PDA

View Full Version : QuickBench, is it any good?



dragon_x
09-20-2001, 09:03 PM
Well I just received and setup my Gurus HD upgrade - one EZQ 80GB HD - Firewire. So far so good. Its a bit smaller than my old 20GB BellStor and the Intech Software is cool - even though I already own it. Not sure why Apple has failed to provide Firewire support with the Drive Setup utitily... o'well - I guess that is what 3rd party development is all bout, right http://macgurus.com/infopop/emoticons/icon_smile.gif

My old 20GB IDE/ATA Quantum (slave drive on ATA/33 bus) gets about 25MB/s in sustained read & writes with QuickBench. My 40GB Maxtor gets about 30MB/s sustained R/W with QB. My EZQ FW drive only gets about 12MB/s Sustained writes, but read performance is in line at about 28MB/s.... I though this had the OXFORD 911? I'm running MacOS 9.1 on a B&WG3/500 yosemite. I have a QPS 12/10/32 burner hooked up to the other FW port. I thought only those old PCI FW cards did bad in write performance.


One other oddity - which is why I question QB - is my 10K 36GB Cheetah is only rating about 7MB/s - that is not write at all... it should do at least 30 in sustained R/W tests. http://www.macgurus.com/ubb/eek.gif

I am running Apple drivers on everything, but I did 'update' the FW drive with HDSpeedTools. Intech says the drive is a '75GB Disk' but shows the other IDE & SCSI drives as '2 MacOS volumes'.

Nice to know the mail and deliveries still come through in this time of trouble.

So any ideas of why QBench is off?

------------------
One Nation, under God; with liberty and justice for all.

marrand
09-21-2001, 10:11 AM
Curious. Isn't it critical to defragment the HD just before running any kind read/write test? If so, couldn't your results be off because file fragments are scattered all over the place?

I am about to get into some serious testing of my HD's, and will acquire QB and ATTO. But also I would like to have the CD for macbench5. I used MB5 as downloaded version, and it doesn't have the sustained read/write tests. Does anyone have any idea where I can buy this CD?

dragon_x
09-21-2001, 11:45 AM
MB5 is very old - I might have a CD somewhere.
Maybe you could download it from the Guru's FTP site and then burn a copy?

Defrag would be important... but
the new drive did not have very many files on it - no need to defrag that.
all the IDE/ATA drives performed as expected - maybe they need to defrag, but what would only help their scores.
Even if the Cheetah is heavily fragmented - it should run MUCH faster - about 5X faster.

I did some simple FINDER COPY tests - a 59MB folder took about 20 seconds to copy on the Cheetah, but on the IDE/ATA drives it only took about 8 seconds. Something is slowing that drive down... http://www.macgurus.com/ubb/eek.gif

marrand
09-21-2001, 12:25 PM
I am still paranoid about defragging before running such tests.
If the file deposited by QB is contiguous, then defragmenting should not improve the scores. How much free space do you have on your old Cheetah? As I understand these tests, the QB takes its own file, then writes it on your HD and reads it. If the written QB file is fragmented, then the head will spend extra time in searching for the pieces. Could this be the reason for slowdown? Also, as I understand it, the seek time and throughput rates are two different beasts. Am I barking up the wrong tree?

Today I don't have CD-RW, so I can't burn anything. I would give anyone a hug and a kiss, with some extra money, if they could send the Macbench5 CD to me. (At this moment, my email addresses is blocked to stop the spam; just give me the word, and I will unblock it to receive your communication should you feel generous enough to give/share the CD with me).

DHB
09-21-2001, 01:51 PM
dragon_x...

Your Cheetah seems to have a broken leg http://macgurus.com/infopop/emoticons/icon_biggrin.gif

Of course, its risky to tweak a drive's hardware setup but I'd at least 'look' at how the Mode pages are configured. Read cache could be off...as well as Write.

There's also somethingelse, as I recall vaguely, that can REALLY slow things down... just can't remember right now.

