i680 RAID 0 Performance - is it bunged?

Updated Jan 9th: As per Post #43, perhaps the problem with RAID 0 performance on the i680 is related to the EVGA boards which both Tom's and Fugger's results (below) are based on? It would be great if others with Striker and EVGA boards could post some RAID HDTach results for us to review.

Original Post:

Both Tom's Hardware: The Southbridge Battle...

and Fugger: eVGA Review seem to have found consistent poor RAID 0 performance on the i680 compared to Intel chipsets.

Why are the Intel SB's so much better? Why is the i680 so bad?

Are others finding this as well with their i680 boards?

Even my nForce4 board does 150MBs HDTach with old 74GB Raptors!

i680...

http://members.shaw.ca/virtualrain/680i.gif

Compared to the ICH8...

http://members.shaw.ca/virtualrain/ich8.gif

Or ICH7R...

http://members.shaw.ca/virtualrain/ich7.gif

Fugger's results...

http://fugger.netfirms.com/680i/hdtach.jpg

image not working

Haha, can't belief they didn't pick that up.

Jokester

Edit: Maybe they haven't flashed to the latest BIOS :D

Sometimes the images do work and sometimes they don't... I can't figure it out... the links are correct.

Anyway, just click here to see the images.

Tom's hardware is a joke imo

Quote:

---

Originally Posted by virtualrain

is Tom's on Crack?

---

Pretty much.

Tom's is about as reliable as The Inq.

What about this ???? (NF4) but it shouldn't perform like this ...

http://www.xtremesystems.org/forums/...d.php?t=128777

Quote:

---

Originally Posted by J-Mag

Tom's is about are reliable as The Inq.

---

Now that's what I call trashing... the Inq ;)
Tom is crap, get over it, what did you expect?

Quote:

---

Originally Posted by J-Mag

Tom's is about are reliable as The Inq.

---

For the most part it is...

wow I was under the impression Toms was pretty reputable (at least back in the day it was). This is a shock to me.
As an owner of the 680 I was unhappy with some of Tom's comments.

Plus they did not even mention the Packet Priority features on the LAN nor the hardware firewall.

ditto, never realized tom might not be telling me the full truth. always thought of the site as fairly reputable.

Years ago tom was pretty much the only game in town and then a very young Anand came along, tom went heavy sponsors and in fact so much so its hard to wade through them all just to find anything.:fact:

http://www.xtremesystems.org/forums/...d.php?t=122354

Seems to match what Fugger found....third to last picture.

Tom's is way crap and has been for 5 years or so...back in the day (mid to late 90's) he was probably the most trusted hw site but they pretty much suck these days...I miss Sharkey Extreme (with the real Sharkey).

how do these results stack up in REAL WORLD applications that benefit from HDD speed :) that would be a more interesting result to check........

Quote:

---

Originally Posted by Tac0

Tom's is way crap and has been for 5 years or so...back in the day (mid to late 90's) he was probably the most trusted hw site but they pretty much suck these days...I miss Sharkey Extreme (with the real
Sharkey).

---

Ross Sharkey does Porsche now:D He also has a monthly column in CPU Mag.

If you want good review sites try X-Bit Labs and TheTechReport. Hothardware and Legitreviews are worth a look too.
Others I monitor to see what they have to say for perspective, in no certain order; Guru3d, Anandtech, Extremetech, Beyond3d, PCPerspective, FiringSquad, NordicHardware, HKEPC, TrustedReviews, Driverheaven.net, EliteBastards, Bit-tech.net, Bjorn3d.com, Gamepc.com, OCWorkbench, Techpowerup!, Hardwarezone, [H]ardOcp, NvNews, Rage3d and Sharkyextreme.

Yup, stopped visiting Tom's a while back due to the excessive amount of advertising (which also seemed to bias some 'reviews' imo).

Having recently owned all 3 chipsets I can honestly say I saw very little, if any, difference in speed or reliability in 'real world apps' using Raid 0.

Quote:

---

Originally Posted by Ic3man

Yup, stopped visiting Tom's a while back due to the excessive amount of advertising (which also seemed to bias some 'reviews' imo).

Having recently owned all 3 chipsets I can honestly say I saw very little, if any, difference in speed or reliability in 'real world apps' using Raid 0.

---

Thank you!!!!:toast:
There is no need to argue about it!

Quote:

---

Originally Posted by wickedld9

http://www.xtremesystems.org/forums/...d.php?t=122354

Seems to match what Fugger found....third to last picture.

---

Yikes... Maybe Tom's is not on crack. It would appear there is indeed something wrong with the RAID 0 performance of the i680. That's terrible and about a third the throughput of the Intel chipsets. Why hasn't anyone been harping on this?

The lousy RAID 0 performance is a feature...:rolleyes:

NVidia makes the best chipsets...ask them, they'll tell you. They have our best interests at heart, and have strongly encouraged (forced) enthusiasts to use their chipsets by not permitting SLI on non-NVidia chipsets. This is to preserve the integrity of SLI (hold enthusiasts over a barrel for more money for an inferior product).

I'm sure they'll release a 690i next month that enhances the performance of the 680i by TRIPLING the RAID 0 throughput!!! Wow...we'll all have to get one of those...;)

I'm interested in knowing if this is a wide-spread problem. Are other i680 users running RAID 0? If so, what are your HDTach results like?

How can the Intel chipsets be so much faster?

My NF4 Mobo can do 150MB/s with old 74GB Raptors in RAID 0... this i680 performance is terrible!

When I get home tonight I will see how my 650i does.

my raid 0 performance is not too good on this chipset , had better numbers with the asus ws pro, but that board was too buggy for me!

This is not reassuring... why would any enthusiast want this chipset if the RAID 0 performance is poor... don't we all run RAID 0 arrays? I guess if you need SLI, you don't have much of a choice, but I can't believe people aren't up in arms about this.

hmm..that's really strange cuz my nf4 results was the same on ich7, in fact more even consistent. I dont know if NV screw up or what but I hope abit resolved this issue by the time it hits the shelf.

Quote:

---

hmm..that's really strange cuz my nf4 results was the same on ich7, in fact more even consistent. I dont know if NV screw up or what but I hope abit resolved this issue by the time it hits the shelf.

---

The fact that they are having serious issues like this in a shipping product is reason enough for me to stay far away from NVidia chipsets. Two major issues already (SATA corruption and now this). I seriously question NVidia's quality control department.

listen you people if any of you have fallen for this then you know absolutely nothing about IDE/SATA/Raid.

it is IMPOSSIBLE, let me repeat that, I M P O S S I B L E to get 300mb/sec on a Sata Raid controller.
i dont care what fereaky god or demon has blessed your hardware its impossible.
and Tom's needs to be ridiculed by every member of their team and ever other website needs to openly critisize their blatant lie's about the nvidia controller.
not even SCSI 320 can do that, well thats not entirely true but a 15k rpm SCSI 320 drive is ALOT faster and ALOT different than a SAta drive.
i dont give a damn who you are, those results a bullsh|t.

the nvidia controller is pulling 100+ which is EXACTLY where it is supposed to be.
the 300mb per second with the ICH8 is a KNOWN FLAW in hdtach, thats why i have earged everyone for as long as hdtach has been around to NOT use it, or to use it in moderation and dont takes its results totally seariously.

hell i had a sata 1 drive, a single drive in non raid pull over 700mb over and over and over in the tests. and this was not a bits and bytes conversion this was that same settings you see everyone else use and this was on an ICH7 controller.

i wish you guys wold stop making a big deal out of this Tom;s has obviously done 2 things

1)
blatantly lied about their results
2)
has no god damn clue at all wtf they are doing.


stop looking at this stupid hdtach results as the end all truth.
its a fvcked up lie and Tom's needs to openly retract their tests and start showing real world test results.

Im curious if all of you are so concerned with stellar raid performance, why arent you using a REAL raid controller with memory on it? You want to see smoking benches, drop $ 300 - 500 on an Areca card! You guys blow $650 on video cards anyways.

Quote:

---

Originally Posted by Haltech

Im curious if all of you are so concerned with stellar raid performance, why arent you using a REAL raid controller with memory on it? You want to see smoking benches, drop $ 300 - 500 on an Areca card! You guys blow $650 on video cards anyways.

---

pssssssst shhhhhhhhhhhhooooosh

people don't know they don't have REAL raid controllers hahahahahah

i want you to take a good look at this picture, ignore the numbers in the bar graph just look at the line graph, the red and blue colores.
those are REAL results.

do you notice anything odd ?
not one of those controllers is reaching 300Mb/sec.
and i will guarentee you the Sil 3112 is a better controller than the ICH8r
this is the NVidia 650i sata controller NON raid using a SATA II Maxtor 300gig drive. in Sata II mode
the other is my 2x 36g Raptors in Raid 0 on my NetCell Revolution SR5000 w/ 64megs of ram onboard.

now i will guarentee you that beyond anything else that that Netcell is a better controller than any onboard raid controller out today, its a hardware raid controller PLUS it has 64megs of onboard Cache for the raid caching.
there are NO drivers that are needed to be installed for this controller to work, it is driverless, there are 5 Raid controller rom chipsets 1 for each port, a raid cpu and 64megs of ram. it dont matter what OS you use no drivers needed. and whates even more funy is that no matter waht OS you use you can not even see the controller, all you see is your raid array,. you dont see the controller card at all, technically it doesnt even use an IRQ thats visible in windows.

what do you see ?

now go back and ask yourself is TOm's results realtime or Peak Burst results, or is it just a bunch of horse crap.

Quote:

---

Originally Posted by dinos22

pssssssst shhhhhhhhhhhhooooosh

people don't know they don't have REAL raid controllers hahahahahah

---

hahaha.

Good posts Lestat.;)

Yeah Lestat good one, bout time somone stuck it where it needed to be stuck. :fact:

I agree on a couple of points...

1... Tom's is a joke.

2... that there is something wrong with TOM's results on the Intel chipsets (thanks to whoever pointed out is is a bug with HDTach).

However, I think there is also something wrong with the i680 results.

You guys forget that Fugger, who is a respected contributor/reviewer here, also got the same results with 4 150GB Raptors running in RAID 0. There is somethign wrong with these results...

http://fugger.netfirms.com/680i/hdtach.jpg

My pair of old 74 GB Raptors on an NF4 chipset has better performance than this!

While I'm not arguing that HDTach is the ultimate RAID bench tool, it does seem to be the tool used by a lot of evaluators (both respectable and not). I'm all for seeing a variety of different HD benching. I'd also like to see the RAID performance of all the recent chipsets (iCH8/iCH7R/i680/RD600) compared with some stand-alone RAID adaptors as well.

At any rate, I think the RAID-0 performance of the i680 merits further investigation.

i have 2xraptor raid0 and 2x maxline 3 300gb raid0 on an evga 680i. my results are about equal to tom's 680i results. upload any of the compare raid0 controllers and they will be slightly under the i680 results. 2x raptor on sil, intel, highpoint, promise or via controllers all are not as fast as the nviida. toms intel results are wrong, not the nvidia results. the nvidia controller is top raid controller in fact. unfortunately, the compare uploads seem slightly dated.




very truly yours,
politenessman

Quote:

---

Originally Posted by politenessman

i have 2xraptor raid0 and 2x maxline 3 300gb raid0 on an evga 680i. my results are about equal to tom's 680i results. <snip>

---

I think this is more evidence that there is something wrong with the i680 RAID...

Here is my HDTach Avg. Read of 128MB/s with 2x 74GB Raptors with 8MB Cache on an NF4 chipset...

http://members.shaw.ca/virtualrain/myhdtach.gif

How can my setup be faster than Tom's/Fugger's tests using 4x 150GB Raptors with 16MB Cache? This makes no sense!

Quote:

---

Originally Posted by Lestat

i want you to take a good look at this picture, ignore the numbers in the bar graph just look at the line graph, the red and blue colores.
those are REAL results.

do you notice anything odd ?
not one of those controllers is reaching 300Mb/sec.
and i will guarentee you the Sil 3112 is a better controller than the ICH8r
this is the NVidia 650i sata controller NON raid using a SATA II Maxtor 300gig drive. in Sata II mode
the other is my 2x 36g Raptors in Raid 0 on my NetCell Revolution SR5000 w/ 64megs of ram onboard.

now i will guarentee you that beyond anything else that that Netcell is a better controller than any onboard raid controller out today, its a hardware raid controller PLUS it has 64megs of onboard Cache for the raid caching.
there are NO drivers that are needed to be installed for this controller to work, it is driverless, there are 5 Raid controller rom chipsets 1 for each port, a raid cpu and 64megs of ram. it dont matter what OS you use no drivers needed. and whates even more funy is that no matter waht OS you use you can not even see the controller, all you see is your raid array,. you dont see the controller card at all, technically it doesnt even use an IRQ thats visible in windows.

what do you see ?

now go back and ask yourself is TOm's results realtime or Peak Burst results, or is it just a bunch of horse crap.

---

This looks like the graph of my HDTach with single Seagate 200GB SATA2 drive on ICH8.

The guys that went out and bougth 4 drives to make a RAID 0 array and think they are getting 300MB/s are really funny.

Well something isn't right. Here's the results on my desktop computer at work.
Two Seagate IDE 80GB Drives in Raid0 on a Gigabyte NF3 board.
http://i35.photobucket.com/albums/d1...r-Face/NF3.jpg

I'm glad I'm not the only one that thinks there is something wrong with the i680 RAID performance.

Just thought I'd throw this out there.
Striker, raid0 on ports 1 and 2, 2x 740adfd's, 16k stripe size.
Seems fine to me for the most part. Or am I missing something?

http://img.photobucket.com/albums/v4...s/680iraid.jpg

if you look at thg, he using 4 x raptors, look @ the dam total
of the volume drives = 600gb = 4 raptors :slapass:

so the nvidia 680 sata is definitely very poor

4 raptors on the nvidia 4 chipset in raid 0 get close to 300mb
http://img251.imageshack.us/img251/153/bench5on.jpg


so toms hardware aint far of

Quote:

---

Originally Posted by ZeroOne

Just thought I'd throw this out there.
Striker, raid0 on ports 1 and 2, 2x 740adfd's, 16k stripe size.
Seems fine to me for the most part. Or am I missing something?

http://img.photobucket.com/albums/v4...s/680iraid.jpg

---

Thanks... This is very reassuring!

Perhaps the RAID 0 problem is related to the EVGA board which both Tom's and Fugger used (as per the OP).

Anyone else with Striker or EVGA boards with RAID arrays care to post your HDTach results? Let's see if there is a trend here.

Note: I've updated the OP to reflect this find.

One other thought... The throughput of the PCI bus is about 1Gbps... which oddly enough is about 120MBps... I wonder if somehow the eVGA BIOS or implementation is limiting RAID throughput to PCI bus speeds?! I know it doesn't make much sense since this is on the SB but what else could it be?

One thing that makes me believe the board, bios or chipset is limiting the throughput on the EVGA i680 is that the HDTach curve is straight across instead of downward sloping (as it is on most other HDTach results) indicating that it is being limited artificially rather than by the mechanical nature of the HD (i.e. where the heads are positioned with respect to the outer/inner edge of the platters).

Quote:

---

Originally Posted by Lestat

listen you people if any of you have fallen for this then you know absolutely nothing about IDE/SATA/Raid.

it is IMPOSSIBLE, let me repeat that, I M P O S S I B L E to get 300mb/sec on a Sata Raid controller.
i dont care what fereaky god or demon has blessed your hardware its impossible.
and Tom's needs to be ridiculed by every member of their team and ever other website needs to openly critisize their blatant lie's about the nvidia controller.
not even SCSI 320 can do that, well thats not entirely true but a 15k rpm SCSI 320 drive is ALOT faster and ALOT different than a SAta drive.
i dont give a damn who you are, those results a bullsh|t.

the nvidia controller is pulling 100+ which is EXACTLY where it is supposed to be.
the 300mb per second with the ICH8 is a KNOWN FLAW in hdtach, thats why i have earged everyone for as long as hdtach has been around to NOT use it, or to use it in moderation and dont takes its results totally seariously.

hell i had a sata 1 drive, a single drive in non raid pull over 700mb over and over and over in the tests. and this was not a bits and bytes conversion this was that same settings you see everyone else use and this was on an ICH7 controller.

i wish you guys wold stop making a big deal out of this Tom;s has obviously done 2 things

1)
blatantly lied about their results
2)
has no god damn clue at all wtf they are doing.


stop looking at this stupid hdtach results as the end all truth.
its a fvcked up lie and Tom's needs to openly retract their tests and start showing real world test results.

---

Toms didnt say anything about it. These knuckelheads on the forum brought it up saying they cant beleive toms didnt catch it. Just more bashing of the 680i chipset by some monkeys who dont even own one.

It appears Evga 680i does have a flaw on the raid performance because ZeroOne's result on the Striker is on point. Also looking at nanoprobe's result on NF4, it's also on point as well. That's why I was wondering how could it got worse from NF4 to 680i. So this crappy raid performance seems isolated only to the reference board design. Anyone with BFG 680i can confirm similiar results?

Quote:

---

Originally Posted by bboy1234

Toms didnt say anything about it. These knuckelheads on the forum brought it up saying they cant beleive toms didnt catch it. Just more bashing of the 680i chipset by some monkeys who dont even own one.

---

I honestly don't think anyone is bashing it. I think there is a legitimate problem with the reference board RAID 0 implementation. If you own one, you should be concerned and/or want it fixed or stay out of the way so that the people who are concerned can get to the bottom of it.

Any updates on the RAID 0 performance on EVGA 680i boards?

here's my eva 680i raid performance, 2x160GB WD sata2's
http://i35.photobucket.com/albums/d1...680i_raid0.jpg

P5N32-E 680
2 x 75 WD Raptor

http://tech-bytes.net/hd.JPG

weird how my sata2's beat out your raptors. wait, are you running raid-1?

i m running them on Raid 0 (stripping)
i will set up my bad axe 2 board later on today after i get back from school and do a comparison

do everyone a favour and run some real world test where raid would benefit performance and also one chipset vs. another

Quote:

---

Originally Posted by neonflx

i m running them on Raid 0 (stripping)
i will set up my bad axe 2 board later on today after i get back from school and do a comparison

---

so how come it says 68.95GB's on your stripe?

Quote:

---

Originally Posted by dinos22

do everyone a favour and run some real world test where raid would benefit performance and also one chipset vs. another

---

like what?

run some 3DMARK PCMARK, VIDEO ENCODING TEST, ANTIVIRUS SCANNING, GAMES, TRANSFERRING FILES............i mean there is a heck of a lot of stuff you can do :)

well hdtach is quick, simple and universal. not everyone uses the same AV or plays the same games or uses the same video encoding software.

My testing ....


http://www.pctuning.cz/ilustrace3/ob...dtach_raid.png

and is UNRELIABLE and does NOT indicate real word performance.......you will see when you do the tests

Abit IN9MAX
2 x 74 GB raptors

http://img406.imageshack.us/img406/961/raiddk1.jpg

Quote:

---

Originally Posted by dinos22

and is UNRELIABLE and does NOT indicate real word performance.......you will see when you do the tests

---

well you seem to be the only one who has a problem with it, meanwhile the thread continues around your noncontributing rants.

sorry

my hd are 34 a piece my mistake

Quote:

---

Originally Posted by scarface

well you seem to be the only one who has a problem with it, meanwhile the thread continues around your noncontributing rants.

---

QFT! (This thread is not about comparing performance but about gaining consensus on whether an issues exists with RAID on 680i)

It definitely seems like something is throttling STR. If the HD Tach curve does not taper off in a natural curve and is flat all the way across, then something is getting saturated and/or throttling data transfer. The curve should naturally taper off as the heads move from the outer cylinders of the platers to the inner.

It seems like the majority of EVGA boards and at least one ABIT board suffer from this problem. :confused:

Some more Striker RAID performance curves would be helpful.

Perhaps there is a potential issue with 680i RAID performance depending on how the manufacturer implements it (BIOS/board layout/etc)?

Quote:

---

Originally Posted by scarface

well you seem to be the only one who has a problem with it, meanwhile the thread continues around your noncontributing rants.

---

look like Lestat pointed out earlier that's what these "raid" controllers are supposed to do....all the peak measurements and other gimicks HD and other software do doesn't exactly translate into mindbogling real life performance hence why i'm telling you to do some real world tests

these raid controllers are not really true raid controllers but i hope you guys know about that right? if you want true raid performance you need to buy yourself a dedicated PCI Express 8x raid controller (EXAMPLE) with 2 or 4 Ultra320 15K SCSI drivers in Raid0. Only then will you actually get the hard drive performance that is good for 300 or greater I/Os per sec and some nice seek times but it will cost an arm and a let and it will be hot and loud like you wouldn't believe...yeah i've run multiple 15K scsi in the dual xeon days and i still build high end workstations with people demand peak performance in HDD speeds particularly

you guys are kidding yourselves with raid setups on any of these desktop motherboards....the only raid useful on them is Raid1 so you can at least protect yourself from a single drive failure and nothing else

what we need are solid state disks.....hopefully 2008 will bring some joy to :banana::banana::banana::banana: HDD performance and main bottleneck in PCs today :(

Dinos22, again, this thread is not about the merits of on board RAID vs dedicated hardware RAID, nor is it a thread about the merits of HD Tach vs. other drive benchmarks, nor is it a thread about the merits of RAID vs. single disks, nor is it a thread about the pros and cons of different RAID levels... sheesh!

This thread is about whether the 680i implementation of on-board RAID by EVGA and/or other manufacturers is actually being artificially bottlenecked, capped, throttled, or otherwise buggered, by something.

2 x 150 raptors on 680i motherboard:

http://www.supload.us/thumbs/default...0212203902.jpg


The trick to getting all the speed is here:

http://www.supload.us/thumbs/default/121.JPG


go into device manager...scsi & raid controllers...right click on nvidia 590/570/550 serial ata controller click properties..make sure primary and secondary channels look like this...reboot and rebench...you will see a difference..:toast: :woot:

Quote:

---

Originally Posted by Grinch

2 x 150 raptors on 680i motherboard:

http://www.supload.us/thumbs/default...0212203902.jpg


The trick to getting all the speed is here:

http://www.supload.us/thumbs/default/121.JPG


go into device manager...scsi & raid controllers...right click on nvidia 590/570/550 serial ata controller click properties..make sure primary and secondary channels look like this...reboot and rebench...you will see a difference..:toast: :woot:

---

Thanks Grinch (BTW, isn't that an oxymoron? :p: )

What are the default driver settings?

Can other EVGA RAID users check these settings and report back?

Quote:

---

Originally Posted by virtualrain

Thanks Grinch (BTW, isn't that an oxymoron? :p: )

What are the default driver settings?

Can other EVGA RAID users check these settings and report back?

---

everything is checked by default...and any 680i based board should be able to do this trick...

okay here's the difference with the Grinch suggestion on my rig.
Before
http://i35.photobucket.com/albums/d1...680i_raid0.jpg
After
http://i35.photobucket.com/albums/d1...ace/680i_3.jpg

Scarface, your RAID 0 on EVGA is not as bad as most... As you've seen, most have a flatline at 110MBps. Since you are one of the few NOT using Raptors, I wonder if that has anything to do with it?

Never the less, the settings did improve your burst and CPU utilization.

i was actually thinking of buying a couple of raptors but i noticed they are Serial ATA150 & not SataII. maybe that's the problem?

Grinch's suggestion improved mine - thanks - it worked! :banana:

Quote:

---

Originally Posted by virtualrain

Dinos22, again, this thread is not about the merits of on board RAID vs dedicated hardware RAID, nor is it a thread about the merits of HD Tach vs. other drive benchmarks, nor is it a thread about the merits of RAID vs. single disks, nor is it a thread about the pros and cons of different RAID levels... sheesh!

This thread is about whether the 680i implementation of on-board RAID by EVGA and/or other manufacturers is actually being artificially bottlenecked, capped, throttled, or otherwise buggered, by something.

---

i realise that and i am trying to tell you that the way you are testing doesn't exactly give you the right answers as people have posted intel matix figures which are more than double the figures of 680i but i bet you anything in real world hard drive intensive tasks you will not have double the performance difference.......i would even be surprised if there was any noticable difference in all honesty across the board......

on a sigle 150 raptor it would take me 2:54 to unrar a 4.7gb file...when using raid 0...it takes about 1:32 to unrar same file..

Quote:

---

Originally Posted by virtualrain

My NF4 Mobo can do 150MB/s with old 74GB Raptors in RAID 0... this i680 performance is terrible!

---

No matter what I try I can never get an average read much over 100MB/s with my Raptors in RAID-0 :(

The rig I had before it (P4 3.2C and Abit IC7) used to destroy the nF4 in HDTach...

Quote:

---

Originally Posted by Sumanji

No matter what I try I can never get an average read much over 100MB/s with my Raptors in RAID-0 :(

The rig I had before it (P4 3.2C and Abit IC7) used to destroy the nF4 in HDTach...

---

let me guess.....your stripe size is over 16K?

Quote:

---

Originally Posted by scarface

i was actually thinking of buying a couple of raptors but i noticed they are Serial ATA150 & not SataII. maybe that's the problem?

---

makes no difference the 150's will be faster than 99% of any SATA II drive simply because of it's rpm's:)

Quote:

---

Originally Posted by Grinch

let me guess.....your stripe size is over 16K?

---

64k yeah. I just tried that tweak though (disabling read caching and command queuing) and average read has gone up to 122MB/s and burst to 238MB/s which is pretty cool!

Before:

http://homepage.ntlworld.com/suman.g...ch_default.JPG


Tweaked:

http://homepage.ntlworld.com/suman.g...h_tweakson.JPG

Are there any implications for disabling read caching though? Surely it's enabled by default for a reason...

Cheers,

Suman

now if you had a smaller stripe size your graph would look more linear....and have higher numbers...but all of that is synthetic anyways...real world testing I dont use any programs at all....I use a stop watch!

BUMP

Can anyone try RAID0 with non-WD disks? I've had a convo with someone and the suggestion of WD firmware came up. In this thread iv seen at least one 680I/Raptor combo unaffected by the saturation issue in HDTach. Still surprises me how many ppl use Raptors :)

Sooo...non-Raptor/non WD users...please can ya have a shot for me/us?

Cheers!

Quote:

---

Originally Posted by Grinch

2 x 150 raptors on 680i motherboard:

http://www.supload.us/thumbs/default...0212203902.jpg


The trick to getting all the speed is here:

http://www.supload.us/thumbs/default/121.JPG


go into device manager...scsi & raid controllers...right click on nvidia 590/570/550 serial ata controller click properties..make sure primary and secondary channels look like this...reboot and rebench...you will see a difference..:toast: :woot:

---

I don't see anything there, what exactly did you do?

Here's my 2 WD caviar RE 250gb drives on DFI's 680i LT

http://i3.tinypic.com/6ksg4kp.jpg

Quote:

---

Originally Posted by Grinch

2 x 150 raptors on 680i motherboard:

http://www.supload.us/thumbs/default...0212203902.jpg


The trick to getting all the speed is here:

http://www.supload.us/thumbs/default/121.JPG


go into device manager...scsi & raid controllers...right click on nvidia 590/570/550 serial ata controller click properties..make sure primary and secondary channels look like this...reboot and rebench...you will see a difference..:toast: :woot:

---

I don't see any images...can you please repost and give us a clue on what you have done to improve the performance or Raid drivers with the 680i motherboard?

Thank you.:confused: :confused: