areca ARC-1220 8 Port PCI-E SATA II RAID 6 Controller (Sorry if i've picked a bad example, it was the only Areca card that i could find) of £300+ is a bit rich for me.:lol:
4 port areca 1210
Scan: £216 out of stock tho
Also sold by http://www.hamiltone.co.uk (where I bought mine) but you need to call for price.
Its an excellent card IMO. I upgraded to it from the equivalent 3Ware Escalade and the performance difference was night and day. Loading times for anything is very very fast.
Areca also seem quite good with drivers. Their x64 driver has been working in Vista since the earliest beta, although they are still to release a digitally signed driver.
That's a pretty awesome card shaithis, that puts you in a league of your own:surprised::)
as i say i'm running cheap & cheerfull using the resources to hand. Depends really what your aiming to do.
Shaithis, is it loud?
The only thing at all noisy with it is one of the doors on the raid bay rattles when I turn the PC on, once i touch it, it stops ;)
It was a very cheap 3 x 5.25" > 4 x 2.5" SATA hot-swap bay converter though which I purchased at least 2 years ago. I am sure there are better available now though (and that's if you even feel the need to use one), the rattle on mine just adds character :innocent:
:lol:Quote:
the rattle on mine just adds character
I have run xp for years on a single drive setup, last year moved to nvidia NF4 raidx0 x2 drives, January intel matrux raid0 x2 & March matrix raid 0 x4. I can tell you now that in day to day operations the os runs with a far better feel, smoothness & response with my current setup.
For me, a RAID is more about resilience and lack of downtime, then it is about saving time. For me the extra performance boost is a very nice extra :)
More so then virtually every other piece of hardware, hard drives develop faults and die. In the last 2 years I have had 3 drives (out of 15) die, 2 of those were part of my RAID array and if I hadn't been RAIDed I would be back to kicking myself for not having an up-to-date backup.
While some people trust their backups, I do not.......I am a bit sloppy with doing them and have had restores fail before. Anything that gives me breathing space before a restore, is a god-send to me :)
The 4 port one shaithis has looks tasty indeed :P
I don't use RAID for it's redundancy, guess you can call it AID :P
I wonder if a maze 4 waterblock would fit on that Areca :laugh:
Yes it is crisper and sharper (similar to the general feeling you get using an old HyperThreading P4 IMO), but it doesn't actually do much unless you give them a specific use. I'll grant you it's free speed (4x80gb instead of 1x320gb), but only if you use a separate drive for data, or use 0+1 or 5.
If I'm honest - my original RAID array (2x40gb 7,200rpm WDs) was built having looked at reviews which never give a clear indication of where the benefits and pitfalls were - you see higher numbers, and so that's the path you go down (well I did!). It was a mistake born through naivety, but the drives are still in use today, and the 36gb Raptors which replaced them are at least in a role where they shine now.
I was in a bit of a hurry earlier, so regarding the 10% CPU usage, that's not correct - I was actually playing with DVDLab Pro on another disk at the time, so the CPU usage won't be accurate. It's normally around 6% or so (thinking back to an old HDTach benchmark), which I guess is about right for a half-decent software RAID controller. I don't think I could justify moving to a hardware one myself, but I have been tempted a few times. Simple RAID0 with multicore CPUs doesn't really warrant a dedicated RAID card.
Yep appreciate that.Quote:
For me, a RAID is more about resilience and lack of downtime, then it is about saving time. For me the extra performance boost is a very nice extra
Touch wood i've only ever had 1 hd failure over the years although i thought i'd come close the other week when the sata support on the data connection came off the drive with the data cable when i unplugged it. It just left the bare metal plug probes. Thought the drive was a gonner but suprisingly i managed to push the lead back on & it worked, (it's going in the wife's , it's not opened up so much:lol:)
:lol: her pc :surprised::lol: :rolleyes:
http://www.thebroadbandchannel.com/t...20RAID%200.png
Is this good or bad for two 74Gb raptors in RAID 0?
See post #7Quote:
Is this good or bad for two 74Gb raptors in RAID 0?
Yeah, I noticed that.
Why would that be then? That Intel raid setup so much faster than the VIA one I have?
Well there can be differences between the raid implementation intel, via , nvidia etc., plus user generated factors like stripe size etc.,Quote:
Why would that be then? That Intel raid setup so much faster than the VIA one I have?
I'm sure gav will be able to pinpoint the reason:)
I forcibly shut down all non-essential tasks with EndItAll (don't think it's available any more, but I'll upload it if anyone wants it), so only had DVDLab Pro running, working on another disk as I was doing it - given that you'll likely have more processes doing their various tasks, I'm sure that's not too bad a result. All the numbers seem within a decent margin compared to mine, other than the CPU usage, which is around 10% too high. Could be driver related, or you have something else running at the same time?
Well, I thought I'd tell you guys.
I just got my wages so I bought 4 x 320 Gb Seagate ST3320620AS Barracuda 7200.10, SATA300, 7200 rpm, 16MB Cache, 8.5 ms, NCQ which means I will have 6 come Friday (including the ones I already have) and they will be running on Highpoint RocketRaid 2320 PCI-E 8Ch Raid5 SATAII Controller. I also got P5K Deluxe etc, so should be good when I overclock the hell out of it :P
I'll let you guys know how I get on, including some benchies.
:lol: that makes it a server doesn't it:)Quote:
I will have 6 come Friday (including the ones I already have) and they will be running on Highpoint RocketRaid 2320 PCI-E 8Ch Raid5 SATAII Controller.
look forward to seeing your results
If I didn't know you'd reply just to save face, as you inevitably will, then no, it won't really be better. It'll shave fractions off your loading times, and that will be it - no advantage anywhere else in the games at all.
If it's RAID5 it'll be a nice setup around the rest of the PC though. :)
Exactly.Quote:
If it's RAID5 it'll be a nice setup around the rest of the PC though.
http://www.derchris.net/hdtune.jpg
Here is mine.
Not so bad
Redundancy + speed + flexibility with regards to failures. Best bet in my opinion unless you really need the space a striped 6x320gb array will give you... and I'll question your sanity or profession if you do :D
or 2 arrays:-
1) smallish partition raid0 for os using the outer faster parts of the drives
2) using remaining space in raid 5
see Enter The Matrix: Slice out and get the best part from your hard drives
this gives imo the best of both worlds, there's nothing stopping you having your documents & settings folder on raid5 for more security.
At the minute my 2nd array has xp64, vista32rc2 & vista64rc2, the vista installs will be time expired shortly , have been used for testing etc., so i'll switch myself to raid5.
Obviously the size of your outer array will be proportional to the size of your drives you'll get the idea from the above link.
Interestingly enough i found this comparison i did feb/mar on my commando when i 1st addded my 2nd pair. all raid 0 64k stripe fast raid is 67gb outer stripe slow is the rest.
http://homepages.nildram.co.uk/~chsm...Comparison.JPG
luck:)
Raid 5 gives you n -1 capacity of your drives .
6 320gb drives would give ~ 1.6Tb of relatively fast and redundant storage.
side note: 6 drives isthe reliablity max for raid 5. after that you should e looking towards raid 6
Hmm, I think RAID 5 looks rather good.
raid 5 is only a good idea if you have a proper hardware. as software based controllers can have pretty piss implementations
Could even be worth setting it up as a 5 drive array + 1 hot drive. You only get the capacity of 4 drives but when 1 dies it will automatically rebuild the data from the dead drive onto the hot spare.
Means you can RMA the dead drive at your leisure, without having to worry about performance or redundancy while the dead drive is awaiting replacement.
I have decided to make a 400GB RAID 0 partition and what is left a RAID 5 partition (if the controller lets me do this, since I have never made multiple arrays before).
If it doesn't let me then :(
If you have 6 x 320GB drives.......and want to cut them up like that, you are going to have to set them up as:
2 x 320GB drives in RIAD0 = 640GB (double chance of losing all data though)
4 x 320GB drives in RAID5 = 960GB
Okay guys, I'm on my laptop right now, seems as though the Vista x64 drivers for the RocketRaid 2320 don't work very well so I'm installing Vista x86 instead which seems to work, I'll be up and running soon, so be ready for some benchies :P
6 x Seagate 7200.10 320GB in RAID 5 on HighPoint RocketRAID 2320
http://craigcrawford.no-ip.org/images/hdtune5.png
Second run:
http://craigcrawford.no-ip.org/images/hdtune6.png
I'll get Diskeeper running in a minute, defrag, then run a 3rd time.
Third run after defrag (strangely enough, the performance is worse):
http://craigcrawford.no-ip.org/images/hdtune7.png
Don't trust these rusults though, everything opens blisteringly fast :)
BTW: That is 6 discs with one partition of 380GB and the rest on another partition.
Have you got write back cache enabled?? It makes quiet a difference.(also seems to be the reason for the weird burst rate scores)
http://homepages.nildram.co.uk/~chsm...Cache_Comp.JPG
Try running HDTach & you'll get a feel for whats happening. speeed decreses as you move away from the fastest outer sectors, the average is taken across the whole raid capacity.
My fast raid benches have been based on the 1st array occupying the fastest part of the drive array.
http://homepages.nildram.co.uk/~chsm...SingleRaid.JPG
compares the 1st fast array with a single raid0 array across alldrives. as you can see by the comparison there is no difference across the space ocupied by the fast array. The average numbers for the full raid are of course lower as the bring in the slower speeds across the remainder of the drive. To be able to get the averages of the faster part , you need 2 arrays, as non of the benching tools are partition capable.
That's why i suggested a raid0 for the os & a raid5 for the rest of the volume.
Your syatem feels fast because i assume your os partition is the 1st on the raid & therefore occupies the fastest part. ??
Here's mine. Same raptor as Clunk's and probably a few others as well. Odd how the transfer rate seems to drop during the test :confused:
http://i116.photobucket.com/albums/o11/Odes1/HDTune.png
read my post above decrease is due to increased seek time as it accesses from outer to inner part of drive/array. This is readily apparent when you see the x axis of the HDTach & the curve.Quote:
Odd how the transfer rate seems to drop during the test
you mean the outer part of rthe array, the 1st partition yes?? whaich is fine & as i say why you feel thatQuote:
But it is partitioned so the fastest part is the main drive.
:lol: which is why i like raid:DQuote:
though, everything opens blisteringly fast
http://craigcrawford.no-ip.org/images/hdtach1.png
RAID 5 with write-through cache setting.
That looks 8~10mb./sec faster than your earlier result is that the quick or long bench??
Here is the long one.
http://craigcrawford.no-ip.org/images/hdtach2.png
Better yet:lol: 117.5Mb/S to 133.9Mb/SQuote:
Here is the long one.
My HDtach results for my old WD drive, and my current seagate 320gb, 7200.10
http://www.zen23322.zen.co.uk/cpuinfo/hdtach.JPG
HDtune for the seagate.
http://www.zen23322.zen.co.uk/cpuinfo/hdtune.JPG
Here's my two seagete 7200.10's in raid 0:
http://www.zen23322.zen.co.uk/cpuinf...dtuneraid0.PNG
writeback cache off
:)
and raid1
http://www.zen23322.zen.co.uk/cpuinf...dtuneraid1.PNG
Nice results Magnets,
Interesting comparing your single 320 v your raid1 results. You have farly closely the same performance but with the added surety of the raid1 redundancy/backup. Nice:D
Does it feel more responsive??
I upgraded my whole PC (from athlon 3000, 512 RAM) to an e6420 and 2GB of DDR25400.
So i can't really say that raid 1 is faster/slower, but knowing that if one HDD dies i won't be up the creek is nice.
I will re-run the test on raid1 in a bit, and see if it's much different.
edit:
about the same
http://www.zen23322.zen.co.uk/cpuinf...ks/hdtune2.PNG
Sorry hadn't realised:)Quote:
I upgraded my whole PC (from athlon 3000, 512 RAM) to an e6420 and 2GB of DDR25400
Sure but i'd be inclined to test recovery so that you have some reassurance it works, before changing your backup strategy. Nothing worse than getting a few months down the line & something goes wrong & you find that it doesn't work.Quote:
but knowing that if one HDD dies i won't be up the creek is nice.
Any idea why HDTach and HDTune show very different results?
http://i204.photobucket.com/albums/b...kcluster-1.png
http://i204.photobucket.com/albums/b...20GB-Samsu.png
My Bench comparison:lol:
Access time & cpu load asre close but average read is 236.2 v 204.1 Mb/S HDTunes Max Read of 222.8 is still below the average for HDTach. So I guess HD Tach may be suspect.
http://homepages.nildram.co.uk/~chsm...bmark_comp.JPG
4x80gb raid0 1st volume 72gb.
burst rate is a controversial figure. intel have been acciused of cooking it. It does seem to be seriously distorted by write back cache.
http://homepages.nildram.co.uk/~chsm...Cache_Comp.JPG
Theres a newer version of hdtach than we are using, so that might be worth a look.
Results are identical on the long test for me.
3rd bench:-
http://homepages.nildram.co.uk/~chsm...id/Everest.JPG
above results are closer to hdtach i think.
Comparison HDTACH Versions
http://homepages.nildram.co.uk/~chsm...HDTachComp.JPGH
Where is the everest benchmark from? I just downloaded the latest everest and there isnt one in it :lol:
I shelled out & got the ultimate:lol:
Its the ultimate that I downloaded, where is the benchmark in it? :lol: