Clevo 5600D P4 3.06GHz HT - Thx Henrik!

Well after some drama I finally got hold of a P4 Northwood 3.06GHz HT 533MHz FSB CPU, here in the UK for my AJP [Clevo] 5600D.

It is running fine on Henrik's custom Sager 1.03.03 BIOS [the one he grafted the 64MB ATi bios module into]

The issues originally found with the USB Controller, invisible cpu usage and modem failure never happened. Whether this is due to WinXP SP2, the new 845 chipset inf drivers or the latest VIA USB 2.0 drivers I'm not sure.

HT appeared to be enabled without a repair install - SP2 again? Two processors show in device manager as expected, but as of now I don't have any HT software to prove the feature, maybe later I will search for something.

The heat doesn't seem to be a major issue, showing an average rise of about 5 degrees above my 2.4Ghz P4. Several hours of playing Day Of Defeat online caused a peak of about 65 degrees, which was back under 50 in a jiffy with maxfan ... Thx again Henrik

Overall performance seems to be pretty impressive. mp3s encode at 10 times playing speed - the track counter looked like a stopwatch
I benchmarked a rise of 300+ to 6800+ in 3DMark2001SE over several runs and an average of 50 fps in the Half-Life 2 Video Stress Test at 800x600x32bit with maximum candy settings [the test comes with Counterstrike Source beta]

Here's to Henrik ... ... "Crazy Swede" indeed

btw, if you haven't taken the plunge, do it now...
3.06GHz Northwood P4s are becoming rarer than rocking horse crap!!!

I did some more homework on this - my bad, I must have been too eager to get it running, I always do more research first

I had 2 cpus in device manager but not the 2 graphs in device manager when I later looked.

I did an OTT install, noting that I should press F5 at the "press F6 to install 3rd party RAID or SCSI driver" prompt at the start of Win XP setup.
This brings up the hidden Computer Type menu, which selects different OS HALs to use.

http://support.microsoft.com/default...b;en-us;299340

I selected "ACPI Multiprocessor PC" as I had read on several sources, did the install and rebooted. I had to go into the BIOS and reset the defaults to get my first boot but after that it rebooted fine.

I then opened SP2 setup, it was then I noticed the drastic system slowdown, 40% cpu usage, but by then I was in mid install, so left it to complete.

On reboot I had the two independent cpu traces in task manager... and 40% cpu use at idle.

I disabled AC97 in device manager and the cpu fell to 0-1% as normal. I also was unable to use the centre pair of USB ports even though they were enabled, only the right side pair were working.

I then reenabled AC97 and disabled the USB components in turn, finding that the cpu fell to 0-1% again when I disabled one of the usb host companion controllers. I still had only the right pair of USB ports in use but disabling the device stopped the 40% cpu load. I had read several useful hints here:

http://www.ntcompatible.com/thread24591-1.html

As far as I could see, Win XP had added 8 new interrupts to the IRQs. My AC97 and USB were on the same one, 17. OK, just reassign them to see what happens... XP had greyed out the manual options. No sweat, change "P 'n' P OS" to "No" in the BIOS... There is no option, in fact there is no IRQ setup menu at all. Guess who has been working on too many desktop PCs lately

These devices are both handled by the 82801 chip, which probably requires them to be on separate IRQs to function without conflict.

Some progress at least then, the problem has been solved by someone, but the solution is not readily available to us because of the primitive laptop BIOS we have

I still have a few drivers to sort out from the OTT install as they were non-logo and refused by WinXP setup.
My modem reinstalled OK, I still have the newer LAN and custom Omega ATi drivers to sort out. I am running HT successfully, but won't see if there are system hits until I reinstall those drivers as my display is running very poorly atm... as if the driver was missing I note.
I am also down one pair of USB ports, but for me this is an issue I would like to solve, as I use them a lot when I am away from home.

I don't currently have the patience to dig out the link, but I read that Clevo are perhaps once again admitting that the 5600D can run a 3.06GHz P4. Maybe a BIOS will surface to solve our problems.

Clevo! Let me at those IRQs

Misc References:
http://support.microsoft.com/default...b;en-us;810231
http://www.geek.com/news/geeknews/20...0110009694.htm
http://support.microsoft.com/?kbid=309283
http://support.microsoft.com/default.aspx?id=817027
http://www.techzonez.com/forums/show...oto=nextnewest

I remembered something that could point to a minor difference in the BIOS.

I flashed the hybrid BIOS before when I still had the 2.4GHz P4.
The same symptoms occurred, but at the time I put it down to not having a HT cpu fitted.
This problem could be the BIOS being subtly different for the 5670's newer USB configuration.

The 5660/5600 is a pot mess of chipsets

845E Core Logic
[a term from the Clevo 5600D blurb]

845MP/MZ Mobile ACPI features
[as identified by Intel's chipid utility]

VIA Rev 5 USB Controller
[probably an early attempt to implement USB 2.0]

As a side-note...
I previously looked at the Intel Application Accelerator [IAA] drivers which are supposed to improve IDE performance, but as I couldn't identify the chipset clearly I never tried them.
Now they are quite old I am not sure if they would bring any performance benefit and might cause conflict with newer technologies.
Anyone tried them? While we are on a quest for performance I thought it relevant to mention them.

Thanks for clearing up the 5620 issue Henrik, I wasn't sure about the older clevo models and meant it to be an incentive for others to do some research like we have with the 5560

D'oh! Do you know how long it took to track down a 3.06 Northwood here in the UK?
Now you tempt me to start hunting for a SL77R 3.2

I will try IAA 2.2.2, especially as it has your recommendation

The one thing that has eluded me so far is the best chipset drivers to use. I am currently using the Intel inf driver 4.10.1012 from the clevo 5600D download page, but the latest inf driver is 6.0.1.1002 from intel.

One last thing for now, I have suffered massive cpu load again. I was playing Half-Life: Day of Defeat online, with TeamSpeak 2 voice software running and my normal 70+fps framerate slumped to ~20fps in-game.
I tried lowering the game sound quality, lowering teamspeak's sound quality, changed it from directsound to wave, nothing had any effect. I quit teamspeak and the cpu load dropped by ~40%.
In non-HT I can run the same game and voice app with no performance hit.
The audio issue seems to be more than a simple conflict, perhaps there are more inconsistencies in the bios.

I will experiment further, but as fascinating as this has been, my need for usb ports and regular gaming habit may just result in a return to a non-HT setup if I can't find a solution

My problem was not heat related I'm afraid.
Also it did not happen with the P4 3.06 fitted before I enabled HT.

The cpu load dropped immediately when I disconnected from the teamspeak server [sudden decrease in audio to process].
The cpu loading again appeared immediately as soon as I reconnected to the teamspeak server [sudden increase in audio to process] regardless of cpu temp.

The extra kernel loading was the same as when I have all USB controllers enabled.

Both Day Of Defeat and Teamspeak are sound intensive. With AC97 they will be relying on the codec, the 82801 and the CPU.

The teamspeak application did not cause the load either, the connection to the server with a sudden increase in audio processing demand caused it.

I believe there is an underlying problem with using the 5670 BIOS on the 5660, possibly with managing the 82801, given that the problem is usb/sound related and the change in specifications between the two machines.

I don't know if you are a gamer, but this kind of requirement for game and comms is pretty normal when playing online. If the AC97 codec interferes with HT, as it apparently does in this case, HT is not worth the effort for me.

I do not get these problems with HT disabled!!
My machine runs just as hot.

As I said, HT is fascinating...
but it definitely has its problems with the 5660 until the usb/audio problem is solved... by the rest of the mysterious 5660 HT BIOS perhaps?

I am almost ready to revert to non HT as the internet is running out of answers for me.

Well would you believe it, I seem to have reduced the problem with Realtek AC97 drivers v3.63 released this week

I'm already using the latest intel 845 .inf drivers, IAA 2.2.2 [Thx m8 ] and was using AC97 v3.62 before.

There is still significant "system" cpu load though.

Yes, I have read that AC97 does not work well with HT and my relatively high demands won't help.
I also read a number of reports that regarded early HT cpus to perform better with it disabled, and to be outperformed in HT mode by non-HT cpus.

I am not trying to diminish your achievement Henrik, you've done a great job. I'm just letting people know of the problems I have encountered with enabling HT.

The BIOS? I'm not trying to hint that I know of one, but searching brings comment by others of a 1.03.03 BIOS that was withdrawn by Clevo, who retracted the claim of the 5600D being HT capable, blaming it on "heat issues".

Was there a 5600D 1.03.03 BIOS? Did anyone download it before it was retracted? Someone must know...

Ohnoes! /me blocks ears ... la-la-la-la-la-la-la [can't hear you]

Good news, after a damned good driver cleanout and reinstall, I have had no more hidden CPU kernel usage