A start..2600K and Intel Burrage board

[retro77]

Maybe Intel is gonna release a patch next week that will really open these up. Watch its only using 10% of the actual power


That would be awesome though!

[Movieman]

You all will excuse me if I'm a bit slow BUT I'm thinking that if Aerou's 2600 NON K will pull 35,000PPD ( and it will) and do that at under 100w draw and comparing that to the 41,500PPD I'm seeing at 4634MHz with 212w draw then maybe the smart thing isn't to overclock these things but to run them at stock..
I know, sounds sacreligeous but look at the elec use vs output.
35,000PPD using 100w
41,500PPD using 212w
Now I need to do some testing and see what I draw at defaults.

[retro77]

See MM!! You don't always have to overclock everything!

[Aerou]

It's certainly far more efficient at stock. Heck, raising the Vcore from stock is always hurting your power-bill hard, no matter what CPU. Keep in mind that I have it setup close to optimal - integrated GPU, efficient PSU, well cooled and efficient circuitry on the motherboard. No addon card. I might save like 2-3W using an SSD perhaps, but the OCZ Throttles I have been using no longer store data well Also, I could lower the stock Vcore by some nudges, but now, I want a decent PPD statistics data before I go into tweaking. Might save up to 5W. Short story though - I am hooked on Sandy Bridge. You can't imagine me dreaming of a sub 100W quad, then being worried it might not make it, when you told me your kill-a-watt was showing Vac instead of W, but here I am looking at a silent and fairly cool cruncher with the muscle for more than 30K PPD and that under 100 VA.

On the other hand, I don't see that machine getting 35K PPD, at least not on Linux. But I won't ever switch to Windows for that. Besides, Linux gets more work done, and that's what matters.

[sierra_bound]

I don't think 35K is possible at stock speed. 30K-32K is more likely. Still very good for stock voltage though.

[Otis11]

Mine at stock is averaging 33,676... But it also runs 2 GPUGrid WUs at a time, so Idk if MM is really that far off with his 35k.

(Don't worry, only at stock till I can get a new HS or backplate - stock one sucks)

[sierra_bound]

Average over how many days? You really need to run the chip for a week or two before you get a realistic average.

[retro77]

Mine has steadily declined on the production since install...


edit:
Image from my 2600K's Task Manager:


Normal or should I reload the OS to see if I get better results?

[slaveondope]

Mine has steadily declined on the production since install...

What clock speed does it run at stock? I mean does turbo run on all cores or is it just 3.4ghz?

[sierra_bound]

There are a lot of variables to daily point production, especially if you are running projects with a quorum system. Right now, I have a lot of Sandy Bridge work units waiting to be validated.

Generally you will get a good daily point total on the first day or two, then the point total drops off a little, then picks up again.

[retro77]

It seems just 3.4:

[sierra_bound]

Turbo boost mode on the 2600K would be 3.8GHz.

[slaveondope]

It seems just 3.4:

Looks like 3.5ghz to me

How low can these be undervolted? .9v maybe

[retro77]

I went into the BIOS and changed the Intel Turbo Boost mode from Auto to Enabled. It looks like it shares the boost mode across all the cores. It had %'s next to each core. Hence why I'm almost hitting 3.5GHz...I guess.

I was going to disable the CPU thermal monitor as it said it would reduce core clock ratio and voltage. But I changed my mind. Seemed too risky. So-Cal has been seeing some hot days this January.


edit: We should start a thread called something like: "The Official Sandy Monitoring Thread" and sticky it! That way we can all come in and post some results and compare. MM must be still testing power consumption...

[Petra]

So-Cal has been seeing some hot days this January.

Really? I haven't noticed... says the guy living at 3k feet, in the high desert to your East.

Which board are you using, again?

[Otis11]

Average over how many days? You really need to run the chip for a week or two before you get a realistic average.

Only did the last 3 days, but I took the points produced divided by the minutes of run time x60 minutes to get hours x 24 hours to get days x 8 threads to get that number so I don't think that should matter... I'll report more numbers back in a few days, then OC it a bit this weekend.

Any idea of OC on stock HS?

[Jaco]

If you want the best efficiency , find the highest overclock with stock voltage.

[sierra_bound]

edit: We should start a thread called something like: "The Official Sandy Monitoring Thread" and sticky it! That way we can all come in and post some results and compare. MM must be still testing power consumption...

Good idea. I nominate Movieman to be in charge of that thread.

30,000+ points at stock voltage is still very good. I don't mind running overclocked at 200W because that's still a lot better than any of my other crunchers. Plus those 5-6K extra daily points add up over time.

[[XC] Hicks121]

Well all I can say is that I am getting hosed in the quorums!

[shoota]

If you want the best efficiency , find the highest overclock with stock voltage.

But how would the best OC on the lowest undervolt fare?

[Aerou]

But how would the best OC on the lowest undervolt fare?

Depends.
IMO stock clocks on anything that is over 3GHz stock and lowest volts is the way to go

But highest OC on stock volts is the most efficient overclock, and is the better alternative if you have limited space for machines or any other limitations.

[WFO]

Holy S guys!!! I know the servers were down for 2-3 hours yesterday but it shouldn't have skewed results this much.

SB2600K-1 @4.609



Uploaded with ImageShack.us

SB2600K-2 @ 4.532



Uploaded with ImageShack.us

Even my 980x had a huge day!



Uploaded with ImageShack.us

[nanoprobe]

Is there a tool that accurately reports the vcore? CPU-Z says one thing, hardware monitor another and the Turbo V tool from Asus another. Thanks

[Bail_w]

Is there a tool that accurately reports the vcore? CPU-Z says one thing, hardware monitor another and the Turbo V tool from Asus another. Thanks

Measure it with a DMM is the only and most accurate way.

[nanoprobe]

If CPUZ is accurate I'm currently at 4.533 Ghz @ 1.288 vcore pulling 188 watts. The temps show a 10c spread. 51c-61c. Hoping the temps come closer together after burn in. Should be interesting to see the PPD in a week or so.