E6750 & Gigabyte P35C-DS3R Overclocking Thread

[Wurmer]

After reading all the discussions about temps and such here and other forums, one thing comes to mind. It's seems freaking hard to get accurate temps read out. You'd think that if satellites can be send to Jupiter, that it would be possible to get an accurate temperature reading for a desktop computer ! Even worst, we need to rely on third party software to get decent readings, it's a shame really. CPU maker and MB manufacturer should work together to provide such tools, specially when you're advertising high OC potential for you MB.

 

[5t3v0]

Originally posted by: random betrayal
I upped the voltage of these patriot modules to +.2 in the bios and speedfan reads it as 2.1, so more empirical evidence of the pour voltage regulation of this board.

random, are those Patriot not rated at 2.2v? In which case you'll need to add +0.4v. You could try +0.3v but dont be surprised if you need the full 0.4. A guy over on the P35C-DS3R forum had a similar issue with his quad on stock settings & that turned out to be undervolting the memory, so crank it up!

Everest et al would suggest that there is a 0.1v over-volting on the vdimm, but I'm not convinced. My ram is rated at 2.1v & I needed to add the full +0.3v but Everest shows it as 2.26v. I dont really care what it is as long as it's stable.

When going above +0.2v, the BIOS will say temps are not optimized in flashing red text but you can ignore this. PC Health Status will also show voltage as FAIL but again this can be ignored.

 

[5t3v0]

Originally posted by: SniperDaws
is there a way to measure the north bridge temp, as like ive ststed since i reseated the heatsink it doesnt get mad hot anymore and im worried its not making contact although if it wasnt making contact im pretty sure my pc would be locking up.

Yeah, you know those thermometers you stick up yer arse? Stick it right up there & put your tongue on the heatsink. Works, honest.

 

[SniperDaws]

Originally posted by: 5t3v0

Originally posted by: SniperDaws
is there a way to measure the north bridge temp, as like ive ststed since i reseated the heatsink it doesnt get mad hot anymore and im worried its not making contact although if it wasnt making contact im pretty sure my pc would be locking up.

Yeah, you know those thermometers you stick up yer arse? Stick it right up there & put your tongue on the heatsink. Works, honest.

lmfao, How rude!


Oh and i found out today that the Vcore does infact drop when using the energy saving options in the bios, with F4 bios regardless of Overclocked or not but CPU-Z doesnt seem to show this, use Everest and the drop at idle and rise at load will be shown.


Open Everest, click Computer and then click Overclock.

 

[mrfatboy]

sniper,

What version of everest do you have? I don't show any Vcore readouts in my version. I just have cpu clock, cpu mulipler (which is wrong), and cpu fsb. CPU FSB & CPU Mulptiplier flip back and forth between two different values every second.

 

[5t3v0]

Originally posted by: SniperDaws
Oh and i found out today that the Vcore does infact drop when using the energy saving options in the bios, with F4 bios regardless of Overclocked or not but CPU-Z doesnt seem to show this, use Everest and the drop at idle and rise at load will be shown.

Mate, that's the VID its reporting, not the actual vcore, same as Coretemp. CrystalCPUID also reports a drop although it says 1.148v instead of 1.175v. It's all bollocks.

If you look in the "sensor" page in Everest, rather than overclock, you'll see the real vcore reported.

Here's my idle readings from the various programs:

cpu-z = 1.408v
Everest = 1.41v (sensor page)
Speedfan = 1.41v

 

[Conjugal Visit]

Originally posted by: random betrayal
I just got all my parts put together and programs installed. Everything is running fine so it seems at least. I am running at stock speed with both the vcore and muliplier modifiers enable in the bios(cant remember the name right now). I idle at 1.152 and load at 1.31 in CPU-Z. I ran orthos just to see how everything was looking and I get rounding errors within 20 seconds. I tried Prime and same thing. Is this due to the vcore fluctuations? Any ideas?

I upped the voltage of these patriot modules to +.2 in the bios and speedfan reads it as 2.1, so more empirical evidence of the pour voltage regulation of this board. Jcossin we have the same memory have you had any issues with it?

As for temps my two cores idle at 9 and 16 C both numbers closely correlate with speed fan and core temp. I DL what said it was the beta of core temp but it shows the tjunction of 85 so I need to add 15 to those reading.

@random...

Hello Random...my two cents: your issue is ram, not vcore fluctuations. I think I've seen some issues in other forums, on this as well as many other boards, running 4gigs of ram, all slots full. I'd be inclined to agree that the ram is undervolted. Find the voltage spec'd by Patriot, and apply it in your bios setup. Run your stress tests. Pass? Try backing the ram down one notch, and test again. BTW - I don't see Patriot listed as an "approved vendor" for this board. You might also try loosening the timings - 5-5-5-etc...and see if this positively impacts your ability to stress test.

Did you really mean to say that your core temps are 9 and 16 C? Maybe more like 42c?

Regards,

 

[random betrayal]

I will bump the volts and or loosen the timings to 5-5-5-15 on the RAM tonight when I get home. Although i would like to run at the tighter timing as this was one of the reason's I bought the memory. They are rated for up to 2.2 volts but I figured that this was for overclocking them. When the board tells me that the voltage on the DIMM has failed that will not have any consequences? You are right they are not approved at this time for the board, but they had a deal on them that I could not pass up recently. They pass memtest for two passes though at these settings, I have not run anything longer as of yet.

As for temps I thought they sounded Fubar as well but both core temp and speed fan gave the same readings. It has cooled off here recently but those just don't sound right. When I first fired them up it was 7 and 13 before it leveled off at 9 and 16. I will have to check tonight what the bios says. I can't tell the load temps as the two stress testers are still giving errors within a minute.

On a side note does having 4 gigs of Ram mess up things in vista? The reason I ask is when I run the auto updater in vista, I can not install half a dozen of the updates for Vista? Maybe a bad install of the OS?

 

[SniperDaws]

Originally posted by: 5t3v0

Originally posted by: SniperDaws
Oh and i found out today that the Vcore does infact drop when using the energy saving options in the bios, with F4 bios regardless of Overclocked or not but CPU-Z doesnt seem to show this, use Everest and the drop at idle and rise at load will be shown.

Mate, that's the VID its reporting, not the actual vcore, same as Coretemp. CrystalCPUID also reports a drop although it says 1.148v instead of 1.175v. It's all bollocks.

If you look in the "sensor" page in Everest, rather than overclock, you'll see the real vcore reported.

Here's my idle readings from the various programs:

cpu-z = 1.408v
Everest = 1.41v (sensor page)
Speedfan = 1.41v

Thanks for pissing on my chips

 

[mrfatboy]

Originally posted by: random betrayal
When I first fired them up it was 7 and 13 before it leveled off at 9 and 16.




@random
Not that it's harming your CPU or anything but I find it interesting why your core temps are so different between the your two cores. I would guess the thermal paste was not put on evenly or the HS is not mounted flush with the chip.

Like I said, don't worry about it. Your temps are really low already.


Just chalk it up to being another victim of the Artic Cooler There seems to be a couple of you on this thread and you know who you are J/K

 

[5t3v0]

random, 2.2v is the rated voltage for running at the EPP 4-4-4-12 timings at 400MHz. If you drop down to the JEDEC SPD (5-5-5-15) then you'll get away with lower voltage but what's the point in that? 2.2v should take you further than 400MHz at 4-4-4-12 if these are any good. The problem with buying memory that's rated at 2.2v is that there's no voltage headroom for overclocking if they're already maxed out.

I presume your core temps using the pre 0.95.4 Coretemp beta & that you need to add 15C to them? Not even the Tuniq fanboys are getting temps that low Even adding 15C, your temps are much lower than mine with the same cooler. Do you live in a cold store?

 

[Conjugal Visit]

@5t3v0 - you're also O/C'd to 3.6 and volting to 1.456 - but point taken, which is why I asked the question in the first place. Even before I "discovered" for myself the right version of Core Temp, and was swelling with pride at idle temps around 26-28C, that still leaves this huge gap between 9C and where I thought I was. Truth be told, we've both browsed a lot of forums, and looked at a lot of folks' specs, and generally speaking, it seems to be 40-45C at idle, and 60-70C at stress. And i mean REAL stress - Prime 95 or Orthos or TAT running for hours on end. I'm not "proud" of my temps, per se - 44C or so idle...68C or so at full stress after a couple of hours - but then I've made some choices too. I don't run 120 cfm Vantecs full bore - don't have seven fans, etc, etc. And these stress temps are temporary, during benchmark stress, and still under spec.

 

[SlicedBread27]

Originally posted by: random betrayal
I will bump the volts and or loosen the timings to 5-5-5-15 on the RAM tonight when I get home. Although i would like to run at the tighter timing as this was one of the reason's I bought the memory. They are rated for up to 2.2 volts but I figured that this was for overclocking them. When the board tells me that the voltage on the DIMM has failed that will not have any consequences? You are right they are not approved at this time for the board, but they had a deal on them that I could not pass up recently. They pass memtest for two passes though at these settings, I have not run anything longer as of yet.

As for temps I thought they sounded Fubar as well but both core temp and speed fan gave the same readings. It has cooled off here recently but those just don't sound right. When I first fired them up it was 7 and 13 before it leveled off at 9 and 16. I will have to check tonight what the bios says. I can't tell the load temps as the two stress testers are still giving errors within a minute.

On a side note does having 4 gigs of Ram mess up things in vista? The reason I ask is when I run the auto updater in vista, I can not install half a dozen of the updates for Vista? Maybe a bad install of the OS?

What on earth are your ambient temps? Are you running your rig inside a walk-in freezer?

 

[SniperDaws]

lmao i was rolling when i read the last few post, well done, Tuniq fanboys lmfao.



I got World in Conflict today, this game is amazing, download the demo and play it online and see what you think.

 

[mrfatboy]

Hey guys,

take a look at this post regarding the voltages on our board. This guys says he measured them at the board with a multimeter. He says Everest and Speedfan matches his multiplemeter findings.

p35c overvolt


what do you think?

:thumbsup: ********** The Tuniq Rules *************** :thumbsup:
:thumbsup: *********Tuniq Fanboys Unite ************* :thumbsup:

 

[5t3v0]

Good find. Also good to know you can trust the monitoring apps. Piss poor that the board over-volts vdimm so badly, especially when you consider the opposite happens on the vcore. Changes nothing though. According to Everest, +0.3v gives me 2.26v which doesn't concern me particularly. If that's what it takes to make my OC stable, then that's what it takes. RAM's cheap.

Very warming to see you Tuniq users uniting behind your cooler. I'll be eating my words when I upgrade to a Penryn quad next year

 

[SniperDaws]

Originally posted by: 5t3v0
Very warming to see you Tuniq users uniting behind your cooler. I'll be eating my words when I upgrade to a Penryn quad next year

OOOOoooooOOOOooo its gonna get a little warmer in bony Scotland next year then

 

[5t3v0]

Aye that it is. We're gonnae need it too ah tell yee. It's as calt as a witch's tit up here.

 

[mrfatboy]

Originally posted by: 5t3v0
Good find. Also good to know you can trust the monitoring apps. Piss poor that the board over-volts vdimm so badly, especially when you consider the opposite happens on the vcore. Changes nothing though. According to Everest, +0.3v gives me 2.26v which doesn't concern me particularly. If that's what it takes to make my OC stable, then that's what it takes. RAM's cheap.

Very warming to see you Tuniq users uniting behind your cooler. I'll be eating my words when I upgrade to a Penryn quad next year

I would be concerned for the unknowning noob that accidently overvolts his ram because he didn't know this board overvolts. We are fine because know the real voltage numbers and can adjust accordingly. Frankly, I don't know if .16v overvolt is that big a deal but maybe it is for cheaper ram. I just hope our boards hold up

 

[Conjugal Visit]

@mrfatboy - noticed that 3DMark '06 number - that's pretty incredible! I'm guessing your vid card (and the 640 mem) really helps the benchies? I can't get close - 10400. And thanks to this thread, I'm playing around with Everest - nice, nice program.

Regards,

 

[5t3v0]

Yeah, I'm a little concerned about the reports of delayed board failures going around. We could be sitting on a time bomb.

 

[mrfatboy]

@conjugal

I'm not sure how much the 640 helps over the 320 in 3dmark06 scores. But overclocking the vid card is a big factor. Here are my scores with different OC's for cpu and vid card.

3dmark 06 (3.2) 9309
3dmark 06 (3.2 & 600/1000) 10655
3dmark 06 (3.6 & 620/1004) 11270
3dmark 06 (3.6. & 690/1090) 12094



@5t3v0,
I have to wonder if there was a bad batch of these boards that got made. If you read the forums, most people love them. Then all of the sudden huge failure rates. You can read the newegg reviews and almost track the timeline when they started failing. Then right before our eyes.. Cleverest's board goes down (or never starts up). But again, he had a bad powersupply to start with. I wonder if these boards are really sensitive to power voltages. Maybe having a bad PS screws you BIOS up. We know that the power regulation on the other parts of the board are scewy.

i'm still waiting for Gigabyte to answer these questions. The have been avoiding my direct questions for a week now. We'll see.....

 

[jcmuse]

hi all,
i just got my e6750 ds3 and 2x1gb ballistix (same specs as OP). I booted straight into the OPs settings and it seemed stable but i dont think i really need the speed and dont like the added heat.
I am wondering what setting you guys recommend @ stock 1.35v. Can i get a slight OC at that voltage? Also, what do i need to set my memory at to get it at its rated timings (@stock speed).
awesome thread btw... great for people like me who dont have time to tinker with settings to see the best way to run things.

 

[5t3v0]

I would shoot for 400MHz with ram running synchronously. Dont know if its possible on stock volts given the big vdrop on these boards. MrFatboy will be able to recommend ram settings but I'm sure he has his on +0.3v with 4-4-4-4-12. If you opt for stock cpu Mhz then you can use a 6:5 ram multiplier to get the ram at it's rated speed. Auto probably tries to do the same thing.

 

[honolululu]

Originally posted by: mrfatboy

I would be concerned for the unknowning noob that accidently overvolts his ram because he didn't know this board overvolts.

I just went back and looked at your bios pics in the initial post and it looks like you're showing +0.4V, and you also say +0.4V is 2.2V

Might be a good place for a little warning message?