Slashdot Mirror
Intel 800 MHz FSB Processor Family Review
David writes "Techware Labs recently had the opportunity to spend some time with Intel's new 800 MHz front-side bus (FSB) processor family. The review includes a overview of the features in this processor family, Intel's new Springdale and Canterwood chipsets, and an analysis of processor scaling within this family. The article focuses on how the relationship between CPU and video card affect various aspects of performance."
The more powerful the chips intel pushes the less effcient the coder becomes, i remember when i used to tweak my programs so they would run optimally on a slower machines
Yeah, I said that too when the PII came out. Sure there is always going to be bloat in code, especially in large projects. But you are more than welcome to go to ebay and get an 8088 or an Apple II and enjoy a machine that fits your computing needs (floppy drive or tape drive your pick).
Me, I would like to have a computer fast enough to do things like audio/video editing, real time ogg encoding, or whatever. I surely would not mind buying a computer today thats 4x faster than these new P4s for about $1000. I'd find a need for it or enjoy the lack of bloat feeling, who cares?
Although I have had 0 formal training in programming, one thing I've read and have incorporated into my coding is early optimization == bad. 1st write good code, then find out where the bottlenecks are (if any) and then optimize those bottlenecks. There are even great profiling tools out there to help you do these things.
What good is benchmarking the new P4-C processors without comparing them to Athlon XPs, or even older P4s? Really, you can just multiply the performance of a P4-C 3.0Ghz by 0.8 to get a guesstimate of the performance differences within the family; what really matters is how they perform in comparison with the competition.
In all my rantings I hardly mentioned the FSB issue when did buss speed become more important than processor speed? but maybe thats just intels way of taking the light off of the idea that they are having a little trouble pushing thier chips faster, and thats to be expected when you cant make it faster make it better. too bad all FSB gets you is 5 FPS in your precious GAMES (yea I play em too!).
oh well I guess I am just too demanding . . .
What is really missing in the article is the comparision betwoon other cpu's running at 533 and/or 400 MHz. How can one interprete the benchmark results if there is no comparison to another product ? It's like saying that something is 600.1 gigaquats without defining a gigaquat.
Some people need all the processing they can get. It's allowing some of us to do things we couldn't before. If you don't need it, don't buy it. And you can always stop buying bloatware and write your own software or optimize your favorite OS program.
Ahh the 266*2 was wishful thinking :). You are right about its 133*4, apologies about that.
:)
My understanding is that xeon mp line is for their
4-way based motherboards. The main advantage is they have a meg of cache on them. But the normal processorshave 512k the same as the new p4's I believe.
The xeon mp motherboards are $2k and the processors are about $2k each (pricewatch 1.6ghz/1meg cache i.e made of gold
In any case the normal xeon dual systems are actually not that much more than buying a 875pe
motherboard and processor. Btw here is the road map I found on the inquirer. Apparantly the xeon mp's are going up to 2.8ghz/2 megs of cache and the normal xeons are going up to 3.06/1 meg of cache and selling for $700.
Here's the weird part, while it looks like intel skipped 667 fbs for the PIV line, the xeon line will "ramp up" to 667 early next year.
In anycase I'm probably going to build a "normal" xeon/iwill running at ~2.66 which comes out to really not much more than a normal PIV/865/875 series. The selling of 800mhz memory/bus speeds on the PIV line while keeping the xeon line at 533/667 makes no sense to me. I was going to wait until a new set of mbs/chipset came out for the xeons but it doesn't look it will happen.
-bloo
Read the article at anadtech. It's the roadmap for Intel. And discusion of all the processor currently in the market. They Discuss why the Xeon isn't getting the nice FSB upgrade even though they need it the most.
Multiprocessor bus speeds and CPU frequencies always lag behind uniprocessor systems. It takes much longer to validate multiprocessor boards when compared to a uniprocessor system. This is because the number of things that can go wrong goes up exponentially with the number of CPUs on the board. The typical customers of multiprocessor systems value this sort of reliability even more than performance.
Pages and pages of pretty graphs and charts all to tell us that yes, higher clock speeds mean higher performance.
Bleh!
This word, "overclocking", I do no believe it means what you think it means.
That was a raw 50MHz chip, no overclocking, that outperformed the 486DX-66 with ease.
Hate to break the news to you, but pretty much every CPU in use today uses exactly the same asymmetrical bus/CPU that the 486 DX2s did, that you are calling "overclocking".
Oh, and a DX/50 would only outperform a DX2/66 in tasks that were bus-bound. If the bus wasn't a limiting factor, the higher clocked CPU would be faster.
The supporting hardware was too expensive, and Intel never bothered working on making it cheaper.
Yes, they did - it's just that by that time (ca. second generation Pentium CPU - P90, P100, P100 and P133) the 486 was obselete.
This is a tougher fight, and it's one Intel is losing, every time someone produces a faster processor or chipset. (Or even just a more reliable one.)
I've yet to see any manufacturer produce a more reliable chipset than Intel. Even the ones that are faster only tend to be so because they lobbed into the middle of an Intel release cycle and aren't outperformed until the next model (in general).