[MW Forums]

Intel shared new details on its Penryn chips and a new system-on-chip plan to save space and energy at its IDF trade show in Beijing. more

[MacGeek1955]

Wasn't one of the bragging rights of the G5 that it ran more than a terraflop with Altivec accelleration? Weren't they calling G5s super computers a few years ago because of this? mybe I was missinformed.
I'm not saying this to dispute any great doings of Intel, but the 64 bit multiprocessor G5s were and are still really fast at certain operations. Also they simply were not an option for laptops and even though they were dumbed down for iMacs they still ran too hot for the design. Like I said I'm not trying to take anything away from Intel. I have a Core Duo iMac and love it. Still when Apple moved from G5 to intel, the selling points for Intel were accentuated and all the BS PC folks were saying about Pentium 4s being faster than G5s was part of the megahertz myth and mostly FUD.

[dcr]

Quote:

---

Wasn't one of the bragging rights of the G5 that it ran more than a terraflop with Altivec accelleration? Weren't they calling G5s super computers a few years ago because of this? mybe I was missinformed.

---

Misinformed, or misremembering. You're probably thinking about the first G4 marketing campaign which threw the terms "supercomputer" and "gigaflop" around. A Teraflop is ~1000x more computation per second than a gigaflop
http://en.wikipedia.org/wiki/FLOPS

[tjustin7]

Quote:

---

Still when Apple moved from G5 to intel, the selling points for Intel were accentuated and all the BS PC folks were saying about Pentium 4s being faster than G5s was part of the megahertz myth and mostly FUD.

---

Actually, not quite.
The "newer" P4's were actually faster "Mhz for Mhz" than the "new" G5s.
Intel made major improvements to the P4 and addressed virtually all the criticisms, although the bad PR that the crappy first gen P4's generated stuck with them.
Outside the marketing dept, however, the writing on the wall was clear.
At the point Intel started talking about going dual core, the CPU wars were over. The only advantage the G5 had was in number of processors (i.e. "Four PPC chips are faster than two x86 chips!").
In other words, the megahertz myth was true when the P4 was 1.4Ghz, but that was't the case when 3.2Ghz P4's were coming out.
(I should also add that it was not just a case of the P4's "Mhz for Mhz" speed going up, but that the G5's "Mhz for Mhz" speed went down due to the boosting of the pipeline length to increase clock speed).
- Me

[doh123]

Quote:

---

Actually, not quite. The "newer" P4's were actually faster "Mhz for Mhz" than the "new" G5s............

---

whaaat... ? I must be reading this really wrong but I keep reading it again and again...
the newer G5s were much faster Mhz for Mhz than the newest P4s. The P4s were faster overall because they went up to 3.8ghz which compensated for their poor efficiency.

Quote:

---

(I should also add that it was not just a case of the P4's "Mhz for Mhz" speed going up, but that the G5's "Mhz for Mhz" speed went down due to the boosting of the pipeline length to increase clock speed).

---

If I remember right, the G5 design barely added anything to the pipeline length, and it was still quite short compared to the P4s. The P4 kept increasing the pipeline length until the last 2 versions had a HUGE 31 stage pipeline. with the netburst architecture, the longer pipeline hindered efficiency, but allowed them to stay stable at much higher frequencies. They kept increasing the L2 cache to offset the poorer performance. I personally ran benchmarks on some of the P4s. the 21 stage pipeline 3ghz 512k L2 cache P4 actually performed about equal, to a bit better than the 31 stage 3ghz 1mb L2 P4.
There is a reason that Apple didnt sell any Macs with P4 processors and waited to put something out until the Core Duo was available... the P4 stunk.