krimkrim

Distinguished
Feb 23, 2006
8
0
18,510
Hi guys,

I have been having hard drives issues. My set-up is the follow
1xASUS P4P800 Deluxe Socket 478 Intel 865PE ATX Intel Motherboard (Oced to 3.0ghz)
1xIntel Pentium 4 2.8C Northwood 800MHz FSB Socket 478 Processor Model BX80532PG2800D
4xCORSAIR XMS 512MB 184-Pin DDR SDRAM DDR 400 (PC 3200) Unbuffered System Memory Model CMX512-3200C2PT
1x80GB EIDE
1x100GB EIDE
2xWestern Digital Caviar SE WD2500JD 250GB 7200 RPM Serial ATA150 Hard Drive
1xeVGA 256-A8-N347-AX Geforce 6800 256MB GDDR3 AGP 4X/8X Video Card (clocked to GT)
1xSound Blaster XtremeMusic
1xPSU Channel Well 450watt max

My problems have been with the 2x250GB hard drives. They sometimes boot up with the computer, sometimes dont. And when they do boot up I can't really copy files into the hd. It would get like cylindrical errors and if you format them, do scandisk, there is no bad sectors. What gives? Should I upgrade my power supply?
 

sturm

Splendid
Are these sata drives, ide drives, or both? Any cd drives?
Try unplugging any non-essential stuff like cd or dvd drives, case fans, neon lights and any hard drive not needed to boot, leave 2 250s plugged in. If they startup fine then I would definitely say to try another power supply. Although 450 should be enough.
 

krimkrim

Distinguished
Feb 23, 2006
8
0
18,510
Hi,

1x80 and 1x100 are IDE drives
2x250 are both SATA drives..

Currently I dont have CD/DVD drives on or any neon lights. The 1x80 and 1x100 never have problems even when 2x250 are acting up...

I will try your suggestion, but the boot up problem only happens once in a while. And as I type right now, the 2x250 are online and I was able to copy files in/out... and sometimes after copying gigs of files, it would have cylindrical error problems. It's just totally random.
 

aBg_rOnGak

Distinguished
Feb 23, 2006
481
0
18,790
if u hav any fren dat hav PSU wit higher max power load, then please ask them to lend it to u,500W or 550 may be?looking @ ur specs, i think ur system really need xtra juice. I only hav 250GB SATA-150 hd, 3 optical drv, 1 agp (ati 9250 8x), 1 neon lite, n i've set in my mind, "dis is it, any xtra device,i'll need xtra power"...
 

TabrisDarkPeace

Distinguished
Jan 11, 2006
1,378
0
19,280
I have been having hard drives issues. My set-up is the follow

P4 2.8(Oced to 3.0ghz)
1x80GB Hd
1x100GB hd
2x250GB HD

My problems have been with the 2x250GB hard drives. They sometimes boot up with the computer, sometimes dont. And when they do boot up I can't really copy files into the hd. It would get like cylindrical errors and if you format them, do scandisk, there is no bad sectors. What gives? Should I upgrade my power supply?

As you are overclocking from 2800 MHz to 3000 MHz your PCI and/or SATA bus may be overclocked 7% or so.... this will cause CRC32 errors between the IDE Controller (Southbridge, or 3rd party controller) and the interface on the HDD.

I suspect your 250 GB HDDs are of the SATA varient, while the others are of ATA100/133 using a more typical IDE/ATA controller / interface. As the SATA bus is very susceptible to even the slightest change in clock speed this is likely to cause problems, as you're overclocking.

eg: [IDE Cntr --- port --- cable --- HDD interface --- HDD]
The I/O between these points has a protocol, and if running overclocked the data / instructions to / from the HDD will become corrupted..... thankfully they implement CRC in the protocol to detect 'most' (but not all) forms of corruption.

Cylindrical Errors = CRC failures

See: http://en.wikipedia.org/wiki/CRC32

If you are overclocking be sure to understand all the theory behind the hardware, and what / which clocks, timings, etc is / are actually being changed, and potential impacts thereof. Trial & Error testing is for chumps and does not guarantee 100% stability. (Heck running at stock settings there days doesn't either --- Looks at Asus & MSI 'performace features').

Suggested fix:

Can you overclock and lock the PCI/AGP/PCIe/SATA host bus speeds, while just running the FSB asynchronously ?

Your HDD may also have being put into PIO Mode 4 (usually) by Windows as a 'protection', this will slow performance and CPU load (interupt thrashing of CPUs, not always shown in OS CPU usage) will be extremely high when accessing the affected HDD(s). You may want to force it back to Ultra DMA Mode 5/6 - Serial ATA 1.5 Gbps (various drivers label the modes differently).

For info see:
http://en.wikipedia.org/wiki/Programmed_input/output
http://en.wikipedia.org/wiki/Ultra_DMA

Note: If you are using a 3rd party HDD controller (eg: some types of integrated RAID) then it is slightly more complex than above, but still similar that the explaination works.

I still encourage overclocking, but only when all the 'risks' are known and avoided.

Sorry to push Wikipedia so much, but hey.... the info is there.

PS: Any data that has been corrupted already (eg: If you performed a defragmentation recently on the HDDs then potentially the entire contents of the HDD are corrupt..... potentially), will be difficult / impossible to 'uncorrupt' said data. Thus, you may need to re-install drivers when not overclocking to ensure stability.... You may need to replace each corrupted file as found on a case by case bases --- or in the case of games just re-install and repatch the entire game.

To Test:
For testing I suggest copying a 8 GB WinRAR archive to/from suspected HDDs, possibly back & forth multiple times, to confirm they work, and once copied, reboot (clears the OS cache), then perform a 'Test' operation on the 8 GB archive to confirm it was not 'corrupted in transit' so to speak. When that works you'll know it is very stable. - I've recreated your scenario here many times, and this test will likely 'fail' in your current configuration, try it before clocking back to stock settings just to see what I mean.

Tip: A similar test to the above is awesome for detecting 'slightly dodgy' ports on switches that rarely show up.. you can then flag each 'risky' port and avoid using them at large LAN events.
 

krimkrim

Distinguished
Feb 23, 2006
8
0
18,510
Wow, that's alot of information. Thanks!

I will read through this more when I get home tonight. I will try your suggestion.

The following is the more descriptive list of hardware i have:

1xASUS P4P800 Deluxe Socket 478 Intel 865PE ATX Intel Motherboard (Oced to 3.0ghz)
1xIntel Pentium 4 2.8C Northwood 800MHz FSB Socket 478 Processor Model BX80532PG2800D
4xCORSAIR XMS 512MB 184-Pin DDR SDRAM DDR 400 (PC 3200) Unbuffered System Memory Model CMX512-3200C2PT
1x80GB EIDE
1x100GB EIDE
2xWestern Digital Caviar SE WD2500JD 250GB 7200 RPM Serial ATA150 Hard Drive
1xeVGA 256-A8-N347-AX Geforce 6800 256MB GDDR3 AGP 4X/8X Video Card (clocked to GT)
1xSound Blaster XtremeMusic
1xPSU Channel Well 450watt max


How would you overclock it then?

While I am not at my home computer, I think currently I have

FSB = 211Mhz
DRAM Frequency = 400mhz
AGP/PCI Frequency = 72.73/36.36





Let me know, thanks!




I have been having hard drives issues. My set-up is the follow

P4 2.8(Oced to 3.0ghz)
1x80GB Hd
1x100GB hd
2x250GB HD

My problems have been with the 2x250GB hard drives. They sometimes boot up with the computer, sometimes dont. And when they do boot up I can't really copy files into the hd. It would get like cylindrical errors and if you format them, do scandisk, there is no bad sectors. What gives? Should I upgrade my power supply?

As you are overclocking from 2800 MHz to 3000 MHz your PCI and/or SATA bus may be overclocked 7% or so.... this will cause CRC32 errors between the IDE Controller (Southbridge, or 3rd party controller) and the interface on the HDD.

I suspect your 250 GB HDDs are of the SATA varient, while the others are of ATA100/133 using a more typical IDE/ATA controller / interface. As the SATA bus is very susceptible to even the slightest change in clock speed this is likely to cause problems, as you're overclocking.

eg: [IDE Cntr --- port --- cable --- HDD interface --- HDD]
The I/O between these points has a protocol, and if running overclocked the data / instructions to / from the HDD will become corrupted..... thankfully they implement CRC in the protocol to detect 'most' (but not all) forms of corruption.

Cylindrical Errors = CRC failures

See: http://en.wikipedia.org/wiki/CRC32

If you are overclocking be sure to understand all the theory behind the hardware, and what / which clocks, timings, etc is / are actually being changed, and potential impacts thereof. Trial & Error testing is for chumps and does not guarantee 100% stability. (Heck running at stock settings there days doesn't either --- Looks at Asus & MSI 'performace features').

Suggested fix:

Can you overclock and lock the PCI/AGP/PCIe/SATA host bus speeds, while just running the FSB asynchronously ?

Your HDD may also have being put into PIO Mode 4 (usually) by Windows as a 'protection', this will slow performance and CPU load (interupt thrashing of CPUs, not always shown in OS CPU usage) will be extremely high when accessing the affected HDD(s). You may want to force it back to Ultra DMA Mode 5/6 - Serial ATA 1.5 Gbps (various drivers label the modes differently).

