[TPU] TIM is Behind Ivy Bridge Temperatures After All - Page 21

Honestly the whole thing is a mess. It's really a crap shoot. Unless you get one of the "golden" chips that use much lower voltages then clock for clock you're power consumption and temps are going to be higher than SB. Now the last link I posted DID say that a 4.8 IB OC is equivalent to a 5.0 SB OC so that's something to consider at least. But really this whole release seems like one big fubar to me. IF AMD's claims of 24% power reduction are true for Piledriver (due to using resonant clock meshing) then the competition is REALLY going to heat up this summer . Either way for all of us that means price wars I think which are always good for us.

Just wrong. I've had four average/weak IB's; one 3770K and three 3750K's, and paired with the Z77 board all have drawn less power than my three SB's did, even at highest clock (~4500MHz).

Besides, almost all the IB's being shown here on OCN are running at much lower voltages. Go read other threads.

I'd love to see the proof on that one because most of the review sites that list power consumption dispute your claim.

http://www.legitreviews.com/article/1924/9/

http://lenzfire.com/2011/07/ivy-bridge-benchmarks-vs-sandy-bridge-first-performance-comparison/3/

Even here with this one (which I would assume is one of the few chips actually operating closer to targeted lower voltages) it's nowhere near "40 watts" less.

http://www.techspot.com/review/523-ivy-bridge-intel-core-i7-3770k/page10.html And that's IB at 3.5 vs SB at 3.4. so at most a 20 watt difference or so.
Edited by Bubba Hotepp - 5/17/12 at 1:31pm

Ivy Bridge Benchmarks Vs Sandy Bridge – First Performance Comparison
posted July 13, 2011 by Arun / No Comments

I stopped reading there...

Seriously, go enjoy Bubba's Bulldozer, it's clear Ivy Bridge is not for you.

Our buddies at Anandtech show that Ivy Bridge DOES have lower power consumption than Sandy.





HardwareCanucks also agrees.



Let's see if anyone else has lower Power Consumption...



I'm surprised, shocked actually about how much you care about Power Consumption, look at where Bulldozer sits on these lists, and you just happen to own one. I'm baffled.

And some people claim there's no difference over Sandy? Sure it's not DRASTIC, but considering Intel lacks competition on the CPU front, they focused their efforts on the GPU and brought a drastic 50% improvement. That means nothing to us enthusiasts, but that's what happens when competition lacks, this was afterall a TICK+ on a brand new process not a TOCK on a mature process.


Edited by 2010rig - 5/17/12 at 3:14pm

Also keep in mind that Ivy Bridge is faster than Sandy Bridge. Comparing their power consumption figures at the same clock speed isn't valid because one is performing more work than the other. You need to compare them at the same performance if you are comparing power consumption to make it valid. So with IB at 3.5ghz and SB at 3.4ghz the IB is already a good amount faster possibly 8-10% WHILE using 20 W less. Other links I provided showed a 50W difference so that is even better. And when you are considering that the increase in power consumption is CPU only that is a big deal because that could mean the CPU is putting out 20-50W more heat due to inefficiencies of SB compared to IB.

Yeah, don't have those SB's anymore but I really don't think you are ready to be proved wrong anyway.

So my lone test was under Prime95 26.6 (5 mins 128K FFT), IGP video, using a Kill-A-Watt EZ and notepad. The max 24/7 clocks I had for the 2500K: 5000MHz 1.36v: 211W at the wall. The highest stable speed at any voltage for the young 3570K: 4600MHz 1.28v: 174W. (It will be running at 4500 now.)

That doesn't even take into account the fact that my 2500K was a badass and this 3570K is more of a lameass. Most people with average Sandy's @ 1.38v+ will see even more improvement with an average Ivy.


Here are the numbers for the Ivy so far (I'm not done yet). Anyone with a 2500K can compare:
IB POWER TESTS (Click to show)
Edited by samwiches - 5/17/12 at 2:28pm

I make no claim that there isn't a difference over SB. Here's the thing though, for every review that posts a lower power consumption there are two that post higher power consumption. And unless the "buddies" at Anandtech post the V used it doesn't necessarily disagree with the other reviews. Look at the IB overclocking club. There are a few users posting very nice numbers (1.09V at 4.5GHz) which falls in line with the claims intel made last year about IB's capabilities and specs. But I'm telling you that it is IMPOSSIBLE for an IB chip clocked at 1.325V to have lower power consumption than a SB clocked at 4.5GHz with 1.4V. The increased current required by the tri-gate design make that theoretically and physically impossible. Oh and .075 to .1 is NOT considered "much lower voltages". It's not even enough to compensate for increased current.

I already posted a review link where they state that an IB at 4.8 is comparable to an SB at 5.0. It will REALLY be a no brainer of IB over SB when the fab problems are solved and the majority (instead of the 10% or so now) of IB CPU'S are operating at the voltage, power consumption, and temperature ranges that they should be at. Then not only will it run cooler and faster clock for clock but it will definately consume less power. That will be a triple win for IB.

edit - you can't point to a few chips within the small minority of "golden" chips and say that is how IB operates as a whole since 90% of IB buyers will not receive a chip that operates as claimed.
Edited by Bubba Hotepp - 5/17/12 at 2:09pm

That's just it, no one using "golden" Ivy's as the basic for comparison. Myself, I've had nothing but lame Ivy's. The important comparison is between the average SB's and the average IB's at equivelent performance levels:

2500K @ 4900MHz w/ 1.38v 200+ system watts, ~75C temps
3570K @ 4600MHz w/ 1.28v 174 system watts, ~85C temps

200, 174 system watts?? At idle maybe. There is NO way your systems are operating at those power consumption levels under load. I'll believe that when you know what freezes over. On top of that I would like you to post evidence that you're systems are drawing that power at those voltages. And even worse let's take a look at what people are posting for V's on the 3570K

Sk1nNy 4499.8mhz 1.320v 14hrs 82-86-87-82 AIR - Thermalright Venomous X 3570k
Majinwar 4501.4mhz 1.212v 13hrs 68-77-73-74 AIR - Hyper 212 Evo 3570k
punceh 4800.6mhz 1.232v 13hrs 72-74-76-70 AIR - Megahalem 3570k
CircuitFreak 4701.3mhz 1.356v 14hrs 80-83-80-76 WATER - EK Supreme HF 3570k
speedy2721 4800.7mhz 1.384v 13hrs 72-80-76-78 AIR - Silver Arrow 3570k
leoxtxt 4301.2mhz 1.212v 12hrs 74-73-70-71 WATER - Antec 920 3570k
Atlas101 4400.6mhz 1.280v 13hrs 86-89-86-84 AIR - Thermaltake Frio 3570k
HardwareDecoder 4400.3mhz 1.200v 17hrs 70-76-75-72 AIR - 212 Evo 3570k
IronWill1991 4500.5mhz 1.320v 12hrs 84-90-87-84 AIR - CoolerMaster V6GT 3570k
PMantis24 4501.1mhz 1.092v 20hrs 62-70-68-64 AIR - CoolerMaster TPC-812 3570k

10 posted results. Out of the 4 at or near 4.5 only 2 are operating below 1.3V and only 1 "golden" chip is operating at below 1.1V which is where, by design, most should be operating at. Out of the whole list there are only 2 that are at the target or near it (the last at 1.09 and the 3rd 1.232 @ 4.8).

Now let's look at SB


Munaim1 5100.0mhz 1.472v 12hrs 61-69-70-67 WATER - EK Supreme HF 2500k
Phaseshift 5000.0mhz 1.432v 12hrs 73-79-79-75 ?????? 2500k
cinnamonbits 4499.9mhz 1.288v 12hrs 63-72-70-69 AIR - Thermaltake Frio 2500k
cory1234 4601.5mhz 1.368v 12hrs 65-73-70-72 AIR - Ultra 120 Extreme 2500k
Lost-boi 4601.5mhz 1.392v 17hrs 58-61-61-58 WATER - EK Supreme HF 2500k
Cilraaz 4600.7mhz 1.328v 12hrs 64-70-68-69 AIR - Xigmatek Dark Knight 2500k
ehpexs 4414.5mhz 1.216v 12hrs 56-59-62-57 AIR - Corsair A50 2500k
BANWELL 4600.5mhz 1.296v 12hrs 54-56-59-57 AIR - Noctua NH-D12 2500k
Kepi 4500.3mhz 1.320v 12hrs 54-61-60-57 WATER - EK Supreme HF 2500k
Atraps003 4513.2mhz 1.404v 24hrs 67-73-73-73 AIR - Zalman CNPS9900A 2500k
phazer11 4800.5mhz 1.440v 12hrs 72-76-80-77 WATER - Corsair H60 2500k
scorpion667 5000.3mhz 1.440v 12hrs 67-75-77-70 AIR - Noctua NH-D14 2500k
bce22 4600.1mhz 1.388v 20hrs 55-61-61-57 AIR - Silver Arrow 2500k
JessyCuh 4800.0mhz 1.352v 15hrs 69-75-75-72 WATER - Corsair H50 2500k
Kaosuonline 4600.1mhz 1.368v 18hrs 67-74-75-71 AIR - 212+ 2500k

It's very easy to take the best and worst from each list and say "at 4.5 SB is using 1.404V while at 4.5 IB is only using 1.09V so IB is significantly lower". But that's not telling the whole truth now is it? Just as saying "at 4.5 SB is using 1.28V and IB is using 1.32V so SB is lower". Both statements are using factual if not limited results and are very misleading. We look at the results as a whole and on average the IB chips really aren't using that much less voltage than the SB chips. Again, add to that the fact that the design uses more current and there is now way that even the majority of IB chips are using less power let alone ALL of them.

Edit - Sources http://www.overclock.net/t/968053/official-the-sandy-stable-club-guides-voltages-temps-bios-templates-inc-spreadsheet/0_20
http://www.overclock.net/t/1247869/ivy-bridge-stable-suicide-club-guides-voltages-temps-bios-templates-inc-spreadsheet/0_20

I can't believe you are looking at those ten IB stable results as if they are suspicious, or as if they disprove any argument. They look perfectly normal---some good, some bad. Now read the Ivy Bridge Owner's thread, which has about a hundred times more results.

As for my reading of wattage numbers, just how do I do it? When the device says THIS MANY WATTS, what am I supposed to think? Again, no video card in use, as stated.

Now what does your meter on your system say? YAH okay.

I don't see why are you trying so hard to go against the obvious flow of info that shows how much less power IB uses, and how much more potential overclocked performance it's got over SB.