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

[LagunaX]

thanks for the bench

ipc won't go down if better batches of ib come out , so ib still could be #1 on my upgrade list,
-still got to figure in a z77 bios more tuned for ib ,lower volts\temps ?

I doubt it but you can ask Don with his z77 Biostar board.

 

[bryanW1995]

I'd think a 2600K would easily be faster than a 3570K because CB uses the extra threads. A better comparison would be 2600K vs 3770K.

Looking at the thermals as yet on IB, IB-E is probably going to be a bloody roaster

They'll probably have the leakage more under control by the time IB-E comes out. If not, they will run the risk of major issues with Haswell, or at least major delays. If intel isn't careful, AMD could creep back to being less than 4 years behind intel!!

Of course, it's also possible that 2500/2600k are really just THAT good.

 

[amenx]

Could the weaker OC, hotter chips be due to a poor initial batch? I recall years ago with an e8400, which was one of the most easily overclockable chips, I and a few others had a tough time reaching 4ghz and that was with excess volts. Most others with chips from different batches did so easily with low volts. I think we need a much wider sample base to get a better picture of IB.

 

[Quantos]

Could the weaker OC, hotter chips be due to a poor initial batch? I recall years ago with an e8400, which was one of the most easily overclockable chips, I and a few others had a tough time reaching 4ghz and that was with excess volts. Most others with chips from different batches did so easily with low volts. I think we need a much wider sample base to get a better picture of IB.

Sounds plausible, but if it's the case, somebody at Intel needs to be fired. If anything, the first batch is the most important one. It's going to leave an impression on consumers that's not very easy to change over time.

Personally, I'm still waiting for proper reviews. It'll be interesting to see what's the consensus. If it ends up that a 3770k needs 4.7GHz to beat a 5GHz 2600k, but that it's much warmer doing so, I might consider the 2600k (or 2700k to get a higher chance of lucky binning). :hmm:

 

[rgallant]

re hotter chips be due to a poor initial batch

been following this one @ ocn, 2 chips 1 good 1 bad
-the good
"
i did both 5hr prime blend and 30ish runs @ max memory linx test on 4.8@1.21V when testing for stability. Like i explained before the temps really spiked during the linx/ibt tests averaging 65-70C load with spikes up to 85C. I expect these to be faulty readings but they might just be actual temperature spikes, who knows tongue.gif"
-post 153
http://www.overclock.net/t/1242313/more-ivy-bridge-benchmarks-sandybridge-comparison-updated/150

so 4.8 stable @ 1.22v where others can't hit 4.8 even.
-can it be said also , that i5-3570k's could be failed 3770k chips.

if we see more like this ib might not be a dud.

 

[Blades]

I wonder what the bottom frequencies/temps/power draw are for both the 2500k and 3550k (or whatever its called). If you think about it, Intel was probably putting a heavy weighting on this type of comparison this time around. Being able to do more with less - less freq, less voltage .. I wonder what the perf difference is at super low freqs and voltages.

This would be a typical scenario a laptop unplugged from the wall might experience.. (unless using throttlestop )

All this said, do you think the 22nm process will be able to live in mobile soc's? Perhaps I'll buy an IVB setup with proceeds from intel's dividend.. Maybe I should just ask for them to pay me in product. I really am jealous of the new HM77 clevos WITH mxm3.0 that use graphics switching.

 

[LagunaX]

"so 4.8 stable @ 1.22v where others can't hit 4.8 even.
-can it be said also , that i5-3570k's could be failed 3770k chips.

if we see more like this ib might not be a dud."

Could you post screens of the 4.8ghz @ 1.22v?

 

[andrew_rs]

I apologize in advance if this has already been discussed (one large thread to wade through 100% of posts), but has anyone recorded power consumption with any of their overclocked IB setups?

 

[SHAQ]

I think Don has a kilo-watt but he hasn't done much testing yet.

 

[psolord]

Could some IB owner provide some gaming temps of their processor with max OC?

Something cpu light like Trackmania 2 and something cpu heavy like Battlefield 3.

 

[Grooveriding]

They'll probably have the leakage more under control by the time IB-E comes out. If not, they will run the risk of major issues with Haswell, or at least major delays. If intel isn't careful, AMD could creep back to being less than 4 years behind intel!!

Of course, it's also possible that 2500/2600k are really just THAT good.

I think AMD will never catch Intel again, unless Intel puts a shrubbery in charge of operations. :awe:

They'll probably iron things out, and have to, before Haswell comes. Isn't that the point of the whole tick tock ? To get experience with the new process before the new arch ?

 

[fixbsod]

Can we see some IBT / linx type runs w/temps? Would love to see that -- my 2700k @ 3.9 all cores tops out maybe like 66/67o C using IBT max w/16GB RAM and H60.

 

[frozencopper]

4.8Ghz @ 1.22Vcore on air must be really good cpu

Heres mine with Spire TherMax Eclipse II cooler

 

[BallaTheFeared]

Looks like 1.248v for 4.9GHz, score of 8.04 I'm assuming it's around 1.248v load since he's not using offset voltage for some reason. I dunno what it actually is under load

On air, 69C highest load.


My i5-2500k @ 5,280MHz 1.528v - which is still the top i5-2500k score on OCN.

On water 57C highest temp.

Pretty decent chip that guy has imo.

 

[LagunaX]

Nice scores but I doubt the 4.8ghz was stable for IBT/linX/Prime95 at 1.24v and load temps of 69c, but just stable enuf for Cinebench.
I looked at the that UK thread and nowhere saw a hard stability test screenshot for IBT/linX/Prime95/etc.

 

[BallaTheFeared]

Possible, but still considerably better than what we consider "golden" for SB.

 

[Tempered81]

still considerably better

Considerably better because it is using less voltage? It looks like your Sandy has the same temperatures and scored higher than that Ivy by 5%. That Puceh fellow is posting on OCN too and the last thing he'll post is a screen of Linx or Prime temperatures. They already asked him a dozen times. Only thing he says is that "it hits 100C too fast". Push that fine chip of yours to TJmax and you'd probably be over 1.56v in IBT or higher and could be reaching 5.5ish before Prochot on IBT/LinX. I'm guessing voltage is why you said 'considerably better'.

I don't think it's better, because they aren't handling the heat well. I wouldn't be scared to rock 1.6v into your gold Sandy 24/7 straight because IMO "death from vcore" = myth. You won't hear anyone tell you that on AT and the reason is? Most of us care about you frying your chip and give you fair warning above ~1.4. But you? You've already passed that mark. Props since you're already pushing 1.52, try to find a report of a recent Intel CPU burn from anything other than vPLL, pin mods, bios, power or spikes. I remember watching folks push 1.8+ core and pll into Penryns and taking bets to see how long before it fried, or which voltage would degrade/kill it faster.. It never died. Not to mention suicides over 2volts all the time. Intel CPUs are boss, bro Keep pushing it. Search for a record of their death. Even those old "ohh noo degredation threads" - the worst thing you ever saw was Joe Blow come in and say, "Well I used to pass an hour of prime at 4.1g 1.36v, now I cant pass 10 minutes at 4.05g, 1.39v." Once again those threads are obscure myths today. Anecdotal, but I've never experienced anything like it on about 8 different intel CPUs, granted never pushing over ~1.6v since I never had mods or cold.

 

[BallaTheFeared]

Not a lot of SB chips are doing nearly 5.3GHz, and yes less voltage is better always has been. My chip isn't even that great in the voltage department it just has a decent top end.

I can't get much more out of it, 1.58v gets me a bit over 5.3GHz and is totally not worth it

You're comparing an air cooled Ivy Bridge chip to a water cooled SB chip with a 240 rad and a 360 rad and no gpu load. Also ram speeds make a difference in Cinebench, I dunno what he has.

Which is exactly why I'm going to sell my SB and upgrade to Ivy Bridge, I have no issues keeping my chip cool, that's why I love my 470s, they run cool as well.. 950 core in the winter gives me 30C load temps on my 470s with a 58% overclock. Summer I tip 45C at 900 core, it's a breeze when you've purchased proper cooling for your parts.

I won't be scared of putting 1.5v down Ivy's throat and going for broke, so long as MHz continue to scale with voltage I'll keep pushing it


This is me: https://www.youtube.com/watch?v=hHaJEjG6qJo

900 core 31C load 470s, 5.2Ghz i5-2500k 1.512v 36C~ It's hotter than my 470s

 

[Tempered81]

Cool cpu loop. Looking forward to your report on 1.5v ivy temps. 3 reports so far of Ivy on water (little H100's and stuff) 25C hotter than Sandy on same water, and 200-400mhz lower clock. Screengrabs at 5.0 and 5.1, but you can screengrab 5.6 on a Sandy.

Ivy looks like garbage to me man. But if you have the clout, get one. I forsee a bunch of returning to your golden 2500, or a similar clocking 3930/2700, or going through a handful of 3770's just trying to hit 5.0 without burning down the house.

Can you do Prime at the same clocks as that cinebench run- 1.54/5280?

 

[BallaTheFeared]

Never tried, I can game all day at 5.2GHz 1.512v though, which is as stable as I need it.

Anything running AVX gets it considerably hotter than normal application temps. Reason being I believe is because Sandy/Ivy uses both the integer and FP pipelines to create the 256bit FP. Since 256 integer isn't even possible with either until Haswell, it's makes AVX temps pretty much worthless to someone like myself who uses his machine primarily for gaming.

Comparing my setup to air isn't far at all though...

 

[Tempered81]

Never tried, I can game all day at 5.2GHz 1.512v

What speed at 1.5v would a E1 3570 or 3770 need to be to be faster than a 5280mhz Sandy with no HT Probably ~4900... hrm you gettin a 35 or 37? I think you'd need at least 4950mhz to even be equal. Unless you upgrade the $ and HT with a 3770. Good luck with 4900+, and 1.5+, sounds crazy with whats out there. First off I don't think you'll get 1.5 under 85C. If you get it game stable at 1.5v with some reasonable temps, it will be impressive. Otherwise a waste of money for what might be a ~0.5% increase or decrease which could be considered within a margin of error for testing. Total wash and waste IMO.
For someone who wants a 4.5ghz Ivy and doesn't game on a 5280mhz Sandy, and likes to spend money, I could see it making some sense... I suppose. Get some pci-e 3.0 action. I've been pumped about Ivy for a long time now - as most of us have. Its disappointing, but not necessarily faster or slower, just definitely more expensive.

 

[PlasmaBomb]

Never tried, I can game all day at 5.2GHz 1.512v though, which is as stable as I need it.

Anything running AVX gets it considerably hotter than normal application temps. Reason being I believe is because Sandy/Ivy uses both the integer and FP pipelines to create the 256bit FP. Since 256 integer isn't even possible with either until Haswell, it's makes AVX temps pretty much worthless to someone like myself who uses his machine primarily for gaming.

Comparing my setup to air isn't far at all though...

Are you chilling your coolant?

 

[BallaTheFeared]

Mother nature with a bit of intervention from me is keeping my distilled + antifreeze right around 0C, any lower and SB won't post to my dedicated gpu's so I have no display output. I heard it wasn't a problem with Ivy which means I might be able to crack 1GHz on my 470s next year since I won't be limited to 0C.

This is of course during the winter, in the summer my rad goes into the basement which sits around 17C.

Add 17C to those temps and you're hitting 60C during summer which is about right, it's actually a bit lower because the antifreeze doesn't dissipate as well as straight distilled and that comes out during spring cleaning.

What speed at 1.5v would a E1 3570 or 3770 need to be to be faster than a 5280mhz Sandy with no HT Probably ~4900... hrm you gettin a 35 or 37? I think you'd need at least 4950mhz to even be equal. Unless you upgrade the $ and HT with a 3770. Good luck with 4900+, and 1.5+, sounds crazy with whats out there. First off I don't think you'll get 1.5 under 85C. If you get it game stable at 1.5v with some reasonable temps, it will be impressive. Otherwise a waste of money for what might be a ~0.5% increase or decrease which could be considered within a margin of error for testing. Total wash and waste IMO.
For someone who wants a 4.5ghz Ivy and doesn't game on a 5280mhz Sandy, and likes to spend money, I could see it making some sense... I suppose. Get some pci-e 3.0 action. I've been pumped about Ivy for a long time now - as most of us have. Its disappointing, but not necessarily faster or slower, just definitely more expensive.

I'd go i5 again, my interest starts and ends with gaming and HT isn't helping where I need it which is in lightly threaded titles like SC2, WoW, and Skyrim.

CPU is of less importance atm because of the additional two 1080p screens I added, but with the state of 28nm price/perf it's going to be a minute before I even consider a gpu upgrade.

That said the cold bug issue with SB is a big problem for me, at least from an overclocking/benching perspective, and a side grade isn't such a bad thing if I can get more from my gpus because of it. I haven't ruled out Ivy just yet, I'll wait to see some actual official reviews and of course get my own to test with

 

[T_Yamamoto]

4.8Ghz @ 1.22Vcore on air must be really good cpu

Heres mine with Spire TherMax Eclipse II cooler

dont block out the important details in CPUZ
for all i know, this can be a i5-3770K

 

[Grooveriding]

Cinebench is worthless for temps and stability judgments. Linx or bust.