AMD Zone tries to slam the G5

[utilizer]

Here's my take

The whole issue here is moot...the way this trio will be at the end of the year is on par. Opteron's will be at 2.2 Ghz by then, then so will the Intel's by November at 3.4 Ghz with faster FSBs, and the G5 will still be at Dual 2 Ghz.

Then the real surprise will be at MWSF. Optimizations galore, Panther updates, third-party updates will magnify this even more so. I'm guessing Dual 2.4 Ghz come January. That along with the optimizations will give AMD and Intel a run for their money.

And about this AMDZone guy: He won't be around for much longer. IBM will buy AMD in October of next year (mark my words, I called it). Why you ask?

IBM will hit a dead end at 3 Ghz by the end of the summer next year with this PowerPC stuff; I see really no reason to continue funneling money into this project when only one computer company is using a chip that requires massive amounts of capital investment that only yields little amounts of profit, compared to what's put into it. Oh, and before you start saying that the IBM Linux Blade Servers will help balance everything out, why are they selling Opertons?!
It's impossible to continue down this inevitable road that will eventually put us, and Apple, in the same place that Motorola did for nearly two and a half years. IBM will buy AMD b/c they see it too and so will Apple, and that's why they're keeping Marklar alive and well.
AMD has 64-bit technology and with all of the technology that IBM engineers learn from the efficiency of the IBM 970/980 chips will be carried over to the effective AMD Operton and subsquent generations of the chip.
But the switch to x86 won't be as heart-wrenching as the "experts" say it will be.
Wait and see...

 

[Rezet]

Well, with the currecnt technology I think Intel and AMD already are hitting the wall. Intel's 3.2 processor already gives out as much heat as 100w bulb... Unless they innovate, they are going straight into the dead end... well kinda like american law system..

 

[Cubeboy]

Intel is switching to Prescott Pentium 5 with 90 nm process in 4Q 2003. Due to the current state of leakage on this process, heat will be a BIG problem although there is little doubt that the Pentium 5 will be able to scale significantly higher than Pentium 4. You have to understand that Pentium 5 is also going to be significantly faster than Pentium 4 clock to clock as well. Simply increasing the trace cache bandwidth (maximum number of instructions the P4/P5 can execute and retire) from 3 uOPs to 4 uOPs will substantially increase IPC. Other improvements include larger data and L2 caches, lower latency on integer multiples (another significant improvement), improved branch prediction, improved hyperthreading, SSE3, larger instruction, store and load buffers as well as others (register files, etc).

Regarding Cinebench optimizations, Richard Kurz, one of Maxon's developers, recently posted on Ars's forum. He predicts that optimizations will improve G5 score by about 25% which although significant won't be faster than the P4 / Xeon / Opteron. Link Below:

http://arstechnica.infopop.net/OpenTopic/page?q=Y&a=tpc&s=50009562&f=48409524&m=5790920455&p=7

Here is the statement, alone:

"This is based on the information we have right, now, there is still a of of work to do and we still have to wait for a new compiler...

With the current CineBench a single G5 1.8GHz scores at about 188, the optimized version will maybe score at about 238...

A hypothetical single G5 2.0GHz could score at about 210 on the old CB, optimized could be 265...

A dual G5 2.0 could maybe score at about 480 with the optimized version of CB....

Depending on the new compilers and our findings (thanks a lot to Apple for being extremely helpful and cooperative) we might even crack the 500 score for the dual G5 2GHz..."

I think the number of unsubstantiated claims around here has gotten a little out of hand. Cinebench is actually very well optimized for the G4 (but not the G5), saying that it it is more optimized for x86 than RISC is just wrong.

 

[Mav451]

Originally posted by Cubeboy

http://arstechnica.infopop.net/OpenTopic/page?q=Y&a=tpc&s=50009562&f=48409524&m=5790920455&p=7

Here is the statement, alone:

"This is based on the information we have right, now, there is still a of of work to do and we still have to wait for a new compiler...

With the current CineBench a single G5 1.8GHz scores at about 188, the optimized version will maybe score at about 238...

A hypothetical single G5 2.0GHz could score at about 210 on the old CB, optimized could be 265...

I think the number of unsubstantiated claims around here has gotten a little out of hand. Cinebench is actually very well optimized for the G4 (but not the G5), saying that it it is more optimized for x86 than RISC is just wrong. [/B]

I really hope that things look better with optimizations. Someone replied (not sure if it was in this thread) whether or not AMD cpus were optimized for cinebench and someone said it was as optimized as its going to be.

With that in mind, my current "budget" AMD rig scored 267CB-CPU--i was looking forward to some "ridiculous" numbers being put out by the G5's

 

[Cubeboy]

Originally posted by Mav451
I really hope that things look better with optimizations. Someone replied (not sure if it was in this thread) whether or not AMD cpus were optimized for cinebench and someone said it was as optimized as its going to be.

That someone needs to do his homework, AMD lacks the resources to develop a custom compiler for their CPUs, unlike Intel, IBM, Sun, HP, and Compaq/DEC/Alpha. AMD cpus aren't anywhere near "as optimized as their going to be" in programs like Cinebench where all cpu-specific optimizations are done by the compiler.

 

[Mav451]

Actually that's wut i meant to say. To put it more clearly, someone earlier had said that it had been 4 years already, and that there weren't any particular optimizations made for AMD.

What i meant is that it doesn't look like it will GET ANY BETTER.

So yeah, i already knew that, but thanks for the clarification.