In case you dropped in from an external link, this page is continued from Page 2 of the Radified [Guide for the Intel Northwood Pentium 4 CPU].
Intel is planning to raise the FSB (Front Side Bus) to 533MHz (quad-pumped 133). The scheduled release date for FSB-533 chipsets is (reportedly) set for May 6th (this date keeps getting pushed back), at which time Rambus modules will run at 1066MHz (referred to as PC1066 .. does your head hurt yet?) Notice that there is still a factor of *2* between FSB clock-rate and the speed/frequency of RDRAM modules.
People who are serious about performance seem more excited about the
prospect of 533-FSB than the (overclockability of the) Northwood
CPU itself. If you can hold off until May, the performance available
at that time should make the wait worthwhile, altho many people are
overclocking and running at these speeds now.
I also read that latency
of PC1066 modules will be cut by 30% (lower is better/faster), making
it roughly the same as DDR RAM .. which would be yet another reason
to wait for 533-FSB. After the recent price increases of DDR RAM, latency
(a measure of 'delay') is the only remaining performance advantage that
DDR RAM holds over RDRAM.
Most PC manufacturers tend to classify (primarily) their systems by the CPU it contains. This practice is misleading, if you agree that the CPU is not the system's single most important component. Choosing your chipset first, before you select a CPU, if often a wise strategy.
For this reason, I'm considering waiting until spring, even tho the Northwood will be here January. But you can also end up waiting forever. I heard that Intel will designate CPUs designed for the 533MHz FSB by replacing the letter 'A' with the letter 'B'.
Don't expect Intel to be very talkative about (marketing) the move to a 533MHz bus speed, as this would give people a reason *not* to purchase the original-release Northwood P4 (based on 400MHz FSB), and wait for spring.
Word on the street is that Intel will release two new CPUs (2.26 and 2.4GHz) in the mid-April to end-of-May time frame that will be the first designed to run on the new 533-FSB chipsets.
There's nothing wrong with AMD CPUs, especially for the average user. I'm all for AMD. I even recently purchased some of their stock (I'm sad to say). The competition helps keep Intel honest, and I readily admit that AMD chips provide better value than Intel's. But value is not my #1 consideration.
I found that, when you try to cram a lot different hardware and software into a system, the potential for compatibility glitches grows proportionately with the installation of each additional component & application.
Personally, I'm far more interested in stability .. than saving a hundred bucks. The headaches aren't worth it (for me).
Without engaging a debate about merits of relative platform stability, I'll simply reference a recent article posted at Tom's Hardware (dated 31oct2001). The article is titled AMD vs Intel. It begins here.
I call your attention to the Conclusion, posted here. Three paragraphs from the end, it says this <copy-n-paste>:
Another factor is the stability and product quality of a system: while all Athlon processors suffered from occasional instability in our tests, the Pentium 4 platform ran without a glitch.
Reasons for this behavior might not lie in the processor itself, but rather in the motherboard design and the chipset used. Future driver updates might not just improve performance but also stability of a platform. </paste>.
While you're there at Tom's, you might as well mosey on up two paragraphs and read where he says <copy-n-paste>:
But one thing should be made clear here - since Intel introduced its 0.13-micron processor, the Northwood Pentium 4, it can turn the MHz dial up higher than AMD can.
High clock speeds are a real burr under AMD's saddle, as has been shown by the developments of the past several months. While Intel has already cleared the 2000 MHz hurdle, AMD has barely scraped over the 1533 MHz one from 1400 MHz. </paste>
If these comments come from a guy who's known to be pro-AMD, what might that tell you about relative platform stability?
Update 04nov2001: I should've known that I'd get a mailbox full of responses to my comments regarding AMD vs Intel. (This is why I was reluctant to address the issue.) For example, one reader writes to say <copy-n-paste>:
This quote: "High clock speeds are a real burr under AMD's saddle, as has been shown by the developments of the past several months. While Intel has already cleared the 2000 MHz hurdle, AMD has barely scraped over the 1533 MHz one from 1400 MHz." .. is of dubious value, because MHz is not everything.
CPU power depends on both MHz *and* instructions-per-cycle, and a few other things. Apple refers to this as the Megahertz Myth. </paste>
This reader references an article posted here, titled Pentium 4: In Depth. I skimmed thru it, but it's too long & too technical for me to read in its entirety. It's blatantly pro-AMD & anti-P4. Here's a quote from the second paragraph:
Articles like this are not surprising, since there is obviously much at $take. <sarcasm> The subtle way in which the author uses the words annoying & miserably </sarcasm> tells me that he is not even trying to appear unbiased.
Note that this article addresses the original Pentium 4, and not the Northwood. Note also that the quote referenced comes from Tom, not me. Finally, note that I never claimed that AMD CPUs represent a poor purchase decision (they don't, especially for the average user).
I agree that the current .18-micron incarnation of the P4 has its shortcomings, especially compared to the bang-for-your-back of Athlon-based system. I merely said that P4-based systems appear to offer superior stability, and that stability is my prime consideration.
Update 27jul2002 - Nathan Pollak writes to say:
If a particular system has problems, it doesn't matter what kind of deal I may've gotten on it, or how well it performs. Again, I have no qualms about paying extra for a system if it buys me stability. (I speak for no one but myself.)
It's not like I haven't been frustrated with many of Intel's decisions. Their product roadmap has left me scratching my head on numerous occasions, making it downright difficult to purchase their products .. even when I *wanted* to!
Ideal configurations appear once in a blue moon. The last time the MHz planets were in alignment was when the P3-700 (cB0 stepping) & Asus CUSL2 arrived on the hardware scene. Hopefully the Northwood Pentium 4 will again find the MHz planets aligned. My fingers are crossed.
The next die-size drop will be to .09-micron (90 nanometers), but those chips won't aren't scheduled to until early 2003.
An article from ZDNet, titled Look P4 you leap, posted here says:
"As soon as Northwood shows up .. its small design architecture an larger cache will obsolete the Willamette chips faster than Malathion kills mosquitoes.
The future is a Northwood Pentium 4 with DDR SDRAM and that's where you're money is invested wisely for the long run. RDRAM systems, if they still exist after Intel's current contract with Rambus expires, will do better with MPEG content and video .."
Benchmarks for the Northwood P4 are posted here (GamePC). I like that they used off-the-shelf chips, and not hand-picked engineering samples. These chips yield results similar to what we might expect.
EETimes posted an article here
claiming that motherboards with Intel's DDR chipset will ship
this December, altho the article mentions nothing specific about supporting
the Northwood P4.
I'll be watching closely to see how things go with Intel's DDR-enabled chipset, and will keep this page updated with pertinent info. If all goes well, we might have a new motherboard user's guide in a few months.
See Intel's support page for its socket-478
P4 processors for info about CPU packaging, heatsink/fan
requirements, chassis/case & power supply requirements, retention
mechanism installation, & other topics related to the 478-pin Pentium
Before closing, I'd like to say that you'll find no advertising on this site. I therefore have no rea$on to parallax my views. I try to present things as I see them, and provide ample references to support my positions.
I'm interested only in what's best for me. I couldn't care less about what anyone else might prefer to put in their system. This is an admittedly selfish position. Yet ironically, you might feel the same.
My first researched-CPU was the (now legendary) C300a, which offered
(at the time) $600 performance (P2-450), when clocked to 464MHz, which
they all did quite effortlessly .. yet cost only $125.
I'd also like to mention <shameless plug> a few other Radified guides that you might find helpful. For example: