Originally posted by: FIFO
Well it looks like icecool83 seems to be awfully short on factual criticism to the point that all he can do is nit-pick about a small slip-up in a term that I used.
And in regards to CrawlingEye..................................
I think that all of my pro intel comments all wrapped up in one little nutshell go like this:
A system based on an i850E and P4 2.53 Ghz with PC1066 RDRAM is the undisputed overall champion of the galaxy. I will not dispute that.
I cannot however STRESS the fact enough that I am speaking SPECIFICALLY ABOUT THE INTERNALS OF THE CPU ONLY!!!!!!!!!!!!!!!!!!!!!!!!!!!
If I jacked the P4 and the AXP to 5Ghz the AXP WILL beat the P4 IF I/O TO THE REST OF THE PC IS NOT AT ISSUE!!!!!!!!!!!!!!!!!!!!!!!!!!!!
AMD is more intent on finding a way to maximize IPC than maximizing Ghz. As a bonus though, AMD has to also make their Ghz higher for marketing reasons. Due to the very high IPC, when the AXP cpus get within even 600 to 700 Mhz of the P4, they match or beat the P4.
Beyond this, if the benchmarks that were done on the 800 Mhz hammer are any indication and AMD can get this puppy to 2 Ghz and beyond, the P4 doesn't stand a chance except in the land of the marketing lemmings.
Like statements I made earlier elude to: Intel may have tried to push the P4 architecture out a little early. It is possible that the P4 may prove longer lasting than some may think, but due to the fact that it is Now and not 2005, that doesn't really matter.
The truth of the matter is that I don't want to dig up sources to support my claims (so that it doesn't sound like I'm pulling it out of my ass). For the record, a lot what you have said so far is correct FIFO, BUT:
A) When you provide information, provide a source as well
B) Unless you have benchmarks that I am not aware of, we don't know how the Hammer will perform (that Quake benchmark counts for nothing since the string in WCPUID was grayed out and the processor was running at 800 MHz). I'll reserve judgement on the Hammer for when I see production-level silicon reviewed by a well respected website.
C) AFAIK, the Pentium 4 has better branch prediction than the Palomino, not positive about this though. pm may be able to confirm this.
D)
The following numbers have been extrapolated directly from the ALU results of SiSoft SANDRA 2002
Athlon = 2.75 x 1Ghz = 2750 mips
P4 = 1.84 x 1Ghz = 1840 mips
For the P4 to "scale" to the same IPS as the Athlon it needs to do the following:
1.84 x 1494565217 hz (1.5Ghz) = 2750 mips
Your basing your comparison of processor performance on Sandra? Please explain what are these numbers that your multiplying together. Remember, the Athlon usually runs applications designed for the P6 architecture better than the P4, but if the program is properly optimized for SSE2 things change quite a bit (check out Anand's Lightwave 7.5 benchy, you'll find that a 1.6 GHz Willy beats the 1.8 GHz Tbred in one of them
). Bottom line, though your general idea is correct, your example is terrible as the IPS changes for every application (and for that matter every platform).
E)
Intel could have VERY easily kept the very good FPU from the P3 and just "tossed" in SSE2 and a bit longer of a pipeline to allow for an acceptable increase in Mhz/Ghz.
Wow, you should definetly apply at Intel, they could definetly use a CPU architecture genius like you.
They wanted back [...] the MHz crown.
LOL, did it ever occur to you that the P4 architecture was designed well
BEFORE AMD took the MHz crown from Intel?
F) You're oversimplifying things a little too much and your "guesses" as to what the performances of future processors is like are laughable since you are essentially basing your assumptions on the scaling of current CPUs and don't (and can't for that matter) predict the effect of architectural enhancements on each processors.
I could probably go on, but I think this proves my erroneous information point well enough.
:frown: I'm done with this thread. Sorry your thread turned into this Spacehead (my CPU suggestion is above).
-Ice
PS: There are certainly flaws in Crawling Eye's arguments, but since he didn't go so far as to imply that I'm a nit-picker who understands nothing, I didn't bother pointing them out.