why apple laptops will stay behind PC's in pure speed...

[gekko513]

The new Pentium-M has 2MB level 2 cache and 533MHz bus. If you put a 2MB level 2 cache and a 533MHz bus on a G4, I'm almost certain that the G4 would kick some serious Pentium-M ass even if the G4 runs at 1,5GHz and the Pentium-M runs at 2,0GHz.

Unfortunately the G4 so far has only got a 166MHz bus and 512KB cache. Freescale better hurry to fix these issues.

 

[Mav451]

(http://www.hwupgrade.com/overclock/cpu/index2.html) So what does it say about "your" credibility that you don't even seem to be aware of ElectroMigration and that it happens much sooner when you overclock? Hmm.

You must be joking rite? Look at the date of that website. 1999. That's the year of the AMD Thunderbirds/early Palominos. *Correction, this was before even the Thunderbirds! This was the time of the original SLOT A Athlons--Socket A wouldn't come out until 2000. Of course overclocking was dangerous then--the technology wasn't there yet.* The Thoroughbred B's (what I have) would not be out for another 3 years! Technology accelerates rather constantly, in the CPU world, so you are aware that knowledge, of even last year, may not extrapolate accurately for technology a year later. Knowledge about the 1st generation Athlons/Pentium does not necessarily translate to the 2nd (and 3rd) generation processors. Things have changed CONSIDERABLY since 1999. 1st generation Athlons ran HOT (1.75vcore). On the other hand, the Thoroughbred B's would come out, at a MUCH higher clock speed (1470-2200Mhz compared to the Athlons 700Mhz-1400Mhz), while running much cooler (1.5-1.65Vcore, even the highest vCore remains .10 less than the original Athlon).

What else has changed since then?
1) Superior Heatsink technology (Copper, fin design of the SLK800/900 weren't even available)
2) Quieter, more powerful AIR cooling technology (e.g. Panaflo)
3) C.O.P. technology (CPU Overheating Protection--Shuts down the system when CPU temperatures reach a critical temperature).
4) PCI/AGP locks--though Intel has had it quite some time, their multipliers are locked, so its a moot point for Pentiums anyway; for AMD this allowed for even HIGHER overclocks while letting the rest of the PCI bus and AGP card to remain running at safe speeds--33Mhz and 66Mhz respectively.

Well, #4 pretty much makes this page outdated and misinformed by TODAY's standards: http://www.pcguide.com/opt/oc/risksRisksBus-c.html

The PCI bus can be LOCKED in a plethora of nForce2 motherboards (available since fall 2002), so if that website chooses to stay up to date, it would remove this severely outdated information before becoming the laughing stock to the overclocking community.

For Intel, prevalanet PCI lock technology was available since the advent of the i845: http://www.sharkyextreme.com/guides/hwGuides/article.php/10709_1380951__3

We should also note at this time that many i845D based motherboards (regardless of FSB speeds) lock their AGP and PCI frequencies at 66 and 33 MHz respectively, so there isn't a great danger to your peripherals when overclocking with many high-end i845 DDR motherboards.

 

[timberfish]

Hector: I know that cooling will make overclocking work without nearly as much risk and damange, but I was just addressing the fact that Mav451 was saying that overclocking doesn't generate any extra heat, nor does it shorten the life of components, which is just not true. So I posted all these articles to prove that overclocking does just that.

Mav451: Processors are marked at their speed because it's determined that that's the speed they can handle safely. Sure, they "could" be marked higher and run higher, I think we all agree on that, but there "are" side-effects. So "yes" I will say that running a CPU at a higher clock than it's rated will shorten its life and be dangerous. I say it and as you can see many reputable sources agree and go into much detail. I can post about 10 other articles from other sources if you like, but I don't want to just dump links on this board.

Hector, I'm glad someone agrees that using overclocked CPU benchmarks only illustrate "possible" future speeds and chip performance and should not be used in a true comparison between common chips on the market. I see that we're now way off topic, so I'll leave it to someone else to bring us back if anyone wants to address what this thread was about.

 

[Mav451]

Hector: I know that cooling will make overclocking work without nearly as much risk and damange, but I was just addressing the fact that Mav451 was saying that overclocking doesn't generate any extra heat, nor does it shorten the life of components, which is just not true. So I posted all these articles to prove that overclocking does just that.

Mav451: Processors are marked at their speed because it's determined that that's the speed they can handle safely. Sure, they "could" be marked higher and run higher, I think we all agree on that, but there "are" side-effects. So "yes" I will say that running a CPU at a higher clock than it's rated will shorten its life and be dangerous. I say it and as you can see many reputable sources agree and go into much detail. I can post about 10 other articles from other sources if you like, but I don't want to just dump links on this board.

Hector, I'm glad someone agrees that using overclocked CPU benchmarks only illustrate "possible" future speeds and chip performance and should not be used in a true comparison between common chips on the market. I see that we're now way off topic, so I'll leave it to someone else to bring us back if anyone wants to address what this thread was about.

I said it does not generate extra heat if you USE THE SAME VCORE. Its clear, after several posts now, that you still don't understand overclocking. Google searches and "10 other articles" does nothing to rebut my point, b/c you still don't understand what I'm saying! Maybe I will give you yet another example:

Ok, you have several 2800+ chips (2.25Ghz). However, demand for 2100+'s is significantly higher, so instead of producing a combination of 2100+'s and 2800+'s (that would be wasteful production), you could simply only produce 2800+'s to cover ALL bases, and THEN REMARK them as demands sees fit. That is the basis for AMD's overclocking success in the first place.

B/c the 2100+ is actually a "true" 2800+ chip (just marked down), that means it is ALREADY CERTIFIED to run @ 2.25Ghz @ 1.65Vcore. However, the 2100+ only runs @ 1.73Ghz @ 1.6Vcore. So, in the factory/FAB, they do some modifications to make the 2800+ seem like a 2100+ (Multiplier set to 13X, FSB to 133). What I'm trying to say is that these chips are ALREADY CERTIFIED, just remarked.

Now, if my 2100+ was actually 2800+ the WHOLE TIME, why should running it at its "true" speed shorten its life? The answer is, it won't.

Read my entire post next time to avoid these rather glaring misconceptions about overclocking.

And please, no more google searches and 3-5 year old articles that are outdated.

 

[timberfish]

LOL, Mav451 you are seriously going to focus on the dates of those articles? I mean they clearly talk about hardware concepts that still apply. Electromigration still exists, but if you insist on this then here are some articles from 2003/2004, but they were harder to find because overclocking isn't as big anymore since Intel's move to prevent it and the fact that processors are just so fast now, it's usually not necessary. Anyhow, here's one from 2003 that talks about reducing the life of your comoponents: http://www.cooltechzone.com/articles/overclockingexplained.html

Here's one from 2004 where they use liquid nitrogen to cool the components so they can overclock to 5Ghz which proves that heat "is" an issue when overclocking. http://www.tomshardware.com/cpu/20031230/index.html

I "do" understand what you're saying about the "fing" vcore. The vcore staying the same is not the only factor that increases the heat or reduces the life of the chip. Look, you find me a source that says you can overclock without touching the vcore and it will be completely safe and "not" increase the heat output or decrease chip life and I'll shut up, okay? If you look on any overclocker enthusaist site (which I visited often about a year ago) then you'll see ads for cooling, warnings about heat and how to cool a cpu and warnings about damage and reduction of component life. These are facts. I'm addressing a couple main things you stated about overclocking: It does not increase heat, it does not reduce processor life. I'm not talking about how to use cooling to reduce heat etc.. The fact is that extra heat "is" generated. If it wasn't don't you think Tom's Hardware guide would have just left the vcore alone and gone to 5Ghz? No, cause they can't without frying the hardware. So knock these articles however you want, but I'm not going to take your "word" over these professionals, sorry.

And I understnad remarking, before you go off on that whole point. I'll bring this up one more time - we are no longer talking about G4 vs P4, we are now talking about overclocking which 95% of the population does not even do.

 

[Mav451]

LOL, Mav451 you are seriously going to focus on the dates of those articles? I mean they clearly talk about hardware concepts that still apply. Electromigration still exists, but if you insist on this then here are some articles from 2003/2004, but they were harder to find because overclocking isn't as big anymore since Intel's move to prevent it and the fact that processors are just so fast now, it's usually not necessary. Anyhow, here's one from 2003 that talks about reducing the life of your comoponents: http://www.cooltechzone.com/articles/overclockingexplained.html

Here's one from 2004 where they use liquid nitrogen to cool the components so they can overclock to 5Ghz which proves that heat "is" an issue when overclocking. http://www.tomshardware.com/cpu/20031230/index.html

I "do" understand what you're saying about the "fing" vcore. The vcore staying the same is not the only factor that increases the heat or reduces the life of the chip. Look, you find me a source that says you can overclock without touching the vcore and it will be completely safe and "not" increase the heat output or decrease chip life and I'll shut up, okay? If you look on any overclocker enthusaist site (which I visited often about a year ago) then you'll see ads for cooling, warnings about heat and how to cool a cpu and warnings about damage and reduction of component life. These are facts. I'm addressing a couple main things you stated about overclocking: It does not increase heat, it does not reduce processor life. I'm not talking about how to use cooling to reduce heat etc.. The fact is that extra heat "is" generated. If it wasn't don't you think Tom's Hardware guide would have just left the vcore alone and gone to 5Ghz? No, cause they can't without frying the hardware. So knock these articles however you want, but I'm not going to take your "word" over these professionals, sorry.

And I understnad remarking, before you go off on that whole point. I'll bring this up one more time - we are no longer talking about G4 vs P4, we are now talking about overclocking which 95% of the population does not even do.

Well, Tom's is a more extreme example . But of course, overclocking will *typically* require more Vcore. I think we are talking about two DIFFERENT kinds of overclocking.

Tom's is more of the "reckless" kind (requires LN2, etc.). I'm talking about overclocking on air, with MINIMAL (if any) Vcore adjustments. And concerning overclocking without raising the Vcore? The Barton 2500+ is a testament of this. It could be changed from 2500+ (1.83Ghz 333FSB) to 3200+ (2.20Ghz 400FSB) with one simple BIOS changes*:

FSB: [166 >> 200].

Since the 2500+ already runs with a Multiplier of 11X, all that is needed to hit 2.20Ghz is a simple FSB change.

*denotes mobo/memory requirements*
Obviously, don't do it with a mobo that is not overclocking friendly...ditto with the memory. Don't expect your overclock to hit 400FSB, easily, if you don't have DDR400 memory. Likewise with the chipsets, if you don't have the Ultra400 chipset (and the older D stepping), then of course, it will be harder.

That is more of the "sweetspot" overclocking, where a chip is KNOWN to run at a given frequency/FSB, at the SAME vcore, b/c it is marked down.

2100+ >> 2800+ is exactly the same as 2500+ >> 3200+

**Of course, if you are overclocking beyond this (e.g. 2.4-2.5Ghz with Bartons), then yes, the Vcore is typically higher; at around 1.8-1.95Vcore. That is the type of overclocking I don't recommend, b/c I know it is much more dangerous.

So again, I think we were confusing 2 types of overclocking. I'm mostly concerned with "sweetspot" overclocking, not "reckless" requiring LN2 or Prometia systems...that's left to the people with EVEN MORE time to waste than me!

 

[wPod]

why didnt you just start a thread 'windows is better than OS X' i dont think anyone buys a laptop to have the fastest processor.

and i dont know about you, but ive never seen a wintel laptop that is noticably faster or even at a compareable speed to my 12" PB (867 for that matter!) maybe its b/c of windows runs slow or whatever, but in the basic opening of apps and running apps i notice bit more speed out of my PB. the only intel machine i know that runs 'just a bit' faster than my PB is the P4 3.0ghz desktop i have at work. even my old P4 1.5 ghz machine i had at work seemed slower than my PB.

so, i dont know what all the 'speed tests' have to say, but from the real life tests i have seen myself the G4 kicks the but of the P4 (especially if you compare Ghz to Ghz. . . )

 

[Soc7777777]

Soc7777777, I'd have to say you're so wrong about so many things here. Your main complaint seems to be that Apple should focus on making a mobile chip similar to the new Pentium-M chips that Intel makes. My question is why? My 15" G4 1.5Ghz laptop has a battery life of 3 to 3.5 hours with Airport on and using it actively. That's roughly the same as the Pentium-M laptops because a co-worker has one and he only outlasts my Mac by about 20 minutes in meetings. Also, the speed of a G4 is amazing compared to most of the P4 line and I'm talking about Desktop machines here. My G4 laptop outpaces my 2.8 Ghz P4 w/HT desktop here at work in most cases. Also, I own a Sager laptop with a P4 2.4Ghz chip in it and my Mac blows it away in speed. Oh, and the Sager has a desktop speed P4 in it, so that's not dumbed down for mobile computing. Besides this whole speed argument which I don't want to get into, you're talking about the need for a mobile chip?

The only reason Intel made the Pentium-M is because the Pentium 4M chips sucked and were exactly what you claim the G4's are - dumbed down desktop chips. Well, guess what? That's what most mobile chips are. The Pentium M was developed with power consumption and efficiency in mind. Yes, it was built separately from the P4, but a lot of the architecture is shared. They did make huge improvements in the architecture to be able to handle processes better and more efficiently while requiring less power, but let me also point out that Pentium M chips are not good for powerful apps like video editing, cutting edge gaming and encoding and so on. They aren't meant as desktop replacements. For those, Intel still makes Pentium 4M laptops and sometimes just puts a desktop CPU right in. Terrible battery life of course and always between 8 and 10 pounds - trust me, I know.

So maybe the G4 in the laptops wasn't built just for laptops, but so what? My Mac laptop is faster than most P4 desktops (except the high end ones of course) and handles battery amazingly well. What more do you want from a mobile machine? Also, AMD already has an Athlon 64M chip in certain laptops. Check out Voodoo and eMachines for starters. They perform well, but are heavy and expensive (in the Voodoo case they're around $3 to $4K and up). Eventually G5's will be in laptops and when that happens great! But Intel and AMD won't be ahead, they'll just be doing things differently. Right now, any comparable mobile machine on the Windows side is roughly the same price and performs the same as Macs. If you go for less-mobile desktop replacement laptops then yes, the Intels win, but at the same time they weigh 8 to 12 pounds, run hot, and suck battery like nothing you've ever seen. That's my take. Oh, and while this isn't scientific, check out this link to see some of my own benchmarking results in Photoshop between 3 machines on Mac and Windows. http://www.grassapple.com/archives/2004/06/apple_vs_window.html

haha im going to ignore the unrelated bs about P4-m chips... because that has nothing to do with p-m chips or g4 chips...

but your attempt at benchmarking is pathetic... haha you are putting 3 machines up against each other in a GRAPHICS INTENSIVE PROGRAM and give one machine (the PB) TWICE the amount of VIDEO RAM... the MAIN factor in graphics like performance...

thats like comparing the acceleration of two cars and having 2 of them go uphill and the other down... thats just not fair..

 

[Mord]

Here's one from 2004 where they use liquid nitrogen to cool the components so they can overclock to 5Ghz which proves that heat "is" an issue when overclocking. http://www.tomshardware.com/cpu/20031230/index.html

I "do" understand what you're saying about the "fing" vcore. The vcore staying the same is not the only factor that increases the heat or reduces the life of the chip. Look, you find me a source that says you can overclock without touching the vcore and it will be completely safe and "not" increase the heat output or decrease chip life and I'll shut up, okay? If you look on any overclocker enthusaist site (which I visited often about a year ago) then you'll see ads for cooling, warnings about heat and how to cool a cpu and warnings about damage and reduction of component life. These are facts. I'm addressing a couple main things you stated about overclocking: It does not increase heat, it does not reduce processor life. I'm not talking about how to use cooling to reduce heat etc.. The fact is that extra heat "is" generated. If it wasn't don't you think Tom's Hardware guide would have just left the vcore alone and gone to 5Ghz? No, cause they can't without frying the hardware. So knock these articles however you want, but I'm not going to take your "word" over these professionals, sorry.

overclocking without changing the vcore dose not generate more heat but it works up to a point and it becomes unstable becuase the voltage can't drive the extra clock speed and when you reach that point an increase in vcore is required and with that better cooling is required hence mav415's 2800+ capable heatsink/fan (out of interest what temperature are you running at?)

overclocking is useful because for example you can buy a 2GHz cpu and a 2.5GHz cpu they cost $200 and $500 respectively now for $50 you can get a good heatsink and a powerful fan to deal with the extra heat produced from overclocking from 2GHz to 2.5GHz.

(it's not exactly like this but it was like this with the early celerys and PII's)

so in essence you are getting $500 performance at a $250 price with maybe a little extra noise.

with my g3 overclock I did not toutch the vcore i just changed the multipliers and it booted fine but was a bit unstable so i poped a giant heatsink onto it and then it proved stable this is because when you cool a chip down the resistance in the die decreases and when you dod that less voltage is required so it may prove to become stable without changing the voltage though if i changed the voltage I could probably run at 550MHz there is little point in such an act

 

[ddtlm]

gekko513:

The new Pentium-M has 2MB level 2 cache and 533MHz bus. If you put a 2MB level 2 cache and a 533MHz bus on a G4, I'm almost certain that the G4 would kick some serious Pentium-M ass even if the G4 runs at 1,5GHz and the Pentium-M runs at 2,0GHz.

Back when G4's had the oversized L2's and L3's I didn't exactly see the situation reversed. Sure a huge memory system on a G4 would help but the core still has "issues" ... first and foremost the solitary scalar FP unit, and lets not forget that it's almost an in-order executing CPU.

 

[patrick0brien]

-One doesn't buy a notebook for speed.

 

[ddtlm]

patrick0brien:

When the battle can't be won, change the battle.

 

[patrick0brien]

-There is no (spoon) battle. This is an invented discussion.

 

[Mav451]

why didnt you just start a thread 'windows is better than OS X' i dont think anyone buys a laptop to have the fastest processor.

and i dont know about you, but ive never seen a wintel laptop that is noticably faster or even at a compareable speed to my 12" PB (867 for that matter!) maybe its b/c of windows runs slow or whatever, but in the basic opening of apps and running apps i notice bit more speed out of my PB. the only intel machine i know that runs 'just a bit' faster than my PB is the P4 3.0ghz desktop i have at work. even my old P4 1.5 ghz machine i had at work seemed slower than my PB.

so, i dont know what all the 'speed tests' have to say, but from the real life tests i have seen myself the G4 kicks the but of the P4 (especially if you compare Ghz to Ghz. . . )

If you knew how bad the 1.5Ghz (well anything <2.2Ghz for the P4 was based on the crappy Williamette core; e.g. the ONLY P4's beaten by the P3's with nearly a 1Ghz deficit in clock speed).

To put it in apples to apples...its like comparing a 1.33G4 vs. a 1.33G5...which would win? Or to put it even clearer, a 1.0G4 vs. a 1.0G5? (in my attempt to illustrate just how crippled/sad the first P4's were in performance).

 

[ddtlm]

patrick0brien:

There is no (spoon) battle. This is an invented discussion.

So how do these "invented" discussions differ from normal discussions?

 

[KingSleaze]

Since it grew over the course of a month, is wildly off original topic. And the original topic is a ridiculous point to begin with. Hey, a Window laptop will always stay behind a Window desktop in pure speed too. What is the point?

Old acronym GIGO.

 

[gekko513]

gekko513:


Back when G4's had the oversized L2's and L3's I didn't exactly see the situation reversed. Sure a huge memory system on a G4 would help but the core still has "issues" ... first and foremost the solitary scalar FP unit, and lets not forget that it's almost an in-order executing CPU.

The G4 never had more than 512KB on-chip. It has had a max of 2MB off-chip which isn't nearly as fast.

Of course the G4 could be better. It is after all a 4-5 year old design. The G4 doesn't need out-of-order execution nearly as much as most other processors because it has a shorter pipeline.

It is still a very good chip. And at speeds at and above 1.5GHz it is still a remarkable performer because it has kept a relatively short pipeline. But like all other processors it needs to be filled with instructions and data fast enough. The current bus and cache system is the main bottleneck. For most tasks a 2nd FP unit would be just as starved for instructions and data as the 1st.

With a 800+MHz bus and dual memory channels a 2nd FP unit would be very nice, but then again, wouldn't the 2nd unit add to the complexity of the design (determining data dependancy and timing and such)? Maybe they would have to add more stages to the pipeline. Which again would require more sofisticated branch prediction and so on and so forth. All in all a completely different design. I've understood that Freescale is also doing work on next generation processors where they address these issues, but wouldn't that come out as a G5, G5e, G6 or whatever?

It seems to me that for the G4 it would be a simpler design task to add more cache and increase the bus speed. More bang for the buck so to speak.

 

[gekko513]

but your attempt at benchmarking is pathetic... haha you are putting 3 machines up against each other in a GRAPHICS INTENSIVE PROGRAM and give one machine (the PB) TWICE the amount of VIDEO RAM... the MAIN factor in graphics like performance...

I happen to like the benchmark. It compares real world computers. And besides .. Video RAM isn't very important in Photoshop. Do you really think Apple would by default use less VRAM than most PC vendors if Photoshop would suffer by it. Graphics design is sort of the core market for Apple.

3D games are driving the need for more VRAM not Photoshop.

 

[patrick0brien]

patrick0brien:


So how do these "invented" discussions differ from normal discussions?

-ddtlm

There is no sharing here, no learning, no knowledge value. It is someone jabbing a fact into the air. And it is a fact that can change in the future. Therefore the only reason to have this conversation is to lock horns.

There is nothing we can carry away from this conversation of any real value.

 

[Mord]

I disagree i think we are having a nice contradiction of opinions on overclocking there is nothing bad about that? it's always nice to have a nice argument to better ones understanding of the subject surly is this not the point of the forums to share opinions and ideas to better oneself and to educate others what would be the point if we all agreed? and we all knew everything?