DFI UT X58-T3EH8 continued - Part 2

[ci2kla]

Sure man. I've found that with 4gb everything works fine, but i have problems with 6gb, it's a cpu problem

If you have time, try to mess with the DRAM rank / channel interleave settings

[Razer(x)]

If you have time, try to mess with the DRAM rank / channel interleave settings

I have time, but i really don't know how to set them

[Peen]

Now running 4.5ghz HT 1.488v stable haha.

1.488v, if it dies...it dies. Not worried about it though since intel says 1.55v max and 100c max im still well within those specs =)

[Sam_oslo]

Now running 4.5ghz HT 1.488v stable haha.

1.488v, if it dies...it dies. Not worried about it though since intel says 1.55v max and 100c max im still well within those specs =)

Got a shot of 4.5GHZ?

If you got proper cooling then Vcore=1.5v can be OK for 24/7, but how much "unsafe" VVT you are pushing for those Uncore>@4200 24/7? VVT may be a bigger problem, not because it will necessarily kill you (even tho this may do it faster than high Vcore), but because it puts too much stress/heat on PWM/system, and the CPU-temp raises too.

The extra stress/heat generated by high VVT and Uncore (x5 RAM multi) was so much that I found it to be far from "sweet spot" for 24/7 use (at least on my setup).

[pangingIII]

BCLK 200 CPU 12~21x (Turbo, CPU EIST enabled, C1E Auto)
Uncore 20x
Memory 8x
12GB RAM 8-8-8-24 1N

BIOS...
CPU VID Auto + 103.30%
IOH 1.13
CPU VTT 1.31
Vdimm 1.62


SuperPi 32M...







LinX and Everest bandwidth & latency...

[Peen]

Got a shot of 4.5GHZ?

If you got proper cooling then Vcore=1.5v can be OK for 24/7, but how much "unsafe" VVT you are pushing for those Uncore>@4200 24/7? VVT may be a bigger problem, not because it will necessarily kill you (even tho this may do it faster than high Vcore), but because it puts too much stress/heat on PWM/system, and the CPU-temp raises too.

The extra stress/heat generated by high VVT and Uncore (x5 RAM multi) was so much that I found it to be far from "sweet spot" for 24/7 use (at least on my setup).

Im running 1.45vtt with 12gb ram. 6gb it needs alot less. I know its alot more then intel recommends but I even ram my old P45 with 1.55v for a year no problems. I have fan's blowing on my PWM's so they don't get too hot under load. I'm not afraid to push voltages at all as long as temps are ok...

On AIR cooling, high uncore and VTT can make a huge temp difference, but water is very minimal so I can run as high Uncore as I want and only have a couple degrees difference.

You just want a CPU-z shot of 4.5ghz?

[Sam_oslo]

Im running 1.45vtt with 12gb ram. 6gb it needs alot less. I know its alot more then intel recommends but I even ram my old P45 with 1.55v for a year no problems. I have fan's blowing on my PWM's so they don't get too hot under load. I'm not afraid to push voltages at all as long as temps are ok...

On AIR cooling, high uncore and VTT can make a huge temp difference, but water is very minimal so I can run as high Uncore as I want and only have a couple degrees difference.

You just want a CPU-z shot of 4.5ghz?

Actually i wanted to see a load-shot with all temps and volts, specially the VVT, but it is OK.

You see, I have been trying to find the sweet spot for my new RAM (1600MHZ CL7) and It has been running stable enough at DDR3-2100MHZ CL9 to consider it, but I have been having hard time to justify the extra stress/heat compared to the performance gains.

I have been experimenting @4.2GHZ(20x210), and
For x4 RAM (Uncore 3800MHZ, DDR3-1688 CL7), I can use VVT=1.21v (BIOS set) and it will end up as:
1.26v (BIOS report)
1.23-1.24 (Idle)
1.24v (Load)
It can raise up to 1.26v under heavy load(LinX, HyperPI), but it is still within the "safe 24/7 Intel recommendations" and runs very cool stable, anyways

For 5x (Uncore 4200MHZ, DDR3-2100 CL9), I have to use VVT=1.32v (BIOS set) and it ends up as:
1.38v (BIOS report)
1.32-34V (Idle)
1.34-1.36v (load)
These are volts reported by SG, but Everest will usually usually report 0.01v more.

Not only these are "unsafe 24/7" values, but also there is a good deal of different between what BIOS, SG and Everest are reporting. Which indicates "unreliable readings" and I'm still wondering how much "real VVT" would show on a multi meter?.

When it comes to temperatures, moving from 4x RAM to x5 has increased the system/PWM a whole 4-5'C on idle, ( with a fan with good airflow blowing right over the NB ), and the CPU-temp raised 1-3'C too. I don't have enough data to conclude about load-temp but it wouldn't surprise me if it would increase even more.

Of course x5 RAM (High VVT and Uncore) shows very good benching numbers (specially on Everest bandwidth and latency), but I have been trying hard to see the effect on my "real life applications" and finding it hard to track the performance gains for 24/7 use.

To make a long history short, even tho it is always fine/fun for testing/benching, but the big question is if all those stress/heat generated by high (and unsafe) VVT and Uncore can be justified compared to the performance gain of x5 RAM for 24/7 use?

[Peen]

Well at 4.51ghz I can't really run 4.3ghz Uncore so its 3.87ghz uncore and 1.32vtt under load.

heres a pic of 4.5ghz running memtest, ran for about 9hrs all 12gb. I only run my ram 1290mhz 7-6-6-16-1t-44. The good news is you can run much lower mem voltage which it doesn't impact performance and your PWM temps and even CPU temps are a touch lower.

I'll get something like prime under load later, busy day today

[Garrett]

Impressive... what cooling do you use for the cpu?

[Peen]

multiPi 32m - 11.297 seconds
superpi 1m - just 9 seconds
superpi 32m was 8min 8s but it deleted the entry when I ran 1m lol




Garret, I have Apogee GTZ and a little Swiftech 220 rad with 4 weak fans push/pull. Really need to upgrade my radiator the rad gets almost hot at the top under load after awhile

[Garrett]

Indeed, 1 core 88c
That's way hotter than you'd want indeed
Your 24/7 speed is 4.4GHz right? But at what voltage?
This DFI board of mine has so many options, and I'm not exactly n00b when it comes to oc'ing but the bios still seems a bit overwhelming at times

[Peen]

Indeed, 1 core 88c
That's way hotter than you'd want indeed
Your 24/7 speed is 4.4GHz right? But at what voltage?
This DFI board of mine has so many options, and I'm not exactly n00b when it comes to oc'ing but the bios still seems a bit overwhelming at times

The 88 was after linX so it never gets close to that actually.

4.4ghz is 1.44v

[Peen]

Heres a pic of my ugly rig w/p the side panel, need to fix the wiring so its not so fugly.

[Peen]

Well something just went "pop" and died, no clue what just died but fans just spin up full blast now and debug LED just shows full on 88

[revogamer]

how much VTT did you have?

I fried my VTT on my first dfi because it burnt itself out, was because to much voltage and not enough cooling, if you keep it cool it doesnt matter~

[Peen]

it was 1.29v vtt

[revogamer]

oh ok, bad luck, definately sounds like motherboard though, might be time for RMA

[Peen]

First and last DFI. Worst board i have ever had, wouldnt recommend it to anyone. Its always made funny PWM squealing sounds even new and at stock speed with a finicky POS bios.

double wow, so I try my cpu in my somewhat dead MSi X58M and the bastards mosfets actually caught on fire it was throwing sparks everwhere and actually a huge flame with it! wtf, did someone cast a spell on me or something? If i wasnt there to unplug the psu it would have kept going and burnt my house down

Can someone on their DFI board take out there CPU and start it see if you get 88 post code? Does the same thing with CPU in or out. I was thinking it would just stay CF post code but right now I have no way to know if the CPU or board died. Both look good CPU caps on bottom and everything like fine and mosfets under heatsink on board look good and no smells. It's probably the board but who knows, don't want to RMA and find out it was the CPU all along.

edit: some edits

[BieBiep]

First and last DFI. Worst board i have ever had, wouldnt recommend it to anyone. Its always made funny PWM squealing sounds even new and at stock speed with a finicky POS bios.

double wow, so I try my cpu in my somewhat dead MSi X58M and the bastards mosfets actually caught on fire it was throwing sparks everwhere and actually a huge flame with it! wtf, did someone cast a spell on me or something? If i wasnt there to unplug the psu it would have kept going and burnt my house down

Can someone on their DFI board take out there CPU and start it see if you get 88 post code? Does the same thing with CPU in or out. I was thinking it would just stay CF post code but right now I have no way to know if the CPU or board died. Both look good CPU caps on bottom and everything like fine and mosfets under heatsink on board look good and no smells. It's probably the board but who knows, don't want to RMA and find out it was the CPU all along.

edit: some edits

88 is indeed standard post code w/o CPU for the board.

but I had it more than once with a CPU in it as well. Usually meant something got screwed so badly only taking out the CMOS battery would work to get it back :p

Worked for me a couple of times already... Ofc, if your CPU is dead it probably won't do anything :P

[ci2kla]

@Peen
You're not going to wait for RMA?...
I've given up this board and heading for i5 now...

[Sam_oslo]

Well something just went "pop" and died, no clue what just died but fans just spin up full blast now and debug LED just shows full on 88

I tried to tell you with a long post, you have been pushing (and rubbing ) it over the "safe 24/7 limits" for too long.

As stated above, the 88 code can appear with a good CPU too. Probably it is your CPU, but it could be your MB too. I would check the PSU too.

Keep as informed, then this can be a really interesting investigation on the long term effect of too much VVT for 24/7 use .

[Peen]

I know but I only had 1.29 vtt under load when it died. Kinda hope its my CPU haha. I like pushing voltages and this is the first time I've actually killed something from too many volts that I can think of. Been OC'ing for long time. But you know, you spend $300 on a enthusiasts board you kinda expect the board would put up with it. Still got my old low end P4 MSI pushing way more then 24/7 safe volts since 2003 and its still going!

The PSU is fine, already tested it

Really wish I had some friends close by that I could test my CPU or something. I really deserve like a war medal for pushing things haha

I tried to tell you with a long post, you have been pushing (and rubbing ) it over the "safe 24/7 limits" for too long.

As stated above, the 88 code can appear with a good CPU too. Probably it is your CPU, but it could be your MB too. I would check the PSU too.

Keep as informed, then this can be a really interesting investigation on the long term effect of too much VVT for 24/7 use .

[sjakie666]

i had 2 boards with code 88. never had used high voltages or something(well within intel spec)...
1st board just did nothing 1day. the other some sort of resistor or something else burned near the NB(after a month). 3rd board my dimmslots(3of6) just died on me(after few days).
replaced psu,vid-card, tested cpu in another pc.
dfi doesn't like me .
Now waiting for another vid(4850x2) to arrive and i'll try my 4th board lol.

[Peen]

That sucks man! I don't think mine killed the CPU because I remember it was just shutting off randomly before it finally popped and stopped working. DFI is going to RMA it, but I might just buy a different board and sell the RMA board. I dunno what I want though, maybe the Rampage II Gene??

[sjakie666]

lol. im getting such a low price on my dfi that i'll try it just 1 more time.
this board has rev.aB01 so maybe i'm lucky.