• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Why does intel raid1 blow?

E

ep0x73

2[H]4U
2FA
Joined
Sep 5, 2013
Messages
2,663
Intel states raid1 is equal to a single drive but they are full of shit.
I've tried 12.9, 13 and 13.2 and all the read only, write back etc with poor results.

v-raptor 500 single
-----------------------------------------------------------------------
CrystalDiskMark 3.0.3 x64 (C) 2007-2013 hiyohiyo
Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]

Sequential Read : 456.202 MB/s
Sequential Write : 179.288 MB/s
Random Read 512KB : 453.702 MB/s
Random Write 512KB : 318.098 MB/s
Random Read 4KB (QD=1) : 1.305 MB/s [ 318.6 IOPS]
Random Write 4KB (QD=1) : 3.754 MB/s [ 916.5 IOPS]
Random Read 4KB (QD=32) : 3.869 MB/s [ 944.7 IOPS]
Random Write 4KB (QD=32) : 3.797 MB/s [ 927.0 IOPS]

Test : 50 MB [C: 46.2% (215.3/465.7 GB)] (x9)
Date : 2014/05/30 16:21:21
OS : Windows 7 Ultimate SP1 [6.1 Build 7601] (x64)


v-raptor 500 raid1

-----------------------------------------------------------------------
CrystalDiskMark 3.0.3 x64 (C) 2007-2013 hiyohiyo
Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]

Sequential Read : 117.365 MB/s
Sequential Write : 127.213 MB/s
Random Read 512KB : 125.745 MB/s
Random Write 512KB : 237.294 MB/s
Random Read 4KB (QD=1) : 2.043 MB/s [ 498.9 IOPS]
Random Write 4KB (QD=1) : 2.628 MB/s [ 641.6 IOPS]
Random Read 4KB (QD=32) : 7.321 MB/s [ 1787.3 IOPS]
Random Write 4KB (QD=32) : 2.953 MB/s [ 721.0 IOPS]

Test : 50 MB [C: 49.4% (230.2/465.7 GB)] (x4)
Date : 2014/09/07 4:19:53
OS : Windows 7 Ultimate SP1 [6.1 Build 7601] (x64)

Not even fucking close.......reads completely blow ass and are only 25% of a single drive.

I like the idea of redundancy but not with such shitty results.

I've used raid1 in the past and it was not this bad.

Z87 with the fastest mechanical drives should be benching much better.

I guess Intel has not tuned their drivers to read off both discs, which ever has the information closest.


About ready to just use raid0 even though it's a gamble and use a spare 1TB drive and acronis to make a backup copy.
 
Last edited:
There is something fishy about the raptor single result. There is no way a single raptor reads at 400MB/s+ (that is more than 2 times the media rate).

Although I can see you are running a 50MB test. That will be very inaccurate.
 
No mention of the actual drives in use or how you have them configured on the board along with the BIOS configurations.
The results you have posted indicate you have them on SATA-II rather than SATA-III???
Have you read all the setup guides on the IRST software?

I have a pair of WD 1TB VR's on a Dell R5400 workstation (Running ICHR9) and I can manage a steady 220MB/sec R/W in RAID-1.....

Oh, your single drive results are wrong. You are seeing caching, google it and learn. The correct speeds for these drives are about 200-210MB/sec R/W.

ep0x73 said:
About ready to just use raid0 even though it's a gamble and use a spare 1TB drive and acronis to make a backup copy.
Click to expand...
That would be a poor decision on your part. Use windows backup tools to avoid the sluggishness caused by Acronis and also have some faith in being able to restore the data if required.
If you must have more speed, then SSD's should have and should be implemented.
 
In retrospect here are a pair of raptors from almost 2 years ago in raid1

-----------------------------------------------------------------------
CrystalDiskMark 3.0.2 (C) 2007-2012 hiyohiyo
Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]

Sequential Read : 107.691 MB/s
Sequential Write : 137.496 MB/s
Random Read 512KB : 111.941 MB/s
Random Write 512KB : 168.174 MB/s
Random Read 4KB (QD=1) : 2.411 MB/s [ 588.7 IOPS]
Random Write 4KB (QD=1) : 0.720 MB/s [ 175.8 IOPS]
Random Read 4KB (QD=32) : 4.359 MB/s [ 1064.2 IOPS]
Random Write 4KB (QD=32) : 0.723 MB/s [ 176.4 IOPS]

Test : 50 MB [C: 62.1% (118.3/190.4 GB)] (x3)
Date : 2012/12/12 0:32:06
OS : Windows XP Professional SP3 [5.1 Build 2600] (x86)

Not exactly mind blowing numbers by any means.
 
Lost-Benji said:
No mention of the actual drives in use or how you have them configured on the board along with the BIOS configurations.
The results you have posted indicate you have them on SATA-II rather than SATA-III???
Have you read all the setup guides on the IRST software?

I have a pair of WD 1TB VR's on a Dell R5400 workstation (Running ICHR9) and I can manage a steady 220MB/sec R/W in RAID-1.....

Oh, your single drive results are wrong. You are seeing caching, google it and learn. The correct speeds for these drives are about 200-210MB/sec R/W.


That would be a poor decision on your part. Use windows backup tools to avoid the sluggishness caused by Acronis and also have some faith in being able to restore the data if required.
If you must have more speed, then SSD's should have and should be implemented.
Click to expand...

They are VR 500's, yes I have been setting up raid1's for many years, yes the board is SATAIII set to raid in the bios.

I know what caching means, I am not a noob but crystal does not always work that way.
If the results were so skewed then professional sites would not use it.

ATTO is more accurate for real world speed and I've benched using those and HDtune.
 
ep0x73 said:
They are VR 500's, yes I have been setting up raid1's for many years, yes the board is SATAIII set to raid in the bios.
Click to expand...
See below!

ep0x73 said:
I know what caching means, I am not a noob but crystal does not always work that way.
If the results were so skewed then professional sites would not use it.
Click to expand...
Your results from the precious Crystal are already giving you dud results, unreal speeds and indicating SATA-II, you should have dropped it already.
Please explain a "professional site"?
Every bloody forum or review site use a bunch of different tools and guess what, they are all synthetics.

ep0x73 said:
ATTO is more accurate for real world speed and I've benched using those and HDtune.
Click to expand...
No, ATTO is known to use rather compressible data and can bias results too.
 
OK then what of the very few benchmarks do you use that are accurate?

As mentioned I tried from 12.8-13.4 and not much difference at all with drivers.

I never said crystal was precious but it's used on many sites, even at [H], Anantech, Toms and a shitload of various sites.

Same drives in raid1 on a P35 and XP 2 years ago.

9uoso8.png


Drive single.

2q8st3l.png


Besides a few more dips and not as smooth it's pretty close performance wise.

Not seeing it on windows 7 though with the same drives.

"Mirroring: Read performance under mirroring is far superior to write performance. Let's suppose you are mirroring two drives under RAID 1. Every piece of data is duplicated, stored on both drives. This means that every byte of data stored must be written to both drives, making write performance under RAID 1 actually a bit slower than just using a single disk; even if it were as fast as a single disk, both drives are tied up during the write. But when you go to read back the data? There's absolutely no reason to access both drives; the controller, if intelligently programmed, will only ask one of the drives for the data--the other drive can be used to satisfy a different request. This makes RAID significantly faster than a single drive for reads, under most conditions."

In theory reads should be better, writes worse but I am seeing the opposite.
 
Last edited:
I'll give it a shot.
Do you run them in safe mode or just turn off AV?

I've done both but only HDtune seems to be impacted and still there are too many drops even though nothing is loaded or should be pinging the drive.

It's hard to get a real accurate reading sometimes.

Besides crystal, hdtune and atto I am not away of other drive bench programs offhand.

