HDD configs and overclocking

[Mikki]

I apologize that this is slightly off-topic, but the simple fact is the folks in GH could care less about this info, and I know my mates here in the OC forum will...

Results of testing I did a week or so ago, sorry to all that it took so long to post! Disclaimer: This is my system, and the results are general, I have no idea if they'll be different on other systems, with different OS's, different drives, etc.

System:
P4B533-E
2xWD 80G SE's
1XWD 80G 7200rpm 2mb buffer
WinXP

HDTach >>> Sandra

Raid vs no Raid:
Raid: 46175.0 >>> 43526
no Raid: 40505.5 >>> 28341
All configs resulted in raid having better performance on my system

1 part vs 2 parts:
1 part Raid: 46175.0 >>> 43526
2 part Raid: 45770.0 >>> 39753
1 part Single: 40505.5 >>>28341
2 part Single: 40468.3 >>> 28332
These tests yielded more of a mixed result, but mostly it was 1 big partition that won here

JB drive vs BB drive
JB: 39292.0 >>> 26921
BB: 33124.8 >>> 23552
No comparison, the BB is a dog compared to JB in all configs...

Single drive on raid channel vs on ide channel
...for those who want to use raid card for more devices...
on raid: 40291.3 >>> 30054
on ide: 40505.5 >>> 28341
and so it went, very mixed results, sometimes HDTach was higher, sometimes lower, sometimes they were the same. sometimes results were better for on raid, sometimes not, I'm drawing no conclusions here. if anyone wants me to post all these results, let me know...

 

[LED]

2 120Gig 8MBers cache in Highpoint RAID 0 with AMD 2000XP 3-Partitions

Win 98SE

Win XP

 

[Mikki]

LED, is that a direct comparison between Win98 and XP, with that setup?

 

[LED]

Yep same set-up with Win98Se in the 1st Partition (3 gig) and XP in the following(7Gig) FAT 32
I found it less favorable on 1 Promise RAID set-up with a Gigabyte MoBo same speed, RAM, Vid card etc.

 

[Mikki]

Nice scores, BTW, and thanks for the info...

 

[Freshbrain]

Hi, I found this thread quite interesting after all, I have two IBM 120GXP in a RAID 0 with an onboard Promise 133 controller, I get about 55k in Sandra at 151FSB, but my scores seem to be lower at lower FSB speeds... could that be right??? or just an error in the sandra test???

What I thought is that maybe the higher PCI frequency makes the Promise controller work a little better, maybe??? thanks

 

[THUGSROOK]

damnit! ~ i was hoping to find my results to shed some more light on the "1 drive 2 parts" config, but i seem to have misplaced them

anyways,
i ran some similar comparisons testing small C: partitions hoping for better C: performance. (1x WD800JB)
what i found was- as the C: partition got smaller the C: became faster :Q
at the 11g mark is where it leveled off but even smaller sizes still showed improvements.

right now im running with a 11g C: partition... Win98SE/IAA/IUMDA onboard
HDtach = max49318 min44217 avg48467 8.8ms
Sandra = 32091

wish i could find those other scores for yas
and yes i realize some of my scores are better then Mikki's RAID scores :Q
Mikki ~ what size partition did you use?

 

[Mikki]

I used a 20g part. I wasn't testing to get the best performance, just the difference between 1 part and 2 parts...

Now that I'm done, and have pondered over my results, I want to know why your results and mine differ so much. If I remember right, your single drive results were better than your raid results (?). If that's the case, why aren't mine? Also, your scores kill mine. Why? We've got similar setups, same drives. and if LED's reults are solid, then my theory that the OS had something to do with it is false. I realize I'm not gonna get the same scores cause I'm not you , but I should be able to get sorta close. I'm gonna mess with this part size and test some more...

 

[LED]

Freshbrain 1st off when running between 133-166 in general your PCI clock is divide by 4 not 3 so @ 151FSB the Promise OnBoard is going @ appx 38 as I haven't seen any HD's that can handle 50FSB and SiSoft is a theoritical benchmark so you it could be correct depending on which one you are using...as you can see in mines...Also the IBM 120GXP series are comparable to the WD's with 8MB cache in RAID. You can read up on it HERE


Here's some RAID info :
"Win98SE likes a 2gig partition to operate from. It likes front of the drive and FAT32 with 16K Clusters on a FAST drive.
WinXP needs about 4 or maybe 5 gigs to operate from and an exact placement of the MFT from the front of the drive or partition. Less than 4gigs and defrag can't work and more than 8gigs is a waste because of my next procedure.
Next> I force all applications/proggies to load to a set aside partition called PROGS and because dual-boot 98SE and WinXP use it; it is roughly 5gigs in size and FAT32 @16K Cluster Size. NOTE for now I make WinXP use FAT32 because of the ease of Cluster sizing. NTFS done "correct" is not simple. I choose simple at this time. "...by RGone

 

[Mikki]

LED, nice info, thanks. I'll have to do some more testing I guess.... Tho I'll admit with all the XP stuff going on, I may go back to Win2000 or even 98...

 

[THUGSROOK]

i dont know Mikki.
im not doing anything special at all.
no custom tweaks, no special drivers, nothing.
just win98se, Intel INF, and IAA - thats all.

i used 2 maxtors for my RAID setup ~ was only slightly faster then a single WD800JB. not worth the trouble at all.

 

[Mikki]

I don't know what to say mate, short of testing the same part size, trying Win98, etc...I just don't get it. It's the same chipset, so I can't see it being the mobo...do you remember even from my first tests (like back in June) you thought something was up with my drives?

 

[THUGSROOK]

honestly Mikki i think something is weird with those P4B533-E boards.
they ARE NOT the same as a regular P4B533.

 

[SupermanCK]

Originally posted by: THUGSROOK
damnit! ~ i was hoping to find my results to shed some more light on the "1 drive 2 parts" config, but i seem to have misplaced them

anyways,
i ran some similar comparisons testing small C: partitions hoping for better C: performance. (1x WD800JB)
what i found was- as the C: partition got smaller the C: became faster :Q
at the 11g mark is where it leveled off but even smaller sizes still showed improvements.

right now im running with a 11g C: partition... Win98SE/IAA/IUMDA onboard
HDtach = max49318 min44217 avg48467 8.8ms
Sandra = 32091

wish i could find those other scores for yas
and yes i realize some of my scores are better then Mikki's RAID scores :Q
Mikki ~ what size partition did you use?

wow...that's a high avg for single jb drive thugs..do you have anything installed on the drive?

 

[THUGSROOK]

yes ~ it is the only hd in my system.

C: 11.1gb / 7.7gb free
my OS, all programs and all games are all on this drive.

D: 63.3gb / 48.8gb free
my storage drive - all files and backups are on this drive

i cant really explain its incredible performance other then the partition size used.

when i was testing different configs and partition sizes (i had 3 hds at the time) i used exactly the same data in all tests. (drive copy)

 

[PlatinumGrill]

Thanks for the info...too bad my motherboard doesnt have raid capabilities.........yet

 

[Mikki]

Thugsrook, I think you're right, I think there's something weird about the -E. I'm gonna order a P4B533 next week to play with...

PlatinumGrill, you'd be better off with a pci raid card anyway, if that's the way you want to go. Check out Thugsrook's results, his raid config was only slightly better than a single drive, so unless you just wanted to tinker like I did, or if you use programs that actually take advantage of raid, I wouldn't bother...

 

[THUGSROOK]

may as well Mikki ~ its not like there is any other board to choose from

youve been putting so much effort into all your testing trying to map out your systems limitations... it would be real nice if it worked out for ya and actually did find that the -E boards are flakey

 

[Mikki]

Yeah, I've got most of my stuff written down or printed out (yes Thugs, most of us have to write our stuff down...hehe ), it'll be interesting to compare. I really wanted to try an Alpha and a chipset cooler on this board to see what I could get, but none of the places I buy from have Alpha's in stock (wth?). I'm done playing with raid, so I don't need the board anymore, and I know I've got a good cpu...

 

[THUGSROOK]

at this point i think a board swap is a good idea.

plus youll be one of the 1st ppl to compare the two

 

[Mikki]

Messed around all night trying to bring my scores up and have had no luck. Tried different partition sizes, different file systems, and IAA/INF. It's gotta be something with the board. I'm at the end of my rope and on my way to a P4B533...

 

[JuicyFruit]

i got the -e just for the raid but i had hoped my 2.53 would hit 3ghz. seems like i'm going down in fsb every time i run prime.

 

[Mikki]

I did the same thing, I just wanted to tinker with the raid. Give that cpu some time, you aren't done yet...

 

[Mikki]

Got some good/interesting news everyone...I changed over to Win98 SE to make sure it wasn't my mobo that was hindering my scores (ruled out everything else), and lo and behold my scores shot way up:
HDTach: 47956.4
Sandra: 31775
Not quite what this drive is capable of, but I didn't bother configuring Win98 all proper cause I don't intend on keeping it. I only installed bare minimum drivers and such to get what I wanted.

 

[THUGSROOK]

:Q