holy moly!300 fsb right off the shelf!

Looks like a 300 fsb aint gonna be a prob with the 875-865 mb's. now that's crankin!!! http://www.vr-zone.com/reviews/EPoX/4PCA3Plus/

Take a close look. It's a unlocked CPU running at 12x instead of 15x. Still good, but don't let the MHz fool ya.

a 2.4 800 fsb will have a 12x

flapperhead said:

a 2.4 800 fsb will have a 12x

Click to expand...

You are correct, sir. The latest rumor is that Intel will release the 2.4C and 2.8C (800 MHz bus CPUs) in the US by May. The 3.0C is already now available (but dang expensive, $530 at newegg.com).

Im drooling at the thought of runnin one of those babys with a 300 fsb...

EPoX 4PCA3+ is able to attain a high FSB of 297MHz with memory running at 238Mhz FSB using 5:4 CPU/DRAM ratio

Click to expand...

The memory is only at 238. The CPU FSB is at 297, or 297 x 4 for 1188.

Also,

Definitely, when Pentium 4 2.4CGhz is available in the market, it is the best choice for EPoX 4PCA3+ board since it has a multiplier of 12x too and FSB above 300Mhz can be attained when good cooling is applied.

Click to expand...

They did not reach 300. They said it is possible though. You need memory that can hit 240 and then you get the CPU FSB to 300 with that 5:4 divider. This is what they are talking about.

youre right but faster memory will arrive... and a 6200 mem score aint chicken feed!!!

I totally agree. The scores are screaming on the new boards.

Hmmm but isnt it a total waste that the mem is at 240 when the cpu is at 300.. i mean thats a HUGE bottleneck

if the mem were running faster, say 340mhz compared to the 300 fsb, then it would have reached its full potential. P4s thrive under conditions where the memory runs faster than the fsb, in contrast to AMDs where its optimal to run them at a 1:1 ratio

Dual channel ddr is a perfect balanced memory designed by intel. A 800 fsb with dual channel ddr at 400 mhz.. But of course us oc'ers aint happy with that, and theres the quandry.As we crank up the fsb the the current memeory just can keep up.. Id love to see what kinda score we'd see runnin the fsb @ 300 with the memory 1 to 1.. But again 6300 bw is still impressive to say the least....

paymon said:

Hmmm but isnt it a total waste that the mem is at 240 when the cpu is at 300.. i mean thats a HUGE bottleneck

if the mem were running faster, say 340mhz compared to the 300 fsb, then it would have reached its full potential. P4s thrive under conditions where the memory runs faster than the fsb, in contrast to AMDs where its optimal to run them at a 1:1 ratio

Click to expand...

Update your thinking. This is dual channel; now you have two memory sticks feeding the bandwidth.

stroligo said:

Update your thinking. This is dual channel; now you have two memory sticks feeding the bandwidth.

Click to expand...

ed,
how would you respond to this?

a 2.4c (800bus)
at 290fsb you'll have a 1160 bus cpu....but you'll have to use the 3/2 ratio on the memory we have now.

that gives you 773 memory bus.....a 33% drop in bandwith.

no thank you very much.

I don't need no 12 multi cpu running with that much of a drop in memory bandwith.

give me a 2.4b that runs 185fsb and full bandwith with great timings....thank you.

18 multi, 740 bus...740 on the mem=3330mhz goodness...
that will kill the 2.4c in real world benchies.

mica

Mica, if you only look at memory speed, then you are right. But there is more to this than meets the eye.

I'm assuming you are comparing a 533 bus CPU and a 800 bus CPU used on the same Canterwood mobo. Things get really confusing if we were comparing a regular DDR mobo with a dual channel DDR mobo.

Total CPU clock speed, system FSB, and memory bus speed all play a role in determining bandwidth and memory performance.

Since I don't have access to a 533 bus and 800 bus CPU yet (I hope to in a couple weeks) it's hard to give you real data for a head to head comparison. But, the new Canterwood chipsets seem to very efficient in the memory bus department and very dependent on high FSB.

Let me just give an example taken from the Abit forum. A guy achieved a 14X240=3360 overclock using a 800 bus ES CPU. The memory ratio was set to 1:1 for DDR480 memory speed. Sandra memory benchmark scores were integer buffered=5449 MB/s and floating point buffered=5354 MB/s (WOW).

http://forum.abit-usa.com/attachment.php?postid=58248

http://forum.abit-usa.com/attachment.php?postid=58249

Next the guy changed the multiplier to 12X and ran the FSB up to 280 to give a final CPU clock of about the same speed as the first test (12X280=3360). The memory ratio was set to 3:2 for DDR375 memory speed. Sandra memory benchmark scores were integer buffered=5411 MB/s and floating point buffered=5397 MB/s (which is almost exactly what it was in the first test).

http://forum.abit-usa.com/attachment.php?postid=58280

http://forum.abit-usa.com/attachment.php?postid=58282

So, same clock speed in both examples, different FSB speeds, and different memory bus speeds, but close to the same memory bandwidth benchmarks. The lower multiplier and higher FSB in the second test made up the difference in having to run the memory bus at a lower speed (DDR480 vs. DDR375). What does this say? The same thing overclockers have been saying for years, high FSB improves performance.

Re: mem speed vs CPU speed. I've done a bunch of posting on this subject @ Anands. Here are some points:

Lets call DC DDR 1:1 "100% BW" and see how other chipsets compare:

875/865/7205 DC DDR 1:1
100%

845PE 3:4
66%

845PE 4:5
62%

845PE 1:1
50%

875/865 5:4
80%

875/865 3:2
66%

As you can see, running a 875/865 @ 5:4 has more BW than any SC DDR setup. Even dropping down to 3:2 has the mem BW of a 845PE running 3:4 and is higher than 845PE 4:5. In real app benches (not Sandra mem), there is ~ 3 - 6 % difference in performance between 845PE 4:5 and DC DDR 1:1. Not that big a deal.

What do you gain by dropping the mem ratio? Large increases in FSB and CPU speed. Unless you have some serious quality memory, you will hit a wall in short order when overclocking 800 MHz P4 "C" CPUs. Drop the mem ratio, lose a tiny bit of mem performance to gain very large increases in performnace due to higher FSB and CPU speeds.

yep. good examples, good logic, well explained

they did with amd , now coming intel.
wow. nice epox