Welcome, Guest. Please Login
 
  HomeHelpSearchLogin FAQ Radified Ghost.Classic Ghost.New Bootable CD Blog  
 
Pages: 1 2 3 ... 10
Send Topic Print
Comparing HDDs - strange results (Read 139488 times)
Christer
Übermensch
*****
Offline



Posts: 1360
Sweden


Back to top
Comparing HDDs - strange results
Mar 4th, 2004 at 5:22pm
 
Hello all!

I have two HDDs, a new 120 GB Hitachi 7K250 with 8MB cache and a three years old 40 GB IBM 60GXP with 2 MB cache.
The 7K250 is Primary Master and the 60GXP is Primary Slave.

The rearmost partition on both (16-18GB) are FAT32, dedicated to Ghost Images.
I have the same Images on both HDDs, created to the target partition on the 60GXP and copied to the 7K250.

When I do integrity checks, the "transfer rates" are bothering me. The latest Image, which is approximately in the middle of both partitions (8-9 GB from the end of both HDDs), is checked at 142 MB/min on the 7K250 compared to 397 MB/min on the 60GXP.

I promptly did a benchmark in AIDA32 with the following results:

Quick Linear Read - 7K250 = 58.3-28.1 MB/s (46.8 avg)
QLR - 60GXP = 38.8-18.5 MB/s (31.2 avg)

Random Read - 7K250 = 57.6-27.9 MB/s (46.0 avg)
RR - 60GXP = 35.2-17.8 MB/s (30.7 avg)

Buffered Read - 7K250 = max 73.9 MB/s (73.7 avg)
BR - 60GXP = max 74.4 MB/s (74.2 avg)
(min readings omitted, low dips due to unknown interference)

Average Access - 7K250 = 11.8-13.4 ms (12.6 avg)
AA - 60GXP = 12.2-13.8 ms (13.0 avg)

The 7K250 should outperform the 60GXP and it does in the AIDA32 benchmark but why the very low performance when running the integity check in Ghost?

I would expect Ghost to create an Image quicker to a different HDD than to a different partition on the same HDD and that goes for restoring too but not for integrity checks which are read only operations.

Ideas anyone?

Thanks for Your time,
Christer
 

Old chinese proverb:
If I hear - I forget, If I see - I remember, If I do - I understand
 
IP Logged
 

El_Pescador
Übermensch
*****
Offline


Thumbs Up!

Posts: 1605
Bayou Country, USA


Back to top
Re: Comparing HDDs - strange results
Reply #1 - Mar 4th, 2004 at 9:08pm
 
Quote:
... The 7K250 is Primary Master and the 60GXP is Primary Slave.

Christer

What is assigned respectively to the Secondary Master and the Secondary Slave?  Also, what is your O.S. and are there any "fine-tuned" settings in the  BIOS re: HDDs, i.e., UDMA On or Off?

El Pescador


 

...
WWW  
IP Logged
 
Rad
Radministrator
*****
Offline


Sufferin' succotash

Posts: 4090
Newport Beach, California


Back to top
Re: Comparing HDDs - strange results
Reply #2 - Mar 4th, 2004 at 9:50pm
 
I have an image stored at the slow end of an 80-gig drive that is several years old. To be honest, I don't recall who made this drive. I think it's an IBM 120-GXP, but I can't tell from the device mgr. But I bought it when 80-gig drives were the ones to get. The drive has 3 partitions: 12+34+34.

Just checked the image & Ghost reports between 500 & 600-MB per min.

So your 60GXP number seems reasonable. The only problem is the low check speed of the master. I suspect the image may have become fragmented while transferring.

I wouldn't defrag it, but check its level of fragmentation to see what you're getting. A fragmented image would slow down the read speed. I had a similar problem once. (I've had just about every problem   Cry  )

Since your benches are good, there should be nothing physicaly wrong with the drive.

I use Diskeeper for frag checks. They have a demo if you need.

http://www.executive.com/diskeeper/diskeeper.asp

Ghost reboots into DOS, so I don't think the OS matters.

I'm using a 2.2-GHz P4 CPU, which will affect the speed of the check. What CPU do you have?

It's not an SATA drive, is it?
 
WWW  
IP Logged
 
Christer
Übermensch
*****
Offline



Posts: 1360
Sweden


Back to top
Re: Comparing HDDs - strange results
Reply #3 - Mar 5th, 2004 at 6:19am
 
Hello guys,
thanks for Your replies!

El Pescador,

Quote:
What is assigned respectively to the Secondary Master and the Secondary Slave?

The Secondary Master is a CD/R-RW running at PIO4 and the Secondary Slave is a DVD-ROM running at ATA33.

Quote:
Also, what is your O.S. and are there any "fine-tuned" settings in the  BIOS re: HDDs, i.e., UDMA On or Off?

The OS is WinXP pro but since Ghost runs from DOS, I agree with Radministrator that it shouldn´t matter.

BIOS settings are checked and in XP's device manager both drives run in UDMA mode 5, which is verified by the benchmark.
I doubt that it would matter because the rates are way below PIO4 which is 16.7 MB/s. The rate of the 7K250 is some 2-3 MB/s but the 60GXP stands out at 6-7 MB/s ......  Undecided ...... !

Radministrator,

Quote:
I suspect the image may have become fragmented while transferring.  

I wouldn't defrag it, but check its level of fragmentation to see what you're getting. A fragmented image would slow down the read speed.

First, I actually created each Image twice, one to each HDD.
The create speed was also much lower on the 7K250 but I thought that might be due to reading from and writing to different partitions on the same HDD.
I have not yet restored from either of the HDDs.
The resulting check speed on the master was the same low rate as when checking the copied Images.

Regarding defragmentation, all Images are split in 650 MB chunks if I in the future should want to burn to CDs. When Ghost creates the Images, it leaves a small slack between each span. These small spaces are actually written to during the next creation, resulting in the second, third and so on, Images becoming fragmented. I had a fragmentation level of ~30% but if 6 out of 19 large files become fragmented, thats the percentage.

I know that Images shouldn´t be defragmented but copying the files also defragments.
I decided to reformat the 60GXP partition and copy the files, one by one, from the 7K250 to it. All copied Images passed the integrity check and not a single file is fragmented.
Next, I reformated the 7K250 partition and copied the files, one by one from the 60GXP to it with the same results from the integrity check and fragmentation level.

Quote:
I use Diskeeper for frag checks.

I used to use Norton Speed Disk but I have come to the conclusion that it fights the XP "on the move" defragmentation and relocation of files.
Speed Disk has one idea of what is the optimum and as soon as XP regains control, it starts moving things around to suit its preferences.
It actually feels like, after running Speed Disk, the computer slows down a wee bit but after a few restarts it picks up normal speed again.
My reaction to this is to quit using Speed Disk and let the native XP tool handle it.
I use the "Analysis" funtion of Speed Disk, though, for a second opinion on fragmentation levels.

Quote:
What CPU do you have?

AMD Athlon Thunderbird, 1 GHz with 266 MHz bus speed.

Quote:
It's not an SATA drive, is it?

Nope, it´s a PATA but it at least has 8 MB cache.

Christer
 

Old chinese proverb:
If I hear - I forget, If I see - I remember, If I do - I understand
 
IP Logged
 
Rad
Radministrator
*****
Offline


Sufferin' succotash

Posts: 4090
Newport Beach, California


Back to top
Re: Comparing HDDs - strange results
Reply #4 - Mar 5th, 2004 at 7:16am
 
2 things:

1. i don't think it's a good idea to mix dma & non-dma devices on the same chanel. i think this forces ALL devices to run in pio mode. i think different DMA devices can run at diff DMA speeds, such as 1 at DMA33 & another at DMA100 on the same controller, but once you put a pio device, i think ALL devices on that channel must run at/in pio mode. i could be wrong, tho.

2. there are different ways to measure fragmentation. suppose a partition contains 10 files, and 1 of them is fragmented into 2 parts. that would be reported as 10% fragmentation. but suppose that same file is broken into 100 parts or 1000 parts, or a million. technically, the partition would STILL be reported as 10% fragmentation (only 1 file in 10). but this would be a much worse fragmentation that the previous example mentioned. so you have to delve deeper into exactly how bad the files are bfragmented.
 
WWW  
IP Logged
 
Christer
Übermensch
*****
Offline



Posts: 1360
Sweden


Back to top
Re: Comparing HDDs - strange results
Reply #5 - Mar 5th, 2004 at 8:18am
 
Quote:
1. i don't think it's a good idea to mix dma & non-dma devices on the same chanel. i think this forces ALL devices to run in pio mode.

That´s a different issue regarding the performance of the opticals which share the Secondary. The HDDs share the Primary and are both ATA100.

I have experimented a bit with the setup of the hardware and mixed as follows:

Primary Master - 60GXP - ATA100
Primary Slave - CD-R/RW - PIO4 (MultiWord 2)
Secondary Master - empty front mounted rack in waiting for delivery of new HDD
Secondary Slave - DVD-ROM - ATA33

I did an AIDA32 benchmark of the 60GXP. The results were the same as when the 60GXP was alone on Primary.
(I saved and still have the results for Linear Read and Buffered Read, with and without the CD-R/RW as Primary Slave. If You or anyone else want to see them, just holler!)

In the early days of IDE, the controller set itself to the lowest transfer rate of any device connected to it.
In recent years they came up with Independent Device Timing which means that when the respective device takes its turn on the channel, it transfers at its own rate. My motherboard is three years old and obviously supports IDT.

Quote:
2. there are different ways to measure fragmentation.

Yes and that´s why I use Speed Disk for a "second opinion".

When there is fragmentation they (XP and SD) always differ in the percentage but now, for my two partitions holding Ghost Images, they agree that the fragmentation level is 0%, the number of fragments per file is 1.00 and the number of "additional file parts" (or whatever it's called in english?) are 0.

I´m 99.99% certain that the files are not fragmented.

Christer
 

Old chinese proverb:
If I hear - I forget, If I see - I remember, If I do - I understand
 
IP Logged
 

Rad
Radministrator
*****
Offline


Sufferin' succotash

Posts: 4090
Newport Beach, California


Back to top
Re: Comparing HDDs - strange results
Reply #6 - Mar 5th, 2004 at 11:27am
 
i am tired, so maybe i'm missing something. however u say "The HDDs share the Primary and are both ATA100", but show the pio4 cd sharing the primary channel.

are u sure Independent Device Timing pertains to pio with dma, and not just different levels of dma? (i.e. 33, 66, 100, 133)

okay, i think you ruled out fragmentation, altho, if that image is not vital, i would be interested to see what happened following a defrag of that drive.

so we still don't know why the low check speed on the new 120-gigger, right?
 
WWW  
IP Logged
 
Ian Dunster
Guest




Back to top
Re: Comparing HDDs - strange results
Reply #7 - Mar 5th, 2004 at 12:47pm
 
Hi Christer.

The 7K250 should outperform the 60GXP and it does in the AIDA32 benchmark but why the very low performance when running the integity check in Ghost?

This may shed some light:

http://linux.tu-varna.acad.bg/~lig/freedos/

I don't think Ghost has a DMA controller - If not, then it will be operating in PIO mode.


Ian.  8)
 
 
IP Logged
 
Ian Dunster
Guest




Back to top
Re: Comparing HDDs - strange results
Reply #8 - Mar 5th, 2004 at 1:01pm
 
Sorry Christer, bit premature in my previous post! Roll Eyes

Ghost can use DMA.

Have a look at your 'Ghost Diagnostic Error File' and see if it is using DMA.

http://service1.symantec.com/SUPPORT/ghost.nsf/pfdocs/1999021911433525


Ian.  8)
 
 
IP Logged
 
El_Pescador
Übermensch
*****
Offline


Thumbs Up!

Posts: 1605
Bayou Country, USA


Back to top
my Re: Comparing HDDs - strange results
Reply #9 - Mar 5th, 2004 at 1:34pm
 
You guys are all operating way above my present level of PC knowledge !!!

With the answers to my questions in hand, I must admit I am "stumped".

El Pescador
 

...
WWW  
IP Logged
 
Christer
Übermensch
*****
Offline



Posts: 1360
Sweden


Back to top
Re: Comparing HDDs - strange results
Reply #10 - Mar 5th, 2004 at 5:10pm
 
Radministrator,

Quote:
i am tired, so maybe i'm missing something. however u say "The HDDs share the Primary and are both ATA100", but show the pio4 cd sharing the primary channel.

I´m sorry for the confusion. I shouldn´t have mentioned my experiments with hardware combinations since they are history. The current setup is:
Primary Master - 7K250
Primary Slave - 60GXP
Secondary Master CD-R/RW
Secondary Slave - DVD-ROM

Quote:
are u sure Independent Device Timing pertains to pio with dma, and not just different levels of dma? (i.e. 33, 66, 100, 133)

No, I´m not 100% sure but the AIDA32 benchmarks indicated that it does.

Quote:
okay, i think you ruled out fragmentation, altho, if that image is not vital, i would be interested to see what happened following a defrag of that drive.

Well, they are all vital in my strategy for sensible "roll back" points but since I have them in duplicate, I could defrag the Image partition on the 7K250. My guess is that the defragmenter will start and then finish immediately but we will not know until I´ve done it. I´m not at home now but will get back Sunday evening.

Ian,

Quote:
Have a look at your 'Ghost Diagnostic Error File' and see if it is using DMA.

I don´t think that I have one. The only place it could be is on the second ghost boot floppy which is in the drive when Ghost runs. There is no additional file(s) on it.

El Pescador,

Quote:
With the answers to my questions in hand, I must admit I am "stumped".

Welcome to the club ......  ??? ......  Undecided ...... !

Christer
 

Old chinese proverb:
If I hear - I forget, If I see - I remember, If I do - I understand
 
IP Logged
 

Christer
Übermensch
*****
Offline



Posts: 1360
Sweden


Back to top
Re: Comparing HDDs - strange results
Reply #11 - Mar 7th, 2004 at 4:41pm
 
I have "defragmented" the Image partition on the 7K250. No defragmentation took place and the time to do the integrity checks are the same.

I noticed that the cluster sizes differ with 32kB on the 7K250 and 16kB on the 60GXP but that shouldn´t make such a difference ......  ??? ...... or should it?

Christer
 

Old chinese proverb:
If I hear - I forget, If I see - I remember, If I do - I understand
 
IP Logged
 
rocknroll
Dude
*
Offline


I love YaBB 1G - SP1!

Posts: 11


Back to top
Re: Comparing HDDs - strange results
Reply #12 - Mar 8th, 2004 at 3:13am
 
Does it make any difference what order you run the integrity checks? Like is the second one faster than the first, or is the 7K250 always slower?
 
 
IP Logged
 
Pelson
Guest




Back to top
Re: Comparing HDDs - strange results
Reply #13 - Mar 8th, 2004 at 3:25am
 
all things being equal, i would expect the drive with 32kb clusters to go faster than the one with 16kb clusters.

since you have both hard drives on the same controller, the pio-dma issue is not relevant because that only pertains to drives/devices on the same controller (primary, secondary).
 
 
IP Logged
 
Christer
Übermensch
*****
Offline



Posts: 1360
Sweden


Back to top
Re: Comparing HDDs - strange results
Reply #14 - Mar 8th, 2004 at 5:47am
 
Quote:
Does it make any difference what order you run the integrity checks? Like is the second one faster than the first, or is the 7K250 always slower?

I have six different Images created at different points of the install procedure. Each one is 2.5-3 times slower to check on the 7K250, no matter which gets checked first.

I have created an additional Image of the system to each of them, as the system is at present time. I checked the Images and also restored using each Image and the one on the 7K250 is 2.5-3 times slower in all operations.

Quote:
all things being equal, i would expect the drive with 32kb clusters to go faster than the one with 16kb clusters.

I would expect that too but it is the opposite!

I had a thought which pertains to the setup of the drives respective firmware. I´ve heard about "normal seek mode" and "quiet seek mode" where the latter should yield lower performance.
The AIDA32 benchmarks doesn´t indicate any difference so, I buried that thought.

Christer
 

Old chinese proverb:
If I hear - I forget, If I see - I remember, If I do - I understand
 
IP Logged
 
Pages: 1 2 3 ... 10
Send Topic Print