i was thinking, if one of you had te money.....

[funnyperson1]

well most of us have seen benchies showing that a PIII wwith DDR is a waste ofmoney because of the low fsb....so iwas thinking, what if someone with extremely good cooling could take say a TUA266 and witha Celly 1.0A to a 166 bus (1666mhz) then the benefits of DDR could proobably start to be felt, and i think such a chip would handily whip a 1700+ in most anything and be almost on par with an XP 1800+......or lets say Ol' Man gets dangerous again:D , takes off his IHS, uses Dry ice and gets a Celly 1.0A on the same DDR board to 1.87 as he did with the Celly 1.2......how do you guys think that would compare to a 1900+, my prediction is WHOOPA**...

[MilkPowder-2]

I agree with u there.. :D

Yeah the 1.0A to 166fsb would be the killer.
If one can manage to OC the 1.0A to 1.87Ghz, which would be CRAZY!! i.e. dry ice cooling the CPU and if the RAM could handle to that insane level of FSB, this would be come a memory bandwidth monster and would rape XP's w/DDR out there. :D

[funnyperson1]

well right now they have PC2700 Kingmax ram that means 166 isnt even overclocking the darn thing, and soon i think theyll have PC3000....

[Yodums]

Thats nuts...

*Pulls his whistle out*

Yells "ol'man where are you???"

[Flu!d]

I don't believe the P3 DDR based boards allow for many overclocking options....Thus limiting the chips potential....

[oldfart]

I remember reading there is something in the PIII architecture that prevents it from benefiting from DDR. Its not bus speed. It was something else.

[funnyperson1]

Originally posted by oldfart
I remember reading there is something in the PIII architecture that prevents it from benefiting from DDR. Its not bus speed. It was something else.

really, im pretty sure it was the 100.133 bus speed...

[ol' man]

Actually the PIII's and cel-t run a dual bus so they could possibly benifit from the DDR ram but I think the 815e is the problems? Powerleap should make a adapter to run the cel-t on a i845D ddr setup :D

[oldfart]

Its not the bus speed. P4's run on a 100 MHz bus. AMD runs on a 100 oe 133 MHz bus. Its the way that memory is accessed that prevents a gain going to DDR.

[pacino]

Hrm, way too tempting LOL! .. I was thinking to wait for the new p4's S-378 along with DDR and see how that would perform with OCing.. but this project looks so cool..

[Yodums]

Originally posted by pacino
Hrm, way too tempting LOL! .. I was thinking to wait for the new p4's S-378 along with DDR and see how that would perform with OCing.. but this project looks so cool..

Not really a project yet no one is willing to try it :p

[pacino]

I would if someone proves to me that p4 S378 sucks (or would suck) in O/Cing LOL!

[Yodums]

Originally posted by pacino
I would if someone proves to me that p4 S378 sucks (or would suck) in O/Cing LOL!

But they are going to be really expensive upon release its going to take sometime for their prices to drop.

Like rambus people were so disapointed when the ram prices of RDRam dropped so much.

[funnyperson1]

Originally posted by oldfart
Its not the bus speed. P4's run on a 100 MHz bus. AMD runs on a 100 oe 133 MHz bus. Its the way that memory is accessed that prevents a gain going to DDR.

the athlons hav e 100 mhz DOUBLE PUMPED bus, just like ddr, its not really running at 200mhz, thats just the effective speed.....with this proposal of mine the bus speed would really be 166mhz....

[theflyingrat]

I'd go and try it - the 1.0As do seem tempting (low multiplier) for this sort of project, and I was thinking about the benefits of DDR on a CPU like this.

The reason PIIIs and Celerons never benefited from DDR is not the memory bus speed, it's the CPU bus speed. Whereas the CPU bus on a Thunderbird is actually 200 or 266, as opposed to just 100 or 133 on the Intels, opening up more memory bandwidth will help the CPU. Same for a P4 - the CPU bus is quad-pumped; the memory bus is NOT!!!! So with basic PC100 memory, the CPU is able to pump four times the information that the memory bus is capable of. Which is why you see such huge gains from using RAMBUS on P4s.

Now, even if you increased the FSB to 166, the CPU bus's peak bandwidth is increased 66%, but so is the memory bus's bandwidth. Since there is PC166 available for this, no more is needed. DDR is still wasted; it's still twice as much as you need.

Assume the Celeron runs at 100 MHz FSB. We can assume for this minute that the memory is running synchronously with the CPU bus...100 MHz also. Now, let's double the memory bus capabilities. The CPU was able to exactly saturate the FSB before, but this does not increase with the CPU's capability to push more to the new 200MHz memory bus. Half of it goes to waste. This is what will happen with any Celeron or PIII design. The CPU isn't able to use a DDR memory bus' capabilities because the processor's CPU bus can not put information there fast enough.

This would be differrent if the Celeron or PIII's CPU was designed in a similar fashion to the Athlon/Durons' EV6 DDR bus (200/266 FSB capability) or the P4's "Quad-pumped" FSB. Coppermine/Tualatin pretty much have what we can call a "single-pumped" FSB. The CPU's bus can't push any more information through it's memory bus than a plain ole' SDR memory solution can provide for.

[funnyperson1]

yup if i didnt have the iwill and memory prices hadnt jumped back to 704 i might of tried this...

[theflyingrat]

From Crucial.com on 1/6/02:

256MB PC133, CL3 - $56.69
256MB PC133, CL2 - $60.29
256MB PC2100, CL2.5 - $71.99

Wow....even at Crucial, SDR is already catching up...It was $34 before the holiday season! :eek:

Well, seems that the days of $20 PC133 and $30 PC2100 are over..at least for now. Looks as if memory manufacturers are getting sick of taking the losses....

[funnyperson1]

Originally posted by theflyingrat
From Crucial.com on 1/6/02:

256MB PC133, CL3 - $56.69
256MB PC133, CL2 - $60.29
256MB PC2100, CL2.5 - $71.99

Wow....even at Crucial, SDR is already catching up...It was $34 before the holiday season! :eek:

Well, seems that the days of $20 PC133 and $30 PC2100 are over..at least for now. Looks as if memory manufacturers are getting sick of taking the losses....

yeah well i got lucky i got both the sticks in my sig for 20$ total at thanksiginving at best buy, and there were micron too....:),....

[theflyingrat]

Hah! A friend of mine did that same thing...his has a Crucial sticker on it! At Best Buy??? Travesty! Looks good...his is running 144 at CL2. Good stuff right there.

Anyway, has anyone actually heard of anybody else using the TUA266? Does it have as good of overclocking options as an ST6 or a GA-6OXE-T? I've only read two reviews on this board, and they were mediocre reviews at best....and, of course, Asus' web site doesn't have poo on it. You are right, though....it would be interesting to find out how DDR would affect it at high FSB speeds. My guess is not much, but at least with the TUA266, you could compare directly PC2100 and PC133 on the same motherboard, to eliminate variables (i.e. different chipsets/motherboard manufacturers' implementations of said chipsets, etc.) If I get the job I'm reviewing for on Tuesday, wait a couple weeks...I might have me a Tually system! :)

[funnyperson1]

anand did a review on the CUV266 and he liked it......so im guessing the TUV266 should be good as well....most modern ASUS boards have very god overclocking options....though the TUSl2-C does not have 1/4 dividers below 133mhz, thats okay since that shouldnt be a problem with the 1.0A

[Phil]

theflyingrat hit the nail right on the head, it doesn't matter what you take the fsb up to, ddr in the gtl bus is a waste of time, at 166mhz the ddr memory would then be running at 33mhz and still be too fast for the gtl bus, it would barely be faster than pc166. With the athlon the fsb used to provide more bandwidth than sdram could provide, and with ddr ram they are level pegged, just like sdr ram is with the gtl bus.