X79 UD5 Gigabyte testing and 3930K

[burt123]

Hi Amurtigress,

Comprehensive reply

Would a later Server OS be stable enough, get away from the XP era of OSes.........

Have you tried AIDA64 for your voltage & temp monitoring, they bring out updates nearly every other week to keep up, works pretty well, and you can have a sidebar gadget, too.

So is there really any need to run a 125 strap ???, I'm using 100...had a quick "fiddle" with 125...didn't like the "odd numbers".

So what do you do with your machine ???

Mine is mainly for video encoding, more power the better.

I also have a 980x on a X58-UD7 @ 4212, and a Sandy 2600K @ 4600 (and quite a few lesser others)

Cheers

[Crazystang01]

Sup guys,

Maybe a stupid question but. Just wondering is it normal on Gigabyte's X79 boards for the cpu mounting holes to still have pcb?

I own a X79-UD5 that I am getting ready and was looking over my board as for placement of my water blocks. I have Heatkillers VRM mosfet and X79 chipset water blocks on top of my Heatkiller cpu.
I was looking at the mounting of the cpu and well I am just use to seeing thru the 4 holes (like in air/light) as in like all my other PC builds. I don't want to go and mess the pcb up with bolts or do they just punch thru?

[sin0822]

it is just a thin plastic layer that blocks the light, you can puncture it with a exacto knife or something, or just screw right through them. I would puncture then screw

1.25x BLCk strap and the BLCk straps in general aren't really needed for 39XX as they have multiplier controller. They do however help with memory speed tuning so you have more selection, but they don't really do much for OCing. They can't change performance either, or shouldn't at least.

Sometimes though on certain boards that are tuned for strap more than without strap, vcore required might be lower with strap.

[Amurtigress]

Burt, (...and sin0822)

a later server would work but they would be nothing but server versions of Windows 7 or vista. And thus pointless.

I just tried Aida 32 bit, and it crashes both Server 2003 x64 and XP. Stop Error 0x8E.

I cannot help but I get the feeling that this problem is having to do with the SMBus. Isn't the new System Management software HECI from intel the same? I would suspect that there is a problem with that under those pre-vista OSses with that driver...
They call it "Software", but I never get to see anything of it after the installation....

I am going to use it for some 3D rendering mostly with Blender, some minor gaming, but overall I am better off with less but higher clocked cores than massive threading on 12 or more threads.

I am going to switch to my secondary BIOS to see if a setup without strap 125 is changing anything, but it looks to me like it won't.

BTW, my X58A-UD7 with i7-950 looked much more mature than this new setup.

Update:

I tested the onboard LAN and CPUID issue with defaults. No change whatsoever.

My Intel LAN's throughput to the internet is seriously reduced as opposed to my PRO/1000 PT add-in card.

On my 50 mbits VDSL line I am getting 47.15 mbits stable throughput on the PRO/1000 card and the onboard 82579V chip tops out at about 25 mbits. Forcibly reducing the link speed to 100 mbits helps a bit, gets me 36 mbits. Tried replacing the router, no change.

In my LAN I am getting 939 mbits/sec between the 82579 and my notebook's Realtek 8168. The same as with the card.

VDSL specs: 50 mbits down, 9.2 mbits up, VLAN tagging enabled for multicast IPTV. Used VLAN tags are 7 (internet) and 8 (IPTV). Router does the untagging...

PS: I managed to make the onboard LAN work on Server 2003. Intel added all the needed device IDs to the driver's INF file but for some stupid reason commented them out by using a "ExcludeFromSelection" section specifically for Dev_1503 and 1502. Deleting that section and selecting the 82579LM network device upon installation makes it work. The previously described problem however occurs on all OSses alike. XP, Win7 and Server.

[zvonexp]

I have X79-uD5 and 3930k.
I had clocked my 3930k at 4.5GHz and after several mins of prime or linx it goes back to 3.2GHz.
C1E/EIST and other options - disabled.
What is the reason?

On Gigabyte board you must go in UEFI / Bios

EDIT : I now saw in your manual and fix for your problem is turning off CPU Thermal Monitor !

[Amurtigress]

Hello people,

after 2 months I've finally gotten the hang of my new setup here. However there are still some details that make me wonder.

One of them is the PCH core voltage. I found it to be stable at 1.20V, but according to intel the max. voltage should be 1.16V. And furthermore, since the bclk is still 100 MHz, shouldn't be the DMI2 and PCIe clock also be 100 MHz, and thus default? Despite having 1.25 strap enabled on my 3820.... 1.10-1-16V cause my mouse to suddenly stop working, and 1.16 causes graphics crashes, which is a known effect of too little PCH voltage on the regular SB systems (BSOD 0x116).

Why is the Graphics card affected by the SB voltage while being connected directly to the CPU's PCIe?

Why do I need more PCH voltage if supposedly the SB isn't overclocked, or am I missing a construction detail of the SB-E platform?


A different thought tho. The SB-E is usually called a power hog. Why? My system at 4.5 GHz takes 60W less under full load (CUDA and CPU stressed, GPU the same as on previous S1366 build) less than socket 1366 and a 950 at 4 Ghz.

That's way more effective.

Testers should be aware tho that they were using 6 cores as opposed to 4 cores on socket 1155. All the power saving (compared to S1366) through the die shrinking can certainly not make up for the two extra cores you have....