Intel X980 on Gigabyte X58 UD7 @ -90c

Printable View

03-16-2010, 11:52 AM
Charles Wirth

Intel X980 on Gigabyte X58 UD7 @ -90c

Intel Q4EG X980
Gigabyte X58 UD7 with 8K OCP
Corsair GT2 2250C8
Nvidia GTX 295 Quad SLI
Thermaltake 1200
Three stage cascade -100c unloaded

Due to OCP tripping at 5.4Ghz, threaded tests run just under to complete.

Cinebench 11.5 on Win 7 64Bit
http://fugger.netfirms.com/ud7/cine.jpg

3Dmark 06
http://fugger.netfirms.com/ud7/3577.jpg

I played more with 3DMark Vantage and scaled up the GPU's for these tests.
http://fugger.netfirms.com/ud7/p34052.jpg

http://fugger.netfirms.com/ud7/p39971.jpg

http://fugger.netfirms.com/ud7/p41284.jpg

http://fugger.netfirms.com/ud7/p41721.jpg

http://fugger.netfirms.com/ud7/p41941.jpg
03-16-2010, 12:06 PM
chew*

So tuning the res to 8k I guess is not fixing the issue of OCP charles?
03-16-2010, 12:08 PM
Charles Wirth

It was fixed to 8k, we did not djust it once the socket was built up.
03-16-2010, 12:39 PM
Linuxfan

Strong CPU at -90! :up:
What did it do without HT?
03-16-2010, 12:47 PM
chew*

Quote:

Originally Posted by FUGGER

It was fixed to 8k, we did not djust it once the socket was built up.

My biggest question is is the res changing once cold can't imagine it would be to much but maybe enough?
03-16-2010, 01:49 PM
zalbard

Very nice results! Are you going to try a different board soon, then? :) Or what's the plan?
03-16-2010, 02:54 PM
miahallen

I did the mod last night, and the VR is extended for easy adjustment during testing ;)

But so far, I've only got a quad to play with. With the quad I had no OCP trouble running at 1.88V and 5.6GHz+ (wPrime).
http://hwbot.org/community/submissio...2min_15sec_0ms
(2nd best global quad score, but currently ranked with the hex cores, trying to get MM or RB to fix now)

I'm planning to pick up a 980x in Tokyo this weekend. I'll test next week under LN2 and let you know how it goes.
03-16-2010, 05:26 PM
vodka3dg

stunning results :up:
03-16-2010, 05:43 PM
Stelios

Nice one Charles !!! :up:

I hope to play with one of these soon .
03-16-2010, 11:22 PM
Harshal

Looking good boss!
:up:
03-16-2010, 11:49 PM
WaterFlex

The chip is my loveliest thing there:) Great work man :up:
03-17-2010, 01:18 AM
andrevieira

Very very nice bench, congrtz :up:
03-17-2010, 04:23 PM
Sinsear

This is getting difficult to resist the urge to buy this chip...
03-17-2010, 06:29 PM
miahallen

Charles, have you done any high bclock testing with this board? I've got a hard wall on mine at 227MHz with X5667 (32nm quad, 105MHz PCIe, 1.45-1.55V VTT) :confused:
03-17-2010, 08:37 PM
HDCHOPPER

More more FUGGER!
I seen em on the egg today but you gotta get a combo deal or no gulfy :(
03-18-2010, 12:28 AM
massman

Quote:

Originally Posted by miahallen

Charles, have you done any high bclock testing with this board? I've got a hard wall on mine at 227MHz with X5667 (32nm quad, 105MHz PCIe, 1.45-1.55V VTT) :confused:

I had 240MHz BCLK with 960 and 980X, 235 with 950.

It's a CPU problem.
03-18-2010, 05:14 AM
Sam_oslo

Quote:

Originally Posted by miahallen

Charles, have you done any high bclock testing with this board? I've got a hard wall on mine at 227MHz with X5667 (32nm quad, 105MHz PCIe, 1.45-1.55V VTT) :confused:

These Xeon DPs seams to have a locked Uncore on x20 and some other "strange" stuff related to it. At least storage for me who have been dealing with SPs until now.
I haven't test the Highest BCLK yet, but my X5650 loves VVT, and the BCLK has been scaling propositionally with VVT so far. Intel is recommending max VVT=1.4v but I'm taking it easy until I find more about how much they can handle for 24/7. I'm using UD7 too.

I don't know much about your setup, but I guess your Uncore is limiting your BCLK wall. We have to learn some new tricks to control the "strange" Uncore on these DPs, for sure.
03-21-2010, 12:41 PM
hicookie

3dmark 06 not hit the OCP easily, if your chip get silmar as vantage clock
then possible its cpu limited. im pretty sure 8K is quiet enough
for 300a of CPU vcore. if your chip hit 300a, first your power supply will
shutdown due to psu ocp at 50a 12v.(most of PSU design OCP 12v at 48a - 50a) besides silverstone doesn't set ocp
til 110a, PSU still work.

hicookie.
03-21-2010, 01:31 PM
Hondacity

300Amps? x 1.8v = 540w :d HOLY COW!!! is that normal for 1.8v GT?
03-21-2010, 02:40 PM
hicookie

Quote:

Originally Posted by Hondacity

300Amps? x 1.8v = 540w :d HOLY COW!!! is that normal for 1.8v GT?

if you hit ocp at 300a, that will like a gulftown chip do 7.5GHz vantage
here is my early test with gulftown@5.97GHz vidoe clip, in the video
you can see 12v is hit 34.3a, input power is around 411.6watts
if you calculate vcore, 411.6*97% / 1.84v , vcore current 217a.
http://www.youtube.com/watch?v=OBQeFNjE42E

if your chip quiet strong, like 1.6v do 6GHz vantage, lets pretend power need
is 380watts. then cpu current around 237a, still far to 300a.

hicookie
03-21-2010, 04:17 PM
rbuass

Can you share the OCP mod?
Here the same
;)

Thanks Fugger
03-21-2010, 05:32 PM
Hondacity

Quote:

Originally Posted by hicookie

if you hit ocp at 300a, that will like a gulftown chip do 7.5GHz vantage
here is my early test with gulftown@5.97GHz vidoe clip, in the video
you can see 12v is hit 34.3a, input power is around 411.6watts
if you calculate vcore, 411.6*97% / 1.84v , vcore current 217a.
http://www.youtube.com/watch?v=OBQeFNjE42E

if your chip quiet strong, like 1.6v do 6GHz vantage, lets pretend power need
is 380watts. then cpu current around 237a, still far to 300a.

hicookie

thats what i was thinking...no ocp trip on psu.(no ocp if single rail)....
woot 217amps on the vregs thanks for info:up:

what is max current rating on the ud7 vregs?

**evga classy can do 600w if 1.9v.....thats more or less 300Amps....
03-21-2010, 05:52 PM
Solid Snake

WOW, Very nice bench :shocked:
03-21-2010, 06:18 PM
Charles Wirth

Not sure who has any of these
http://valid.canardpc.com/show_oc.php?id=1091831

I used a different board and reached 150Mhz higher all cores without any shutdown going on.

Hicookie, show us a B1 past 5.4Ghz all core vantage please.
03-21-2010, 06:49 PM
Movieman

Quote:

Originally Posted by FUGGER

Not sure who has any of these
http://valid.canardpc.com/show_oc.php?id=1091831

I used a different board and reached 150Mhz higher all cores without any shutdown going on.

Hicookie, show us a B1 past 5.4Ghz all core vantage please.

What sspec on the xeons?:D

Wait! Something is wrong!
Me posting in this section?
Is there a shift in the universal plane?
Is the sky falling?:rofl:
03-21-2010, 07:16 PM
hicookie

Quote:

Originally Posted by FUGGER

Not sure who has any of these
http://valid.canardpc.com/show_oc.php?id=1091831

I used a different board and reached 150Mhz higher all cores without any shutdown going on.

Hicookie, show us a B1 past 5.4Ghz all core vantage please.

i didn't record a video of vantage, but i do have a clip with 3dmark 2006
B1 1.719v 5.66GHz 3Dmark 2006

http://www.youtube.com/watch?v=sJjLRTJaRKU
03-21-2010, 11:34 PM
Raja@ASUS

Quote:

Originally Posted by hicookie

if you hit ocp at 300a, that will like a gulftown chip do 7.5GHz vantage
here is my early test with gulftown@5.97GHz vidoe clip, in the video
you can see 12v is hit 34.3a, input power is around 411.6watts
if you calculate vcore, 411.6*97% / 1.84v , vcore current 217a.
http://www.youtube.com/watch?v=OBQeFNjE42E

if your chip quiet strong, like 1.6v do 6GHz vantage, lets pretend power need
is 380watts. then cpu current around 237a, still far to 300a.

hicookie

If you are taking VTT from EPS 12V as well (which I think you are), VCC draw will be less than 217 amps at this load.
03-21-2010, 11:46 PM
massman

06 is easier than Vantage in my experience. I can get through 06 at 5.6G 12t, but need to go down to 5.45G for Vantage.
03-21-2010, 11:50 PM
Raja@ASUS

VTTA/D combined at stock is spec'd at around 28 amps on the 4c/8T chips, so VCC at 5.6GHz looks to be around 180~190 amps; nowhere near enough VCC OCP - could be one of the other rails (or an MSR related problem in BIOS).
03-22-2010, 12:51 AM
Sam_oslo

Quote:

Originally Posted by FUGGER

Not sure who has any of these
http://valid.canardpc.com/show_oc.php?id=1091831

I used a different board and reached 150Mhz higher all cores without any shutdown going on.

Hicookie, show us a B1 past 5.4Ghz all core vantage please.

I have one of these, and this is a really Interesting findings.
What other MB could give 150MHz more?
These new beasts have a nasty locked Uncore that I need to find a way around it. If you could give some details, it will be useful for adjusting MB-related settings for Xeon DPs.
03-22-2010, 01:11 AM
Charles Wirth

I can also run vantage at a much higher speed due to lack of all cores being used. iirc it uses 6 cores.
03-22-2010, 02:10 AM
Movieman

Quote:

Originally Posted by Sam_oslo

I have one of these, and this is a really Interesting findings.
What other MB could give 150MHz more?
These new beasts have a nasty locked Uncore that I need to find a way around it. If you could give some details, it will be useful for adjusting MB-related settings for Xeon DPs.

You don't have one of those.
That's a X5680 B1:D
03-22-2010, 02:46 AM
Sam_oslo

Quote:

Originally Posted by Movieman

You don't have one of those.
That's a X5680 B1:D

Yeah I know, but I was hoping to see what could make it run 150MHz faster on a new MB.
My X5650 B1 should act pretty much the same, hopefully. But he is not telling what other MB, or what MB-settings would make the differences. ;)
03-22-2010, 04:41 AM
zalbard

FUGGER, what batch is this? :)

All times are GMT -8. The time now is 08:06 AM.

XtremeSystems