Gigabyte EX58-UD5/Extreme Discussion Thread

[rge]

So, if the extra cooling isn't that good, the best choice would be UD5?
Because, UD5 is like 50€ cheaper than extreme...

Yep, what Jcool said. If you want to watercool NB then extreme, otherwise UD5.

With NB temps 47C with spot fan and IMC on cpu, Im not going to bother with wc NB. On other hand extreme was only $30 more when I purchased, not that much diff. in price here.

[ReverendMaynard]

just got the GTZ installed, now maybe I can open this beeyotch up

[safan80]

just got the GTZ installed, now maybe I can open this beeyotch up

since your not using turbo why are you hiding the 4th thread?

[ReverendMaynard]

since your not using turbo why are you hiding the 4th thread?

Hiding? give me an effing break lol. And what does turbo have to do with it? Are you implying core3 was failed?

[freshy]

hey guys, i got a question, first i got 4ghz 14 hours prime stable with 1.375v vcore , qpi/vtt 1.315v

then i flashed to f4n and i needed more vcore so i flashed back to f4k and i still needed more vcore so i ended up doing 21 hours of prime @ 4ghz @ 1.395v vcore

anyway i was folding with 2 smp clients and when i got up this morning the pc had crashed, but when it crashes it doesnt just restart , it kinda just hangs and my gpu fan revs flat out until i push the restart button,

anyone else getting that when it crashes ?

and why is it crashing after 21 hours of prime small ftt, i also ran memtest86 ver 2.10 to test memory and that was ok

[safan80]

Hiding? give me an effing break lol. And what does turbo have to do with it? Are you implying core3 was failed?

It looks like you were trying to show load temp with prime. If you want to show cpu load temps with prime run Small FFts.

[GAR]

It looks like you were trying to show load temp with prime. If you want to show cpu load temps with prime run Small FFts.

Also enable all 8 threads, as it puts more load onto the cpu.

[ReverendMaynard]

dudes, it was the first minute that I fired it up, just comparing my temps from air. Jesus lol.

Blend gives hotter temps on the i7...at least what I've tested so far.

[dinos22]

dudes, it was the first minute that I fired it up, just comparing my temps from air. Jesus lol.

Blend gives hotter temps on the i7...at least what I've tested so far.

lol why are u giving him a hard time
btw if you do want to test for highest temps with prime smallFFT is the hardest usually

[AcUaViVa]

I am flashing with usb key and think all others are too, you must see it but need be fat 32, if formatted ntfs q-flash cant see the partition

its in fat32...

any suguestion?

[jcool]

lol why are u giving him a hard time
btw if you do want to test for highest temps with prime smallFFT is the hardest usually

Dinos,

usually SmallFFT gives the highest temps all right, but I noticed something different with i7 as well.
Like I said in a previous post, I ran prime blend for 53 hours on a customer rig and started off with around 50C on all cores after 30 minutes.
At the end, it stabilized at around 60C, while watertemp didn't change that much (was with 8 threads enabled though). Strangely I saw that high temp on a larger FFT (4096k I think it was) than smallFFT

[X.T.R.E.M.E_ICE]

its in fat32...

any suguestion?

Just format your usb flash drive to fat32, That's what i did.

[dinos22]

Dinos,

usually SmallFFT gives the highest temps all right, but I noticed something different with i7 as well.
Like I said in a previous post, I ran prime blend for 53 hours on a customer rig and started off with around 50C on all cores after 30 minutes.
At the end, it stabilized at around 60C, while watertemp didn't change that much (was with 8 threads enabled though). Strangely I saw that high temp on a larger FFT (4096k I think it was) than smallFFT

yeah if you prime for longer than an hour any test will have roughly the same temps

initial start...first say half hour SmallFFT and LargeFFT tend to be hotter

[fornowagain]

Given the i7's IMC I guess a higher VTT could well make prime blend hotter than just small or larger ffts.

[jcool]

Sounds like a solid enough theory to me.
Running 12GB of ram, 170BCLK and 1,45V QPI/VTT like I said...

[Xello]

Well remember the description for small FFT "data fits into l2 cache, not much ram used", so i'd assume that means it doesn't go into the l3 cache either, which is the largest cache. Blend will put load on the uncore aswell so not surprising to see temps higher.

[ReverendMaynard]

Given the i7's IMC I guess a higher VTT could well make prime blend hotter than just small or larger ffts.

Well remember the description for small FFT "data fits into l2 cache, not much ram used", so i'd assume that means it doesn't go into the l3 cache either, which is the largest cache. Blend will put load on the uncore aswell so not surprising to see temps higher.

Exactly. Different architecture different test methods.

I was always a small fft guy with previous l2 cache dependent cpu's, but with the i7 I found blend gave the hottest temps the fastest because it is working that IMC.

[Nilos]

This is my 24/7 setting for so far. 2 hrs prime stable is more than enough for me. I did it with Bios F4r and the following bios settings:

http://users.telenet.be/nils-marleen/BEST

Now i want to push it a little further to 4.2 Ghz also with HT On but prime 95 fails in 2 minutes, even with 1.392 Vcore. I also tried some additional cooling to the north bridge but nothing helps. Can someone plz give me a tip how to achieve 4.2 24/7 stable?

Thanks.

CPU Batch is 3837B194

[ReverendMaynard]

Still priming, but most likely my 24/7 settings minus HT becase well...I'm a gamer and HT doesn't give me anything but heat.

Nilos, I have the identical cpu batch as you. I'm using 1.4 in bios, which gives 1.376 idle and returns with 1.344v load. QPI/VTT is at 1.335. To keep these levels air isn't going to cut it. I was capped at 200x20 on air.

[Splave]

killer rev! what is vcore in bios, as well as your qpi and pll? 4.2 w/HT sick

[Hornet331]

more voltage... 4ghz is just the wall for most Ci7s. After that frequency voltage requierments to get the rig stable make big jumps.

[rge]

Still priming, but most likely my 24/7 settings minus HT becase well...I'm a gamer and HT doesn't give me anything but heat.

Nilos, I have the identical cpu batch as you. I'm using 1.4 in bios, which gives 1.376 idle and returns with 1.344v load. QPI/VTT is at 1.335. To keep these levels air isn't going to cut it. I was capped at 200x20 on air.

Thats a pretty sweet improvement in temps and OC from air to water, can you list difference in vcore needed on air vs water per any OC when you find difference. Interested to see for example if need notch or two less vcore at same OC or if just requires same with water vs air.

[ReverendMaynard]

Thats a pretty sweet improvement in temps and OC from air to water, can you list difference in vcore needed on air vs water per any OC when you find difference. Interested to see for example if need notch or two less vcore at same OC or if just requires same with water vs air.

Thanks man. So far voltages are identical, but I'll confirm that after a few more tests.

killer rev! what is vcore in bios, as well as your qpi and pll? 4.2 w/HT sick

haha, you know this Vcore in bios is 1.4v qpi is 1.335v and pll is 1.3v.

[Xello]

I wouldn't think different cooling would make a difference to volts needed for stable clocks, unless you were temp-bound ie reaching tjmax and experiencing throttle.

Does the pencil mod work for anyone? I applied it but still seeing the same vdroop. Is it supposed to cut it out altogether or reduce it? I'm still setting for example 1.462v, getting 1.44v, and dropping to 1.42v under load.

[rge]

I wouldn't think different cooling would make a difference to volts needed for stable clocks, unless you were temp-bound ie reaching tjmax and experiencing throttle.

Yeah, just wishful thinking on my part, when i take my rig outside, drop temps 20-25C, I need less vcore for a given mhz to load windows...and can go 90mhz higher for max clock....just did not know if get maybe 1 notch lower for smaller drop in C