Quicktest Retailstepping Gulftown under dice

[@rne]

Hello Folks,
just tested another little Gulftown under dry ice.

System:
ES i7 980X B1 Step 2
GA-EX58-UD4P
Mushkin Redline 6GB old Hyper
ToPower 1100W

I benched ~30 Minutes, reached 5,2GHz 3DM06 with all cores at 1,6vCore under load.



Northbridgefrequency was very relaxed. Maximum NB Freq with 1,45vTT/QPI was about 4,3GHz. Scales further, but i was more focused on a stable corefrequency in order to push some old gfx.

Aquamark with HT enabled (just a little test with setfsb ) run fine at 5,3Ghz. Validation about 5,4GHz.

http://valid.canardpc.com/show_oc.php?id=1101576

After 30 minutes, giving 06 a full first try with all tests enabled, this happened in Canyon Flight:



Seems like my sample wasn't even able to handle a gulftown with only 1,6v under fully load. Embarrassing!

Conclusion: Retailstepping of Gulftown is good under dry ice, with more time and a working board, the results can be improved significantly. The difference between Q3FE and Q3QP becomes larger with lower Temps and higher vCore. If you need something for dry ice, go for a retail one

[thijs]

wow, really looks like this gulftown is pushing them old mobo's to the max.
time for a new generation motherboards to take over for the benching it seems.

[SoF]

Gigaburn

[FatAlbert]

@rne, so in you opinion Gigabyte EX58-UD4P is motheboard for extreme Gulftown overclocking ? In my not and I'm not surprised that mb died with GT.

anyway, waiting for scores with this cpu and ln2.

[@rne]

From where do you pull this assumption? Why are you not surprised? Which technical facts leads you to this conclusion?

Only because one mobo died, this means nothing. Could be a bad one though. Anyway, i wouldn't bench this model again, because of some bios and mem oc issues

[cdolphin]

RMA ?
Nice clock @ voltage.

[FatAlbert]

I just asked and I wanted clear answer. In my opinion motherboard for gulftown should has rather strong PWM as we can see from the previous benches with GT. I'm not surprised because Gigabyte has better motherboards for extreme overclocking, like X58-UD5P or X58A-UD7. It's just that I never believed that the board which aren't the highest model can go high, with some exceptions of course and as you wrote this could just bad sample and this board has weaker bios support (also as you wrote about some issues) and that's why I never believed in lower models

[SoF]

back in the days I used DS3P and DS3R (so low-end models) to reach some insane E6850 scores with nearly 2V and the boards did very well. I think that board should have shut down before burning at least if it is not designed to take relatively easy voltage on GT...additionally it is the 2nd time I see a Gigaburn...same part of another (different model) burned under stock voltage & clocks while one of our crew-members was watching TV. So I somehow get the feeling there is some kinda issue on the lower-end gigaburn models with the pwms...
and everybody knows I used to be quite giga-pro but some things got me thinking.

[Hondacity]

no heatsink on the vregs... lol.

[FatAlbert]

you're right SoF it should shut down before but it didn't for the bad luck of @rne.

[massman]

Gigaburn

GigaBANG!

[miahallen]

I just asked and I wanted clear answer. In my opinion motherboard for gulftown should has rather strong PWM as we can see from the previous benches with GT. I'm not surprised because Gigabyte has better motherboards for extreme overclocking, like X58-UD5P or X58A-UD7. It's just that I never believed that the board which aren't the highest model can go high, with some exceptions of course and as you wrote this could just bad sample and this board has weaker bios support (also as you wrote about some issues) and that's why I never believed in lower models

This was the same model which we used at the GOOC North American qualifier last April...while we did have a couple mobo issues at the event, I don't recall a single issue caused by PWM failure, they all seemed to be BIOS issues. We were all pushing i7 975s to 5.2GHz+ under LN2. PWM on this board should be fine for GT IMHO, especially at these low voltages

Like the OP said, probably a bad board...or maybe some condensation melting durning the heavy load?

[Brian y.]

Nice benching Arne, too bad about the board. It just happens sometimes...

no heatsink on the vregs... lol.

Have not had any on my ASUS board from day one and it works fine with a small fan

[FatAlbert]

This was the same model which we used at the GOOC North American qualifier last April...while we did have a couple mobo issues at the event, I don't recall a single issue caused by PWM failure, they all seemed to be BIOS issues. We were all pushing i7 975s to 5.2GHz+ under LN2. PWM on this board should be fine for GT IMHO, especially at these low voltages

Like the OP said, probably a bad board...or maybe some condensation melting durning the heavy load?

Didn't know that, anyway as I wrote, I don't believe in not the highest models (with some exceptions). Maybe for Bloomfield it's good board, will see how older boards goes with GT when more overclockers get it and test it subzero

[Hondacity]

Have not had any on my ASUS board from day one and it works fine with a small fan

try it on gt cpu 12threads...1.7v ...cpu vantage damage

[Brian y.]

try it on gt cpu 12threads...1.7v ...cpu vantage damage

hahaha......1.95v GT vantage run good enough?

remember now I don't play on air

[Hondacity]

asus doing good on pwm ...thats good

[@rne]

This was the same model which we used at the GOOC North American qualifier last April...while we did have a couple mobo issues at the event, I don't recall a single issue caused by PWM failure, they all seemed to be BIOS issues. We were all pushing i7 975s to 5.2GHz+ under LN2. PWM on this board should be fine for GT IMHO, especially at these low voltages

Yep. I used this model also at GOOC in Eindhoven, Netherlands as far as i remember. The PCIe Mod wasn't public this time, so we were limited to bclk.
Other than that, the board worked very well. I also think the pwm design is powerful enough for GT, maybe there are some issues with the quality of components on some boards...

Nice benching Arne, too bad about the board. It just happens sometimes...

THX!

[SoF]

GigaBANG!

And if we put the dead boards together we have a GigaGANGBANG

[@rne]

Drunk again ? j/k

[SoF]

Drunk again ? j/k

Just woke up, not yet

Man I hoped to see some more tests but so far specially the 06 CPU test is /was really close to what my Q3QP does under DICE. Just that you stuck at 5.4 must be something else, I can't imagine 5.2 CPU Test and 5.4 to be upper limit, maybe nb-freq at that point? but it was too short to judge about the upper limit, but I am pretty sure you would have squeezed out some more without ht and maybe only 2 cores (helps me quite a bit under DICE).

[@rne]

Yeah prolly it was the NB Freq / to less vTT

But even after this very short test it becomes clear that retail and q3qp under dice are more equal than under ln2.

[zalbard]

I'm sorry, where is the promised retail stepping from the topic title?

In any case, doesn't look like a bad chip, the voltage requirement for this clock is far from being extreme.

[@rne]

Because Retail has the same stepping.
Spec Code SLBUZ, have a look in here:
http://ark.intel.com/Product.aspx?id...ec-codes=SLBUZ

[Raja@ASUS]

Yep. I used this model also at GOOC in Eindhoven, Netherlands as far as i remember. The PCIe Mod wasn't public this time, so we were limited to bclk.
Other than that, the board worked very well. I also think the pwm design is powerful enough for GT, maybe there are some issues with the quality of components on some boards...



THX!

Looks like a 4~6 phase PWM (2 inductors per phase) with each phase looking to be 30 amps - do the math for the approx max current draw (depending upon how OCP is implemented).

GT overclocking really needs PWM circuits with 200 amp+ handling (+ a good 40~50 amp overhead if you want the board to last).

later
Raja