Core i5 650 Reaches 4.70 GHz on Air-Cooling

"Intel's socket LGA-1156 quad-core processors are closely trailed by the Intel's first processors based on the 32 nm manufacturing process: Core i5 and Core i3 "Clarkdale" dual-core processors. Engineering samples of these processors were evaluated as early as Q2 2009, but one of the first attempts to show the processors' overclocking potential using air-cooling was made very recently. Even prior to that, a low-voltage overclocking feat by Coolaler showed how engineering samples didn't particularly struggle reaching clock speeds close to 4.00 GHz with vCore as low as 0.832V. Romanian tech community Lab501. Lab501 community leader "Monstru" tested the overclocking headroom of a Core i5 650 LGA-1156 dual-core processor (engineering sample) with air-cooling."

http://www.techpowerup.com/img/09-10-21/68a.jpg

LINK

lab501.ro

32 nm from Intel looks to be really good for us. :yepp:

whau, nice !

Thats rad :eek:

Niceeeeeee

Now i need another CPU upgrade :(.

32 nm I7 please.

I'm more interested in a Quad too, not dual core.

Oh my god thats insane, 5ghz benchstable on air?

8K prime doesn't really rock my world :p

I dont get it my E8500 did 4.8 on air stable?

We have nearly 4.8 Ghz i7 920 stable already. On water, though.
But not bad I guess.

4.5 GHz should be easy on air 24/7, with some getting to 5 GHz. Good stuff. My E6600 machine is due for an upgrade when these babies are out.

Just make sure you get a mobo that doesn't have the foxconn socket...

looks like its 17C ish air.
Looks good.

How's the performance? Any Pi screens?

I too can stick my computer outside or in the garage on one of those cold Canadian nights and claim high clocks on "air cooling". The min temp of 20°C means an ambient temp of what? 15°C? 17°C?

what kind of power consumption is found on these 32nm chips? im interested in the 4ghz at .8v

Quote:

---

Originally Posted by Manicdan

what kind of power consumption is found on these 32nm chips? im interested in the 4ghz at .8v

---

You actually think that 4Ghz is achievable and stable clock with 0.8V?:confused:

i clocked my 860 with only 2 cores to 4.2 like it was nothin but shortly after board would not boot
thank gigabut for using a foxconn socket i counted around 20 pins not making contact or barely lol

Quote:

---

Originally Posted by SKYMTL

I too can stick my computer outside or in the garage on one of those cold Canadian nights and claim high clocks on "air cooling". The min temp of 20°C means an ambient temp of what? 15°C? 17°C?

---

:welcome:

Why make such allegations my friend? Why would I deceive readers like that? Simply showing Clarkdale is good enough. Why would I start some rumours that will be put down in 1-2 months if they are not true?

No my friend, I am sorry to set the record straight like this but this is done inside my apartament, with 24 oC temp ambient, a normal retail cooler and a normal retail fan. I am trying to help Kevin to adjust RealTemp for this CPU's, because right now it's obvious the TjMax is a little bit too optimistic. Also I am interested in how Turbo works on this one, how CPU-Z reads the mem latencies and so on. Why the heck would I do that at 15 or 17 oC??? =))

Intel's 32nm process looks absolutely astoneshing, but this is not the first time people are seeing this. Weird enough, I don't remember anybody saying that all those Gulftowns stable at 4,5 on air are done in Siberia, or Canada, or wherever :)


mrcape - patience, we will get there :)

manicdan - well, I haven't seen 0.8 at 4GHz, so this might not be the brightest puppy in the house. But the power consumption with the CPU priming at 4,7 and GTX 295 in idle was something like 260W (roughly).

4.6 - 4.7GHz isnt a problem for latest Clarkdale samples..i had one last month...

also with air cooling

http://www.abload.de/img/4.64ghz-primey0x2.png

With much better volts even!!! Woot, this 32nm is a killer!!!!

Quote:

---

Originally Posted by CoolZone

I'm more interested in a Quad too, not dual core.

---

no 32nm quads apparently just these dualies + gulftown :down::(

5.0ghz on water should be sweet... :)

Well I think more is possible on air too, at the moment I am fighting with 190 BCLK, so the frequency and the volts were not the actual limit in the screenshot, but the BCLK. So, who knows. :)

Quote:

---

Originally Posted by Jamesrt2004

no 32nm quads apparently just these dualies + gulftown :down::(

---

That sucks:down:

Does anyone know when 32nm CPUs are hitting the shelfes?

Why do you need quad if it has 4 threads that perform equally well?

Quote:

---

Originally Posted by zalbard

We have nearly 4.8 Ghz i7 920 stable already. On water, though.
But not bad I guess.

---

huh? Average 920 do 4-4.2 max with reasonable volts stable

Quote:

---

Originally Posted by kiwi

Why do you need quad if it has 4 threads that perform equally well?

---

this has better overclocking, with less power, but its only going to be used for the cheap chips, kinda sounds wrong.

i think with the quads you might start seeing limitations with the motherboard power

Quote:

---

Originally Posted by ArneMS

That sucks:down:

Does anyone know when 32nm CPUs are hitting the shelfes?

---

ill take a guess at feb~

AFAIK January. But you can never be sure what Intel is up to in the last moment, they allways surprise us in a good way :)

Quote:

---

Originally Posted by Monstru

:welcome:

Why make such allegations my friend? Why would I deceive readers like that? Simply showing Clarkdale is good enough. Why would I start some rumours that will be put down in 1-2 months if they are not true?

---

I'm not saying its a rumor or anything else. I am trying to put things into perspective before the thread gets out of hand with unrealistic expectations for Clarksdale.

And yes, the 32nm process looks very promising. :up:

Quote:

---

Originally Posted by Ozzfest05

I dont get it my E8500 did 4.8 on air stable?

---

show us?

Quote:

---

Originally Posted by SKYMTL

I'm not saying its a rumor or anything else. I am trying to put things into perspective before the thread gets out of hand with unrealistic expectations for Clarksdale.

And yes, the 32nm process looks very promising. :up:

---

you could have worded it better if that was your intention :rolleyes:

Quote:

---

Originally Posted by Monstru

Well I think more is possible on air too, at the moment I am fighting with 190 BCLK, so the frequency and the volts were not the actual limit in the screenshot, but the BCLK. So, who knows. :)

---

try subzero CPU
i am at 240 bus

but my chip doesnt like clocking high for some reason...i'm thinking it might be the bios donno

Quote:

---

Originally Posted by dinos22

try subzero CPU
i am at 240 bus

but my chip doesnt like clocking high for some reason...i'm thinking it might be the bios donno

---

Which mobo and BIOS?:D

same as OP :D

awesome! :toast:
4G at .8v sounds a lot more interesting than 4.7 on air... :D
who needs that much cpu power besides for benching anyhow, having 4g at only .8v sounds great! thats 50% less volts than my 920! :eek:

still dont get why intel would call this i5... this deserves i7 if not i8! :D

Quote:

---

Originally Posted by saaya

awesome! :toast:
4G at .8v sounds a lot more interesting than 4.7 on air... :D
who needs that much cpu power besides for benching anyhow, having 4g at only .8v sounds great! thats 50% less volts than my 920! :eek:

still dont get why intel would call this i5... this deserves i7 if not i8! :D

---

perhaps for a new stepping or process refining. it might be like intels 65nm steppings. now that would be impressive. am i the only that is not impressed by 4.7GHz? back in the day doubling frequency was the norm for new processes. intel said the would be at 20GHz by the time they got to 45nm and now they are at 4.7. i miss frequency scaling.:confused:

Quote:

---

Originally Posted by dinos22

try subzero CPU
i am at 240 bus

but my chip doesnt like clocking high for some reason...i'm thinking it might be the bios donno

---

Hmm...your likes high BCLK but no high clocks, mine loves high clocks but not high BCLK. Pretty much sure it's the bios, it's to early for this :)

Quote:

---

Originally Posted by Chumbucket843

perhaps for a new stepping or process refining. it might be like intels 65nm steppings. now that would be impressive. am i the only that is not impressed by 4.7GHz? back in the day doubling frequency was the norm for new processes. intel said the would be at 20GHz by the time they got to 45nm and now they are at 4.7. i miss frequency scaling

---

Well, in terms of Netburst performance, this is 20GHz :)

Quote:

---

Originally Posted by Monstru

Hmm...your likes high BCLK but no high clocks, mine loves high clocks but not high BCLK. Pretty much sure it's the bios, it's to early for this :)

---

youngpro has the same chip as me and his is stuck at 190 bus air cooled as well

That's not cool, the chip could do more even on air. :(

Quote:

---

Originally Posted by kiwi

Why do you need quad if it has 4 threads that perform equally well?

---

No,Quad performs much better in most cases .....


http://img.photobucket.com/albums/v4...08269cb61e.png

http://img.photobucket.com/albums/v4...kdale-GTA4.png

http://img.photobucket.com/albums/v4...rkdale-RDG.png

http://img.photobucket.com/albums/v4...rkdale-FC2.png

http://img.photobucket.com/albums/v4...-Cinebench.png

http://img.photobucket.com/albums/v4...e-PaintNET.png

http://img.photobucket.com/albums/v4...-Truecrypt.png

http://img.photobucket.com/albums/v4...kdale-x264.png

Ys, 2 cores / 4 threads is not exactly a quad-core, it is something between quad and dual, as performance. But that one tested above it's an i3 540. Normally those ones should not have HT.

Quote:

---

Originally Posted by Monstru

Ys, 2 cores / 4 threads is not exactly a quad-core, it is something between quad and dual, as performance. But that one tested above it's an i3 540. Normally those ones should not have HT.

---

i3 has HT,but no turbo.....

I see that from your graphs :)

If Intel will keep HT on i3 too, this will be veeeery interesting :)

Quote:

---

Originally Posted by kiwi

huh? Average 920 do 4-4.2 max with reasonable volts stable

---

One chip against one chip... :shrug: Not like they tested a hundred and all clocked that good. Not too impressive anyway in this case. I am expecting more from 32nm.

Quote:

---

Originally Posted by dinos22

show us?

---

http://i191.photobucket.com/albums/z...vSP2M48GHZ.jpg

SP2M 4.8GHz 1.55v Vcore

metro.cl on XS

So, ya not that impressed but the volts are nice, i would like a i7 based on 32nm tough.

Did anyone notice that i3 540 at stock speed outperformed Phenom II X3 720 in the majority of cases, according to benchmarks posted by mindfury ? It looks like those i3 can compete very well against Triple core CPUs, and lower end Quad cores

Also, I wonder why they disabled Turbo mode in i5 750 and i7 860 ? It is unfair to compare a CPU to others when one of them have a feature disabled.

Quote:

---

Originally Posted by Monstru

Ys, 2 cores / 4 threads is not exactly a quad-core, it is something between quad and dual, as performance. But that one tested above it's an i3 540. Normally those ones should not have HT.

---

Pentium will have 2 cores.
Core i3 will have 2 cores with HyperThreading
Core i5 will have 2 cores with HyperThreading and Turbo boost

Quote:

---

Originally Posted by dinos22

but my chip doesnt like clocking high for some reason...i'm thinking it might be the bios donno

---

or maybe its the socket ^^
snif snif... whats that smell? :smoke:

:lol:

hey! we need a new emoticon on xs, a cpu socket smoking or catching fire! :D

Quote:

---

Originally Posted by Chumbucket843

i miss frequency scaling.:confused:

---

well, i wouldnt mind if wed see a netburst to core ipc jump instead of a mhz jump all the time... but ipc boosts are very scarce and are only notable in some applications that make use of new indstructions... and there are almost no mhz boosts anymore, yeah... its all about more cores now... more and more cores... :/

Quote:

---

Originally Posted by mindfury

No,Quad performs much better in most cases .....

---

define MUCH better please :P 5fps?
better, yes, of course, MUCH better? not really...
take into account price/perf and that we will be able to push 32nm dualcore chips higher than 45nm quads and they are at least even, if not the 32nm dualcore chips beating 45nm quads...

Quote:

---

Originally Posted by saaya

define MUCH better please :P 5fps?
better, yes, of course, MUCH better? not really...
take into account price/perf and that we will be able to push 32nm dualcore chips higher than 45nm quads and they are at least even, if not the 32nm dualcore chips beating 45nm quads...

---

You need read graphs again...

Dont compare i5 750 at stock with i3 540@3.8g.....

i5 750 at stock is mush stronger than i3 540 at stock....

Quote:

---

Originally Posted by ajaidev

SP2M 4.8GHz 1.55v Vcore

So, ya not that impressed but the volts are nice, i would like a i7 based on 32nm tough.

---

Your chip is a pretty special one and pretty rare among 45nm based ones. The 32nm cpus will go a lot easier above 4Ghz and with less voltage. Its called progress. :)

4g with 0.8V
:shocked::shocked::shocked::shocked::shocked:

Pi 2M has nothing to do with 30 minutes prime, but that is indeed a very good E8500. :)

Quote:

---

Originally Posted by mindfury

You need read graphs again...

Dont compare i5 750 at stock with i3 540@3.8g.....

i5 750 at stock is mush stronger than i3 540 at stock....

---

Not that I fully don't trust their graphs but please wait retail boards/bios and chips :)

Also look at GTA4, how can 750 beat 860 by almost 20%? Is it just because they turned HT on? Why did they not test 860 without HT?

waiting on this cpu.. will it beat Q6600 in threaded apps is my only Q as it being 2core 4 threads??


and "close to" 4gz at only 0.84v is incredible.. amazing for htpc's with a corsair H50 or something

I may even switch my bad sample i7-860 for a 32nm i5 if average max overclock (on air) is high enough, say 4.6~4.70GHz at ~1.4v. I don't care that much if I'd lose 2 cores cuz like 95% I use doesn't offer any improvement over a dual core today still and then this still has HT support too.

Quote:

---

Does anyone know when 32nm CPUs are hitting the shelfes?

ill take a guess at feb~

---

Actually even sonner... January 3rd... I hope that Intel will count the holidays and hurry the launch to 20th december...I mean samples are for 2-3month aviable so ...:up:
And the prices might be a little lower than suggested until now

Quote:

---

The slowest of them all will end up branded as Pentium G6950 and this dual-core will have only two threads enabled. It runs on 2.8GHz and has 3MB memory, while LGA 1156 is the socket of choice for it. When it launches it will sell of $87.
The slowest Core i3 530 works at 2.93GHz has two cores and four threads in hyperthreding, 4MB cache and will sell for $113 when it launches on January 3rd.
The faster iteration is Core i3 540 that shares the same spec but 130MHz boost will cost you additional $20 or $133 total.

---

http://www.fudzilla.com/content/view/16049/1/
These sems to me like E8xxx ad E7xxx prices. i3 530 has a very good price.
I expect i5 650 to cost 166$(E8400), 660 186$(E8500) and 670 266$(E8600).
The difference between i3 and i5 will be turbo and AES instructions(i3 will not have them)....AES are some new instructions for winrar, winzip an may be some encoding. With these instruction i5 will almost catch X4 in these aplications.:up:

Quote:

---

Originally Posted by Jamesrt2004

no 32nm quads apparently just these dualies + gulftown :down::(

---

I dont understand why intel are not releasing 32nm quads before sandybridge. You would think that they would be desperate to get all its range shunk to 32nm.

Probably Sandy bridge is more than 32nm shrink for budget quads, it has more performance tweaking like Conroe-Perynn.

Looks promising!

Quote:

---

4g with 0.8V

---

4 ghz with .84 but 4.7 at 1.43 volts I will wait for more results before drawing conclusions
I have seen E8400's at 5ghz on air

I think the bigger implications are for the mobile segment.. I mean 4Ghz on .84... is ridiculous, I ll take a 3.5ghz dualcore with smt at .7v / .65v with speed stepping on a laptop any day!!!

Quote:

---

Originally Posted by Jamesrt2004

waiting on this cpu.. will it beat Q6600 in threaded apps is my only Q as it being 2core 4 threads??


and "close to" 4gz at only 0.84v is incredible.. amazing for htpc's with a corsair H50 or something

---

lol curious too cause that would be amusing why chips can't beat a what 3 year old chip?

with laptops, it feels like the best 65nm were not much worse than the best 40nm, 32nm better bring a whole new level of perf/watt or ghz/watt

Quote:

---

Originally Posted by ajaidev

http://i191.photobucket.com/albums/z...vSP2M48GHZ.jpg

SP2M 4.8GHz 1.55v Vcore

metro.cl on XS

So, ya not that impressed but the volts are nice, i would like a i7 based on 32nm tough.

---

mate
PRIME stable and 2M are different things.................

show me ONE Wolfdale that could prime at 4.7GHz ..... i remember one guy cheating here opening his windows in winter in Canada somewhere or US saying he could get his chip 4.6GHhz stable from memeory

4.7GHz prime stable on "normal" air with Wolfdale is unheard of

Quote:

---

Originally Posted by Smartidiot89

Pentium will have 2 cores.
Core i3 will have 2 cores with HyperThreading
Core i5 will have 2 cores with HyperThreading and Turbo boost

---

hmmm mine is an i3 iirc but it has HT & Turbo :confused:
i gotta double check

Well, I actually benched E8400 at 4,9 on air, and E8600 at 5,something. My colleague Poparamiro had an E8600 that did some Pi 1M at 5,3 if I remember that right. However, none of this chips would Prime at 4,6-4,7 on air. And those were good chips. I can't say that this is a "gem" i650 or anything like that. I think that once more mature bios-es hit the market, and more and more people start showing results with these babies, we will be able to know for sure what is an "average" Clarkdale stable frequency on air. For now, the start is very promising and I have to take my hat off to Intel :)

Quote:

---

Originally Posted by dinos22

mate
PRIME stable and 2M are different things.................

show me ONE Wolfdale that could prime at 4.7GHz ..... i remember one guy cheating here opening his windows in winter in Canada somewhere or US saying he could get his chip 4.6GHhz stable from memeory

4.7GHz prime stable on "normal" air with Wolfdale is unheard of

---

i guess you dont hear much lol:ROTF:
i been @ 4.6 24/7 prime stable all summer and soon to be @ 4.8 for winter on air lol

^^^^ oh yeah this is that guy :rolleyes:

Quote:

---

Originally Posted by Monstru

Well, I actually benched E8400 at 4,9 on air, and E8600 at 5,something. My colleague Poparamiro had an E8600 that did some Pi 1M at 5,3 if I remember that right. However, none of this chips would Prime at 4,6-4,7 on air. And those were good chips. I can't say that this is a "gem" i650 or anything like that. I think that once more mature bios-es hit the market, and more and more people start showing results with these babies, we will be able to know for sure what is an "average" Clarkdale stable frequency on air. For now, the start is very promising and I have to take my hat off to Intel :)

---

me too i've had CPUs capable of doing 3dmark06 at 4.9 but priming was a different story....

Quote:

---

Originally Posted by zsamz_

i guess you dont hear much lol:ROTF:
i been @ 4.6 24/7 prime stable all summer and soon to be @ 4.8 for winter on air lol

---

Me too.
http://img3.imageshack.us/img3/9666/460val.jpg
That's my 24/7.
Ran it 5.0 on the old I45 - that's around here somewhere.

Quote:

---

Originally Posted by chuck4456

Me too.
http://img3.imageshack.us/img3/9666/460val.jpg
That's my 24/7.
Ran it 5.0 on the old I45 - that's around here somewhere.

---

well thats easy


some now know :ROTF::ROTF:
but they they satisfied with a slow laptop for 24/7:rofl:

i only primed @ 4.8 i will try higher

Quote:

---

Originally Posted by drjunk

I dont understand why intel are not releasing 32nm quads before sandybridge. You would think that they would be desperate to get all its range shunk to 32nm.

---

well as saya explained to me, intel has no need for 32nm quads there 40nm's are selling well the fab is paid off so there making the chips for little to nothing atm :up:

Hey guys, thhose are some sweet valids and Pi SS. I can play that, even with my old C0. But that does not mean this is done on air and it does not mean this is prime =))

http://lab501.ro/wp-content/uploads/2009/10/E8400.JPG

Quote:

---

Originally Posted by Monstru

AFAIK January. But you can never be sure what Intel is up to in the last moment, they allways surprise us in a good way :)

---

Yep, Jan 3 for Clarkdale and Arrandale.

Quote:

---

Originally Posted by drjunk

I dont understand why intel are not releasing 32nm quads before sandybridge. You would think that they would be desperate to get all its range shunk to 32nm.

---

(1) Just because 32nm Westmere quads have not been ANNOUNCED by Intel on any roadmap does not mean they won't make them... why crap all over current i5 quad sales when they don't need to?

(2) 32nm will only ramp so fast, and there'll be a LOT of Clarkdale and Arrandale volume, plus Xeon (tho that last is a small part).

(3) It could be that Sandybridge is looking good, and will make a late Q3 intro, in which case, together with (2), might not be a large enough window to justify 32nm westmere quad.

(4) They want to use the 45nm fabs for *something*, and with AMD non-competitive, why NOT keep selling quad 45nm desktop parts until Sandybridge? Higher material cost per die, BUT the 45nm fabs are "paid off" vs 32nm, so it is unclear which actually benefits them more.

So... a bunch of possibilities :)

Quote:

---

Originally Posted by Glow9

lol curious too cause that would be amusing why chips can't beat a what 3 year old chip?

---

And why would that amusing..?

You can't beat more cores by brute force in multithreaded apps.

4 physical cores always will be better the 2 physical + 2 virtual cores in apps that use all available threads.

In a bit older games i3 will just whipe the floor with a Q6600, in encoding and other multithreaded apps it will get close sometimes, and other times the Q6600 will still be faster. Especial when we talk about non stock clocks.

Can this cpu compares to i5 750 ? or is it way lower performance ?

Quote:

---

Originally Posted by dinos22

hmmm mine is an i3 iirc but it has HT & Turbo :confused:
i gotta double check

---

Strange, is it a ES chip? I know Core i3 isn't supposed to have Turbo:shrug:

Quote:

---

Originally Posted by jooelhoglund

Can this cpu compares to i5 750 ? or is it way lower performance ?

---

depends on how threaded the application is

I believe that once overclocked it might even be faster than a i5 750 in games

The 650 should be a Core i5 processor. Not a Core i3.

but it's dual core?

6.6GHz on air would be nice

Sure, and without Heatsink fan :P

Quote:

---

Originally Posted by Ozzfest05

I dont get it my E8500 did 4.8 on air stable?

---

http://i191.photobucket.com/albums/z...vSP2M48GHZ.jpg

SP2M 4.8GHz 1.55v Vcore

metro.cl on XS

So, ya not that impressed but the volts are nice, i would like a i7 based on 32nm tough.[/QUOTE]

Quote:

---

Originally Posted by dinos22

mate
PRIME stable and 2M are different things.................

show me ONE Wolfdale that could prime at 4.7GHz ..... i remember one guy cheating here opening his windows in winter in Canada somewhere or US saying he could get his chip 4.6GHhz stable from memeory

4.7GHz prime stable on "normal" air with Wolfdale is unheard of

---

Exactly right Dino, 2M is nothing compared to Prime, so be careful what you call 'stable'. Further, 4.8 GHz 2M is nothing special for 8500/8600 even on air.

Here is my 8600 at 5011MHz on air (TRUE Copper + 2000rpm scythe) 32M stable.

http://i720.photobucket.com/albums/w...01MHzPL7-1.jpg

No way this would run Prime at 4.7 though, although perhaps the very best 8400/8500/8600 could get close, I have seen guys at 4500 MHz at <1.30v Prime stable.

32nm duals certainly look promising for those who don't want/need quads/6 cores.

@Monstru

Good come back@response
I honor your virtue :up:

Quote:

---

No way this would run Prime at 4.7 though, although perhaps the very best 8400/8500/8600 could get close, I have seen guys at 4500 MHz at <1.30v Prime stable.

---

My e8600 is 13 hours prime stable 4.5ghz 1.29 volts 9x500 havent tried for more because it is high leakage gets way too hot I ran TRUE for a while now its on water.
Now how about putting that 32nm on LN2!!!!!!!!!!!!!!!!!!!!!!!!!!!

dro - thank you

Ozzfest05 - this is the next project, putting thise baby on LN2. I will have to wait untill I get back from TweaKING though :)

you got pm , romanian god of beer

=)) And a good one it was =))

^^ Now I'm curious lol