Sign in with
Sign up | Sign in
Your question

P4 overclocked to 8 GHz!

Tags:
Last response: in CPUs
Share
January 22, 2007 11:27:23 PM

link here. I wonder if it can beat an X6800 in benches :D 

More about : overclocked ghz

January 22, 2007 11:34:01 PM

This May only be able to match a Athlon FX-62(stock). Or a E6400(stock) at best.
January 22, 2007 11:39:33 PM

Quote:
This May only be able to match a Athlon FX-62(stock). Or a E6400(stock) at best.


Could probably do better than that. I wonder how high they can get it and have it be benchable.
Related resources
Can't find your answer ? Ask !
January 22, 2007 11:42:23 PM

Quote:
This May only be able to match a Athlon FX-62(stock). Or a E6400(stock) at best.


FX-62 seems to be degraded more and more every day ... first it sat next to an 6700 then 6600 and now 6400? I wonder how much until a Celeron ( the Core version) will beat the crap out of it ...
January 22, 2007 11:44:52 PM

Quote:
This May only be able to match a Athlon FX-62(stock). Or a E6400(stock) at best.


FX-62 seems to be degraded more and more every day ... first it sat next to an 6700 then 6600 and now 6400? I wonder how much until a Celeron ( the Core version) will beat the crap out of it ...

Hehe. No, it trades blows with the E6600. Pretty much hovers there.
January 22, 2007 11:45:19 PM

Holy crap thats high! I bet it isn't stable enough to bench though. I think it should be able to match a (if there were a such thing) single core conroe at 3 ghz.
January 22, 2007 11:46:37 PM

Umm.. The AMD has never "degraded" the Core 2 Duo just seem to outshine it from the aggressive marketing and Intel fans pushing them as getting better by the minute :?: :?:
January 22, 2007 11:50:37 PM

Pretty Impressive 8O
January 22, 2007 11:56:05 PM

It's practically 10GHz :D 
January 22, 2007 11:56:20 PM

I would like to see how far they can get with a E4300. Since they are real pros...
January 23, 2007 12:01:39 AM

Anyways here's the topic the news item links to on Xtreme Systems

More picks of and screen shoots. I think I should have put that in the original post as it's a little obscure to see
January 23, 2007 12:29:05 AM

Quote:
Anyways here's the topic the news item links to on Xtreme Systems

More picks of and screen shoots. I think I should have put that in the original post as it's a little obscure to see


What this does is bring into sharp relief the difference between the quality of the process technology between the two major players.

When going for extreme OCs like this, using LN2 setups and modified boards remove the 'power' ceiling variable from the equation and drive straight to the max (Fmax) potential of the process technology regardless of architeture and power walls.

I would love to see a similar experiment done on an AMD 65 nm process for comparision, as AMD's process is not power limited but rather transistor speed limited (opposite of Intel's problems), I am pretty certain I know what the result would be....

That might very well change in the future as AMD has shown at IEDM 2006 they have improved they transistor by quite a bit, also we haven't seen their second iteration of 65 nm transistors for which they also have announced improvements at IEDM 2005.

On the other side Intel's 45 nm high-K manufacturing is coming which looks very good. The only issue might be transistor size (correct me on this if I'm wrong)
January 23, 2007 1:10:21 AM

Quote:
They will need to do some massive (and I mean massive) stress engineering to get any real clock gains going to 45 nm.


I thought a smaller process implied higher clock speeds? Are you saying that 45nm shouldn't allow intel to increase the clock speeds over 65nm?
January 23, 2007 1:27:46 AM

Quote:
They will need to do some massive (and I mean massive) stress engineering to get any real clock gains going to 45 nm.


I thought a smaller process implied higher clock speeds? Are you saying that 45nm shouldn't allow intel to increase the clock speeds over 65nm?

Smaller process does not guarantee higher clock speed, at constant voltage scaling, all things being equal, smaller simply means the same as the prior but lower power.

What I am saying is that both Intel and AMD transistion to a new node and re-engineer the processor with a new target clockspeed in mind... hence new materials, new stressing technologies usually go with that.

If you noticed, within the power constraints, clocks did not go up very much 130 nm to 90 nm and again, not much 90 nm to 65 nm (EDIT: I should add for either AMD or Intel) .... this is because the gate thickness has reached it's physical scaling limit, it can go no thinner.

If the rumors are true that Intel is using high-K for the gate material, then expect a massive jump forward in clocks on the same given archtiecture.
Jack

Right, but lower power = less heat = higher clocks, no? Granted heat is the limiting factor, as it generally is with Intel architectures...

Isn't intel using high-K now at 65nm?
January 23, 2007 1:32:21 AM

[quote="JumpingJack
I would love to see a similar experiment done on an AMD 65 nm process for comparision, as AMD's process is not power limited but rather transistor speed limited (opposite of Intel's problems), I am pretty certain I know what the result would be....[/quote]

Remember that AMD's sometimes catch the cold bug and won't OC very high at all when using LN2. I remember reading Oppainter/FUGGER had better results with an FX55 they were benching on Dry Ice than they did with LN2.
January 23, 2007 2:59:09 AM

I am unsure how they could come to such a conclusion. Drop the CPU multiplier and see the highest FSB the mobo will reach and if it happens to be the 513.xx they acheived then we can venture a guess that it is mobo limited. The theory is sound at lower clock speeds and more normal operating conditions, but too many variables are changing to really know if the chip has more in it or not.
January 23, 2007 4:45:34 AM

Quote:
It's practically 10GHz
actually if I remember my white papers correctly the trace cache in the P4 runs at 2X the clock rate which would equal 16ghz for at least a small part of the core.

please correct if wrong btw.
January 23, 2007 6:31:36 AM

That is an interesting font and font color within the text boxes on that CPU-Z screenshot.
January 23, 2007 2:10:14 PM

Quote:


{...}

Not necessarily, the speed at which a transistor can switch ultimately determines how fast the clock speed it can run.... that is why when I say take the power variable out of the equation and compare the max possible clocks.

I.e. 8 GHz (ok we do not know if it is stable, but it is enough to get a CPUID shot) is likely pushing the absolute max -- however, the reviewers pointed out that it was limited by the board before the CPU. How they determined this I do not know.... anyway, take an AMD CPU and do the same and it's max clock is much lower --- take away all power variables.

This is related to gate delay, td=CV/I and C, I are products of the process capability. Now, add the power variable back in and yeah, you are correct --- if I set an upper limit on power to be X that I will not exceed, then there will be a clock that reaches X and I can go no higher on the self imposed power envelop.

For AMD's 65 nm process though, haven't you noticed --- lower power did not help them get higher OCs - why?? Because their process, SOI process, is gate delay limited to just over 3 to 3.2 GHz on 65 nm, and 90 nm is slightly higher.... some good a 2.6 GHz 65 Watt 65 nm chip did against a 2.6 GHz 89 Watt 90 nm chip eh???

It's physics man.... :lol:  :wink:

EDIT: Also, I should make a point here --- the long pipeline of the P4's certainly have allowed the processor to take advantage of the short gate delay process.... K8 short pipeline retards this.

Jack


Thanks :)  Just trying to get my facts straight.
January 23, 2007 4:36:08 PM

For those that didn't go through the original thread, this is something that needs to be seen...





January 23, 2007 4:48:23 PM

since the C2D roughly performs double the Pentium D at the same clock speed, in single threaded applications where a Pentium 4 and pentium D perform the same, an 8.0GHz P4 would perform the same as a 4.0GHz C2D. In multithreaded apps, I would think it would be, at worst, equivalent to an E6400, and at best an E6600. An AMD K8 core would have to be around 5.5 to 6GHz to be equivalent. In mulithreaded apps, it would probably be close to an FX-62.

Anyway, an 8GHz P4, while unrealistic, is hard to beat.
January 23, 2007 4:53:24 PM

Quote:
since the C2D roughly performs double the Pentium D at the same clock speed, in single threaded applications where a Pentium 4 and pentium D perform the same, an 8.0GHz P4 would perform the same as a 4.0GHz C2D. In multithreaded apps, I would think it would be, at worst, equivalent to an E6400, and at best an E6600. An AMD K8 core would have to be around 5.5 to 6GHz to be equivalent. In mulithreaded apps, it would probably be close to an FX-62.

Anyway, an 8GHz P4, while unrealistic, is hard to beat.


Did you read this thread?
I posted benchies showing it outperform a E6600 at 3.8Ghz which is WAY faster than an x6800 @ stock.

So when you say it is at best equivalent to an E6600 you are incorrect. It is much faster than a stock E6600. Though this is a single benchmark and a single data point it is the best judge of performance that we have thus far. I wish they would do more benches!
January 23, 2007 5:13:06 PM

Quote:
link here. I wonder if it can beat an X6800 in benches :D 


I aint impressed, Intel showcased a 1thz CPU over 5 years ago now.
January 23, 2007 5:18:10 PM

a multi billion dollar company and what appears to be some guys in their basement are very different...
January 23, 2007 5:22:09 PM

Quote:
a multi billion dollar company and what appears to be some guys in their basement are very different...


You're chatting *expletive deleted*
January 23, 2007 5:25:06 PM

Quote:
a multi billion dollar company and what appears to be some guys in their basement are very different...


You're chatting *expletive deleted*

What?
January 23, 2007 6:29:45 PM

Quote:
a multi billion dollar company and what appears to be some guys in their basement are very different...


You're chatting *expletive deleted*

What?

It was an attempt at a joke, albeit a very bad one.

Really it's meant to be "You're chatting *insert rude word for excrement here*", quite a common phrase among Chavs/Hoodys in the UK.
January 23, 2007 6:32:10 PM

I believe you need you reread my post and apologize for jumping to conclusions.
January 23, 2007 7:12:11 PM

Such hostility in this thread...
January 23, 2007 9:22:37 PM

Quote:
a multi billion dollar company and what appears to be some guys in their basement are very different...


You're chatting *expletive deleted*

What?

It was an attempt at a joke, albeit a very bad one.

Really it's meant to be "You're chatting *insert rude word for excrement here*", quite a common phrase among Chavs/Hoodys in the UK.

ahhh I get it! :D  It makes more sense if its in an english accent ;) 

Quote:

I believe you need you reread my post and apologize for jumping to conclusions.


Hahaha, yeah I did mis read it. And you want me to apologize for saying that you were incorrect for saying it was at best equivalent to an E6600? haha, it wasn't a personal attack of any sorts, simply trying to keep the facts straight.

Good day
!