Pentium D 805 OC to 4.0+ GHz w/ ECS C19-A SLI

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
I was one of the people that got sucked into the idea of having such a thing (Pentium D 805 @ 4.0+ GHz) when the articles came out. This would be the second time I've overclocked a CPU, so I'm still really new at this.

Has anyone gotten up to 4.0 GHz or higher using this ECS C19-A SLI motherboard? I can't manage to get my computer to post at this speed. I've gotten it up to 3.94 GHz, but when my CPU gets under stress the entire computer locks up. I have no clue what I'm doing wrong. I have researched possible solutions and have done trial-and-error attempts for a few days now, with no luck.

Mainly I ask just to see if it's even possible. But if you can spare some assistance, please do so. Thanks!
 

ikjadoon

Distinguished
Feb 25, 2006
1,983
44
19,810
Well, I need some more information...What memory are you using? Timings? Memory divider? FSB? Multiplier? Stock Voltage? Current Voltage? What cooler are you using? What are the temps for idle/load? Not all chips will hit 4.0Ghz, try 3.8~3.9...

~Ibrahim~
 

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
Right now, the CPU is at 3.6GHz
Memory: 1GB Kingston (ValueRAM?) DDR2 PC4200
Memory Divider: Unsure, never touched it.
Timings:
memorytimings4lh.jpg

FSB: 180 MHz
Multiplier (Locked): 20x
Stock Voltage: 1.248V (double checking)
Current Voltage: BIOS set to AUTO right now (1.248V at 3.6 GHz, 1.5125V at 3.94 GHz)
Cooler: Cooler Master AQUAGATE Mini R120 Liquid Cooling System
Temps: Idle hovers around 40-41C, not sure about load (EDIT: I can't get an accurate reading, it's jumping in values from 3C to 49C in no logical order)

Edit: Bunch of stuff.
 

ikjadoon

Distinguished
Feb 25, 2006
1,983
44
19,810
Well, your memory divider is 4:3...Look at the FSB:RAM thing above the timings..That is quite a big jump in voltage for 3.94Ghz. What power supply are you using? Is it 100% stable at 3.6? Hmm..The memory seems fine, but I'm not so good with DDR2. I would go for a max of 3.8Ghz because those temps are a little high for idle, even with water cooling. Slightly bump the voltages, slightly, and inch FSB...1.5V seems too high above stock. That may be why it won't POST.

~Ibrahim~
 

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
Power supply is 500W. I researched the CPU a bit: the maximum safe (keyword safe) voltage is 1.55v. Looking at something like the article about the Pentium D 805 at Tom's Hardware, they upped the voltage to 1.56v to allow it to run at 4.1 GHz, which is just above the safe range, but they report that it's completely stable.

My temperatures do seem to be high for liquid cooling, but I didn't spend hundreds of dollars in it. Plus, if it's anything relevant, I live in Texas and it's summer, so it's really hot in my house, this room in particular. I might check about reapplying the thermal paste, I may have put a bit too much or not spread it enough.

It is rock solid stable at 3.6 GHz. I've been doing the Prime95 Torture Test (I heard something about this only stressing one core, has this been addressed?) and playing some modern games. Again, at 3.94 GHz my system locks up when put under stress. I would think that I might need more voltage, but I'm not too sure.

Please tell me if you need more information, and thanks for your assistance and quick response. :)
 

ikjadoon

Distinguished
Feb 25, 2006
1,983
44
19,810
OK, that seems like enough for the PSU. Well, it is still water cooling, which is better than most air setups...How does it perform at 3.8Ghz?

I'm a little confused about DDR2, as I've never overclocked with it...What is it, you multiply by four instead of two to get the "true" Mhz? I think so, as that seems more correct. If it is by four, you have 540, which is hardly above stock. So we can effectively rule out the memory.

Next is the chipset, which can hit 250 stable:

Featuring a chipset without overclocking protection, ECS C19-A SLI achieved an excellent result, surpassed by only 3 MHz by DFI LanParty 925X-T2 (Intel i925X) and achieving a better overclocking than all other motherboards based on nForce4 SLI Intel Edition chipset that we reviewed. With this motherboard we could put our CPU running externally at 250 MHz (1,000 MHz QDR), making our CPU to run at 4 GHz internally, an outstanding result (25% increase on the CPU internal clock).

That was with a Pentium 4 640, which's stock clock is 3.2...I mean, I would expect a little better than 3.6 with water cooling. Try 3.8, just up the FSB. Do it slowly, not in a big jump. Maybe 5Mhz at a time, less if you'd like. The temps seem higher than expected, but they are a little more room. Make sure you test and test and then test after each increase; we wouldn't want a burnt CPU chip. I always thought on Dual-Cores, you'd always stress both cores at the same time...Why only one? I'm not trying to be mean, just wondering.

Try with the thermal paste, and check the fittings on the water cooling. It might be because of the hot ambient temp...Here in Kentucky, we hit 90 degrees most of the day...Must be toasty in Texas...

~Ibrahim~
 

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
I don't remember at 3.8, but temperature wasn't a problem at 3.94 I believe. It's just the issues with locking up.

As for the memory, close, but I believe it's 533 MHz. (Says so on the page I bought it from)

What's interesting is that in the BIOS I can only change the bus speed (after multipliers etc., is this called "QDR"?), not the actual FSB itself, so it's hard to relate to guides that talk in increments in terms of FSB. I've been trying little steps at a time to ensure safety of my hardware. I got up to 790 (3.94 clock speed), but again it's unstable.

For the stress test, I'm only relaying information I read on other forums. That's just something I heard. I don't think that Prime95 would *intentionally* only stress one core. I understand your question on this.

Where I live, it's not unusual to have temperatues of 100+ F every day. I'm literally sweating in my room right now.

I suppose I'll look into upping the voltages at that 3.9, but I'm still completely puzzled as to why I can't break the barrier to get to 4.0.

Thanks again for your time!
 

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
Tested a few more things. I got the RightMark CPU Clock Utility and found out that my CPU constantly throttles down from 3.8 GHz because of heat. I just ordered a new case and a new fan, so I'll see in 4-5 days to see if this solves anything.
 

ikjadoon

Distinguished
Feb 25, 2006
1,983
44
19,810
Ahh, I see. Well, I'm not that bright, lol, but thanks... I think the new case/fan will help alot! Whew, 100? Damn, I'd die! I've been to hot places, i.e. Pakistan, Mexico, and I think I've had my share of heat, lol. How you can live, I'll never know...Good luck, tell me how it goes!

~Ibrahim~
 

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
I got the new case and transferred all the parts. While I did this, I found out that my fan for the radiator was blowing the WRONG WAY. When I switched out the fans (for a better one I bought) I put it on the right way. Right now I'm running the torture test at 3.6 GHz and it's not throttling down at all!

I'll have to see if this solves my problems with overclocking later, but for now everything's taking a turn for the better.
 

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
I got it up to 3.99 GHz. Even then, it takes several resets to get it to post for some reason. I simply cannot get it to post at 4.0 GHz. Just today I had a power surge which turned off my computer. I couldn't get it to post so I underclocked it to 3.94 GHz again. I'm confused as to why it sometimes works and sometimes doesn't, and why I can't break that dreaded barrier.
 

ikjadoon

Distinguished
Feb 25, 2006
1,983
44
19,810
3.99! So close, it isn't fair! Argh!!!

Hmmm...How are the rails? They should be tightish, a few .1V increments either way.

What temps do you get with it now, at 3.94Ghz? Was there a dramatic increase of temperatures when you increased the voltages eariler? If not, I would up the voltage, the smallest increment possible, and go for 198/199/200.

I can't see why it would work sometimes and sometimes it won't. The chip can't be going bad, no where near critical temp ranges...

Sorry!

~Ibrahim~
 

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
I had it on the maximum voltage the CPU range allowed without it being overvoltage. The temps could be better, but it doesn't throttle down so I don't know if that's an issue.
 

ikjadoon

Distinguished
Feb 25, 2006
1,983
44
19,810
$@^($#)&!!! Yeah, the temps are fine, I think dangerous is 70~80+C...

Hmm... How are the rails? Were they OK?

I am at a lost to what it could be: we ruled out the chipset, the memory, the CPU...Why won't it go a little further?

Is there any way we can figure out what is the temperature of the NB/SB? Maybe those could be the cause, but not very often.

~Ibrahim~
 

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
If you can explain to me what rails are, I'll be glad to check them. :p

Not sure about the chipset temperatures, but Everest has a motherboard temperature reading if that's of any help.

Right now it's at 36C, but I've seen it up up to 38-39C.

EDIT: On another note, I saw that every time I booted up my computer, the date and time weren't saved... I thought maybe I had put the CMOS in the reset position and left it there or something stupid like that, but it turns out that every time I push the reset button when my computer freezes/doesn't post/whatever, it resets the date and time. Joy... My computer is playing games with me.
 

ikjadoon

Distinguished
Feb 25, 2006
1,983
44
19,810
D'oh! lol, I'm sorry! I checked with your motherboard's manual. In the BIOS, go down to the 7th thing listed on the left-pane side, "PC Health Status". Then see where it says "+3.3V, +5V, +12V"? Write those numbers down, and go ahead and write the temperatures.

Yeah, the reset button is with the battery, which makes sure the time is the right time, so when you turn off your computer, the clock doesn't stop.

Motherboard temperature should be good.

Alternatively, you can use nTune in Windows. Link Put some load on, a little, and see if the number's change dramitically.... Thanks!

~Ibrahim~
 

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
+3.3V - 3.3V
+5V - 5V
+12V - 11.20V

CPU Temp - Idle 50C
System Temp - 36C

It's not the reset button, it does this with the power button, but only after exiting the BIOS... It doesn't seem to post after I exit (at least without changing anything).


Sorry for the wait!
 

waylander

Distinguished
Nov 23, 2004
1,649
0
19,790
Sorry to butt in here but that 11.2v is out of spec for ATX psu's. They should be +/- 5%.

You say the psu is 500w but what maker is it? Is it generic? How many amps on the 12v rail?
 

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
Spend the extra cash and get the Pentium D 930; at $173 it's one hell of a chip and can overclock much better than the Pentium D 805 thanks to the 65nm process.
http://www.newegg.com/Product/Product.asp?Item=N82E16819116238

I wish I did have the extra cash to spend. I may later on though.

The PSU is generic, also cheap. That's just something else to buy (a better PSU).
 

ikjadoon

Distinguished
Feb 25, 2006
1,983
44
19,810
Yes. There isn't enough power going to the 12V rail, which is the CPU's primary rail.

How many amps are on that particular PSU? You can tell by looking on the side of the power supply, there we be a chart similar to this:

17-103-457-03.JPG


Your PSU, however, may only include one 12V rail, which isn't nessecarily bad. We need about 20A total; as that is the about the amps needed by a 965, which is somewhat like you have now. But the 12V powers the GPU and the drives as well, so we want 25A or more. Even that is cutting it close. What GPU are you running?

I forgot to explain what a rail was, earlier. I'm not exactly sure, but they are the amount of power that is supplied to the computer. We need to get the processor around 241W or more. To calculate the amps needed to power that, on the 12V rail, we divide 241 by 12. That is about 20 Amps. But the 12V rail powers other things; we'll say your GPU takes about 60 watts. That is 60/12 = 5. So we need 5 more amps, so we want 25A+. I know this may be confusing, this link should clear it up:

Power Supply Wattage

If you look at the Enermax, it has 32A, which is more than enough for a processor, but when you add some GPUs, it gets finicky; hard drives take up watts as well, and there are other things, too, but GPU + CPU + Drives take up the beef of the power needed.

To see how much you need: What GPU are you running? How many drives? Anything else that draws power, minus motherboard, PCI cards, fans/lights? Does the Water Cooling have it's own power or does it draw from the PSU? That would be a big factor, sometimes they can take a large amount of watts.

~Ibrahim~

~Ibrahim~
 

chameleon493

Distinguished
Jun 23, 2006
32
0
18,530
GPU - XFX NVidia 7900GT
Drives - 1 DVD Drive (don't know specs, I salvaged it)
Liquid cooling connects to the mobo, so I guess it draws from the PSU.


Also, I just had my first BSOD on this computer, "IRQL_LESS_OR_EQUAL" or something. I'm going to run memtest86 sometime soon to check my hardware. I'm getting upset that nothing's turning out right.

Considering this PSU:
http://www.newegg.com/Product/Product.asp?Item=N82E16817103931

EDIT: Second BSOD this night, this time about some .sys file. Didn't get time to read it, my computer automatically rebooted.