Interpreting the decreasing CPU speed grades

[Leo V]

Many years back, we had these:

486/25
486/33
486/50
486/66
486/75
486/100
486/120

Each newer 486 CPU was on average clocked 30% faster than its predecessor ([7-1])th root of [120/25]). A few years later, we had this:

Pentium 60
Pentium 75
Pentium 90
Pentium 100
Pentium 120
Pentium 133
Pentium 150
Pentium 166
Pentium 200
Pentium 233

Each next Pentium was on average clocked 16% faster than the previous. Today, we get:

Tbird 700
Tbird 750
Tbird 800
Tbird 850
Tbird 900
Tbird 950
Tbird 1000
Tbird 1100
Tbird 1200

Each next Tbird is on average clocked 7% faster than the previous. The same trend holds with P4:

Pentium4 1300
Pentium4 1400
Pentium4 1500

Somehow, having a Pentium 75, 80, 85, 90, 95, and 100 doesn't feel as different as the Tbird 750, 800, etc etc. even though the percent change is identical. A psychological conspiracy, or a result of fervent Intel-AMD competition? Are we seeing covert signs of a general slowdown in technological progress?

 

[Impact55]

It's all about the $$$.

 

[Mr.IncrediblyBored]

Very true.

Besides, perfecting die processes takes much more time now adays. Yields on higher grade chips are usually horrendous, causing clock speed increases to be nominal and slow.

 

[BenSkywalker]

New processors are released much more frequently then they were in the 486 days. AMD for instance went from a 500 MHZ K6-2 to a GHZ Athlon in about a year(inlcuding all the various K6x chips over a dozen releases in about a year). Outside of the jump from 33MHZ 486 to the 66MHZ variant we haven't seen this kind of ramp up of processors that I can recall.

The thing is now there are more smaller steps introduced, gives OEMs a CPU within pretty much any $30 price range, that certainly wasn't the case back in the day when a "budget" system with a 14" monitor would run you $2500.

 

[Degenerate]

interesting

Use a bit of maths and find a different story.

let say the speed was going like think
10
20
30
40....

the fist step was a 100% increase, the last was a smaler one

Its a bit of clear maths which shows that the speed is increasing still rapildy.

With all these competition, do you think no companies will try and ramp up thier speed fast as possible??

not me

 

[Leo V]

Degenerate, I'll argue that the sequence you've presented (10,20,30,...) actually *does* demonstrate a slowdown of progress! Imagine that a new CPU is released every other month for a total of 6 every year, like so:

1980: 10,20,30,40,50,60MHz
1981: 70,80,...
... (I'm making no pretense at realism here!!)

This year, we'd be looking at the following six new CPU's:
2001: 1270,1280,1290,1300,1310,1320MHz

Can you seriously claim that we're *still* making any progress in this situation? I'll make my point even stronger:
2021: 2470,2480,2490,2500,2510,2520MHz

At this constant pace, we've just doubled our performance in a short 20 years! 😀 Performance *has* to increase exponentially (not linearly) to be considered level! When performance fails to increase exponentially, progress is slowing down. I'm not saying this is necessarily the case right now.