For info see:
http://en.wikipedia.org/wiki/Programmed_input/output
http://en.wikipedia.org/wiki/Ultra_DMA

Note: If you are using a 3rd party HDD controller (eg: some types of integrated RAID) then it is slightly more complex than above, but still similar that the explaination works.

I still encourage overclocking, but only when all the 'risks' are known and avoided.

Sorry to push Wikipedia so much, but hey.... the info is there.

PS: Any data that has been corrupted already (eg: If you performed a defragmentation recently on the HDDs then potentially the entire contents of the HDD are corrupt..... potentially), will be difficult / impossible to 'uncorrupt' said data. Thus, you may need to re-install drivers when not overclocking to ensure stability.... You may need to replace each corrupted file as found on a case by case bases --- or in the case of games just re-install and repatch the entire game.

To Test:
For testing I suggest copying a 8 GB WinRAR archive to/from suspected HDDs, possibly back & forth multiple times, to confirm they work, and once copied, reboot (clears the OS cache), then perform a 'Test' operation on the 8 GB archive to confirm it was not 'corrupted in transit' so to speak. When that works you'll know it is very stable. - I've recreated your scenario here many times, and this test will likely 'fail' in your current configuration, try it before clocking back to stock settings just to see what I mean.

Tip: A similar test to the above is awesome for detecting 'slightly dodgy' ports on switches that rarely show up.. you can then flag each 'risky' port and avoid using them at large LAN events.
 

krimkrim

Distinguished
Feb 23, 2006
8
0
18,510
I just checked with my mobo settings and altered a few things:

CPU speed is now back to 2.8ghz

FSB = 200Mhz
DRAM Frequency = 400mhz
AGP/PCI Frequency = 66.66/33.33

I forced within BIOS to make all HDs to be at UDMA Mode 5

Rebooted and used winrar to rar 11GB worth of uncompressed photos (i shoot photos) into one big 11GB rar file. This is done on one 1x250GB SATA drive only.

I then copied the rar file from the 1x250GB SATA drive to the 2nd 1x250GB SATA drive. The copying failed at around 50-60% and I received cyclic redundancy check error.

For 2nd test, I copied the same file to the 1x100GB EIDE drive. Got the same error.


Any more suggestions? Should I get a better PSU?
 

TabrisDarkPeace

Distinguished
Jan 11, 2006
1,378
0
19,280
Is there anything in the BIOS not set to a default, or rather a typical 'industry standard' setting ? (or within an acceptable range of 'typical settings' that comes to mind) ?

Windows / IDE drivers can ignore the BIOS setting for the HDDs aswell, suggest checking IDE Controllers in Device Manager instead.

Must say you've got a very uncommon fault there, Slim chance of encountering it under normal conditions, Could be memory or power supply or HDD unit related....

No custom mod lighting, or heaps of fans running off the PSU or anything similar ?

I'll download the manual for your mainboard just to check a few things over and refresh my memory a little.

Is it possible it is only affecting the 'target' HDD of the copy ?


Got 2 approaches you can take, thankfully at the same time (God bless multi-tasking operating systems) - Combining these two tests will place the system under a fair load, and thus the PSU aswell:

================================================

Download a tool to monitor / read the SMART statistics from your HDDs:
Google: "hddhealth v21 b159 beta.exe", Recall it was a very useful tool in a similar situation I encountered some time ago, certain stats indicate it is the drive itself, others could be IDE controller, dodgy cable, bad cable connection, or interface on the HDD itself playing up.

Instead of searching for it just go here and get latest version: http://www.panterasoft.com/ ; the SMART standard doesn't change often (well defined) so it is still up to date despite its age.

Once installed copy the file again, see what it reports.

================================================

Try 4 x Prime95 instances at once:

Win32 / x86 - http://mersenne.org/gimps/p95v2414.zip
Win64 / x64 - http://mersenne.org/gimps/p64v2414.zip


Use these settings:
Options, Torture Test;

Min FFT size (in K) 1024,
Max FFT size (in K) 8192,
Run FFTs in-place [Unchecked],
Memory to use (in MB): [25% of your memory (for each instance, but run 4 at once).]
Time to run each FFT size (in minutes): 15

If it is memory / MCH (Northbridge of chipset) that will find the fault far faster than MemTest86/+

================================================
 

TabrisDarkPeace

Distinguished
Jan 11, 2006
1,378
0
19,280
Checking: http://au.asus.com/products4.aspx?l1=3&l2=12&l3=31&model=178&modelmenu=1

Indicates the board does support AI NOS, and other similar dynamic clock changing features - (Non-endorsed ones, unlike Intel SpeedStep or AMD Cool'n'Quiet / PowerNow!) -, consider disabling AI NOS if it is somehow enabled (even though it might appear off).

The board also sports 4 x SATA ports, 2 are off the Intel ICH5R, and 2 off the Promise RAID / SATA controller on the board.
- Are you able to move the SATA HDDs to the opposite device (eg: from Promise SATA to Intel SATA, or vice versa ?). - Warning: Do not do this if you are booting from the SATA controller(s) though.
 

krimkrim

Distinguished
Feb 23, 2006
8
0
18,510
Oh yeah, I just tried a 2nd test, I copied the same file from the same 1st 1x250GB SATA to the 1x100GB EIDE drive. Got the same error.

There is no custom mod lighting.

1xCase Fan running currently

If it were a problem with memory chip, wouldnt that affect my 2xEIDE drives as well?

By the way, I used to have a DVD+-R reader/burner. And it used to burn just fine and at certain point, it just decided to give, again, random cyclic redundancy error while burning or reading discs. I have then removed the drive altogether and have gone w/o a DVD drive for a long time.



I am going to try the tests right now. will report back.
 

TabrisDarkPeace

Distinguished
Jan 11, 2006
1,378
0
19,280
If it were a problem with memory chip, wouldnt that affect my 2xEIDE drives as well?

Good point, so likely not memory, (athough, also noticed it supports Asus HyperPath, which may force PAT on, using a i865PE - not an Intel sanctioned setup so it may lead to problems).

By the way, I used to have a DVD+-R reader/burner. And it used to burn just fine and at certain point, it just decided to give, again, random cyclic redundancy error while burning or reading discs. I have then removed the drive altogether and have gone w/o a DVD drive for a long time.

Was this while you were overclocking or not ?

Because of this I am leaning very slowly towards power supply. (Although 450w max, 350w sustained is ample for your current hardware.... unless the PSU is very old or dodgy that is :p )

Since clocking / changing back to stock settings, do the HDDs still have trouble spinning up sometimes, maybe their motors are defective ?

Using HDDs that support NCQ, Native Command Queuing, on a SATA controller that does not support NCQ, thus NCQ is off, sometimes can cause very similar issues to this.
 

krimkrim

Distinguished
Feb 23, 2006
8
0
18,510
I couldn't get the memory test to run 4 instances... but so far so good, with one instance.

HD Health only reports back that ATTRIBUTE seek error rate changes from 200 to 100

and RAW Read Error rate changes from 61 to.... i cant read the log because i can't scroll nor export..

I don't think ihave had problems with drives spinning up and down, but i did have problems with drives just not coming on at all. I would have to switch off the computer entirely (not just power button in front, but also the switch on PSU itself) and then power it on to get the 2x250 sometimes one of the 2x250GB to be online.
 

TabrisDarkPeace

Distinguished
Jan 11, 2006
1,378
0
19,280
HD Health only reports back that ATTRIBUTE seek error rate changes from 200 to 100
This is normal be haviour.

and RAW Read Error rate changes from 61 to.... i cant read the log because i can't scroll nor export..
More concerned about the HDDs themselves now.

Have you tried powering the SATA HDDs from a different molex / SATA power connector (if possible) on the PSU - As the other drives appear fine.

Western Digital Data-Life-Guard Diagnostics are available from:
http://support.wdc.com/download/

As for 4 x Prime95 sessions, you can unzip it 4 times into 4 different folders. Normally just double clicking the EXE from WinRAR does this automatically but to 4 seperate temp folders.

Good Hunting.
 

krimkrim

Distinguished
Feb 23, 2006
8
0
18,510
HD Health only reports back that ATTRIBUTE seek error rate changes from 200 to 100
This is normal be haviour.

and RAW Read Error rate changes from 61 to.... i cant read the log because i can't scroll nor export..
More concerned about the HDDs themselves now.

Have you tried powering the SATA HDDs from a different molex / SATA power connector (if possible) on the PSU - As the other drives appear fine.



By the way, I got the CRC error(copying 11GB file) while HDHealth was running and HD Health did not report any error on any of the drives other then the rate change --- which were already happening before the copying.

I will try different SATA power connector. Need to dig around a bit when I get home tonight. Oh yeah, just realized, my Geforce 6800 is a 'special edition'. It shouldve been an Ultra but most likely failed some stability test at Ultra settings and was sold at regular 6800 speed. It needs two molex power connectors.

I really appreciate your help, by the way.
 

krimkrim

Distinguished
Feb 23, 2006
8
0
18,510
Coming back to update.

I have swapped the molex connector. The 2x250 are now connecting to the same molex connector that's providing the 2xEIDE hd's power.

So far so good.

I rared a 8gig file on 2nd 250gb drive and copied over to 1st 250gb drive -- no issues what so ever. I then unrar the files, again, no error.

I am now in the process of raring another 8 gig file on the 1st 250 gb drive to do another round of tests. Hopefully this will end the mystery. If it does, thanks in advance, man!