I'll let you know if it comes to me.

kaye
09-21-2001, 02:22 PM
d_x,

Your broken leg Cheetah, 68-pin and hooked to what card? And I have yet to see a problem with QuickBench. And version 1.5, I think it is, will even do a 10MB file size test. k

[This message has been edited by kaye (edited 21 September 2001).]

marrand
09-22-2001, 05:22 PM
dragon mentioned an odditity. I just saw one on my system. Ok, I put aside all my reservations, and plunged into some quick testing - not rigorous, but just a quick-see. In my case, I was checking two IDE drives.
The results below by FWB using default settings, although ATTO confirmed the numbers. I will try Intech later.

1. IBM 75GXP 45GB - my main drive, Apple driver. Sustained read/write 15 MB/s. Avg. access - 9.2 msec

2. IBM 60GXP 60GB - via Sonnet ATA/66, FWB driver. Sustained read 33 MB/s. Sustained write 55 MB/s !!!!
Avg. access - 8.0 msec.

I got the same results no matter which drive I was booting from.
Question #1 - 55 MB/s - is that possible on my G3 rev. 1???? Even with G3 500 ZIF?
Question #2 - why the huge difference between the two drives? Yes, yes, 75GXP is an older model,
bought last year, while 60GXP was acquired only several months ago.

Does my 75GXP have a broken leg too? Can this difference be explained by saying IBM improved their drives?

dragon_x
09-22-2001, 07:57 PM
Kaye,

I am using QB 1.5 - and have run it from the TINY files sizes up to the slightly less tiny sizes of 10MB. It is all about the same. The Cheetah is slow with just about any file size.

I verified this in simple finder duplications. A 55MB folder would duplicate about 1/2 to 2/3 the speed as it would on an IDE/ATA drive.

I do not have a specific model number for the Cheetah, but it is fairly new (9 months old?) 68pin LVD 36GB model. It is connected to the Adaptec2940U2b (from you guys) for greater compatibility with X. I have a FOXXCOMM cable with terminator connecting the two and I am fairly sure that the 1.2 firmware is running on the SCSI card.

The legs may not be broken, but it sure seem that I have a lame Cheetah.

I remember MaddDog (or was it DogStarMan, I can not keep them straight http://macgurus.com/infopop/emoticons/icon_wink.gif) Had issues with some benchmarks - I think he was using QuickBench and ATTO Tools. It seems for some odd reason QB would slow things down.... Maybe I should search for that one?

The one thing that gets me though is the Finder tests... the 10K Cheetah should be WAY faster than the IDE/ATA drives (at least the older ones) with a file/directory duplication. I can initialize the drive with SoftRAID and maybe tweak it for better performance - if possible. Maybe the page mode stuff is slowing it down?

kaye
09-22-2001, 08:17 PM
Marrand,

Review for us what computer, what OS, etc.

d_x, I can't remember which dogxxx either, but I do remember that after he ran ATTO EPT, he had to reboot to attain normal drive speed. I always reboot after finishing with one test utility and before going to another utility. But I did try EPT tests, quit, then run another utility without the reboot. I could sometimes duplicate his problem. The ATTO EPT does something temporary sometimes to the drive that requires a reboot to fix. I have no idea what. I just accepted the fact that it does. k

marrand
09-22-2001, 10:57 PM
kaye,
my computer: G3 desktop rev.1, 384RAM, G3 500 Sonnet ZIF, built-in ZIP, Tempo Ultra ATA/66, two external SCSI's.
The "faster" internal IDE (IBM 60GXP) upper bay (above ZIP) connected to ATA/66 (profiler says SCSI bus 1).
The "slower" internal IDE (IBM 75GXP) in original place (were the orignal 4GB used to be).
Also have Orange Micro Firewire/USB2 card, with a scanner plugged in.
All HD's bootable, each with OS 9.0.4.
I hope that is enough info to answer my questions in previous post.

