"ASUS P5Q Deluxe discussion".

[Rylo]

I wonder if anyone got the following problem or a solution for this:

I get very very very low repair speeds when using Quickpar to fix broken archives.

When parring speeds peaks to 600MB/s for a few seconds, then it keeps @ 100 MB/s for some time, slowly dropping to 60-70 MB/s. When i open 'My Computer' or any other small handling, speed drops to 5-25 MB/s and i get hicups/ system hangs.

I'm using a WD 6400 AAKS which ran fine in my previous system, i'm using this with a E8400 and 2GB ram.

If i need to be more specific, tell and i'll give more info.

Hope someone knows whats wrong.

[truehighroller]

Also Asus removed the 1003 bios from the P5Q dlx website!!

I wonder why lol

[CrAsHnBuRnXp]

I wasnt aware that the P5Q Deluxe had a performance level option. Is this new with BIOS 0803? It never used to show up in Memset. Mine is at 13 right now.

My FSB strap is set to auto, do you think if I made it 333 or 400 that it would help get my RAM out of the 9xxMHz range and into the 1xxx?

[R@ptor =]]]

I tried with both yellow and black slots and yellow slots were actually better ocing. It may not be the case for all tho.

Nobody more?

[truehighroller]

Nobody more?

I will when I get home from work. I am interested to see how this works. So I should be able to reach a higher speed with my CPU right?

[truehighroller]

I have mine plugged into the second ones right now hell I didn't know it made a difference.

[Cob]

I wasnt aware that the P5Q Deluxe had a performance level option. Is this new with BIOS 0803? It never used to show up in Memset. Mine is at 13 right now.

My FSB strap is set to auto, do you think if I made it 333 or 400 that it would help get my RAM out of the 9xxMHz range and into the 1xxx?

PL has always been there.

[CrAsHnBuRnXp]

PL has always been there.

Must be blind then.

[R@ptor =]]]

I have mine plugged into the second ones right now hell I didn't know it made a difference.

I don't know too, I have not tested it.
but, like i said, in the asus manual, say to put in the yellow slots to better oc, but i don't know if there was any difference.

[coolpain]

I wasnt aware that the P5Q Deluxe had a performance level option. Is this new with BIOS 0803? It never used to show up in Memset. Mine is at 13 right now.

My FSB strap is set to auto, do you think if I made it 333 or 400 that it would help get my RAM out of the 9xxMHz range and into the 1xxx?

if u are using ram modules that are supposed work over 1000mhz(8000, 8500, 9200,so on) but you are having problems in achieving such speed, straps 333 or 400 might help. However, if u r using 800mhz stock speed modules, there's no promise that they can achieve over 1000. Maybe they will since many 6400's can be OCed over 1000 but maybe not.

[CrAsHnBuRnXp]

if u are using ram modules that are supposed work over 1000mhz(8000, 8500, 9200,so on) but you are having problems in achieving such speed, straps 333 or 400 might help. However, if u r using 800mhz stock speed modules, there's no promise that they can achieve over 1000. Maybe they will since many 6400's can be OCed over 1000 but maybe not.

Ive had these sticks at 1095 before. So they can get over 800Mhz speeds.

Which strap would be better?

[IanB]

I wonder if anyone got the following problem or a solution for this:

I get very very very low repair speeds when using Quickpar to fix broken archives.

When parring speeds peaks to 600MB/s for a few seconds, then it keeps @ 100 MB/s for some time, slowly dropping to 60-70 MB/s. When i open 'My Computer' or any other small handling, speed drops to 5-25 MB/s and i get hicups/ system hangs.

I'm using a WD 6400 AAKS which ran fine in my previous system, i'm using this with a E8400 and 2GB ram.

If i need to be more specific, tell and i'll give more info.

Hope someone knows whats wrong.

I have a "professional" interest in QuickPAR, as a few years ago I wrote my own PAR/PAR2 client and corresponded heavily with Peter Clements, the QuickPAR author to gain pointers into the algorithm and techniques to speed up the processing. http://www.i-asm.com/yDecode/ - unfortunately my ground-up rewrite stalled badly due to to time and life pressures.

While I am not entirely sure of Peter's approach to file and memory handling in QP, I am very familiar with the PAR2 algorithm therefore. I can tell you that QP doesn't do anything different to any other data-processing utility, it uses the same memory and file-handling API calls, nothing special or undocumented, but it does A LOT more of them than most other apps. PAR2 processing is VERY intensive on memory and disk access, and QP ruthlessly exposes deficiencies in your overall system, probably more so than WinRAR and other similar utilities. I have never understood why it is not used as a benchmark or stability tool, actually...

Without detailed information (the size of the dataset you are trying to repair and the number of missing blocks, the block size and the total number of blocks in the set) it's hard to make a specific diagnosis, but I can make some general comments. First, to repair a set, EVERY file in the dataset must pass through memory, and a "window" into every block must exist in memory at the same time to allow the parity reconstruction. This means extremely large numbers of disk accesses, both read and write (as the repaired data is written into a new file). It's possible that either the SATA controller or the disk itself is a bottleneck here. NCQ might or might not help. Sadly, QP (at least the version I am aware of from a few years ago) was only able to repair files into the same folder as the original source dataset, so it relies on a single disk interface for all data accesses and cannot use a second disk as target to reduce the data/file access bottleneck on a single disk.

In addition, QP relies on some potentially very large lookup tables being in memory as well as the source data, one of which can reach 4GB in size if you had a set split into the maximum number of blocks and a large number of missing blocks (and enough repairs, thankfully an extremely unlikely combination). This means that the size of your available physical memory is crucial to good repair speed, as otherwise large amounts of data needed to be in memory at the same time might be being swapped out to disk and you end up with disk thrashing. I suspect that QP is sensitive to processor cache size as well.

So I don't think this is likely to be a disk incompatability issue, but it might be throwing up a deficiency in the mainboard or simply in the combination of hardware you are using to do the repairs you want. More RAM should definitely help and will reduce the need for paging. Try adding more and see if it helps. For my own software, I will be interested to follow your results, so keep me posted privately if you wish on whether and how you resolve this.

[coolpain]

Ive had these sticks at 1095 before. So they can get over 800Mhz speeds.

Which strap would be better?

try both 333 and 400 and see which one gives you a reasonable/achievable frequency setting. It mainly depends on ur fsb.

[ThugsRook]

e8200/P5Q/Mushkin PC6400

CPU Margin Enhancement:
OPTIMUM = highest FSB
PERFORMANCE =
COMPATIBLE = lowest FSB

actual performance only varies slightly between settings, at the same fsb of course.

[khaydin]

I wonder if anyone got the following problem or a solution for this:

I get very very very low repair speeds when using Quickpar to fix broken archives.

When parring speeds peaks to 600MB/s for a few seconds, then it keeps @ 100 MB/s for some time, slowly dropping to 60-70 MB/s. When i open 'My Computer' or any other small handling, speed drops to 5-25 MB/s and i get hicups/ system hangs.

I'm using a WD 6400 AAKS which ran fine in my previous system, i'm using this with a E8400 and 2GB ram.

If i need to be more specific, tell and i'll give more info.

Hope someone knows whats wrong.

Are you using BIOS version 0506 that came with the board? When i first got my board i was having weird issues where sometimes i would go to my computer and the computer would totally freeze up, then 5 seconds later it would unfreeze. I updated to a new BIOS and the problem went away.

[Rylo]

I'm using 0704 BIOS;

I tried several cables and sata ports, a quick error scan on the WD6400AAKS turned out negative (no errors). Still getting the same thing tho', verrryyyyyyyy slow repair rates, seems something can't keep up. Which test can i do to exclude key components, like RAM or HDD ?

@Ian; pmd

[khaydin]

I'm using 0704 BIOS;

I tried several cables and sata ports, a quick error scan on the WD6400AAKS turned out negative (no errors). Still getting the same thing tho', verrryyyyyyyy slow repair rates, seems something can't keep up. Which test can i do to exclude key components, like RAM or HDD ?

@Ian; pmd

Are you using AHCI/RAID or IDE? I also changed to RAID at the time i upgraded the BIOS, so maybe it was actually the RAID/AHCI that fixed it.

[Rylo]

IDE - i've heard/read AHCI can lead to BSODS (?), isn't raid setting more something for when using multiple disks?

noticed that unrarring a big archive is painfully slow aswell - and systems hangs/hicups etc.

[khaydin]

IDE - i've heard/read AHCI can lead to BSODS (?), isn't raid setting more something for when using multiple disks?

noticed that unrarring a big archive is painfully slow aswell - and systems hangs/hicups etc.

Use AHCI if you don't want to use raid, it enables new features like NCQ that the normal IDE doesn't. Yes it will cause BSODs when windows is loading if you don't install the driver in windows before changing the setting in the BIOS.

http://en.wikipedia.org/wiki/Advance...ller_Interface

If you install the Intel Matrix Storage Manager/Console before changing the setting in the bios you should be fine. But, if you do get a BSOD you can always go back into the BIOS and change it back and try again later. I think if you're using Vista you don't even need to install the driver since it's supported out of the box.

Check this out also if you try it with Vista: http://support.microsoft.com/kb/922976

[Polizei]

I'm torn between this board and the Biostar... which do you think would net me the best clocks on air and water?

Which is better on quads?

[truehighroller]

I say this.

[Rylo]

Use AHCI if you don't want to use raid, it enables new features like NCQ that the normal IDE doesn't. Yes it will cause BSODs when windows is loading if you don't install the driver in windows before changing the setting in the BIOS.

http://en.wikipedia.org/wiki/Advance...ller_Interface

If you install the Intel Matrix Storage Manager/Console before changing the setting in the bios you should be fine. But, if you do get a BSOD you can always go back into the BIOS and change it back and try again later. I think if you're using Vista you don't even need to install the driver since it's supported out of the box.

Check this out also if you try it with Vista: http://support.microsoft.com/kb/922976

I might try this tomorrow if i have some time - is this BAD performance the cause of IDE mode then? Shouldn't it be fine in this mode aswell, ill try a hd tune error scan tomorrow aswell

[rusty]

use extra HD w/ extra OS. I corrupted a few installs playing.

Adjust each way until you find stability is lost, up and down. mid-point will be stable.

If adjusting 2 GTL values(quad), only adjust one at a time.


Personally, I found .61/.63 good for 45nm, .65/.67 for 65nm, quads, of course. .61 for E8500, .67 for E6600.

I wonder why so low .61 hmm
my core 3 always fail in prime so I assume it's gtl ref needed more volts

[cadaveca]

Dunno, it's gonna vary from cpu to cpu anyway, too bad we didn't have more exact coices, but then that'd only further confuse things.

[truehighroller]

I say that yes it is easier to overclock with them in the yellow because I could never reach the following,