DFI UT X58-T3EH8 continued - Part 2

[Razer(x)]

Did you find some kind of spec sheet for your ram to set the timings or did you try around for a bit? Would be nice to know if something like that exists for OCZ memory.

What bothers me the most is that one boot is stable, the next is not.

P2501 i'll continue to test. If you have some news please contact me via PM, we're on the same boat

[ChaosAD]

I noticed using CPU Tweaker that my sub timings did not match the XMP for my Dominator 1600CAS8 stuff.... I manually set some stuff like TRFC, did the TRRD set to 14 so it shows up as 6 and some other timings. I also +1 to Channel C on RTL it seems to be doing a little better.

Since i also use the same Dominators 1600C8, is it possible to write down the sub timmings that you run stable?

[Peen]

Kinda worried about this board now, read all the pages. I also have OCZ 1600 6GB ram.

Thanks for all the links Eva on the first page, will have to read all those first

[The-Fox]

cheapseats, what ram and Dram voltage are you running there ?

Eva, killer timings, what Dram voltage are you giving those Elpidia ?

[Kimi.Rk]

My PC

DFI x58 te3h8
920d0
OCZ platinium cas 7 12800 @ 1600 7-6-6-18 1t (200x20) 6x1 giga
VTT1.39
RAM 1.72
Bios 508

Stable LinX,Prime95-259,OCCT

NO PROBLEM!!

[ChaosMinionX]

Did you find some kind of spec sheet for your ram to set the timings or did you try around for a bit? Would be nice to know if something like that exists for OCZ memory.

What bothers me the most is that one boot is stable, the next is not.

Since i also use the same Dominators 1600C8, is it possible to write down the sub timmings that you run stable?

Ok as you guys can see in the SPD window of CPU-Tweaker 1.1, the XMP#1 profile for the Corsair Dominator 1600C8 3x2GB kit has the profile for 800mhz (DDR3 1600).

The subtimings you will be interested in are

TRC-
TRFC-
TRRD-
tWR-
tWTR-
tRTP-

For this particular set of ram it looks like this....

TRC- 41 (Cannot set this in BIOS, max TRC = 31 in BIOS, is this an issue?)
TRFC- 88 (in BIOS this auto's to like 72)
TRRD- 6 (this was covered in a couple of the begin topics listed in eva's post, setting TRRD to 14 in BIOS results in actual TRRD of 6. This Thread for more info)
tWR- 12 (set AUTO in BIOS to 10, not sure how much of an effect this has)
tWTR-6 (again not sure on its effect on stability)
tRTP-6 (again not sure on its effect on stability)


I have also messed around quite a bit with RTL (round trip latency) settings, however its quite a daunting task considering every time I change a setting on VTT, it seems like RTL is changing? I read in the same thread discussing the TRRD that this was occuring.

I am finding more stability with 20x200 HT on and running RAM @ 1600mhz when I bump the IOH/ICH 1.1v voltage to 1.21, and have found IOH Analog Voltage to be of little to no use past 1.15v. (I am not sure how safe it is to run IOH/ICH @ 1.21v, but it certainly has gotten me a little longer before BSODs, so I will chalk it up to progress! Can anyone shed some light on the safeness of that voltage?)


EDIT: Going to try the new 6/19 BIOS posted in the BIOS thread, says

# 06/19 beta

1. Enhance the system performance

EDIT2: This new BIOS looks promising, my memory bandwidth has dropped a bit, however looks like Prime95 LargeFFT / 5300MB Priority 9 is passing further than it did on the 5/08 BIOS. Will report back fully when I see what else might have changed, I am gonna compare some stuff.

[miahallen]

Forgive me if this has already been pointed out, I didn't read the whole thread:

So I've been using a Gigabyte EX58 EXTREME for a few months now, and I've been very happy with it. But I do not have an 8th expansion slot in my case to run tri-SLI with it. All of the GB boards have their PCIe slots arranged so that it you run tri-fire or tri-SLI using dual slot cooled cards...the 8th expansion slot would be required.

So I wanted to upgrade to tri-SLI, and was looking for a new board. I'd used the eVGA before, and thought it was pretty nice, but I wanted to try something new, and went for the DFI instead...oops
Link to the DFI board for reference.

The onboard sound option requires mounting in one of the rear expansion slots (the normal I/O area is used for cooling). Since I was going to be running tri-SLI, the only spot available is the top spot...PROBLEM IS THE RIBBON CABLE ISN'T LONG ENOUGH TO REACH THE TOP SPOT!!! Not only that, but since there isn't an electrical slot in that top position, adding an aftermarket sound card isn't an option either.

Bottom line, if you want to run tri-SLI on this board, the only option for any audio would be to buy a USB audio solution!!! WTF??? :screwy: Who the hell designs this crap?

[Peen]

I would try to find a longer cable, or email them to mail you one. This board is $300 bucks they should just give you one I would think...

Well mine gets here friday, cant wait to fight it with my OCZ ram

[ChaosMinionX]

I keep getting this BSOD......

A clock interrupt was not received on a secondary processor within the allocated time interval.

Does anyone know what this directly correlates to? In terms of what voltage is too low or what might not be set properly?

[Peen]

I had that happen on my MSI X58M alot, needed more vcore.

[Sam_oslo]

My OCZ133MHZ CL7 (non Animal), would crash exactly after 4.49sec LinX each time, unless I set DRAM PWM switch to 15% decrease. Then it will pass LinX easily with exactly same settings on everything else.
My RAM can't run at 8x but the Uncore is at 16x.

[Praz]

I will ask that the "DRAM PWM Switch Frequency" be looked at. One should be able to leave this at Default for most overclocking. Also of concern is the reported memory bandwidth increase some are seeing when changing this setting. It is not possible for the switching speed of the supply to impact memory bandwidth. Something else must be changing behind the scenes when altering this value.

[Sam_oslo]

I will ask that the "DRAM PWM Switch Frequency" be looked at. One should be able to leave this at Default for most overclocking. Also of concern is the reported memory bandwidth increase some are seeing when changing this setting. It is not possible for the switching speed of the supply to impact memory bandwidth. Something else must be changing behind the scenes when altering this value.

That's great, this setting seams to be affecting LinX, mostly. Everything else including Prime Belnd, Memetest 3.8, windows memory test, and all applications that i have tried would run without any problem with "DRAM PWM Switch Frequency" at Default.

It sames to be mostly related to something that would make LinX unstable (at least on my setup, and as far as i could catch the effect of this setting).

[ci2kla]

Does anyone tried 0619 beta?
It seems that 26 reboot issue is improved.
Can anyone confirm this?

[cheapseats]

Great result!
Good to see you here Cheapseats

thanks ci2kla any update for 230bclk Prime ? a long way from that here

nice work cheapseats pretty spot on

thanks for posting this screen eva, helpful stuff

1) i'm ~4-5s slower @ x16 uncore + same mem timings. looks like my OS is too slow atm for efficiency runs.
need x18 uncore + tighter mem timings just to keep up with your x16 efficiency.

2) second very useful thing is vTT/vDIMM for 32M.. noticed that your vDIMM was less than 0.1V > vTT.
so realised i could also reduce my 0.235V gap between vTT/vDIMM to run single 32M.
good news for my vDIMM levels

cheapseats, what ram and Dram voltage are you running there ?

was 1.8V now 1.7V

[ci2kla]

thanks ci2kla any update for 230bclk Prime ? a long way from that here

Here you go mate

That's a direct setting from BIOS, direct cold boot to windows

[p2501]

ChaosMinionX, thanks for your post. I hope I find the time today to mess with the timings. IIRC at least tRFC was set automagically far tighter for me, will report back on that. Concerning that BSOD, it's core related I think, most likely VTT. That is was fixed these for me at least.

[Razer(x)]

ChaosMinionX, thanks for your post. I hope I find the time today to mess with the timings. IIRC at least tRFC was set automagically far tighter for me, will report back on that. Concerning that BSOD, it's core related I think, most likely VTT. That is was fixed these for me at least.

Mate let me know

[ci2kla]

Oh one more thing...my machine stopped running PI recently as it keeps hard locking me....any ideas? LINX+MEMTEST+ORTHOS = PASS....

[Praz]

Oh one more thing...my machine stopped running PI recently as it keeps hard locking me....any ideas? LINX+MEMTEST+ORTHOS = PASS....

You're talking single instance? If so good luck. The board I have here has never been able to run a single instance of SuperPi at 190 or higher BCLK. Can run 8 instances just fine. I have asked that this be checked out.

[cheapseats]

very useful bclk ci2kla..

single 32M lock/freeze windows for me @ 220bclk was pcie.. < 110 = lock / 110 = 32M ok. but only tested 32M.

[ci2kla]

Seems I've found something interesting. Once I proved that I'll share to you guys

[Star_Hunter]

Hardocp just reviewed this board and gave it a silver but you can tell he hates DFI, he only seems to like those who pay for ads. (mentions no free review board and no ad support in the forums along with the bad customer support (I never personally dealt with their support but many have said its not the best)).

[eva2000]

Seems I've found something interesting. Once I proved that I'll share to you guys

232bclk prime95 - do tell!

[ci2kla]

okay seems the freezing issue is pointed to the relation between VTT and vDIMM.
Here's the test.
vDIMM 1.65v VTT 1.28v CL8-8-8-20 2000Mhz -> 32M PASS
vDIMM 1.62v vTT 1.28v CL8-8-8-20 2000Mhz -> 32M FREEZE
vDIMM 1.65v VTT 1.34v CL8-8-8-20 2000Mhz -> 32M FREEZE
vDIMM 1.65v VTT 1.28v CL9-9-9-20 2000Mhz -> 32M PASS
vDIMM 1.62v VTT 1.28v CL9-9-9-20 2000Mhz -> 32M FREEZE

I'll say from this test my chip doesn't like VTT to be too close to vDIMM...