** Official ** Gigabyte UD3 P45 Series -- EP45-UD3 / EP45-UD3R / EP45-UD3P

[Don_Dan]

I'm sorry I don't know what you're referring to

Thermalright Ultra eXtreme or Noctua NH-D14 cooler.

[In-Fluence]

EIST enabled

Just wondering about the EIST thing, as it's something I'm considering when I finally get my cooling setup runing semi-passively. I know on my old opteron/dfinf4, I had to test the CPU at the multi and voltage it would be during c&q mode in order to guarantee stability if under mild load. Is the step down in voltage with C1E/EIST enabled at any risk of going too low whilst running at 6x multi with high fsb? What did you use to test it?

[gurusan]

Here's my effort so far on getting my rig prime stable.

[Musho]

Here's my effort so far on getting my rig prime stable.

Those volts are insanely low for that speed! Well done!

[OC Nub]

Very nice gurusan! Great vcore.

[gurusan]

thanks. the weird thing is it's 1.2625V VID, highest I've ever had on a wolfdale.

[In-Fluence]

Nicely done, as usual, some great clocks & voltages on this board.

Jor3lbr, do your MIPS mosfet blocks help a lot with stability at >4.5GHz? I know mine run hot, and I want to push further but I think these are now the cause of instability after managing to get CPU temps more reasonable (and I'm still p95 stable at 4.5GHz).

[Jor3lBR]

Nicely done, as usual, some great clocks & voltages on this board.

Jor3lbr, do your MIPS mosfet blocks help a lot with stability at >4.5GHz? I know mine run hot, and I want to push further but I think these are now the cause of instability after managing to get CPU temps more reasonable (and I'm still p95 stable at 4.5GHz).

Guys I never used the blocks... I ended up selling them all.

All I use in the mosfet area are Fans. One on top of each mosfet area.

For stability above 4.5Ghz:
- Q9650 below 65oC;
- Right amount of memory tweaking (not too tight not to lose);
- NB below 70oC.

[In-Fluence]

Fair enough, thanks for your summary.

With the cpu I have trouble getting below 70°C during p95 8k on the first core. the other 3 are sitting comfortably at 65°C at up to 1.4375v.
The 4th (weakest) core seems to increase in stability between 1.40-1.4125 and lasts around 10 tests during 8k at 1.4125, but it still fails, and gets worse when I go over that voltage. I know there's a fine line between temps/voltages here, but as the 4th core isn't too hot, this is the reason I'm starting to suspect the mosfets. But then it's a bit of a risk to splash out on some blocks if it's not going to help much more, and I'm simply reaching the cpu's limit.

[Songok]

Are there not any DDR reference voltage adustments on the EP45T?

It does have reference voltage. Its not something i considered! I have been leaving it on auto when going for a 1600 oc on ddr3 1333. Should it typically be raised above or below auto?

[In-Fluence]

I only asked because I didn't see them in the ET6 screenshot, just curious!

the DDR reference voltages should be half of the memory volts. I would start from there and work your way up or down

Perhaps skews are worth a try at those speeds too, they only had an effect when I was pushig my DDR2 beyond 1300, but in your case it might be more crucial

[Songok]

Just wondering about the EIST thing, as it's something I'm considering when I finally get my cooling setup runing semi-passively. I know on my old opteron/dfinf4, I had to test the CPU at the multi and voltage it would be during c&q mode in order to guarantee stability if under mild load. Is the step down in voltage with C1E/EIST enabled at any risk of going too low whilst running at 6x multi with high fsb? What did you use to test it?

Does it step down voltage with EIST on? I thought it just lowered multi on the cpu when loads are low. Cpuz reports the voltage slightly higher (vdroop) when it is at x6 idleing as when at full load.

[Songok]

I only asked because I didn't see them in the ET6 screenshot, just curious!

the DDR reference voltages should be half of the memory volts. I would start from there and work your way up or down

Perhaps skews are worth a try at those speeds too, they only had an effect when I was pushig my DDR2 beyond 1300, but in your case it might be more crucial

I'm half asleep! It doesnt have ram ref, i was thinking of term!!
Cheers for the tips though, Im determined to get this ram oc'ed within an inch of its life!

[shiver_8]

Thermalright Ultra eXtreme or Noctua NH-D14 cooler.

Ahhh... neither. Using my good old $30 Xigmatek HDT-S1283.

[shiver_8]

Also before I call it a night just a quick question... Is there a trick that anyone knows of to decrease vcore but maintain stability? I tweaked my reference voltages (using a ud3p) and am prime stable, but my vcore is too high now and decreasing it results in errors. Does PLL help at all? Thanks

[Songok]

Here's my effort so far on getting my rig prime stable.

Great stuff Gurusan, Very nice indeed! I cant put vcore above 1.325 or IBT hits 76/77. My cooling is a noctua NH-U12P. All the reviews said it was great, turns out its not!! What cooler are using? Temps?

[Hoss331]

Are there not any DDR reference voltage adustments on the EP45T?

The P and R T boards do have ref voltages, more than the ddr2 boards, but the LRs do not. So far the only one that seems to be of benefit is DRAM Term. when on 4.00B above 450fsb.

[In-Fluence]

I see, good to know, thanks for the reply. I see you've got a 'voltage II' tab as well - nice . The ram refs seem to have minimal effect on stability, although when I ran my reapers over 2.1v, the values did benefit from a notch above 0.5xvdimm, and that got me 1295MHz memtest #5 - 1 hour stable (plus a few passes of full test). As for the skews, on the MIIF, tuning the skews with one stick at a time in its own slot, and pushed to the limit was the best way to do it (and relaxed PL). Both sticks together makes it harder. Some settings will not boot at all, others will give you more stability, much like the CPU/NB skews.

[gurusan]

Great stuff Gurusan, Very nice indeed! I cant put vcore above 1.325 or IBT hits 76/77. My cooling is a noctua NH-U12P. All the reviews said it was great, turns out its not!! What cooler are using? Temps?

The NH-U12P is great, some chips run hotter than others though, usually lower VID chips will run hot, I had a 1.165V VID chip once that ran hotter than the sun.

I am only using a Scythe Mine 100mm cooler that I picked up used for 10 quid Temps are not a problem as you can see, ambient is 20C. Also I don't usually use IBT because it heats the chip up so damned much, it's quite unrealistic IMO. I might do an IBT run later on to get in that 4.5GHz LinX stable club though

[Songok]

The NH-U12P is great, some chips run hotter than others though, usually lower VID chips will run hot, I had a 1.165V VID chip once that ran hotter than the sun.

I am only using a Scythe Mine 100mm cooler that I picked up used for 10 quid Temps are not a problem as you can see, ambient is 20C. Also I don't usually use IBT because it heats the chip up so damned much, it's quite unrealistic IMO. I might do an IBT run later on to get in that 4.5GHz LinX stable club though

I didnt know VID was a factor with heat. I thought all chips would give the same temps at a given Vcore no matter what the VID? So lower VID=better oc but more heat? I had a mates E5200 at 4Ghz in my machine and it never passed 55@full load with the the NH-U12P now that i think of it. Maybe my E8400 is just a scorchio chip! It has a 1.175 VID btw. Agreed IBT does kick things up to a new level but ive had chips run small fft's for 24hrs+ and need a small vcore bump to get 25IBT passes without errors. I am open to the idea that IBT is so unrealistic as to be unneccesary though.

[gurusan]

In my experience chips with lower VID run hotter and OC well. Chips with higher VIDs usually run much cooler but are not necessarily bad clockers (my current chip as an example).

1.175V is a good VID, I would keep pushing for 4.4GHz or maybe 4.5GHz

[Songok]

In my experience chips with lower VID run hotter and OC well. Chips with higher VIDs usually run much cooler but are not necessarily bad clockers (my current chip as an example).

1.175V is a good VID, I would keep pushing for 4.4GHz or maybe 4.5GHz

I See! Learning something new here everyday, literally!
I can get small ffts stable at 4.4 with 1.248@CPU-Z (llc disabled) but temps are borderline at 75. 4.5 seems to need two notches of vcore higher (at least) and temps hit 78 so i havent primed that for more than couple of minutes. Have tried AS5 instead of the noctua compound but temps jumped 5c at load so im just gonna stick with 4.25 for now and start reading up on watercooling!!

[gurusan]

Well I'm usually fine with low 80s for prime as normal system use the temps won't get quite as high. Watercooling is nice though as you really don't have to worry about temps much at all. I sold my WC setup over the summer and miss it

[In-Fluence]

Mine's a 1.2 vid and I've hit 5GHz with it. Easily stable at 4.5 too. WC setup is more than likely to be needed for those kind of speeds though, particularly once you go over 1.35v on air.

[bluestang]

I finally got memory stable at DDR3 1333 (stock!). The problem was the ocz mem (or all DDR3 on this board?) doesnt like volts at all. I think hoss posted similiar few pages back. I couldnt get mine stable at rated 1.8v so after much i lowered to 1.78 then stable at 1.76. Now i can oc it and hopefully get 1600 stable. Also gonna try for 4.5 but think my cooling might limit me as am hitting 74 on IBT.


EIST enabled

I still think you MCH ref might be 1 or 2 steps too low. It would be great if you could get your PL down to 8 or 9. And your right more volts isn't always the answer. On my DDR3 setup more volts made unstable; less made stable, especially with the RAM. I always keep a voltage monitor program open so I can see what under/over volts from BIOS settings.