Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MasterX (OSiX)

macrumors 6502
Original poster
Sep 3, 2001
310
0
With the SOBs at Intel ripping off the PPC more than ever do we have anything to fear? No of corce not, but don't you think Intel's actually catching up?

I mean we all know the G4 is the fastest PC CPU money can buy, especially for those AltiVek tasks. But look at this:
Systems: G3/G4/G5/Pentium4/Itanium
Pipeline: 4/7/6?/20/8
CPU-bits: 32/32-128/64-256?/32/64
Ram: SDR DIMM / SDR DIMM / DDR DIMM / RAMbus / ?
Speed*: 100/100-300/200-600/50/200
Instruction type: RISC/RISC/RISC/CISC/RISC rip-off

*Speed per Mhz relative to processors in it's claas (PPC/Intel)

- represents AltiVec results
-----------------
According to my poorly constructed and researched analysis we can see that the Itanium is at least a threat, wheras the P4 was a POS. I mean the concept that Intel will be playing their coimputers in the feild with Mhz performance simmilar to ours again (remeber P3 vs G3) is scary to me. Luckily the OS is key, Macs will only go faster running OSX, while PCs will mostly go slower running XP.
 

spikey

macrumors 6502a
Apr 26, 2001
658
0
64 bit chip PCs have a problem with XP, in that it isnt as good as OSX.

Also the flops ratings of the itanium are nearly beaten by the dual800 G4. Also the G5 is supposed to have nearly 2 or 3 times the speed per clock on the itanium aswell as having higher expected clock speeds.
I doubt competition will be the itanium, hold on to see if AMD launch their "clawhammer" processors, laugh if they make a balls up of it like intel have.
 

MasterX (OSiX)

macrumors 6502
Original poster
Sep 3, 2001
310
0
True, true

Originally posted by spikey
64 bit chip PCs have a problem with XP, in that it isnt as good as OSX.

Also the flops ratings of the itanium are nearly beaten by the dual800 G4. Also the G5 is supposed to have nearly 2 or 3 times the speed per clock on the itanium aswell as having higher expected clock speeds.
I doubt competition will be the itanium, hold on to see if AMD launch their "clawhammer" processors, laugh if they make a balls up of it like intel have.

Glad to see Intel's bigger than MSFT, engineering ahead of their time.

As for G5 Mhz see the other thread (I suggest a 2Ghz G5 release, yes it's possible, not PROBIBLE, but possible)

AMD makes me angry, just when Apple clearly has Intel beat a new competetor arrives. (not recently, but you know what I mean)
 

ThlayliTheFierce

macrumors regular
Jul 31, 2001
248
0
San Luis Obispo, CA
Apple may have the faster processor but the P4's sales outstrip the G4's by 10 to 1. However, as the saying goes, "the enemy of your enemy is your friend". AMD competes more with Intel than Apple, thus weakening Intel's position.
 

MasterX (OSiX)

macrumors 6502
Original poster
Sep 3, 2001
310
0
Yeah

I thought about AMD weakening Intel's market share, and in fact Intel looks pathetic, they now have to fight to stay x86 king. But I don't think of it as you're enemy's enemy is your freind; I'm thinking divide and conquer (gulp)
 

iJed

macrumors 6502
Sep 4, 2001
264
10
West Sussex, UK
Complete crap

Nobody on this forum seems to have a clue about what they are talking about!

Itanium
Itanium is based on a new VLIW parallel architecture that is fundamentally a generation ahead of PPC. The successors to Itanium, which is held back by x86 compatibility, will quickly become the most powerful processors in the world.

Pentium 4
The current P4 smashes the MPC7450 in nearly every task. If you don't use selective benchmarking (eg optimized vs non optimized Photoshop filters) then the new P4 instructions hammer AltiVec into the ground. It is unfortunate the PPC has fallen so far behind in this case.

Before everyone starts flaming me - I am a Mac user and I will never consider switching to Windows (XP or otherwise). I have programming experience in many languages and APIs, including Mac (Classic, Carbon and Cocoa in C, C++ and ObjC), Java, BeOS, Ada95, Basic, ASP (in JScript), PHP, X86 asm, and probably a few more. I know this does not qualify me as a hardware expert but it at least shows that I can give an objective view based on facts.
 

MasterX (OSiX)

macrumors 6502
Original poster
Sep 3, 2001
310
0
Thanks, I was gonna get the VLIW thing but I can't get to runors right now, and it's my dictionary. As for the Itanium being ahead of the PPC? I don't know about that, does that include the G5 running Puma and it's true 64-but CPU w/256-bit altivek?

Plus keep in mind the optomization for AltiVek will be the same problem for VLIW, for it to work apps will need re-programming right? Or am I way off? I remember the PPC code had to be re-written for RiSC a while ago before I can clearly remember.
 

iJed

macrumors 6502
Sep 4, 2001
264
10
West Sussex, UK
Originally posted by MasterX (OSiX)
As for the Itanium being ahead of the PPC? I don't know about that, does that include the G5 running Puma and it's true 64-but CPU w/256-bit altivek?

Its not so much Itanium being ahead of PPC but the fundamental architecture being superior to that of PPC. I don't think Apple has anything to worry about from IA64 though, as Itanium is currently only aimed at the very high-end of servers and workstations. I also have some reservations about the G5. Can its new AltiVec unit use longer data (64 and 128 bit). The current one is limited to 8, 16 and 32 bit data which is unsuitable for high precision floating point stuff. However, I don't think that there will be any OS out there that can beat Mac OS X 10.1 on the majority of tasks, and Cocoa is easily the best API I have ever developed anything for.

Plus keep in mind the optomization for AltiVek will be the same problem for VLIW, for it to work apps will need re-programming right? Or am I way off? I remember the PPC code had to be re-written for RiSC a while ago before I can clearly remember.

Yes, all code will require recompiled for IA64 just like 68K apps needed recompiled for PPC. I assume the completely new FPU in the Pentium 4 also requires recompiled code.
 

Megaquad

macrumors 6502a
Jul 12, 2001
817
1
i-jed

i completely agree with you.. too bad that you can't prove that to many mac users..
 

spikey

macrumors 6502a
Apr 26, 2001
658
0
Very true, but that is not the point.
The point is Apple will bring in a 64 bit chip to the consumer market, while the itanium is aimed at small servers.
The itanium will be held back by software not taking advantage of its power, i.e 64 bit XP not being fully 32 bit compatible, and software companies sticking with programming for the x86 architecture.
Might be partly why the itanium can only reach something like 13 gigaflops, the dual 800 G4 can reach about 11 i think, god knows what the G5 will reach.
The point is we are moving into a new era of chips (going 64 bit), and this is where compatibilty comes in for both software and hardware. PCs are really a bunch of components which are not really made to be compatible with each other, while Macs have components made for each other.
and with chips changing design right now, PCs are going to be held back by microsoft and Macs will thrive in OSX.

In the long run then maybe it will be different, but you cant deny that from what we are hearing the G5 looks impressive.
In the long run i doubt the itanium will succeed anyway, intel have been caught out by their own marketing strategy.
whether or not the itanium is a better design is irrelevant because in this situation it is going to be held back by software and the fact that it is aimed at a different market.
 

spikey

macrumors 6502a
Apr 26, 2001
658
0
And from my experience Macs and pentium 4's are just good at different things, i find that it evens out in the end. Only PCs are better value for money, and Macs dont crash half the time, and they also dont have compatibilty issues PCs do, like cpus not running at full potential, or geforce 3 cards not wanting to increase clock speed.
 

spikey

macrumors 6502a
Apr 26, 2001
658
0
If you look at the design of the pentium 4 by the way, it is a really **** chip, just held up by its Ghz.
Intel have a habit of ****ing things up and getting away with it.
 

Megaquad

macrumors 6502a
Jul 12, 2001
817
1
spikey,spikey..

you dont know nothing about performances of cpu's,for example,atlon 1.4 ghz is like p4 1.6 ghz or so,nothing spectacular,even amd says that,and,look at the benchmarks..
perhaps spikey works in AMD?
and please dont answer me with "shut the **** up you ****in' ******** i'll blow up your head"
please show me some benchmarks,facts!
 

spikey

macrumors 6502a
Apr 26, 2001
658
0
No, not true.
again you seem to have missed the point, Athlons are much better at certain tasks than pentium 4s of a higher clock speed, also they are equivalent of a 1.6 Ghz Pentium 4 at some tasks. It just depends what you do with your PC.
For example, because the pentium 3s design had not changed since the pentium 166, at certain tasks the 166 beats it! although it is very rare.
and besides, if you think differently why dont you come out with some benchmarks?
Gaming wise the pentium 4 is good, not great but good. But anything that likes floating point operations it sucks in.
I will not swear at you today because i am resonably happy with things.
i wish i did work for AMD, but i dont. i just know that the athlon is a good chip, i also know that the every pentium since the 166 has been a cheaply produced chip just backed up by alot of marketing.
 

iJed

macrumors 6502
Sep 4, 2001
264
10
West Sussex, UK
Originally posted by Megaquad
please show me some benchmarks,facts!

What shall we benchmark then? Lets try MIPS (million instructions per second)... Intel once claimed a high (for the time) MIPS rating on one of their early RISC chips. Unfortunately for anyone stupid enough to buy the chip, it could realistically only achieve less than half the claimed MIPS! Why? Well Intel gave its MIPS rating executing NOP (no operation) instructions one after another! One of my computing science lecturers once said that MIPS should stand for Meaningless Indication of Performance for Salespeople!

OK, so MIPS is rubbish, what shall we try now? How about some Photoshop tests? We run the same tests on both G4 and P4. G4 wins easily, then we optimize for the new P4 FPU instead of x87. Now P4 vaporizes G4. It always works this way.

Benchmarks are all selective. They never show realistic results.
 

spikey

macrumors 6502a
Apr 26, 2001
658
0
And if you are asking me to refrain from insulting you, why dont you do the same instead of saying

"you dont know nothing about performances of cpu's "

I would appreciate it if you took your own advice.
 

spikey

macrumors 6502a
Apr 26, 2001
658
0
This is another agree to disagree situation, it makes me sick.

I think this convo has strayed from the original, point about the G5
 

MasterX (OSiX)

macrumors 6502
Original poster
Sep 3, 2001
310
0
Spikey

I agree with spikey this time too. The co-op between Intel and MSFT goes way back, where they both used their monopolies to their advantage, and they always kept backwards-compatibility as top priority. I think Intel has been forced to show it's hand by AMD, the Itanium seems like a good product, for a PC. So as a server solution I think the dual 800 G4 running OSX server will probably serve quite well, and it's a modified pro-PC. As a consumer CPU Intel will need to use dual processors at least, to advertise 1.6Ghz, to look formatable.

As for the AMD model number thing, I think it seems more like a continuation of their own Ghz, even though they list it as a compasison to the P4. Keep in mind Apps with SSE2 (i think that's the new acceleration scheme) are actually quite efficent per Mhz, thus the tests at MacWorld are more convincing then they look. Most tests on websites are FPS tests or video tests, codecs can be SSE2 (like when the MPEG4 codec was reprogrammed to show the P4 as faster), but I don't think OpenGl is (although OpenGL now uses AltiVek)

My new disclaimer: I'm only 80% shure on everything I say
 

ThlayliTheFierce

macrumors regular
Jul 31, 2001
248
0
San Luis Obispo, CA
Actually...

Photoshop is already optimized for the P4, and the G4 still beat it. But before I get flamed, I agree that that still is a pretty narrow test. I'm sure if Intel were showing off their test would show the same thing, just in their favor. The gap between PPC and x86 used to be much greater. Now it's narrowed considerably, and they beat each other in specific tasks, instead of one being the hands down winner. Oh, and as for the Itanium...of course it's a generation ahead, it's the next generation! The G5 will be on par with the Itanium in this respect, but like Spikey noted, it will be aimed at us, not high level servers and the like. This is where it will really beat the Itanium. If the reports I've been reading are true, it will absolutely smash its competition (the P4, probably not Itanium for a while yet), and the gap will widen yet again.
 

MasterX (OSiX)

macrumors 6502
Original poster
Sep 3, 2001
310
0
I agree with Fierce, Adobe is pretty good at staying up to date with new CPU/FPU features. The current copy of PhotoShop 6 uses all the Pentium4's features. Keep in mind the G4 was specifically designed as a Photoshp/Codec/FPU tool. The AltiVek engine is the key factor separating the G3 and G4 (oh that and memory architecture) the AltiVek unit is specifically designed to utilize both swappable and 128-bit wide registers. The ability of the G4 to run up to 4 32-bit tasks simultaniusly results in a 3 fold increase in real-world performance. Thus a 12 Gigaflop G4 can only use 3 gigaflops in Photoshop 5.0, because it only runs instruction through the standard functions, not the AltiVek processes or the 128-bit structure.

My new disclaimer: I'm only 80% shure, and welcome any KIND fixes to my explanation.
 

Catfish_Man

macrumors 68030
Sep 13, 2001
2,579
2
Portland, OR
Per Mhz, the P4 is not even as efficient as the P3. Intel has carefully trained consumers to look at Mhz, so when they release an extremely high Mhz chip that doesn't go that much faster everyone will think it flattens the competition. Unconfirmed Info--- Apple actually toyed with the idea of releasing a 200 mhz chip ( This was back when everything was <100mhz) that didn't actually go any faster than it's <100mhz competitors; It just looked impressive on spec sheets. This is pretty much what Intel is doing. (The early P4s are actually slower than the latest P3s.)
 

evildead

macrumors 65816
Jun 18, 2001
1,275
0
WestCost, USA
Originally posted by Catfish_Man
Per Mhz, the P4 is not even as efficient as the P3. Intel has carefully trained consumers to look at Mhz, so when they release an extremely high Mhz chip that doesn't go that much faster everyone will think it flattens the competition. Unconfirmed Info--- Apple actually toyed with the idea of releasing a 200 mhz chip ( This was back when everything was <100mhz) that didn't actually go any faster than it's <100mhz competitors; It just looked impressive on spec sheets. This is pretty much what Intel is doing. (The early P4s are actually slower than the latest P3s.)


Thasts my understading too... it's not very efficent. Performance is petter with WinXP (or I have been told) but its still a slug of a CPU
 

MasterX (OSiX)

macrumors 6502
Original poster
Sep 3, 2001
310
0
Ironic

Funny though, the Pentium4's high Ghz isn't enough, I think it's dell that adds 3000 to it:

1.3Ghz=Dell 4300

Is intel trying to screw themselves over (Itanium 3800) no they're probably sign an agreement with Dell to use a 8 so:

800Mhz Itanium=Dell 8800 (wow over 2x faster based on BIG impressive numbers…)

PC Consumers are smart, just like man's best friend...
 

Capt Crunch

macrumors 6502
Aug 26, 2001
486
14
Washington, D.C.
The Itanium is a very good processor. It simply kicks everythings ass. The company my dad works for got a couple to test with, and they are super fast. And they are not stuck at 800mhz.

Intel finally got something right, however, now the software end needs to be revamped. The Itanium closely competes with the G4 (and from what I saw will beat it) but if windows is still the mainstream OS, I cant see myself buying one.

BTW the stuff the itanium was running on was UNIX, and it was doing some number crunching, and damn fast. It's very good for signal processing. Spanks the hell out of a PIII.

 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.