hard drives & raid - benchmark and compare!

[Angry_Games]

ok some oddities...i just got started testing...instead of going and lying down and being sick i am intrigued so will just get more sick haha

 

Hitachi 80GB SATA II drive

i used the tool:

http://www.hitachigst.com/hdd/support/download.htm

 

Feature Tool (v1.97)

 

to enable only 3.0Gb/s transfers. I did NOT enable Spread Spectrum Clocking yet.

 

take a look at my results:

 

 

 

 

 

and compare them with the same 2x Hitachi SATA II drives in the same raid config, same partition size, same EVERYTHING (just only at default 1.5Gb/s transfer before I used the tool)

 

 

hrmmm...what the...i seem to have lost my 2x Hitachi SATA II RAID-0 benches (or did I forget to do them???)...

 

i will have to redo them grrrr

 

anyway...its not apples to apples cuz im gonna compare them to the 3x Hitachi SATA II drives:

 

 

 

 

dont worry, i will of course run the 3.0Gb/s test on 3x and 4x drives, as well as redo my 1.5Gb/s test with 2x drives...got to always have consistent tests to compare!

 

back to testing for me...!

[Angry_Games]

ok, i used the util to enable Spread Spectrum Clocking on the hdd's:

 

 

 

 

 

 

 

 

and then i re-did the 2x RAID-0 drives with all defaults (1.5Gb/s + Spread Spectrum Clocking disabled)

 

 

 

 

 

so looks sorta the same...except look at the burst-transfer speeds when 3.0Gb/s is enabled!

[soundx98]

your 2x36G raptor ATTO http://www.dfi-street.com/forum/showpost.p...03&postcount=44

KILLS the 2xHitachis screenshot

 

same NV Raid/motherboard bioses?

 

Hope ya feel better

 

EDIT: Oops - is that the Sil 3114 controller?

[Angry_Games]

but wait, there's more!!!

 

3x Hitachi SATA II drives, 1.5Gb/s enabled + Spread Spectrum Clocking disabled:

 

 

 

 

 

versus

 

3x Hitachi SATA II - 3.0Gb/s enabled + Spread Spectrum Clocking enabled:

 

 

 

 

 

only real difference I can see is the burst times are incredibly fast compared to 1.5Gb/s mode

 

otherwise...looks about the same to me. I'd do 4x drives, but the 4th drive is a SATA I drive and i doubt it would be a valid test with only 3x SATA II and a single 4x SATA I (but you never know i suppose...when i feel well i might try it)

[Angry_Games]

ps:

 

ALL BENCHMARKS (except the 2x Raptors) WERE RUN WITH THE FOLLOWING SPECS:

 

DFI NF4 SLI-DR (3/16/2005 BIOS, 4.82 NVRAID ROM)

Athlon64 3500+ @ 2750Mhz (250x11)

2x256MB Buffalo BH-5, 2-2-8-2, 3.4v vdimm, 1.450v + 110% vcore

Skyhawk 570w power supply

LiteOn 52x CD-ROM

Floppy drive

2x 6600GT Chaintech SLI, 76.41 beta drivers

 

Nvidia 6.53 chipset drivers

 

all tests used the NV IDE driver installed

[red930]

Looks like there are two different drives available at the egg.

 

$63.00 Hitachi 80GB 7200RPM SATA Hard Drive, Part# 0A30356, OEM Drive Only

 

Model# 0A30356

Item # N82E16822145082

 

http://www.newegg.com/app/ViewProductDesc....-145-082&depa=1

 

and

 

$66.99 Hitachi 80GB 7200RPM SATA Hard Drive, Model HDS728080PLA380 Part# 0A30358, OEM Drive Only

 

Model# HDS728080PLA380

Item # N82E16822145075

 

http://www.newegg.com/app/ViewProductDesc....-145-075&depa=1

 

The HDS728080PLA380 part# 0A30328 drive is the one that I have and I'm pretty sure AG has. You can find it listed on the Hitachi site.

 

The part# 0A30356 drive is nowhere on the Hitachi site. I sent an email to customer service at Hitachi to see what they say.

 

ExRoadie

 

 

EDIT

 

NEW INFO

Response (RefNum:508-147-854-0)

1 message

 

--------------------------------------------------------------------------------

[email protected] Tue, Apr 5, 2005 at 11:26 AM

To: [email protected]

ExRoadie,

Thank you for contacting Hitachi Global Storage Technologies.

 

The two part numbers are for the same exact drive.

Hitachi Global Storage Technologies sells the same drive model under

different part numbers, depending on which sales channel they are sold

through.

 

Regards,

Matt

 

_______________________________

Hitachi Global Storage Technologies

US toll-free: 888.426.5214

Fax: 507.322.2419

[email protected]

www.hitachigst.com

_______________________________

 

Looks like you can save the $3.99 and get the cheaper drive.

[soundx98]

Thanks for that COMPLETE explanation Angry!

 

Gotta ask, is that the same Buffalo BH-5 you toastered with so long ago?

 

so much RAM, so little time:)

[dpadmore]

Ok looking at the screenies AG posted it looks to me as if 3Gb/s can have a measurable effect as the line graphs were nearer 140-160MB/s with 3 drives and 3.0Gb/s enabled and nearer 120MB/s with 3 drives and 1.5MB/s. The average read value is also noticably higher with 3.0 enabled (~140 vs~120). As for the burst speed increase, I wonder if this is caused by the interfaces increased speed when reading the cache memory but is obviously limited in sequential read by the mechanical speed limitations of the drive. It seems that SCC has little or no effect and so there is no point in using it likewise with just 2 drives both 3.0Gb/s and SCC has little effect and are probably not worth the risk of BSOD.

 

One additional point, can anyone explain to me where the 3.0Gb/s value comes from as my understanding was that SATA1 is 150MB/s=1200Mb/s and SATAII is therefore 300MB/s=2400Mb/s so where does 3Gb/s come in?

 

EDIT: ***Q for AngryGames (when he feels a better of course)***

 

Do you think the upgrade to NvRaid 4.82 that you have due to the alpha 3/16 bios you are running has an effect? Is 4.81 bug free or if not does a public beta have 4.82?

 

Sorry to bug you when you don't feel great. Get well soon and all that.

[red930]

One additional point, can anyone explain to me where the 3.0Gb/s value comes from as my understanding was that SATA1 is 150MB/s=1200Mb/s and SATAII is therefore 300MB/s=2400Mb/s so where does 3Gb/s come in?

Your math is based upon a perfect 8bit data word. That doesn't happen even in a perfect world since most data communication has some type of error correction. You should use a 9bit data word at the very least.

 

You have to include the data overhead such as error correction in the total throughput. SATA is 1.5Gb/sec raw. SATA II is 3.0Gb/sec raw. Actual data throughput will always be less due to lost and re-transmitted data packets. Add in the provision for NCQ data and you have the "lost" bandwidth accounted for.

[dpadmore]

Your math is based upon a perfect 8bit data word. That doesn't happen even in a perfect world since most data communication has some type of error correction. You should use a 9bit data word at the very least.

 

You have to include the data overhead such as error correction in the total throughput. SATA is 1.5Gb/sec raw. SATA II is 3.0Gb/sec raw. Actual data throughput will always be less due to lost and re-transmitted data packets. Add in the provision for NCQ data and you have the "lost" bandwidth accounted for.

 

Got it. Thanks for the clarification, had forgotten about parity etc.

[red930]

I edited the response from Hitachi into the original post...

 

http://www.dfi-street.com/forum/showpost.p...1&postcount=113

 

ExRoadie

[Angry_Games]

all of my SATA II drives are:

 

HDS728080PLA380

 

my SATA I drive is a VSLA380