Gigabyte EX58-UD5/Extreme Discussion Thread

[Chri$ch]

• Bios F4l for GA-EX58-UD5 (beta)
  - improved M.I.T. setup

[dinos22]

1.6V to reach 4ghz?

no i am saying that QPI/VTT(uncorevoltage) is key to RAM overclocks with 12GB install jcool is talking about

[jcool]

how do you know that

Because Intel max spec is 1,35V for QPI/VTT voltage..
And I'd say 1,5-1,55 is absolute max for running 24/7 without damaging anything. Of course I can only guess, but running 1,6-1,7V QPI will probably degrade CPU and/or mem over time, at least probability is high.

[dinos22]

yeah man i'd probably be ok with 1.55 for 24/7 but not so sure about more

[fornowagain]

Because Intel max spec is 1,35V for QPI/VTT voltage..
And I'd say 1,5-1,55 is absolute max for running 24/7 without damaging anything. Of course I can only guess, but running 1,6-1,7V QPI will probably degrade CPU and/or mem over time, at least probability is high.

Yeah and default VTT is 1.1v, 1.7v does seem quite an overvolt for normal use. I guess time will tell.

[Lukee]

For me, I can't get 210 Bclk unless I hit 1.7+

For now, I'll stick with 195 Bclk and 1.3 vtt

[Xello]

What are people finding their max stable bclk to be? With all everything else turned down, cpu / memory / uncore multi etc. I'm at 222 and going for 230 next i think.

[CodeRush]

Xello, 222 is hard line to step over.
Benny's word record on i7-920 ES (229+) is greatest result.
There is some kind of hardware bug in C0/C1 stepping, IMHO.
---
Guys, what is your max NB freq?
I have only 4200 Mhz SuperPi 32M stable on Vqpi = 1.555v and Vqpipll = 1.44v, cold water.
I see dinos22's SuperPi 32M i7-920 WR taken at 4400+ Mhz on SS.
Is extreme cooling significantly improves this freq or it is total processor-dependent?

[Xello]

You ain't kidding, suffice to say 230 was a nopost Even after going from 1.38v to 1.5v VTT and upping QPI PLL a bit, no luck.

Strange though to go from 100% 9hr prime stable to nopost in just 8mhz!

I also realized i have QPI in 'slow mode' and that i'll want the lowest normal multi (36x) max bclk if i want to base an actual overclock on it. @ 222 that puts QPI at like nearly 8ghz, i'm guessing that won't be stable

Brings me to another question, is there any general idea of what VTT volts are needed relating to QPI / uncore clock? ie we have good idea of min-max voltages needed for the cpu's between 3ghz and 4+ghz. What about the new clocks? I know some of you guys are saying VTT up to 1.65v is safe for 24/7 no problem but what kind of clocks does that give you? (I know it varies from chip to chip/board, generalise )

[rge]

There was an article on either toms/anandtech...but have to go back to work now, so cant look for, but VTT/QPI voltage scales not only on uncore mhz, but also on amount of ram, as more ram places more demands on mem controller...in addition to diffs in chips, matters a lot whether using 3x1, 3x2 or 12 gb of ram. I am using 3x1 since still on xp 32bit, which is why once I am at 1.30-1.32 VTT, using 18x uncore, raising it more does nothing for me. Those with 6 and 12gb probably on average, benefit more...we need a big chart somewhere of everybodies details and overclocks would be interesting to learn from...and no I am not offering to do it.

[jcool]

Okay guys,

figured the whole 12GB deal out a lot better I guess. It all depends on QPI/VTT like dinos and others have already said, question is, how much is safe..
I managed to get 170 BCLK linpack stable with 1,45V QPI using the 8x mem divider. Using 10x I get no boot at 1,45 QPI, so I tried 1,55 and it works. 1,515 again no boot with 10x mem divider, so I'm guessing it won't be stable either with 1,55V. I am not willing to go higher than 1,55 so I backed it down again to 8x mem divider... runs CL7 tho, CAS latency doesn't seem to be too hard on the IMC

[Nocturno]

Okay guys,

figured the whole 12GB deal out a lot better I guess. It all depends on QPI/VTT like dinos and others have already said, question is, how much is safe..
I managed to get 170 BCLK linpack stable with 1,45V QPI using the 8x mem divider. Using 10x I get no boot at 1,45 QPI, so I tried 1,55 and it works. 1,515 again no boot with 10x mem divider, so I'm guessing it won't be stable either with 1,55V. I am not willing to go higher than 1,55 so I backed it down again to 8x mem divider... runs CL7 tho, CAS latency doesn't seem to be too hard on the IMC

Max Vcore : 1.55v
Max VTT/QPI : 1.35v
Max Vdmim : 1.85v (for 1.35v max VTT/QPI)
Max PPL : 1.89

[Xello]

There was an article on either toms/anandtech...but have to go back to work now, so cant look for, but VTT/QPI voltage scales not only on uncore mhz, but also on amount of ram, as more ram places more demands on mem controller...in addition to diffs in chips, matters a lot whether using 3x1, 3x2 or 12 gb of ram. I am using 3x1 since still on xp 32bit, which is why once I am at 1.30-1.32 VTT, using 18x uncore, raising it more does nothing for me. Those with 6 and 12gb probably on average, benefit more...we need a big chart somewhere of everybodies details and overclocks would be interesting to learn from...and no I am not offering to do it.

YEah i forgot memory comes into it aswell. That would be a big job, i'll settle for just posting my final overclocks when they're eventually done

Max Vcore : 1.55v
Max VTT/QPI : 1.35v
Max Vdmim : 1.85v (for 1.35v max VTT/QPI)
Max PPL : 1.89

Right, but how conservative is the 1.35v limit? There's a thread around here somewhere with long-term VTT tests on core 2 chips and i think they showed intel's recommended max to be a little low. I don't think i'm going to go above 1.38v personally, simply as so far there seems to be no need for it to get a 4ghz+ stable setting - currently 1hr into prime blend test for 219bclk on 36x QPI, with 1.38v (might actually need less). Maybe in the future!

[CJ145]

I seem to be having a problem with my EX58-UD5 and the Virtualization Technology. The setting is enabled in the bios and I am running 64bit vista. VMware says that VT is disabled in the bios and does not allow me to run 64bit guest OS's. Is anyone else having this problem or gotten it to work? Currently on F4I bios.

[theonlybabyface]

Does the Gigabyte have control over all fan speeds? In other words, if I enable fan control in the bios for the cpu (only option) does the bios adjust the other fans speeds accordingly?

I like my pc fast AND quiet....

Anyone?

[Xello]

I'm not sure. I have 7 fans connected to psu and only 2 to mobo. One of them is running at its full 2000rpm, the other is at about 1300rpm. I could control them with my old blackops, using Aegis Panel i could choose speed myself, or set Smartfan to control them in the bios. This doesn't seem to have that, i just downloaded speedfan too and that can't seem to adjust them either.

From the description of the Smartfan in GB bios, it seems to do cpu fan only.

edit:

Checked manual, there are 5 fan connectors including cpu fan (plus 1 for NB fan). Only cpu fan and sys_fan2 have speed control, they both have 4 pins, the others have sense but no control and have 3 pins. sys_fan2 is the one at bottom of board beside front panel connectors.

I'd like to use sys_fan2 to bump my 1300rpm fan up to 2000, but the fan connector is only a 3 pin

[FuriousSalesman]

Has anyone successfully ran memtest with 12gb of ram on this board? I am running 12gb of 1600 corsair dominator and memtest won't boot.

[theonlybabyface]

I'm not sure. I have 7 fans connected to psu and only 2 to mobo. One of them is running at its full 2000rpm, the other is at about 1300rpm. I could control them with my old blackops, using Aegis Panel i could choose speed myself, or set Smartfan to control them in the bios. This doesn't seem to have that, i just downloaded speedfan too and that can't seem to adjust them either.

From the description of the Smartfan in GB bios, it seems to do cpu fan only.

edit:

Checked manual, there are 5 fan connectors including cpu fan (plus 1 for NB fan). Only cpu fan and sys_fan2 have speed control, they both have 4 pins, the others have sense but no control and have 3 pins. sys_fan2 is the one at bottom of board beside front panel connectors.

I'd like to use sys_fan2 to bump my 1300rpm fan up to 2000, but the fan connector is only a 3 pin

Thanks for that. I have this board sitting here in it's box but looks like it may go back as I really like having fan control without using a fan controller. I guess it's the Asus RIIE instead or the Evga x58. I have about an hour to decide. Anyone care to "sway" me either way?

[Hornet331]

Has anyone successfully ran memtest with 12gb of ram on this board? I am running 12gb of 1600 corsair dominator and memtest won't boot.

Im just about to try.

edit:
works without a problem. What version did you use?

[dinos22]

Xello, 222 is hard line to step over.
Benny's word record on i7-920 ES (229+) is greatest result.
There is some kind of hardware bug in C0/C1 stepping, IMHO.
---
Guys, what is your max NB freq?
I have only 4200 Mhz SuperPi 32M stable on Vqpi = 1.555v and Vqpipll = 1.44v, cold water.
I see dinos22's SuperPi 32M i7-920 WR taken at 4400+ Mhz on SS.
Is extreme cooling significantly improves this freq or it is total processor-dependent?

yup havent tested higher yet but i think it wont have issues probably past 4500

i will be testing a 965 and if the CPU is decent i'll go for crazy uncore voltage and crazy NB clocks

[dinos22]

Max Vcore : 1.55v
Max VTT/QPI : 1.35v
Max Vdmim : 1.85v (for 1.35v max VTT/QPI)
Max PPL : 1.89

intel is very conservative
we break the rules all the time heheheh
i realise the concept of degradation too but we are yet to really see the effects of it for many users

if you havent read BenchZoner's thread about high VTT testing you should......we were being scared into ideas that a CPU will die overnight pretty much with 1.5vTT and BZ stressed the out of CPUs for months without any ill effects heh....not only did he test VTT but also vcore and PLL

[flopper]

The fear itself is more scare than the actuality of the cpu to go down.
If people run stuff as they do here and seldom get issues, a more common overclocker will rarely have any issues.

[FuriousSalesman]

Im just about to try.

edit:
works without a problem. What version did you use?

I figured out what the issue was. I was running two x-25m ssd's in raid. Memtest wont boot with the raid array active. I turned off my raid, and then ran Memtest. But when I tried to start my array again it failed. Is that supposed to happen?


Edit. Now that I can memtest I am seeing a problem. If I set Performance Enhance in the MBIT to Turbo or Extreme, I get many many errors in memtest. Did you have that issue?

Edit V2. I just burned my hand on my ram. I placed a yate-loon fan over the ram, It is running much cooler now. Does anyone know of an elegant ram cooling solution?

Edit 3. Even when cooled. Running the Corsair Dominator at 1600mhz with 12gbs still has too many errors when running memtest. Does anyone have any tips?

[namegt]

BCLK 221.53

[ReverendMaynard]

awesome man. Benchable?