Ivy Bridge 3570K Testing, Opinions, Results, New Bios, 4.5Ghz At 1.236v

Page 12 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

LagunaX

Senior member
Jan 7, 2010
716
0
76
This is an example of the benchmark I would like ran in Truecrypt:

http://s9.postimage.org/gal1givlb/bench.png

Here you go, stock settings:

TCstock.png

ysRhTZywWBAAAAABJRU5ErkJggg==
 

BD231

Lifer
Feb 26, 2001
10,568
138
106
IMO you'll find out that voltage required for under 4.5 Ghz stability is less than SB (think 1.1V vs. 1.2V+ that most SB require), and that power consumption is accordingly much lower. That's the strength of IB, moderately high clocks with a significant reduction in power.

It's not all about maximum frequency. A 4.4 Ghz IB running at 1.1V will outpace a 4.5Ghz SB running at 1.25V while using 30-40% less power (augmented from 20% stock vs. stock power savings).

LagunaX, try a more reasonable overclock under 4.5 Ghz, and find the lowest stable voltage. I think that will make IB look much more interesting.

So you're saying a chip with an 18 watt delta at stock is going to net you 40% less power overclocked? ... even though the results show you still need about the same amount of voltage to get the same clocks as SB? That's some seriously wishful thinking.


It's hard to believe one 3570k would run at 85 C and another would max out at high 50's...................... If you know what I'm sayin

Don ran his chip for all of two minutes on cinebench, other guy ran 30 minutes of prime so it's pretty evident Don will have the same issue's.
 

JAG87

Diamond Member
Jan 3, 2006
3,921
3
76
So you're saying a chip with an 18 watt delta at stock is going to net you 40% less power overclocked? ... even though the results show you still need about the same amount of voltage to get the same clocks as SB? That's some seriously wishful thinking.

No, the premise is that it requires less voltage. i.e. 4.4 Ghz with 1.1V vs. 1.2V for SB.

Power savings would be greater than stock vs. stock.
 

Tempered81

Diamond Member
Jan 29, 2007
6,374
1
81
No, the premise is that it requires less voltage. i.e. 4.4 Ghz with 1.1V vs. 1.2V for SB.

It's taking 1.248 for the LagunaX's IB, and 1.304 for 4.6, and 1.36+ for 4.7 which isn't stable. There are lots of Sandies out there that are much better than this.
 

BallaTheFeared

Diamond Member
Nov 15, 2010
8,115
0
71
Not a large enough sample base yet, there are a lot of SB chips that people tap out at 4.4GHz 1.3v.

I'm sending this Max IV back tomorrow and getting a Z77 board instead, if my i5-2500k still doesn't work I'm going to risk IB over another SB because the chances of me getting another 5.3GHz SB is quite low. At least I'll have something new to play with even if it's not quite as fast.

Cooling for me is a non issue, and I hear these don't have the cold bug like SB... Which is good for me since that means I won't have to bust out the blow dryer during the winter!


Edit: I was one of the first people on OCN to prime95 blend my first i5-2500k @5GHz, and it took 1.52v to do it. As time went by people started getting better chips, really good ones could do it slightly below 1.4v, but most were over 1.4v for 5GHz in prime95 anyways. Point is, the sample base needs to be much higher than what it currently is before we rule out IB. Plus some people just have more balls than others and will run higher voltage with better cooling.
 
Last edited:

Alaa

Senior member
Apr 26, 2005
839
8
81
4.0ghz stock volts, nice load temps 50's:
img
4.2ghz stock volts, nice load temps 50's:
img
4.5ghz 1.24v, ok load temps low 70's:
img
Thanks so much for doing that. Now can anyone tell me if intel turbo overclocking is still on in these situation? Is the chip going to overclock further by itself?
 

alex_123_fra

Junior Member
May 4, 2011
11
0
0

Again, this is a tri-SLI setup. The vast majority of gamers and low level enthusiasts are:

a) Not using tri-SLI but are mostly on one single or 2-way
b) Most gamers who have bought GTX680s/HD7970s (i.e. PCI 3.0 cards) only buy one card largely due to the cost at present
c) I would guess that for the vast majority of games in the next 1-2 years, PCI-E 3.0 bandwidth will not be necessary at 1080p.

Upgrading to ivybridge to make use of PCI-E 3.0 is again not a compelling reason to upgrade for the average low level enthusiast/gamer. Not saying that other features of Z77/Ivy are not worth it but pretending the masses need the graphics bandwidth is where I'm sceptical.
 

BD231

Lifer
Feb 26, 2001
10,568
138
106
At least we know Intel was expecting lower voltage's and ended up closer to sandy bridge numbers. This confirms it, tri-gate " at the moment " is a more power hungry process.

Edit: I was one of the first people on OCN to prime95 blend my first i5-2500k @5GHz, and it took 1.52v to do it. As time went by people started getting better chips, really good ones could do it slightly below 1.4v, but most were over 1.4v for 5GHz in prime95 anyways. Point is, the sample base needs to be much higher than what it currently is before we rule out IB. Plus some people just have more balls than others and will run higher voltage with better cooling.

I say keep the motherboard and buy another, that way you'll have 3 boards waiting to confirm 1.5v kills Ivy bridge chips.
 

meloz

Senior member
Jul 8, 2008
320
0
76
for the entusiast market, Ivy seems to be a very big fail

The enthusiasts are well looked after, they have their Sandy Bridge-E thank you very much. And there will be a large socket IVB in another 8-12 months, so the performance crowd is not being abandoned or anything.

How can you declare IVB to be a 'very big fail' when no one has yet sampled its socket 2011 SB-E equivalent? The only failure I see is people who still do not understand the two very different roles small socket and large socket play in Intel's lineup.
 

Quantos

Senior member
Dec 23, 2011
386
0
76
Not a large enough sample base yet, there are a lot of SB chips that people tap out at 4.4GHz 1.3v.

Agreed. Although I would say things aren't pointing in the right direction so far, we've heard of what... less than 10 chips so far? If you looked randomly at 10 SB chips, you can't guarantee they'd all overclock at 5GHZ stable. I'm waiting for actual reviews to provide a higher sample size. Also, you'd think that if at any point chips are binned, it's when they're given to reviewers, so we should see the best of this chip (with first retail stepping of course).
 

psolord

Platinum Member
Sep 16, 2009
2,125
1,256
136
I wonder how Ivys would perform if they were made at plain 22nm ie without 3D xtors.

Would they be better or worse in terms of overclocking?
 

Imouto

Golden Member
Jul 6, 2011
1,241
2
81
I wonder how Ivys would perform if they were made at plain 22nm ie without 3D xtors.

Would they be better or worse in terms of overclocking?

We would see the same problems with cooling since the CPU runs very hot but heatsinks are just lukewarm. No matter how good your cooling is, everyone is reporting this.

We don't know how much area the IGP ate with its 33% EUs increase. The actual area set for the CPU may be really small.
 

Tempered81

Diamond Member
Jan 29, 2007
6,374
1
81
I wonder how Ivys would perform if they were made at plain 22nm ie without 3D xtors.

Would they be better or worse in terms of overclocking?

Probably like a sandy, except problems shrinking planar tranny below 32 was the main reason to create 3d trany. They arent going to make traditional trannies, here on out will be tri gate as they continue to shrink, 22,21, 16,11,8,4 Any nodes/half nodes pitch changes will be finfet or tri-gate afaik. Unless they deem it a fail and shoot for new tech like algae or something. Maybe wait for some respins and new steppings to see if they make some improvements. Its supposed to have substantially less voltage... supposed to. I don't wanna say planar 22nm doesnt exist, but i dont think its part of their plan right now. Its going to boil down to them improving leakage of 3d
This is from 3 years ago:
Intel_Technology_Outlook_Sept2009.jpg
 

dma0991

Platinum Member
Mar 17, 2011
2,723
1
0
I think the heat issue should pretty much take care of itself since I'm already in the process of doing a custom WC to cool it. Will probably do a conservative 4.5GHz 24/7 overclock instead of going all out but I'd might just push it beyond 4.5GHz temporarily to see how much heat my loop would be able to hold up. Aiming for 70-80C or less at full load with a Core i7 3770K.

Even the H100 could not keep up, only to be made worse with my 30C ambient temps. Affordable phase change coolers for mainstream users perhaps? :biggrin: