More Consideration

[FatBurger]

I've seen Crucial PC2100 Cas2.5 at 160MHz Cas2. It could probably get 166, I think the board or AGP was holding it back. That was a while ago.

<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>

 

[lhgpoobaa]

bah... cas 2.5 is for whimps.

i run at the hardest possible memory timings or i get new ram.
i did some 'cure for cancer' number crunching benchmarks a while ago, and the difference between 3-3-3-6 and 2-2-2-5 on my sdram board was spectacular.


Overclocked athlon 1200C @ 8.5 x 166FSB + PC2700 = GOOD!

 

[AMD_Man]

Crucial RAM can handle some pretty aggressive settings but at 166MHz, I doubt it'll handle the maximum.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor

 

[eden]

Hmm 160MHZ at C2 is still pretty nice, I mean it's only 6 MHZ less!
If AMD_MAN can do that, it'll be well worth it!
Also how is the P4 handling asynchronous FSB with DDR if it ain't like RDRAM and starts at 133MHZ instead of 100MHZ? Is it that why it doesn't couple well?

--
For the first time, Hookers are hooked on Phonics!!

 

[FatBurger]

That could be, but the biggest problem with asynchronous memory is that the memory is giving more bandwidth than the CPU can use.

<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>

 

[AMD_Man]

Actually, at 333MHz, the P4S333 is on par with RDRAM on most tests.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor

 

[eden]

Yes but those tests were most likely not perfectly at Cas 2, I am sure of it...?

--
For the first time, Hookers are hooked on Phonics!!

 

[eden]

Why so?
Isn't that when Bus and Ram are on par? If so what is the problem in it?

--
For the first time, Hookers are hooked on Phonics!!

 

[AMD_Man]

No, 333MHz CAS2.5 is roughly equal in performance to RDRAM.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor

 

[eden]

Well if that's so why are you still rebutting for no reason man? Get the Corsair and you got yourself Cas 2, an ever better alternative for high performance in case you reached 2.1GHZ and above! Besides, RDRAM becomes a royalty case when you move to 512MB sticks and above.

--
For the first time, Hookers are hooked on Phonics!!

 

[AMD_Man]

True, true, but the only good overclocking boards that support DDR that I've read about are the ABit BD7 (which only supports up to PC2100, but includes all the overclocking features of the ABit TH7-II) and the Asus P4S333. Fugger, doesn't seem to like the P4S333 for some reason. I don't know if that's cause enough for concern or not.

Slightly off topic, but for now, I'd like to quiet down my PC. My GlobalWin WBK38 is an excellent cooler but it's too loud. I want something that can cool just equally well but quieter. I've considered the Alpha PAL8045 + a low-speed Sunon, but if I finally decided to continue going the Athlon route, not all motherboards support it. I'm looking at either getting the Volcano 7 (with the thermal pad that's already on it), or the Volcano 6CU plus some Arctic Silver 3. Which do you think will cool better?

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor

 

[eden]

Heh, read my post in the topic Athlon XP's Heat Controls. I had answered you but I think you didn't read it. It will explain to you all about my experience with the Volcano 7, and it is an indeed awesome HSF.
On a side note, the new V7+ allows you to control the fan speed from HI-6000RPM, Med-4800RPM and Low-3100 RPM. Personally I would like a more customizable LOW speed, as 3100 RPM is way too low for cooling and although it's silent, I'd like some temps that show how much hot it will be. It's nice to see TT making some really helpful HSFs like these. The V7+'s best advantage besides ultra-thin Heatsink fins is also that it can convert itself from Socket 462 to 478 and vice-versa!! Meaning a versatile 2-in-1 combo that will satisfy any Intel-AMD user like you in the near future! I wonder how its price will be here in Canada, but I paid only 39$ for the V7, and the V6CU+ is 39$ too! How odd...

--
For the first time, Hookers are hooked on Phonics!!

 

[AMD_Man]

I wonder how its price will be here in Canada, but I paid only 39$ for the V7, and the V6CU+ is 39$ too! How odd...

That's extremely expensive! You live in Canada right? At <A HREF="http://www.canadacomputers.com" target="_new">Canada Computers</A>, they sell the Volcano 7 for $29 and the Volcano 6CU and 6CU+ for $19. Arctic Silver 3 there is $12.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor

 

[eden]

Wow!
I didn't think they were this low!
I dunno man, I'm not much into HSFs to know which place sells lower prices!

--
For the first time, Hookers are hooked on Phonics!!

 

[FatBurger]

Synchronous is when things are together, asynchronous is when they aren't. An example of asynchronous is the P4 at 100MHz and RAM at 166MHz. Synchronous would be the P4 at 100MHz and the RAM at 100MHz.

<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>

 

[juin]

Slightly off topic, but for now, I'd like to quiet down my PC. My GlobalWin WBK38 is an excellent cooler but it's too loud. I want something that can cool just equally well but quieter. I've considered the Alpha PAL8045 + a low-speed Sunon, but if I finally decided to continue going the Athlon route, not all motherboards support it. I'm looking at either getting the Volcano 7 (with the thermal pad that's already on it), or the Volcano 6CU plus some Arctic Silver 3. Which do you think will cool better?

It also off topic

The alpha MXC478 is out for Intel P4 socket 478

http://service.madonion.com/servlet/Index?pageid=/orb/projectcompare&projectType=6&projectId=2310900

 

[Raystonn]

A 2.5GHz overclock of a 1.6A would require the PC800 RDRAM to be overclocked to 1250MHz keeping the memory synchronous.

Normally you run with a 4X RDRAM multiplier, set in BIOS. When running on an external clock that is 133MHz or greater you can flip this to a 3X multiplier and not lose any memory bandwidth. Your RDRAM would be running at 469MHz, which is greater than the default 400MHz and is attainable with any old 400MHz DRCGs. The memory bandwidth offered by such a solution would be 3.752GB/s. The memory bandwidth offered by any DDR SDRAM solution for the same external clock and FSB would be less.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =

 

[Raystonn]

Oh really now, cause raystonn swears up and down ALL samsung ram will hit 1066.

This used to be the case. Samsung has since introduced new versions of memory that are high density single-sided modules. Being a newer technology, and more dense (which means more heat), these particular modules will not overclock very well. If you want PC1066-capable memory, make sure you get Samsung's 128MB or 256MB double-sided PC800 modules. The low density (double-sided) modules are the more plentiful kind at the moment, and used to be the only kind available at all.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =

 

[Raystonn]

just like how Intel and AMD rate their procs - test a sample of a batch, see how high it will reliably operate and mark a speed on the whole batch.

This is not how it works at all. Every processor is tested individually and has passed at the speed at which it was rated.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =

 

[ath0mps0]

Normally you run with a 4X RDRAM multiplier, set in BIOS. When running on an external clock that is 133MHz or greater you can flip this to a 3X multiplier and not lose any memory bandwidth.

But, at this point you lose synchronosity which drastically reduces the efficiency of the memory bandwidth and memory bandwidth starving the processor. You will still see performance increases with major overclocks, but on a smaller scale than with synchronized CPU & RAM. This problem is exaggerated with the P4 as an asynch config has the FSB faster than the RAM, so the CPU ends up stalling while waiting for RAM. With Athlons you have the opposite effect in asynch situations, RAM faster than the CPU can use. Either way, you lose performance per clock.

Your RDRAM would be running at 469MHz, which is greater than the default 400MHz and is attainable with any old 400MHz DRCGs.

But to maintain synchronocity you would need to clock the RDRAM to 625MHz (1250 effective - yeah, RDRAM is DDR as well) to attain 2.5GHz on the proc as I stated in my previous post. This is not attainable by any RIMMs I have ever seen and probably won't be for a year.

The memory bandwidth offered by such a solution would be 3.752GB/s. The memory bandwidth offered by any DDR SDRAM solution for the same external clock and FSB would be less.

Correct, with current chipsets and motherboards (and why a NW 1.6A RDRAM based system makes good sense for the moment).

Dual-channel DDR-SDRAM (yeah, yeah, I know - beating the dead horse) would provide more bandwith. PC2100 would do 4266MBps and PC2700 would scale to 5333MBps memory bandwidth all without even stressing the RAM - allowing the memory to stay synchronous. Imagine the higher benchmarks and frame rates!

If you don't believe the performance hit you're taking because Intel doesn't want to admit they are wrong with RDRAM, try it:

Push your RDRAM based P4 system's FSB to 133Mhz (533 effective) and run your benchmarks. Now change your 4x FSB/memory multiplier to 3x and run them again. How much performance did you lose - 10%? More? That's about the same percentage as what you're losing by going asynch to reach 2.5GHz. Yes you still gain performance, so it's worth it, but wouldn't it be better to get the full potential of the chip?

I thought a thought, but the thought I thought wasn't the thought I thought I had thought.

 

[Matisaro]

This is not how it works at all. Every processor is tested individually and has passed at the speed at which it was rated.

-Raystonn

ray is right, each proccessor is tested individually.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!

 

[ath0mps0]

Actually we are both right. The batch is sampled and slotted for a specific speed as I stated. Then, each processor is stress tested to verify stability at this pre-determined speed before being locked in. Each processor is not stressed to determine its maximum stable speed.

At least this was true the last time I toured Intel's plant here in Hillsboro, OR.

I thought a thought, but the thought I thought wasn't the thought I thought I had thought.

 

[AMD_Man]

ath0mps0, you have to ask yourself: Are you really <i>losing</i> performance? No, you're not; you can't lose something you don't already have, in fact, you'll gain performance by this type of overclock. The FSB alone will provide a marginal performance boost and the Northwood will provide a significant performance boost. The Northwood has 512KB cache now which means less cache misses and less access to RAM, so as you increase clock speed, the difference in performance will continue to be significant for quite some time. A 2.2GHz P4 runs with PC800 and 400MHz bus, right? A 2.133GHz P4 runs with PC800 and a 533MHz bus? Both will probably be roughly equal in performance due to the higher clocked FSB. Let's not forget that the FSB is the connection between the northbridge and the CPU. The Northbridge has other things to do than just manage memory traffic. It has to relay data from the southbridge to Northbridge as well, among other things. Take this scenerio: memory traffic is at it's maximum, and the FSB is running at the same speed as the RAM. And then, a PCI care or AGP card requests to send data to RAM or to the CPU. It'll have to wait until free cycles are available for the FSB, the CPU isn't the limitation here because it can process information significantly faster than the main memory can provide it. With a faster FSB, wouldn't the FSB be able to manage saturated memory traffic as well as be able to relay messages from the southbridge to the CPU, and thus wasting less cycles.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor

 

[scamtrOn]

i can think of one thing that i can lose that i don't have.

and no not that.

<A HREF="http://www.geocities.com/scamtron2000/Lochel.html" target="_new"> <font color=red>go to my site</font color=red> </A>

 

[AMD_Man]

i can think of one thing that i can lose that i don't have.

and no not that.

And what is that?

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor