PCA Forums

Following AMD's marketing change in it's processor labeling, Intel decided to mark P4's with a new PR rating. Just like AMD compares it's Athlon XP processor with the old Athlon processor line, Intel decided to compare the performance of the Pentium 4 with Pentium III chips. Changes are follows:

P4 1500Mhz = "PR 1100-"
P4 1600Mhz = "PR 1166-"
P4 1700Mhz = "PR 1233-"
P4 1800Mhz = "PR 1300-"
P4 1900Mhz = "PR 1466-"
P4 2000Mhz = "PR 1533-"
P4 2200Mhz = "PR 1700-"

The new system should help consumers understand IPC and deep pipelines according to Intel's press release.

Man...that was funny.

Greg

LMAO sad but probably pretty close to the truth

Man, that's like the car manufacturers changing from cubic inches to liters. I will never understand it.

Imagine how fast a 1 Ghz 386 would be! I bet it would be as fast, if not faster, than any modern
AMD/Intel chip at 1 GHz.

Look at what happened with the G4. Everyone makes fun of it because it is 'slow'. Nobody seems to notice it less latency than either AMD or Intel. It also gets more done every clock cycle.

Imagine how fast a 1 Ghz 386 would be! I bet it would be as fast, if not faster, than any modern
AMD/Intel chip at 1 GHz.

well, not really. you're comparing a non-pipelined, in-order CPU (386) to modern out-of-order, deeply pipelined CPUs.

The 386 can't reorder instructions for more efficient use of resources, and since it's non-pipelined it can't begin work on an instruction until the preceding one is finished.

A 386 is indeed pipelined. The 386 has a 4 stage pipeline, but it is not 'super-pipelined'.

Even without out of order execution, clock for clock, I bet a 386 could probably hold its own against a P4 or Athlon.

Originally posted by poop
A 386 is indeed pipelined. The 386 has a 4 stage pipeline, but it is not 'super-pipelined'.

Even without out of order execution, clock for clock, I bet a 386 could probably hold its own against a P4 or Athlon.

I highly doubt it, one reason being L1 and L2 cache. The 386 has or would have niether. And lets not forget that the 386 wouldn't have a FPU or MathCoprocessor built in.... Did the "DX" versions of the 386 have MathCo's built in like the 486's did? Or did the DX for the 386 only mean the data and address bus's were 16bit? I can't remember now, its been too long.

I wouldn't go as far as to say a 386 at the same clock speed would hold up against a P4 or Athlon, but I would damn sure put money on a 486@1GHz kicking the shit out of a P3/P4.

I'm running RC5 here and some of these 486's can do a unit 1/3'rd the speed of a P3 700.
1/3'rd the speed isn't impressive, but the fact that 3 of them, running at 133MHz could 0vv3n my P3 is.

Yeah, I forgot about cache, too. So... let me get back to my point.

Clock speed doesn't mean anything on its own. Processors used to get much more done per clock cycle, now latency is overwhelming.

Did the "DX" versions of the 386 have MathCo's built in like the 486's did? Or did the DX for the 386 only mean the data and address bus's were 16bit? I can't remember now, its been too long.

SX had the 16-bit buses, DX meant they were 32-bit. In both cases the FPU was the separate 387 chip. the 486 DX was the first to have an integrated FPU.

Oh, and the 386 was not pipleined. The 486 had the distinction of being the first pipelined x86 CPU.

from here:

The 486 bought a number of mainframe techniques into the X86 world for the first time: internal cache, rudimentary branch prediction, integrated [F]PU, and a five-stage pipeline.

Even if it were possible for the 386 to run at 1 GHz, it would get wasted by an Athlon or PIII, it took several clocks for the 386 to complete an instruction and could not begin another one until the preceding one was finished.

'Think Athlon'
'Think Athlon'



Took me a minute, but :lol :laugh

i dont get it.....?

I just upgraded to a p4 1.6A running at 133FSB to 2.13 GHZ and I must say that I am completely underwhelmed. I don't know, maybe I was expecting too much from a chip running at 2 Gigahertz. You know something like a real noticable difference in everyday apps instead of synthetic Becnchmaks. When I didn't have it overclcocked my 1.2 Tbird was actually faster in some apps.

If I were Intel I wouldn't make to much of a commotion because if they do the truth will get out that the Athlon clock for clock kicks the P4's ass...It also makes for a nice Space heater. Thourobred hopefully will address that issue.