iYAN inside! Online                                                                                  Index Hardware

 

The K7 hype

AMD Athlon

The rumours and speculations about AMDs upcoming K7 are growing everyday. The K7 will be a beast, a killer, according to many. And it probably will be. A 200MHz bus, based on the EV6 Alpha protocols is ahead of its time, while Intel still uses 100 and 133MHz FSB. But... the question is whether this highspeed bus will highly increase performance. And... will there be PC200 equivalent RAM?

Honestly, I just don't like all of the hype revolving around the K7.  But don't get me wrong... I am not an Intel loyalist, I actually own an AMD system as well as stock in the company (albeit a very small amount).  I would like nothing more than the K7 to do well.

Many are making a really big deal about the 200MHz bus used for the upcoming K7 system.  The EV6 bus is superior to current offerings, I use an Alpha at work and wouldn't trade it for any PC.  However, most people hype the bus speed as opposed to the actual bus protocol  The actual increase of the bus speed won't effect the performance of a system based on the K7 processor.

Everyone likes to point to the increase in performance of the K6-2 based systems when they moved from the 66 to 100 MHz bus speeds.  However this is irrelevant with the K7 because its cache is already on the daughtercard of the processor and runs at a fixed rate ( 1/3 the clock speed on initial release, upto 1/1 the clock speed later on).  The peripherals are also going to run within their normal specs.  To most components in a system it doesn't matter if the bus is 100 or 500MHz.  The only part of the system that benefits from the faster bus speed in this case is the ram.  However, RDRAM more than likely won't be available at the time of the K7's release, therefore it will use the standard PC100 SDRAM.   These systems may be well built enough to pair SDRAMS so that the system can still read memory at 200MHz, but that is a big maybe. 

Another thing to consider is how much of an impact increasing the bus speed has on the overall performance of the computer.  When the BX chipset first came out, people with 300MHz PIIs used the 100MHz x 3 setting to compare the results.  In most cases the gain was less than 2%.  Increasing by 100% may give a maximum gain of 4%, but that is not a significant real world performance increase.  The K7 will be an excellent processor, however, I just don't like people using insignificant details to hype its debut. 

How many times have people been disappointed when testing a new product that had been hyped for months (does TNT ring a bell).  It will be an excellent product, but it won't be the giant killer everyone is hoping for by June.  Remember the Katmai will already have been out for a few months and support for KNI will be at least partially implemented.  The K7 will also initially ship with a 1/3 divider for the cache speed (K7--1/3x500=167MHz , Katmai--1/2x550?=275, both MHz ratings are estimates, the K7 may ship with a higher MHz rating and the Katmai might be higher or lower by next summer). 

Another thing to consider is cost.  The K7 won't be the consumer friendly product everyone is used to from AMD.  It will cost at least as much as the Katmai.  AMD caters to the needs of the low-cost solution.  However, Intel is finally beginning to catch on to what consumers want (which could be quite scary for AMD, as intel has the resources and power to give the consumers what they want, they just haven't had to before now).  Take for instance the Celeron 300A.  Right now you can buy that chip for the same price as the K6-2 300.  Intel has finally decided not to take it lying down. 

I expect some major number jumping for the Celeron line of CPUs.  Why?  Because they can handle it.  Look at hardware tech sites at the overclocking articles on the 300A.  The malaysian chips have almost a 100% success rate at 450MHz.  A faster cache and a die shrink could put the celeron up to 600MHz with ease.   

I hope Intel does just roll over and give up.  My 200 shares of stock in AMD would skyrocket and I would really enjoy it, but I doubt that will happen.  And let me end with my biggest fear...

"Too much hype = Too little performance"