Just had a thought....... Is OCing over with the P4???

[Windogg]

As anyone who has tried to OC a Athlon or Duron via the FSB has known, you can't go more than a few Mhz before you get an unstable system. The DDR bus just doesn't tolerate OCing. To get around this the wonderful Golden Fingers and Golden Bridges have allowed tweakers to play around with the multipliers. To the joy of many, 800, 900, and 1Ghz can be acheived with the "lowly" Duron.

Chipzilla on the otherhand clock locks the processor so adjusting the FSB is th eonly way to go. The P4 will be running at quad data rate. If history has been any indication, we will not be able to OC more than a few Mhz before BSODs and lockups.

If this is true than long live AMD. I don't plan on getting a P4 but it will certainly make me thing twice about getting one.

Anyone else have thoughts on the issue? I could be way off here.

Windogg

 

[ViRGE]

I would think Intel would just totally lock the P4 anyways, so I doubt we have much to lose. Then again, the P4 will be running so close to it's max anyways, it's possible Intel may leave it unlocked knowing the core won't go much higher.

 

[Hulk]

Unless we learn something new about the P4 and supporting motherboards it looks like you may be correct.

 

[compuwiz1]

Windogg, there is always that kind of speculation, but I can remember back when people thought the original PIII's and Celerons beyond the original 300a's would be unoverclockable. Time and again, it keeps being proven otherwise. Typically, though, the chipmakers have left some headroom in the silicon. I can only hope that doesn't change.

 

[Windogg]

Hope you're right Virge.

Compuwiz1, I think its guaranteed that there is plenty of headroom with the P4s. The big concern is the clock quadrupled bus. AMD's DDR bus has made even attaining 105Mhz FSB a challenge. A clocked locked P4 would maybe only yield a few Mhz via the FSB.

Windogg

 

[Bryan]

Right, because they will have much lower multipliers than the recent P3's

3.5x400 (100MHz quad pumped)= 1400
3.5x448 (112MHz quad pumped)= 1568
3.5x532 (133MHz quad pumped)= 1862, BUT... can it handle it?? Most likely not.

Not a huge improvement by today's standards on the 112 example...

 

[AndyHui]

But presumably a quad pumped bus and a double pumped ALU would be far more sensitive to any changes.

As Windogg said above....if a DDR bus doesn't tolerate overclocking too much, a QDR bus would be even less so. Strictly speaking, the bus really isn't QDR....it's double pumped double width. Even then, the room for adjusting the FSB is very small.

As for the ALU, all I can say is that I hope the thing is stable.

 

[Windogg]

I'm thinking more like

3.5 x 103mhz (412Mhz) = 1,442Mhz.

Ugh.... not much there.

Windogg

 

[Bryan]

<< 3.5 x 103mhz (412Mhz) = 1,442Mhz >>



Not even worth the risk, is it?

 

[ViRGE]

Like I said in my previous post, I really don't think we should be worrying about this. At more than 200mm for the die size, and supposedly requiring a humongous heatsink, I think it's a safe bet to say that the P4 @.18 is being built near it's max specs.:Q

 

[Sephiroth_IX]

<< again, the P4 will be running so close to it's max anyways, it's possible Intel may leave it unlocked knowing the core won't go much higher >>

I think by multiplier OC'ing you could probably get 1.8 from it with a good HSF, this is not a chip that is close to its limit.

 

[BigToque]

Can someone help me out a bit here...

With AMD, the CPU bus speed is 200 MHz, and its memory bus speed is 100 MHz. thus needing only pc100 sdram. correct?

With Intel P3 (newer chips), the CPU bus speed is 133 MHz, and the memory bus speed is 133 MHz. thus needing pc133 sdram.

Now my question starts here...

With Intel P4, the CPU bus speed is 400 MHz (?), and its memory bus speed is ??? MHz?, thus needing pc??? ???dram.

....


shit, never mind. i just answered my question myself while writing out the question.


**************

new question... is the cpu frequency and the ram's frequency now asynchronous? (ie, ram speed doesnt equal the fsb)

 

[WoundedWallet]

If history repeats itself, the P4-1400 will probably be worth to be overclockable by the time the 2GHz or so are out.

I haven't seen their roadmap(I don't read fiction ), but I have this feeling that we will have to wait till the .13Micron wafers are out.

That will give AMD a good time to increase their production to satisfy all the overclockers.

 

[Windogg]

WoundedWallet: The big problem is still the muliplier lock and how double/quad pumped busses not overclocking well. Like Virge said, we may get lucky if Intel does not clock lock the early P4s (I'm not getting one though) but if/when they do lock down the multiplier, the FSB will be the only way to go. Who knows, the inablity to OC a DDR FSB may just be an AMD thing. Unfortunatly, I feel that its not a problem that only affects AMD processors.

Windogg

 

[RGN]

um, I don't think you guys have it quite right...

Even with Rambus, the bus speeds are still running @ 100 or 133 Mhz bus. Nothing changes. The &quot;quad Pumped&quot; bus is only like the AMD design, you know 100Mhz ram and &quot;200 mhz&quot; bus.

So the 1400Mhz P4 that I saw running at the &quot;Chip N Tips&quot; was 133Mhz bus with a 10.5 multiplier. I confirmed this with the intel guy there.

intel also stated that they were going to release a SDRAM and DDR ram chipset also, just not at launch. &quot;But to get the most performance, you must use RDRAM&quot; (my a$$)

 

[Methodical]

I think AMD will save us here... If P4 has no O/C room which is something i hadnt even thought about, then i think AMD will be here to save the day....

I love the concept with PowerNOW for AMD's mobile chips, with is on-the-fly multiplier changes... anand's tests have proven this to be a sweet-no-downfalls technology for mobiles.

err..nevermind, not sure what i was trying to say.

AMD's current prices almost make me thing that overclocking is pointless...When you overclock, stability is your number 1 concern.

OMG i just checked pricewatch... I mean... christ... how far down can the prices go?

A thunderbird 900 costs 170 bucks...

what the hell..
Heheheh...

'nuff said..

if amd keeps up like that, i dont care at all what intel is doing with the P4... All hail thunderbird + 760 chipset (Duallies!!!!)

 

[Muerto]

Another problem with overclocking high speed chips is will it be worth it. If you can overclock a 600 MHz Duron to 900 then you're getting a 50% speed boost. What are the chances of getting a 1.4 GHz P4 up to 2.1 GHz? Not bloody likely! As processor speeds get higher and higher any speed increases that we can do will have less of an effect.

A 300 MHz overclock on a 600 MHz Duron is 50% where as that same overclock on a 1.4 GHz P4 is 21%. And as CPUs get faster that increase will get smaller. For now overclocking is a very good thing, but I'm not sure how long it will be practical.

Or, on the other hand, I could just be full of sh!t. We'll let the benchmarks and reviews decide.

 

[AbRASiON]

Isnt it the AMD chipsets which have the problems at over 100mhz (ddr) from the chipset -> cpu and in actual fact they could do this no problem if they wanted to?

I ONCE heard of a 115mhz fsb o/c on an AMD site.

Check www.amdzone.com and www.overclockers.com - they often discuss this.

I think FSB overclocks were just a screwup with the athlon should be ok with the P4.
Who cares anyhow the 1'st ones will be so slow in 32bit apps etc - that it's not worth considering one for at least 12+ months when they revise chipset, socket and the cpu itself (as in their roadmap) then perhaps the 1.7ghz &quot;p4-B&quot; (shrug?) will be capable of 2ghz or some such?

Have to wait and see.

 

[MikeyP]

Actually hitting 115 FSB for Athlon's is not rare at all. In special circumstances they can get to around 120mhz. I would put the average at around 110. It has been reported that AMD has intentionally crippled the EV6 bus, so that it is difficult to get above 110mhz. This seems to be supported by Japanese overclockers whom heavily modified the board and clock generator. I believe they were able to hit around 155 mhz DDR. I believe AMD crippled the bus also. How could they immediately release a chipset that runs at 133mhz when nothing the currently have can do it reliably? Simple, &quot;uncripple&quot; it What was the point of this rant? I don't know, but I don't think this is the end of overclocking.

Oh yes:


<< A 300 MHz overclock on a 600 MHz Duron is 50% where as that same overclock on a 1.4 GHz P4 is 21%. And as CPUs get faster that increase will get smaller. For now overclocking is a very good thing, but I'm not sure how long it will be practical. >>



Unless companies intentionally lower yeilds, I don't believe it will hold true. I mean, look at the celeron 300a. It was amazing because it could hit a 50% overclock. After this came the 366. How many chips are on the market that can easily do the same thing? Currently, there are several: Duron 600, Celeron 533, Celeron 566, Celeron 600. And in some cases the Duron 650, and Celeron 633's do the dance as well.

If companies do not intentionally cripple the overclocking effort, it should remain a valid resource for the &quot;more bang for your buck&quot; crowd.