**Gigabyte GA-X58A-UD7 OC-Tips-Bios-Discussion Thread**

[Lsdmeasap]

Cool, just wondered how it was since not many people have them.

It's sandforce right?

I just got a Gigabyte 5870 SOC myself last night! Will be posting a thread on it next week and I will be doing a review of the 5870 SOC vs 4890 once I have time

[radaja]

the P128 is the with the older samsung controller i think?

[Amurtigress]

Hicookie,

thanks for postiung your BIOS template on the 980...

Something seems odd to me. The VTT/Uncore voltage. My "old" C0 920 is known to need higher uncore voltages, mine runs fine at 1.435V/3600 MHz QPI clock. The trademark of the D0 steppings is their much lower uncores...

And there is your 32 nm chip, needing 1.415V VTT like my C0 920?

Is that a common thing on the hexcores?

[radaja]

hicookie is running memory at 2300Mhz cl8.thats probably why vtt is at 1.42v

[Sam_oslo]

BIOS-F7f has unlocked the locked Uncore on my X5650, and now it is hitting BCLK221 with safe VVT The new microcode seams to do great wonders on Westmere

I've posted the good mews here: http://www.xtremesystems.org/forums/...d.php?t=251511

Just a teaser, and a quick Prime-run to test the temp, and it looks promising, but I have to test and optimize more.

X5650 @4.2GHz(210x20) 1.32v , VVT=1.30v

[Orion24]

BIOS-F7f has unlocked the locked Uncore on my X5650, and now it is hitting BCLK221 with safe VVT The new microcode seams to do great wonders on Westmere

I've posted the good mews here: http://www.xtremesystems.org/forums/...d.php?t=251511

Just a teaser, and a quick Prime-run to test the temp, and it looks promising, but I have to test and optimize more.

X5650 @4.2GHz(210x20) 1.32v , VVT=1.30v

And what's the problem with some higher VTT? I currently run at 1.54V to stabilize 4.0 GHz uncore and I'm thinking of trying to sync it with the core at 4.2 with about 1.6V or more.

[Sam_oslo]

And what's the problem with some higher VTT? I currently run at 1.54V to stabilize 4.0 GHz uncore and I'm thinking of trying to sync it with the core at 4.2 with about 1.6V or more.

High VVT is usually a bigger killer/degrader than high vCore, but i7 920 (and other 45nm i7) has proven to take a lot of torture, including high VVT. These 32nm are hopefully as resilient, but we don't know the long term effect of high VVT on these yet.

I think VVT=1.54v is too much for your 920, even if it doesn't kill/degrade your CPU, it will add a lot of heat in you CPU and VRM. You could save those heat and get a higher OC with the same cooling. You can reduce the VVT with using x16 Uncore and x8 RAM with thigh timing, and you will get the same performance for 24/7 with less heat, and a long life for your CPU.

[MikeMK]

Guys, quick question - hope it's not too n00bish but it's bugging me. Ever since I had i7 I've overclocked with the rule that the mem multi has to be half the uncore multi. However I have seen people posting stable results indicating they aren't following this rule. Are there any exceptions where u do t have to follow this?

[Sam_oslo]

Guys, quick question - hope it's not too n00bish but it's bugging me. Ever since I had i7 I've overclocked with the rule that the mem multi has to be half the uncore multi. However I have seen people posting stable results indicating they aren't following this rule. Are there any exceptions where u do t have to follow this?

On 45nm i7, the Uncore has to be at least 2 times of the RAM.
Uncore>=2xRAM, it meas x10 RAM needs at least x20 Uncore. But it can be bigger too, like x8 RAM and x18 Uncore.

on 17 32nm, the Uncore has to be at least 1.5 times the RAM. I think, but I've never tried it, correct me if I'm wrong guys.
Uncore>=1.5xRAM.

[MikeMK]

Ahh, that would make sense then. Didn't realise it was different for the hexacore/32nm i7s.

[Lsdmeasap]

On 45nm i7, the Uncore has to be at least 2 times of the RAM.
Uncore>=2xRAM, it meas x10 RAM needs at least x20 Uncore. But it can be bigger too, like x8 RAM and x18 Uncore.

on 17 32nm, the Uncore has to be at least 1.5 times the RAM. I think, but I've never tried it, correct me if I'm wrong guys.
Uncore>=1.5xRAM.

That is correct, and with the changes in 980X running 2000+ memory is now MUCH easier

[Fritz_the_germ]

could you try if your kit could do 1.66v 2.4GHz for 24/7?

GSKill's XMP 2.3GHz CL8 no problem with Gigabyte X58.
i though my kit is only max 2350MHz CL8 for 24/7

Intel 980X
Intel 980X stock Fan
X58A-UD3R
Gskill F3-18400CL8T-6GBPIS (DDR3-2300C8)








>Advanced CPU Feature------------------------------[Press Enter]
CPU Clock Ratio --------------------------------------21X (4.03GHz)
Intel(R) Turbo Boost Tech.---------------------------- [Disabled]
CPU Cores Enabled------------------------------------- [All]
CPU Multi-Threading---------------------------------- [Endabled]
CPU Enhanced Halt (C1E)-----------------------------[Disabled]
C3/C6/C7 State Support------------------------------ [Disabled]
CPU Thermal Monitor--------------------------------- [Disabled]
CPU EIST Function--------------------------------------[Disabled]
Bi-Directional PROCHOT------------------------------ [Disabled]
Virtualization Technology----------------------------- [Disabled]
QPI Clock Ratio------------------------------------------X36 (6.91GHz)
>Uncore & QPI Feature---------------------------------[Press Enter]
QPI Clock Ratio -------------------------------------------[X36]
Uncore Clock Ratio--------------------------------------[X19] (3.648GHz)
Isochronous Support-----------------------------------[Enabled]
Base Clock(BCLK) Control-----------------------------[Enabled]
BCLK Frequency(MHz)---------------------------------[192]
>Advanced Clock Control-----------------------------[Press Enter]
Base Clock(BCLK) Control-----------------------------[Enabled]
BCLK Frequency(MHz)---------------------------------[192]
PCI Express Frequency(MHz)-------------------------[100]
CPU Clock Drive-----------------------------------------[800mV]
PCI Express Clock Drive----------------------------[900mV]
CPU Clock Skew-----------------------------------------[0ps]
IOH Clock Skew------------------------------------------[0ps]
Performance Enhance---------------------------------[Turbo]
Extreme Memory Profile(X.M.P)--------------------[Profile1]
System Memory Multiplier(SPD)--------------------[12.0]
>Advanced DRAM Features-------------------------[Press Enter]
Peformance Enhance---------------------------------[Turbo]
Extreme Memory Profile(X.M.P)------------------[Profile1]
System Memory Multiplier(SPD)------------------[12.0]
DRAM Timing Selectable (SPD)---------------------[Quick]
Channel Interleaving-----------------------------------[6]
Rank Interleaving----------------------------------------[4]
Channel A 8-11-8-28-auto-auto-auto-auto-auto-8-auto-auto-auto-auto-Auto-Auto
Channel B 8-11-8-28-auto-auto-auto-auto-auto-8-auto-auto-auto-auto-Auto-Auto
Channel C 8-11-8-28-auto-auto-auto-auto-auto-8-auto-auto-auto-auto-Auto-Auto
>Advanced Voltage Control
Load-Line Calibration------------------------------------[Level 2]
CPU Vcore -------------------------------------------------[1.27500v]
QPI/VTT Voltage------------------------------------------[1.415v]
CPU PLL-----------------------------------------------------[Auto]
PCIE----------------------------------------------------------[Auto]
QPI PLL-------------------------------------------------------[Auto]
IOH Core-----------------------------------------------------[Auto]
ICH I/O-------------------------------------------------------[Auto]
ICH Core-----------------------------------------------------[Auto]
DRAM Voltage----------------------------------------------[1.660v]
DRAM Termination---------------------------------------[Auto]
Ch-A Data Vref--------------------------------------------[Auto]
Ch-B Data Vref--------------------------------------------[Auto]
Ch-C Data Vref---------------------------------------------[Auto]
Ch-A Address Vref--------------------------------------[Auto]
Ch-B Address Vref---------------------------------------[Auto]
Ch-C Address Vref-----------------------------------------[Auto]

Good work!
But don´t you think the QPI/VTT is a little to much English for the 24/ use?

@Lsdmeasap,
this is for you: http://www.youtube.com/watch?v=3oRKvpZ7PjE

[Sam_oslo]

You guys with 980x or x56xx should flash to BIOS-F7f, and have a new look at your previous OC. You will probably be able to improve it further.

This new microcode patch for 32nm has made several big improvement. The Uncore of my X5650 has improved dramatically, a totally new world, and much more stable and manageable. The LLC has changed a good deal too, now I get the same Vdroop with LLC=level1 and standard. Overall the CPU feels much better and more stable.

[hicookie]

Good work!
But don´t you think the QPI/VTT is a little to much English for the 24/ use?

@Lsdmeasap,
this is for you: http://www.youtube.com/watch?v=3oRKvpZ7PjE

1.42v still be safety vtt, even 980x no need to follow formula(ddr*2=<uncore)
but i try 2.66, 3.2ghz uncore isn't stable at ddr3-2300.
vtt is all about uncore speed's need.

with 3.2(uncore)GHz just needs 1.255v
3.6GHz uncore needs more, i just do a quickly run, not sure if lower than
1.42v stable with 3.6GHz uncore.

[mbreslin]

I feel like I could get so much more out of my chip (I was able to run pcmv at 4.8ghz I just mean in terms of doing 4.4-4.5 with lower vcore) The problem is I don't really know what I'm doing I just stick my 2000mhz c7 triple kit on xmp, set dram voltage to 1.66 pick an odd multi that gives me like 4.2ghz then start raising bclk and vcore/vtt and try to get something high.

This can't be the best way to go about it.. Maybe someone could give a few pointers? I've read i7 oc guides but the 980x seems to be a bit different or at least that's what I read on the forums from people who have pushed 920s and now are pushing 980s.

[SteveRo]

GA-X58A-UD7 - F7h

-Fix Gulftown(Vdimm auto voltage) -Special JZ BIOS
-23.Apr 10

Good morning, does the F7h (special JZ bios) contain the new 32nm microcode patch that was in F7f?

[Sam_oslo]

Just a little 15min prime. Not testing 24/7 stability fully yet, because still trying different options with x19, x20 and x22 multi.
This is not final yet, but looks like a promising candidate for 24/7 use.

X5650 @4.1GHz(216x19) 1.28v, VVT= 1,32v 15min Prime

[MadHacker]

I just got the UD7 board. and at first it worked fine.
got windows installed and then on reboot, had a problem.
network refused to connect.

I flashed the bios from F5 to F7 but that didn't help either.

someone else had a similar problem on the Overclockers.com board.
http://www.overclockers.com/forums/s....php?p=6447538

and their suggestion to fix it was to reset the CMOS and then load bios defaults.

how ever i tried this and it didn't help
quoted from thread above since i have Identical problem.

What it does... with a cable plugged in, I found that the NIC (both of them) is showing in quick succession... "enabled" then "network cable unplugged" and keeps repeating that one after the other.

I have tried different cables / connecting straight to the router rather than the switch / updating the BIOS from F3 to F4 / every time it is the same.

I went into the BIOS and used the smart-lan (I think it's called) feature, and it is showing 187m on all four fields if that is helpful.

anybody experice this before as well and is there a solution besides RMAing the board?

[mbreslin]

Disable them in bios, boot into windows. Reboot and enable them, boot into windows. (not saying it's for sure going to work just that it's an idea) I use an intel pcie nic and recently I thought it was acting up so I switched to internal and both ports showed cable unplugged, took a couple reboots but they're working fine now. I honestly think it's an OS issue.

[MadHacker]

ok i'll give it a try

[MadHacker]

OK i went into the bios , disabled the NICs rebooted into windows, went back to the bios and re enable them. Same effect.
then deleted the devices from device managers, rebooted, disabled NICs in bios , booted to windows, back to bios, enabled NICs rebooted to windows. same affect.
doing widows update and it states there is a update for the NIC driver.
cross my fingers.

[Amurtigress]

I just found this...new beta BIOS for the UD7:

http://www.fileden.com/files/2009/12...58aud7.f7h.zip

No clue how different it is. I am prolly going to flash it... I found F7F to cause more fluctuations in Linx' GFlops results than F6 on my C0 920.

Forwarded from Tweaktown.

Edit: Eeps, sorry, too late...NVM.

[SteveRo]

Good morning, does the F7h (special JZ bios) contain the new 32nm microcode patch that was in F7f?

bump - Good morning, does the F7h (special JZ bios) contain the new 32nm microcode patch that was in F7f?

[Amurtigress]

bump - Good morning, does the F7h (special JZ bios) contain the new 32nm microcode patch that was in F7f?

Not like I had insider information, but I think the microcode update fixes severe enough problems to justify a new BIOS revision. Hence it would be the trademark and purpose of the F7 BIOS in general.

I'd say yes.

[Sam_oslo]

bump - Good morning, does the F7h (special JZ bios) contain the new 32nm microcode patch that was in F7f?

I don't know when they attempted to add the new microcode, but I don't think it was exactly successful before. I couldn't see any effect of the new microcode before F7f. The F7f has made dramatic improvements to my Uncore, and noticeable effect on LLC=standard. I hope they keep it in all future BIOS, including the F7h .