When I ran these tests (ATTO and FWB), I was booting from three different drives, and results were about the same.
However, these tests were not rigorous, just quickies to practice with them. I didn't even defrag the disks.
(But all my HD's are not severly fragmented anyway, since I optimize them often).

[This message has been edited by marrand (edited 22 September 2001).]

[This message has been edited by marrand (edited 22 September 2001).]

[This message has been edited by marrand (edited 22 September 2001).]

kaye
09-23-2001, 01:41 PM
marrand,

I'm in an area I know little about but I'm thinking the Beige G3 internal is ATA/33 while your Tempo card is ATA/66 which could very well account for the difference. k

dragon_x
09-23-2001, 03:37 PM
The IDE/ATA bus on the BeigeG3/gossamer boards is ATA/3 - or 16MB/s. I have installed ATA/100 drives in Gossamer boards and they work great - fast but are probably somewhat limited by the stock bus. It still blows away the stock drive though.

I got my PTProG3/500 up and running (I think one PCI card was bad). In any even the SCSI drives and cards are working as they should. I ran QB on these and the scores are as expected.
I ran them with MacOS 8.6 - VM off, 480MB of RAM - interleaved.
Off the MilesU2W bus - my Atlas V, 7200rpm 2 drive RAID-0 achieved about 50MB/s reads and 40MB/s write - sustained. My old Viking2's on the bus achieved about 20MB/s. On my BlueNote - both RAIDs (Barracuda 2 drive setup- RAID-0) both hit the max for the bus - about 18~19MB/s. I am using SoftRAID-2-2-2.

My B&W is also setup to run X - so if SoftRAID is the solution for the slow bench/performance then I hope there is an update - or maybe some built-in function of 10.1

marrand
09-24-2001, 09:59 PM
dragon writes: "My old 20GB IDE/ATA Quantum (slave drive on ATA/33 bus) gets about 25MB/s in sustained read & writes with QuickBench."
lasvegas writes: "the G3 Beige's built in IDE is ATA3 or 16MB/s"

dragon gets more than the port allows? what have I missed?

kaye
09-24-2001, 10:54 PM
marrand,

d_x's Quantum is on an ATA/33 bus. The Beige has an ATA/3 bus. They are not the same. The ATA/33 bus is faster. k

dragon_x
09-24-2001, 10:59 PM
Well the IDE/ATA ports on my B&WG3 are ATA/33, not ATA/3 - thus 25MB/s is not more than is allowed. http://macgurus.com/infopop/emoticons/icon_smile.gif

I just ran QB again - on my PTPro now with
MacOS 9.1, VM on, FileSharing on, normal ext and normal disk cache.
The extended test show everything is in line. The drives on the UltraNarrow bus get from 14MB/s to 19MB/s. The main boot partition gets about 50MB/s in sustained reads and 35MB/s in sustained writes. I am going to hook my old RAID box up to the AdaptecU2b and see how those drives do - if they do as expected, then its either an internal cabling issue or a drive issue http://www.macgurus.com/ubb/eek.gif

Maybe Gurus should setup a general drive forum? I mean this really does not fall into SCSI, IDE, FW or USB exclusively. Maybe some forums should be consulidated? http://www.macgurus.com/ubb/eek.gif

------------------
One Nation, under God; with liberty and justice for all.

dragon_x
09-25-2001, 09:31 PM
OK - I fired up the old RAID box - connected to my YOSE/G3-500 and the Apple OEM 2940U2b. For some odd reason now - QB is showing about 33MB/s sustained read & writes on the extended test... my SoftRAID volumes are also scoring fairly well - even the lowly Medalist pro 4Gx2 RAID setup gets about 23MB/s. Even the Finder duplication works quicker - copying the 50MB folder in about 5 seconds - before it took about 25 seconds.

WEIRD. The only thing different that I know of is the RAID box is connected - but that should not affect the speed of the internal drive? Maybe its that SUPER GRANITE TERMINATOR?
http://www.macgurus.com/ubb/confused.gif

Maybe it was the reboot?