If I had $500 laying around I'd hit up two evo drives in raid1, least in theory with it set up it's just a matter of swapping out a drive/built, repeat.
 
ep0x73 said:
"Mirroring: Read performance under mirroring is far superior to write performance. Let's suppose you are mirroring two drives under RAID 1. Every piece of data is duplicated, stored on both drives. This means that every byte of data stored must be written to both drives, making write performance under RAID 1 actually a bit slower than just using a single disk; even if it were as fast as a single disk, both drives are tied up during the write. But when you go to read back the data? There's absolutely no reason to access both drives; the controller, if intelligently programmed, will only ask one of the drives for the data--the other drive can be used to satisfy a different request. This makes RAID significantly faster than a single drive for reads, under most conditions."
Click to expand...

That is exactly why you are not seeing any benefit in RAID1. HDTune probably reads large sequential blocks with a QD of 1.
This is the only way to properly test sequential reads and those will go to only drive at a time.
You could try two parallel HDTune instances. Or use a benchmark that tests higher QD workloads.
 
omniscence said:
That is exactly why you are not seeing any benefit in RAID1. HDTune probably reads large sequential blocks with a QD of 1.
This is the only way to properly test sequential reads and those will go to only drive at a time.
You could try two parallel HDTune instances. Or use a benchmark that tests higher QD workloads.
Click to expand...

+1

Only raid1 setups I have seen that will do faster sequential read speeds on a single threaded (qdepth 0) read is areca controllers. None of the software or other brands do this from what I have seen.
 
Intel on-board RAID was able to give RAID-0 like reads if the volumes were good and initialised, this was touchy depending on drivers and chip-sets though.

As for the request of what "Benching" apps I use that I hold "Accurate": NONE

They are all synthetic, I use my gut for the general use of a system. I install OS, all drivers and updates. I install my software and use it. If the system feels good, then I stay with it, if not, then I tweak or use trial-n-error testing.

I should make it clear, I own/run a successful IT business and build systems daily thus I have more hands on than a good chunk of members on forums and also why I don't bother with forums much.
Here is a tit-bit for those who need the latest and greatest:
I had a Dell R5400 WS (Google it) with a pair of WD 1TB VR's in RAID-1 (ICH9R) and the system was filled with 16GB FBDIMM along with 2x 3.0GHz quad-core Xeons. The rest of the guts don't matter.
I have currently, a latest generation 1150 quady with 16GB DDR-III 1866 and a pair of 240GB Intel 530 series SSD's in RAID-0 (on-board) and it peels off 1000+ MB/sec R/W all day, no worries. I also had a huge s2011 system that also had balls to the wall.
The R5400 can chew out 75Gflops average, the s1150 can spit out 100 and the s2011 system did 100 as well.
But after all the e-penis swelling, I am going back to older R5400 as we speak. Why?
The old girl didn't have any specs going for it but it starts, run and works flawlessly. Boot times are fast as the SSD's and having the RAID-1 gives me and what I use the system for, better security (I use the system in 1000+ seat theatres for audio production and playback, BSOD mid-show not good).

What the moral to the story?
Stop getting caught up on benchmarks, use the maggot or turn it off at the wall and go outside to get some fresh air.
 
I live in the country so I get plenty of fresh air....:)
Seriously though I guess I am just surprised the benches are somewhat slow but the system is not exactly a dog either, it's the fastest system to date I've built.

Still not 100% onboard SSD's, they are good unless you have to flash the firmware.

Nothing wrong with raid1 with samsung 850 pro's if you want top speed but a mirror, granted you can justify the cost.

Maybe someday Intel will include in their driver a way for raid1 to have split seek so you actually get faster reads then a single disk as Intel keep falsely claiming.
 
I prefer to use the older hardened workstation kit rather than going for bleeding edge or 'non-standard for the sake of it!"

Plus Raptors aren't all that anymore when a $50 1TB HDD can push 200MBps round the outside.
 
daglesj said:
Plus Raptors aren't all that anymore when a $50 1TB HDD can push 200MBps round the outside.
Click to expand...

Sadly, they don't handle the IOPS and general feel that comes from faster drives.

I could always go SAS but while still on SATA-II, the VR's are doing well.
 
With WD/HGST pushing out current tech the raptor line is due for a refresh.
I am sure they could increase the cache to 128 and increase areal density beyond the current 333 per platter.
The current Seagate V4 drives are actually faster then the 10K raptors, least on paper.
 
You must log in or register to reply here.
Back
Top