Sandy Bridge vs (your older rig) Kill-A-Watt readings

Here's some preliminary numbers comparing an i7 920 with a 2600K:


Unfortunately, the WUs are different, not sure how much that would translate into watts usage...

If you have a Kill-A-Watt meter please post some comparisons between your Sandy Bridge, and a typical folding rig that you have.

My place holder... I'll have to get the readings from the 930 rig and the idle number for the SB. Just started on a bigadv unit... don't want to stop it.

I dressed up the format a little bit Tim.

Try P95 on both. Not SMP_FAH but at least you would be comparing the same fruit and give us a good idea of the difference. Looks like the power savings will pay for these layers before their lifetime is over.

OTOH you can just duplicate your WUs and run several frames on each system of each test project.

harlam357 said:

My place holder... I'll have to get the readings from the 930 rig and the idle number for the SB. Just started on a bigadv unit... don't want to stop it.

I dressed up the format a little bit Tim.

Code:

5 Series System
---------------
Mobo:   Gigabyte X58A-UD3R
CPU:    i7 930 @ 200x19 - 3.8GHz 1.23v
Memory: G.Skill - 3x2GB @ 1600MHz 1.5v
GPU:    Gigabyte GTX460 768MB @ 850MHz 0.987v
WU:     p6900
Idle:   ???w (APC UPS)
Load:   ???w (APC UPS)

6 Series System
---------------
Mobo:   Gigabyte P67A-UD4
CPU:    i7 2600k @ 100x45 - 4.5GHz 1.33v
Memory: G.Skill - 4x2GB @ 1600MHz 1.5v
GPU:    Sapphire ATI 2400XT stock
WU:     p6900
Idle:   ???w (Kill-A-Watt)
Load:   170w (Kill-A-Watt)

Click to expand...

Yes, looks much better Ryan! I took the liberty of editing my post, using your much cleaner looking template.

Audioaficionado said:

Try P95 on both. Not SMP_FAH but at least you would be comparing the same fruit and give us a good idea of the difference. Looks like the power savings will pay for these layers before their lifetime is over.

OTOH you can just duplicate your WUs and run several frames on each system of each test project.

Click to expand...

I'll do that after the current P2686 completes, don't want to miss that preferred deadline.

I don't have per WU wattage numbers for my X58 system, but the numbers were always floating around the same area anyway. I also don't have an idle wattage, but I'm pretty sure it was around 200w.

This is a pretty good comparison, I think, because the only thing changed was the mobo and CPU. I didn't even reinstall the OS.

Bump... I need to fix my readings but I'm still having Sandy Bridge stability issues. Thus, Tim is gaining on me... FAST!

I'll clean up this thread if we can get some more numbers going.

Subscribed. Very interested in this data.

harlam357 said:

Bump... I need to fix my readings but I'm still having Sandy Bridge stability issues.

Click to expand...

Ryan,

Trying to keep CPU volts down? 1.35v +(offset= -0.005) is working for my Asrock P67 @4.5ghz. Are you 'recycling' some DIMMs from a X58 rig? I've read that these P67 boards don't like DIMMs volted over 1.5v.

harlam357 said:

Thus, Tim is gaining on me... FAST!

Click to expand...

I thought I had 4.5GHz stable. Did a 12hr run of P95 Small FFTs on it a few days ago. Now that OC doesn't want to seem to hold. I've tried and tried and tried different settings, voltage, bios, etc. but the 4.5GHz OC just doesn't want to hold no matter what I do. I dropped it down to 44x multi and it's been running strong for over 6hr (P95 Blend).

Yes, I'm trying to keep it under 1.35v under load. With LLC that voltage equates to ~1.33v under load and that's where I'd like to keep it. Temps on the hottest core just reach into the 70C range.

Using the same kit you are I believe. Don't quote me on exact version, but it's the G.Skill C9 1600 (red heat spreader)... rated at 1.5v and I'm running it at the default voltage the board is applying; which is actually around 1.52~3 volt.

EDIT: should also note that I'm now working with 4x2GB dimms. When I tested the 4.5GHz OC it was with 2x2GB... although I wasn't able to replicate that run again even when removing two of the dimms. This board and cpu was my planned "media server" replacement with RAID array (no GPU Folding) as that is my last box in the house running a C2Q. Not sure if the platform is ready for that level of stability. Can't have my movie box going down... I'll be seriously PO'd if I can't watch Stargate after work.

Ryan, where's your PLL voltage set? PLL voltage of 1.7 to 1.75v is reportedly to be favorable for a stable OC. For now, I'm leaving the rest of the voltages @stock, except for the memory and CPU.

Looks like 4.5ghz is going to be it, for this Asrock p67pro.

ihrsetrdr said:

If you have a Kill-A-Watt meter please post some comparisons between your Sandy Bridge, and a typical folding rig that you have.

Click to expand...

As you wish...............

Energy meter: BluePlanet Model: EM100
Numbers are based on cpu speed of 4.2Ghz
To load systems, OCCT was used. Test Mode: Med Data Set
These numbers are from bare bones folding layers consisting of board, cpu, memory, video card, hard drive and power supply only. When testing the same hard drive, power supply and video card was used on each system.
Also be aware no power saving options were enabled.

X58 Series System
-----------------

Mobo: Biostar T-Power X58
CPU: i7 920 21X200 4.2Ghz
Volts: Under load as reported by Cpu-Z 1.56.3 = 1.36
Memory: OCZ - 3x2GB @ 1600MHz
GPU: ATI 4670
Idle: 150w
Load: 325w


P67 Series System
-----------------
Mobo: Biostar TP67XE
CPU: i7 2600k 100x42 4.2Ghz
Volts: Under load as reported by Cpu-Z 1.56.3 = 1.26
Memory: Corsair - 2x2GB @ 1866MHz
GPU: ATI 4670
Idle: 75w
Load: 165W




Worst case P67 Series System
----------------------------
Mobo: Biostar TP67XE
CPU: i7 2600k 100x48 4.8Ghz
Volts: Under load as reported by Cpu-Z 1.56.3 = 1.45
Memory: Corsair - 2x2GB @ 1866MHz
GPU: ATI 4670
Idle: ----
Load: 225W

Click to expand...

Updated my 5 series numbers... finally!

Sub'd, would like to see how this measures up.

cuda340 said:

ihrsetrdr said:

If you have a Kill-A-Watt meter please post some comparisons between your Sandy Bridge, and a typical folding rig that you have.

Click to expand...

As you wish...............

Energy meter: BluePlanet Model: EM100
Numbers are based on cpu speed of 4.2Ghz
To load systems, OCCT was used. Test Mode: Med Data Set
These numbers are from bare bones folding layers consisting of board, cpu, memory, video card, hard drive and power supply only. When testing the same hard drive, power supply and video card was used on each system.
Also be aware no power saving options were enabled.

X58 Series System
-----------------

Mobo: Biostar T-Power X58
CPU: i7 920 21X200 4.2Ghz
Volts: Under load as reported by Cpu-Z 1.56.3 = 1.36
Memory: OCZ - 3x2GB @ 1600MHz
GPU: ATI 4670
Idle: 150w
Load: 325w


P67 Series System
-----------------
Mobo: Biostar TP67XE
CPU: i7 2600k 100x42 4.2Ghz
Volts: Under load as reported by Cpu-Z 1.56.3 = 1.26
Memory: Corsair - 2x2GB @ 1866MHz
GPU: ATI 4670
Idle: 75w
Load: 165W




Worst case P67 Series System
----------------------------
Mobo: Biostar TP67XE
CPU: i7 2600k 100x48 4.8Ghz
Volts: Under load as reported by Cpu-Z 1.56.3 = 1.45
Memory: Corsair - 2x2GB @ 1866MHz
GPU: ATI 4670
Idle: ----
Load: 225W

Click to expand...

Click to expand...

I like the fact that you are testing with the same video card in the setups. But instead of testing with F@H, load it up with P95/OCCT/IBT for the cpu and Furmark on the gpu and look at power draw with each. That would definitely be a "worst case" load on the systems.

Good thread guys.

To load systems, OCCT was used. Test Mode: Med Data Set

Click to expand...

I did use OCCT Mudd

Med Data Set test mode was used because it loads system close to what folding does....IMO

That video card was used because it doesn't draw much juice. Wanted the numbers to be based on the platform.

Hey, good job then, cuda!

I wonder how it compares to something more directly comparable, like an i7 875k system? Something based on the 1156 platform.

muddocktor said:

Something based on the 1156 platform.

Click to expand...

I would gladly look into that for you, but i skipped the 1156 platform altogether.