e2140 and gigabyte P35-DS3L

[superstition]

I have been having consistent trouble going beyond 360 MHz (2.88 GHz) with my setup. I thought temperatures were the problem so I lapped my Tuniq to a hazy mirror finish and lapped the processor to a dull mirror (copper). I also didn't use much AS5. The spreader and sink are flat now and temperatures are well within safety ranges.

If I go to 3 GHz I get errors in Orthos or spontaneous reboots in Orthos. I get reboots or black screens more quickly if I quit all other programs, including CoreTemp. I've tried turning the voltages up for every part. The only thing that seems to bring stability is Gigabyte's automatic voltages setting, but that's much too hot. It certainly shouldn't be necessary to go from 1.25 to 1.39 volts just to go up 200 MHz, eh? I also worry about how many volts that setting is pushing into the chipset. I've been able to run the processor at 3.2 GHz with auto voltages but heat was a problem and stress testing with Orthos wasn't viable. I haven't tried anything over 3 GHz since the lapping. I don't think temperatures are a problem anymore. Is it my power supply being too weak? It seems really strange to be able to run everything at stock voltage (as well as the processor under stock) and then suddenly have no luck with much higher voltages after just 200 MHz of increase.

Fans:
I have a 140mm fan pointed at the northbridge/ram/Tuniq, plus a 120mm exhaust fan right next to the Tuniq. I have a 120mm intake fan and no drives in the upper area of my Antec P-180. I even removed the lower drive panel, the plastic covers on the top two panels, and the middle cage and plastic slide holders, to let the 120mm and 140mm fans suck in more air. I have a Seasonic 120mm 380 watt PSU and a 130mm silent wheel fan blowing on it. So, I have a total of 5 fans, plus the GPU fan (which is a Zalman 700 copper).

Parts:

Seasonic 380 watt 80 plus PSU. PSU fan is at 715 RPM, according to Speedfan, I think.
e2140 MO (currently 1.3v in BIOS 2.88 GHz. Says 1.248 in CPU-Z)
Gigabyte P35-D3SL 2.0
Two 1 GB sticks PQI Turbo CAS 4 DDR-2 800, dual channel running at 1:1 ratio 4, 4, 4, 12, 2T
eVGA 6800 GS PCI-e 256 MB DDR-2 stock
250 GB Seagate IDE HD 7200 RPM
160 GB WD SATA HD 7200 RPM (idles at 54C unless put right in front of a 120mm fan!)
Tuniq tower cooler with internal 120mm fan. Set to 1023 RPM.
Cheap firewire PCI card. Two port.
Antec P-180 case.

Speedfan gives me the following:

Vcore1: 1.25V
Vcore2: 1.87V
+3.3V: 3.25V (Speedfan) 3.248V (BIOS reading)
+5V: 5.00V
+12V: 4.22V
-12V: -16.97V
-5V: -8.58V
+5V: 5.13V
Vbat: 3.22

Right now, with Orthos running 8K FFT (priority 7):

Temp1: 30C
Temp2: 37C
Temp3: -2C
Local: 29C
Remote: 37C
HD0: 35C
HD1: 36C
Core0: 41C
Core1: 40C

When I had the system at 3 GHz with a voltage of 1.375 (or something similar), heat wasn't a problem but the system went to a black screen as soon as I quit CoreTemp while running Orthos with priority 9. I have tried giving the RAM more voltage and relaxing the timing to CAS 5. I get reboots, errors, and black screens. Every time I think there's a pattern, there isn't one.

 

[nyker96]

There's something amiss here, my E2140 M0 runs 3.1Ghz at about 1.328v in CPUZ on IP35E. I doubt this will have trouble running 3ghz with 1.375v. Try test the RAM separately with memtest86+ just to rule it out.

Another thing is maybe your Tuniq isn't seated correctly since when I run ortho blend on that 3.1ghz E2140 M0 I get 39C (speedfan) using a ACFreezer 7 pro which is definitely not as good as Tuniq.

 

[superstition]

Originally posted by: nyker96
There's something amiss here, my E2140 M0 runs 3.1Ghz at about 1.328v in CPUZ on IP35E. I doubt this will have trouble running 3ghz with 1.375v. Try test the RAM separately with memtest86+ just to rule it out.

Another thing is maybe your Tuniq isn't seated correctly since when I run ortho blend on that 3.1ghz E2140 M0 I get 39C (speedfan) using a ACFreezer 7 pro which is definitely not as good as Tuniq.

There's definitely something amiss. Why would a hard drive idle at 54C? It's not even mounted, because it's HFS+ (Hackintosh). I read that a dying hard drive may heat up, but SpeedFan's SMART test gave it a 96 and 100 rating which is higher than my IDE drive. Is my power supply too weak, perhaps? I'm going to try Memtest, although I've used Memtest on the past when testing an unstable Shuttle system (bad RAM) and it never found the problem.

As for the processor...perhaps I'm using too little thermal paste? I've been considering that gallium/indium Liquid Pro stuff, since I've lapped down to copper, but I don't want it to run off past the copper and cause corrosion. I'm not sure how the Tuniq could be improperly mounted, unless it's possible to overtighten it. I have it mounted so the fan pushes the hot air to the exhaust fan which is very close to it. I also have a 140mm fan blowing on it.

 

[superstition]

I'd try a BIOS update, but the most recent is a beta BIOS that's been up since December.

Is having only 3.248 on the 3.3V a big problem?

 

[lopri]

If you can boot at frequency A and mildly use Windows, your CPU/board doesn't have a 'hole' at that FSB. Usually with FSB hole, you either fail to boot or system is very unstable without much stress. Ruling out FSB hole, what's left is the usual suspects;

- more NB voltage
- more memory voltage
- cooling on NB
- etc.

Can you touch your HDD? If it's not overly hot (you will know), I wouldn't worry about Speedfan reading. Not only there is calibration issue, but also sometimes stupid firmwares on HDD can send wrong information as well.

 

[superstition]

The hard drive was extremely hot when I removed it. Speedfan's temp reading appearently was accurate. What could cause it to idle at 54c? It seems to be working normally when I boot into OS X from it. To keep the drive cooler, I took it out of the P-180s hard drive case and put it circuit board side up directly in the path of the 120mm intake fan (I don't have the middle cage installed).

I have increased my RAM voltage by +2 and relaxed the timing to CAS 5, and it doesn't seem to do anything. Why would the RAM need more than +2 with CAS 5 for an increase from 360 FSB to 375? It's crazy. I've tried increasing the FSB voltage by +1. I've tried increasing CPU voltage. The machine seems to be rock solid at 1.300 in BIOS for the processor and "normal" motherboard voltages for everything else?360 FSB with 1:1 CAS 4 RAM undervolted to 1.8xx (spec is 2.1). The only thing I can think of is putting the voltage for the RAM all the way to stock, which is 2.1. It really seems bizarre that such a large voltage increase would be necessary for a 15 MHz increase.

As far as NB cooling goes, I have a 140 mm fan blowing on it. The Tuniq blocks the flow to some degree but I've rigged the fan to get a fair amount of air on it. Again, it seems wierd that a 15 MHz increase would be such a deal-breaker. Maybe the NB heatsink just isn't doing its job, even with such a large amount of air blowing on it? If so, why would Speedfan not show a high chipset temp? The irony is that the highest temps in my setup tend to come from the hard drives! (Unless I use the auto voltage setting in BIOS, which I think overdoes the voltages by quite a lot.)

 

[VirtualLarry]

What was your vcore at auto settings at 3.2Ghz?

My E2140 M0 chips need 1.425v (BIOS) to reach 3.2Ghz stable. My load temps are 85C (CoreTemp 0.96)

 

[superstition]

85C??? At 3.2 GHz, my reading in speedfan is 1.39 with auto voltages, although that's with droop.

Well, I think I finally got things working. I bumped the RAM voltage up +3 (2.00-2.02V in speedfan), set the FSB to 375, and put the CPU voltage at 1.350. I did an Orthos blend for 1 hour and 40 minutes with priority 9 and nothing other than CoreTemp running (for half the time). After all the overclocking I've done since lapping the heatsink and processor, I haven't gone over 50C, and didn't even reach that since I repositioned my case on a more oblique angle to keep hot air from bouncing back in. Now it's maxing at 48C, which isn't bad for 3 GHz.

I'm not certain why I was having so many problems and now things are working. The RAM is probably stable at +2V, but I haven't lowered it because I'm glad to finally have something that seems stable and the RAM's spec is 2.1V, which is still more than it's getting now. I'm happy to keep everything at stock voltage (except the processor which is getting 1.30-1.31 after droop, which doesn't seem bad). I don't think I can do 3.2 GHz without raising the FSB voltage, which I'm not keen on.

 

[Duvie]

If you want to test the fsb hole.....

Drop the multi down one notch and bump the fsb up to 365...then jump to 400.....it will be lower overall clock but will test the fsb theory.....

The passive cooling on that board is similar to the lower end cooling that was on my S3 mobos....I was able to hit 500fsb on both of my S3's...They however required active cooling at that point....I could keep them good at 470-480 with just a case fan directed at them....they did up to 445-450 without any direction of fans on them, but just decent air cooling....


I would say pop the hs of the NB and put some AS5 on that as well.....

What kind of voltage boost are you giving the NB?
What do your rails look like at those speeds? 12V,5V, etc.....

 

[Tempered81]

will the e2xxx's do 400mhz fsb?

 

[sjwaste]

Originally posted by: jaredpace
will the e2xxx's do 400mhz fsb?

My E2200 does. 400x8 stable, 1.3v. I may try to push higher to see if it'll go. I ended up with the 2200 because I wanted the higher multiplier in case I ended up being FSB limited, but it turned out not to be the case so far.

 

[VirtualLarry]

Originally posted by: sjwaste

Originally posted by: jaredpace
will the e2xxx's do 400mhz fsb?

My E2200 does. 400x8 stable, 1.3v. I may try to push higher to see if it'll go. I ended up with the 2200 because I wanted the higher multiplier in case I ended up being FSB limited, but it turned out not to be the case so far.

My E2140s are also at 8x400. I could do 410, but 425 was too much.

 

[milesl]

I could only get my e2140 m0 to 2.6ghz (333x8) 1.3v stable on a P35-D3SL.No amount of voltage seems to be able to get it higher.It may be my ram holding me back(it is rated for 333mhz 5-5-5-15 at 1.8v) ,but I don't think so because I am running the timings at 333 mhz 4-4-4-12 at 1.8v and its passes memtest forever and is stable.

I tried setting it to 5-5-5-15 at 2.0v and still cant get the chip any higher.Any higher and it crashes soon after it gets into windows.System won't even boot at 400fsb.I am using the stock heat sink ,but the temps at 2.6 1.3v are 30c idle 50 load so they aren't that crazy for me not to be able to get a little more out of the chip.

I think either my chip is a dud or the mainboard is.

Congrats to everyone who gets 3.0GHZ + out of their setup.

 

[superstition]

miles, try these settings: RAM +2V. Processor 1.350V. FSB speed 375. Turn every Gigabyte automatic overclocking feature to its lowest setting (standard, etc.). I have been stable at 3 GHz for quite some time now with these settings.

Also post your Speedfan voltages when running Orthos (priority 9) small FFTs. Also tell us what your power supply brand/model/rating is.

 

[Tempered81]

wow i could never get my e4300 over 380