New Pentium 4 Monday

[Mike]

Intel keeps cranking out the new products don't they?

[vesther]

Unless Intel find ways to increase the efficiency per clock cycle, I don't think I would ever want to cough up $600 on a single Intel Pentium 4 Processor alone. Intel has a bad habit of persuading us that Megahertz equal performance, but it's more than that. What Intel needs to do is to improve the architecture of their Pentium Processors, increase the efficiency per clock cycle, and admit that they've been disappointing a lot of customers because it appears that for some reason, ever since the Pentium III, Intel has been rushing their newest processors onto the market.

[mstubble]

And why do they feel the need to rush their products to market? They have this little annoying creature on their tail called AMD

[plucky duck]

Is this the new processor that supports the new hyperthreading technology? SMP-able?

[vesther]

I feel that instead of increasing the amount of Megahertz in a particular processor (I don't know if Intel Northwood 2.2GHz will use Socket 478 or Socket 423 or even a brand-new socket), Intel would think about going back to the drawing board and see what they did to rush the processor into the market.

There are many reasons why this chip ISN'T REALLY Pentium 4 to my tastes. Here are my suggestions to Intel:

* The most obvious is the use of Megahertz. Just increasing the amount of Megahertz inside a Chip isn't enough. You also have to take a look at the overall architecture of the Chip.

* Another key area is that Intel does not take the time to give each clock cycle maximum effect. The more instructions a clock cycle can produce the better. Unknowingst, Intel Processors usually perform less per Clock Cycle than other processors of slower speeds. In order for Intel to compete better against AMD would mean that Intel will have to start spending money on R&D in order to find out how to maximize the effect of each Clock Cycle.

* Another noticeable flaw is the unforgiving Pipeline. Larger Pipelines might cause the processor to take longer to process instructions. Intel doesn't spend enough time to know the disadvantages of having a 20-stage pipeline. Intel really needs to work on trying to reduce the Pipeline of their Processors.

* On-Die Cache. This is less noticeable, but I'll leave On-Die Cache up to Intel.

* Lack of Optimized Instruction Sets inside the Chip. Intel needs to spend time on optimizing before releasing a chip.

[Tomteriffic]

The real Achilles Heel for the P4 thus far has been the depth of the pipeline, which increases the latency of the processor, and therefore negatively affects the efficiency. Which is why the AMD at a lower clock speed outperforms it. In theory, this was done to optimize it for programs using the SSE2 optimizations, which are precious few at present. Perhaps when those applications start to surface, we'll see some advantage, but not until then, I would imagine.