Printable View
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.
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.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.
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.
pssssssst shhhhhhhhhhhhoooooshQuote:
Originally Posted by Haltech
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
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
I think this is more evidence that there is something wrong with the i680 RAID...Quote:
Originally Posted by politenessman
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!
This looks like the graph of my HDTach with single Seagate 200GB SATA2 drive on ICH8.Quote:
Originally Posted by Lestat
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
Thanks... This is very reassuring!Quote:
Originally Posted by ZeroOne
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
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?
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.Quote:
Originally Posted by bboy1234
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