Testing / comparing : Intel D975XBX2 / Asus P5B DX ***56K WARNING***

[gtj]

Did u get s-ata driver to work with 64bit Vista?

I just recived my BX2. Installed Vista Business 64. So far I have probelems with the message; s-ata driver did not instal... I also have the issue with mobo cant locate my other 4 harddrives.

My last probelm is, I can`t update Bios, using the .exe file. And I have no floppy

If you're using the Marvell SATA Controller (red/blue ports) there's a new driver posted today at Intel's site. It supports Vista 32/64.

[stmckin]

IDCC has never worked stable-ly for me to make changes, my suggestion would be to try the bios method.

so you booted with the maintenance jumper on, looked at the watchdog, and confirmed it's actually not there or that it's off? cause coincidentally 280 is where watchdog kills me too...... I haven't tried back flashing and then checking, and it doesn't make a lot of sense, but it seems awfully coincidental....

mine works fine on 2663 with watchdog off, 1.45, 400fsb.. etc.

[minsc_tdp]

IDCC has never worked stable-ly for me to make changes, my suggestion would be to try the bios method.

so you booted with the maintenance jumper on, looked at the watchdog, and confirmed it's actually not there or that it's off? cause coincidentally 280 is where watchdog kills me too...... I haven't tried back flashing and then checking, and it doesn't make a lot of sense, but it seems awfully coincidental....

mine works fine on 2663 with watchdog off, 1.45, 400fsb.. etc.

Yeah Watchdog doesn't exist on the jumpered Maintenance menu. I guess I'll try this without IDCC but I'm not optimistic. I tried 300 FSB via the BIOS and got nowhere.

Do I need to bake all these new parts in at their higher voltages using Orthos for a few hours/days before attempting higher clocks?

[argylesocks]

Yeah Watchdog doesn't exist on the jumpered Maintenance menu. I guess I'll try this without IDCC but I'm not optimistic. I tried 300 FSB via the BIOS and got nowhere.

Do I need to bake all these new parts in at their higher voltages using Orthos for a few hours/days before attempting higher clocks?

as gtj states:

Make sure you've completely removed power from the motherboard whenever you make a voltage change. I.E. Pull the plug and wait for the little green LED on the MB to go out. If you don't the VRM doesn't pick up the new voltage.

[minsc_tdp]

Make sure you've completely removed power from the motherboard whenever you make a voltage change. I.E. Pull the plug and wait for the little green LED on the MB to go out. If you don't the VRM doesn't pick up the new voltage.

What's VRM?

I did not pull the power when changing the voltage, I see it does the double-boot thing, and IDCC in Windows shows the voltages where I set them. 1.5 for the CPU.

Is it lying?

I think something is seriously wrong to not be able to push 1 MHz past 279 MHz exactly, with bios 2333 and the latest bios from a few days ago, at any strap, with voltages higher than what most people use for 400 FSB.

This is all on a brand new CPU, board, and RAM that hasn't been used too much. Maybe a week of on and off usage, and the high voltages only got maybe an hour or so of time in Orthos to make sure they were stable.

[gtj]

What's VRM?

I did not pull the power when changing the voltage, I see it does the double-boot thing, and IDCC in Windows shows the voltages where I set them. 1.5 for the CPU.

Is it lying?

I think something is seriously wrong to not be able to push 1 MHz past 279 MHz exactly, with bios 2333 and the latest bios from a few days ago, at any strap, with voltages higher than what most people use for 400 FSB.

This is all on a brand new CPU, board, and RAM that hasn't been used too much. Maybe a week of on and off usage, and the high voltages only got maybe an hour or so of time in Orthos to make sure they were stable.

VRM is Voltage Regulator Module. IDCC is NOT your friend. Do all your changes from the BIOS and do the complete power cycle after changing voltages.

Don't try to overclock your memory and cpu at the same time. Use the memory calculator in my sig keep your memory underclocked while you raise the FSB.

[minsc_tdp]

VRM is Voltage Regulator Module. IDCC is NOT your friend. Do all your changes from the BIOS and do the complete power cycle after changing voltages. Don't try to overclock your memory and cpu at the same time. Use the memory calculator in my sig keep your memory underclocked while you raise the FSB.

I will avoid IDCC but I did already try and duplicated the 279 to 280 MHz=Crash condition via the BIOS.

This may sound stupid but I'm a little unclear on how to do a complete power cycle immediately after changing voltages. There's no option to "save and shut down" only save and reboot basically. I don't want to force a power off in the middle of the reboot. So do I have to use Maintenance mode, since when I exit that, it tells me to immediately shut down?

I've kept the memory at ref 200 and mem 400 (=1:1), isn't that severely underclocked vs. the automatic mode of ref 266 and mem 667 (=4:5)? It makes no difference, 280 FSB = crash.

I'm still having trouble understanding how the ref freq relates to the FSB potential, for example, what's the safest ref and mem freq to keep the ram as underclocked as possible while ramping FSB?

133/200 is the lowest but that's a 2:3 ratio, so maybe that's not good.
266/667 is the default which is a 4:5 ratio.
333/400 is the lowest RAM ratio at 5:3 and I don't think I've tried this. Is *that* the lowest setting for the RAM?
What about timings? I've loosened them AS FAR AS THEY GO (6-5-5-18) for RAM rated 5-5-5-4.

Does the memory setting recommendation change as you go from 279 to 280 MHz? Then 300? Then 330? And so on? Or is the lowest ref and mem frequency the lowest no matter what the FSB is that?

I've been reading this thread for days. I'm on page 60. And it's still unclear how to set the memory to be the most stable while doing FSB clocking.

Does the CPU need to be broken in with Orthos for a few hours at 279 before it can go higher? It seems like a ton of people here are taking this stuff fresh out of the box and putting 1.45 volts and 380 FSB *average* with no problems, and here I am stuck in the 200's...

I'm thinking it's the Watchdog On in the latst BIOS that's hurting me, and I'm just unlucky with the 2333 bios. I think I'll flash back tonight to the latest, load defaults just to be sure, set the volts and watchdog off in maint mode, shut down fully, and start clocking and using a boot CD with memtest before trying to jump straight into Windows to keep my disk safe.

I started OC'ing after reading only about 40 pages of this thread, I hadn't gotten to the Watchdog part yet, heh

[gtj]

I will avoid IDCC but I did already try and duplicated the 279 to 280 MHz=Crash condition via the BIOS.

This may sound stupid but I'm a little unclear on how to do a complete power cycle immediately after changing voltages. There's no option to "save and shut down" only save and reboot basically. I don't want to force a power off in the middle of the reboot. So do I have to use Maintenance mode, since when I exit that, it tells me to immediately shut down?

I've kept the memory at ref 200 and mem 400 (=1:1), isn't that severely underclocked vs. the automatic mode of ref 266 and mem 667 (=4:5)? It makes no difference, 280 FSB = crash.

I'm still having trouble understanding how the ref freq relates to the FSB potential, for example, what's the safest ref and mem freq to keep the ram as underclocked as possible while ramping FSB?

133/200 is the lowest but that's a 2:3 ratio, so maybe that's not good.
266/667 is the default which is a 4:5 ratio.
333/400 is the lowest RAM ratio at 5:3 and I don't think I've tried this. Is *that* the lowest setting for the RAM?
What about timings? I've loosened them AS FAR AS THEY GO (6-5-5-18) for RAM rated 5-5-5-4.

Does the memory setting recommendation change as you go from 279 to 280 MHz? Then 300? Then 330? And so on? Or is the lowest ref and mem frequency the lowest no matter what the FSB is that?

I've been reading this thread for days. I'm on page 60. And it's still unclear how to set the memory to be the most stable while doing FSB clocking.

Does the CPU need to be broken in with Orthos for a few hours at 279 before it can go higher? It seems like a ton of people here are taking this stuff fresh out of the box and putting 1.45 volts and 380 FSB *average* with no problems, and here I am stuck in the 200's...

You can safely power off as soon as you press "Y" after F10 in the BIOS.

I think your problem is that your using a 200 MHz reference frequency. It makes the 975X MCH internal timings VERY tight. So even though 200/400 is a 1:1 ratio, the MCH is operating as though it's running at a relatively slow 800FSB.

Go to 266/533

[minsc_tdp]

266/533

That was one of the first alternate settings I tried and no luck. But it was early in the process. I'll let you know what happens with the latest bios and watchdog disabled.

[Blacklash]

Would someone with Intel's ear please tell them to make a BIOS express update .exe that works under a 64 bit OS?

It didn't work under XP Pro x64 and doesn't work under Vista x64 either. Get with the program guys. I am disappointed.

[Speederlander]

How is the on-board sound? If I ditch my creative X-Fi will I notice a difference in movies or games?

[minsc_tdp]

Well, regarding this pissy board bombing at exactly 280.... I went back to the latest bios, restored defaults just in case, disabled Watchdog this time (and EIST/C1E, Slope on all the usual stuff), started at 266/533 and pulled the voltages down a bit (1.475 vcore, 1.65 MCH, 1.35 FSB, 2.2 RAM.) then did 279 FSB for about 30 minutes in memtest (booting from a CD this time to avoid raid damage.)

Then I went to 285 and it booted up clean , just started memtest and it's 10 minutes in with no problems.

I think the problems were any number of possibilities:

1. Too much vcore too early, not breaking the CPU in enough before pushing it
2. Too much volts on the RAM - rated at "2.2 - 2.4" and 2.36 was maybe too much, 2.2 now and good.
3. Ram at 200 MHz probably bad, 266/533 actually does seem pretty good
4. Watchdog going berzerk
5. *Maybe* not shutting down entirely between volt changes, but I doubt it since all utilities were reporting the new volts and there's been much downtime since initially changing that

I'm still afraid to try 280, I'm leaving it in the dust and will be targeting 334 fsb to start (gotta break 3 GHz

Advise please, how long should I run memtest before proceeding, and what increments should I ramp the FSB at? I have infinite patience but don't want to do anything unnecessary. I figure 10 MHz increases with 1 pass of memtest (~50 minutes or so) until I get errors, then pull back and memtest 8-24 hours, then orthos etc.

Thanks to everyone who has posted here or ever posted anything on XS. 100 pages of posts about one motherboard is truly awe inspiring.

I should add the following about my cooling:

I've got a SilverStone FM121 at full blast inside the Tuniq 120, and it's about 2 inches from the rear case fan hole - and that's all that's there, a hole, no case grill at all (cut it out to reduce turbulence and noise, definitely noticable improvement!). Seemed like 2 fans there wasn't really doing much, but now I'm wondering if I'm not cooling the mobo enough with only the fan in the Tuniq. Think I need it? If so, I'd put the FM121 on the case rear since it's strongest, and put the original from the Tuniq back in the Tuniq.

I have another 80mm Zalman fan above the mobo in the rear, next to the PSU, which pulls air through 3 hard drives before exhausting. The Seasonic 700W PSU has a giant 120mm in it, and a tiny fan inside to assist exhaust. This fan is also pointing directly at the hard disks next to the PSU above the mobo.

Beyond that, I have the old Tuniq fan 120mm in the front that is blocked somewhat by two hard disks, but to assist with general intake I have a big case and five 5.25" bays wide open in the front so the CPU, Rear, and PSU should have no trouble pulling air in. This concept was used on an Antec case I saw where they dumped the front fan completely, instead opting for a single rear 120mm fan to do all the work.

All of the fans are at around 2000-2500 RPM when full blast and it's pretty deafening. At stock settings the temps were virtually unaffected by reducing the fans to 1600 nearly silent running, so I hope I can still do that when I finish OC'ing.

Sorry, I should really just take a picture.

[gtj]

Advise please, how long should I run memtest before proceeding, and what increments should I ramp the FSB at? I have infinite patience but don't want to do anything unnecessary. I figure 10 MHz increases with 1 pass of memtest (~50 minutes or so) until I get errors, then pull back and memtest 8-24 hours, then orthos etc.

memtest is great for diagnosing problems but I wouldn't use it as a long term test tool. Save the time and go right to Orthos blend. This also allows you to monitor temperatures during the test.

I tend to use FSB increments that give me a .1GHz change in processor speed. 10MHz FSB should be fine.

[minsc_tdp]

Save the time and go right to Orthos blend. This also allows you to monitor temperatures during the test. I tend to use FSB increments that give me a .1GHz change in processor speed. 10MHz FSB should be fine.

Thanks. I'm about to run Orthos, checking idle temps. How long on blend before bumping another 10 MHz?

What's the best tool with the latest BIOS to measure temps?

Core Temp 0.94 says 24C and 27C for cores, and Tjunction is 85C ?? After 3 minutes of orthos blend, cores are 35/39C, and TJunction hasn't moved. I think that's broken and I should ignore it.

Other idle temps:
RMClock says 18C & 22C on the cores.
IDU says 34C proczone, 34C syszone1, 35C syszone2
IDCC says 34C for all (same as IDU i would imagine)
TAT says 28C for both cores.
Everest 2006 Ult. (not latest) says 22C for each core.

I have no idea where each of these are getting their values, but there seems to be three sets: The IDU/IDCC vs. CoreTemp/RMClock/Everest vs. TAT.

[SiGfever]

I am having an issue with my new BadAxe2 board. When I shut down the system sometimes when I restart the computer the window will appear requesting you to "CTRL-ALT-DEL" to sign in and then the cursor and the keyboard freeze making me have to reboot. When it comes back up I can sign in fine, it only appears to happen from a cold boot.

[mezcal]

Thanks. I'm about to run Orthos, checking idle temps. How long on blend before bumping another 10 MHz?

What's the best tool with the latest BIOS to measure temps?

Core Temp 0.94 says 24C and 27C for cores, and Tjunction is 85C ?? After 3 minutes of orthos blend, cores are 35/39C, and TJunction hasn't moved. I think that's broken and I should ignore it.

Other idle temps:
RMClock says 18C & 22C on the cores.
IDU says 34C proczone, 34C syszone1, 35C syszone2
IDCC says 34C for all (same as IDU i would imagine)
TAT says 28C for both cores.
Everest 2006 Ult. (not latest) says 22C for each core.

I have no idea where each of these are getting their values, but there seems to be three sets: The IDU/IDCC vs. CoreTemp/RMClock/Everest vs. TAT.

Yeah, I can't get any sort of consistent temp measurements either. TAT is always 6-10C higher than coretemp-even when they're running side by side. Don't know what to believe

[minsc_tdp]

I am having an issue with my new BadAxe2 board. When I shut down the system sometimes when I restart the computer the window will appear requesting you to "CTRL-ALT-DEL" to sign in and then the cursor and the keyboard freeze making me have to reboot. When it comes back up I can sign in fine, it only appears to happen from a cold boot.

I think this board has lots of difficulty with cold boot. At stock settings and a 27C CPU, I get a thermal alarm every few cold boot attempts. I've actually found you can wait it out and it'll actually continue booting, and then all the sensors are dead.

By the way, I'm at 295 MHz orthos stable for 15 minutes so whatever barrier kept me at 280 is gone. I'm moving slow but thanks again to everyone here.

To any newbies, you *really* need to read all 100 pages before you do anything

[gtj]

Thanks. I'm about to run Orthos, checking idle temps. How long on blend before bumping another 10 MHz?

What's the best tool with the latest BIOS to measure temps?

Core Temp 0.94 says 24C and 27C for cores, and Tjunction is 85C ?? After 3 minutes of orthos blend, cores are 35/39C, and TJunction hasn't moved. I think that's broken and I should ignore it.

Other idle temps:
RMClock says 18C & 22C on the cores.
IDU says 34C proczone, 34C syszone1, 35C syszone2
IDCC says 34C for all (same as IDU i would imagine)
TAT says 28C for both cores.
Everest 2006 Ult. (not latest) says 22C for each core.

I have no idea where each of these are getting their values, but there seems to be three sets: The IDU/IDCC vs. CoreTemp/RMClock/Everest vs. TAT.

IDCC and IDU are lousy for reading temps because they don't measure the cores. I've found absolutely no use for either program.

Coretemp, RMClock and TAT are all getting their temps from the Digital Thermal Sensor located on each core. Those sensors are VERY sensitive and can update as fast as 30 times a second so unless the programs are all reading at EXACTLY the same instant, they'll probably not agree. They should be close though. RMClock averages the core temps over 5 seconds by default so it's going to be even more slightly off. Pick one and stick with it.

Speedfan and Everest report both the DTS core temps and the standard cpu and motherboard temps from the onboard sensor chip. You HAVE to use the latest versions of each program though. Everest versions before 3.80.880 or so can cause the board to lock up and prior to Speedfan 4.32, our board wasn't supported.

Tjunction is the reference temp for the processor. If the temp goes above that point, throttling should occur to prevent damage. It can be either 85 or 100 depending on the processor you're using. It's important to report because the processor reports the core temps as a number relative to tjunction. If the display program is using the wrong tjunction as a base, the temps will be 15c off.

[minsc_tdp]

Thanks for the info. Is the Tjunction for an E6600 85C correct then?

By the way things are flying. Just hit 305 MHz FSB and temps are barely moving at all. Crazy, loving it. SuperPi 1M in 18.453. Stock is 21.594, and the Athlon 64 I came from was 32.25.

I wonder how high I need to go before the core temps really start to move above 40C under full load!

I wonder about my NB temp though, seems there's no way to monitor it in software?

[mezcal]

Tjunction is the reference temp for the processor. If the temp goes above that point, throttling should occur to prevent damage. It can be either 85 or 100 depending on the processor you're using. It's important to report because the processor reports the core temps as a number relative to tjunction. If the display program is using the wrong tjunction as a base, the temps will be 15c off.

Ah, have to check that tjunction info when my bx2 box is back up. I wonder if my budget e4300 is different than the prior c2d chips somehow and is therefore not in their databases (or whatever) yet.
If I recall, coretemp was reporting temps below ambient at idle so I was tending to believe TAT. I only had the system assembled to test for a day prior to tearing it down to fix the ugly p180 cablegami/airflow issues so I don't have much time in with it yet. Still have a ways to go before it's back up. Once it is, I'll check if the discrepancy for my application is indeed a consistent 15 degrees.

[SiGfever]

Ah, have to check that tjunction info when my bx2 box is back up. I wonder if my budget e4300 is different than the prior c2d chips somehow and is therefore not in their databases (or whatever) yet.
If I recall, coretemp was reporting temps below ambient at idle so I was tending to believe TAT. I only had the system assembled to test for a day prior to tearing it down to fix the ugly p180 cablegami/airflow issues so I don't have much time in with it yet. Still have a ways to go before it's back up. Once it is, I'll check if the discrepancy for my application is indeed a consistent 15 degrees.

I never had this issue with an Asus board, now I am wondering if this BadAxe2 was my best choice.

[stmckin]

Just pointing out: minsc_tdp had
1-what appears was a watchdog stop
2-backleveled BIOS to pre watchdog
3- still had watchdog 'like' stop
4- went forward to latest BIOS, turned off watchdog and it works...

inferring that once you go to a BIOS that is watchdog, there's no going back....

[gtj]

Thanks for the info. Is the Tjunction for an E6600 85C correct then?

By the way things are flying. Just hit 305 MHz FSB and temps are barely moving at all. Crazy, loving it. SuperPi 1M in 18.453. Stock is 21.594, and the Athlon 64 I came from was 32.25.

I wonder how high I need to go before the core temps really start to move above 40C under full load!

I wonder about my NB temp though, seems there's no way to monitor it in software?

85C for E6600 is correct. When you start raising the voltages, you'll see the temps rise. At 305 though, you should have plenty of room before you need to do that.

Don't worry about the NB. There's no way to measure the temp from software but it won't be a problem for a while. Just feel the heatsink. If it burns your fingers, stick a fan on it.

[gtj]

Ah, have to check that tjunction info when my bx2 box is back up. I wonder if my budget e4300 is different than the prior c2d chips somehow and is therefore not in their databases (or whatever) yet.
If I recall, coretemp was reporting temps below ambient at idle so I was tending to believe TAT. I only had the system assembled to test for a day prior to tearing it down to fix the ugly p180 cablegami/airflow issues so I don't have much time in with it yet. Still have a ways to go before it's back up. Once it is, I'll check if the discrepancy for my application is indeed a consistent 15 degrees.

I believe the e4300 has a 100c tjunction. There are a few threads in the info and support sub-forum that talk about this. You might take a look.

[gtj]

I never had this issue with an Asus board, now I am wondering if this BadAxe2 was my best choice.

The temp issue has nothing to do with the board. It's solely a processor problem in that Intel provides a way to read the core temp sensors but not a way to read the tjunction temp on which they're based. You need both to determine the absolute temperature.