Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
G4 vs. Pentium M
- Thread starter xgsrpg
- Start date
- Sort by reaction score
Hector said:
Try this page:
http://barefeats.com/al15b.html
For the Unreal Tournament 2003 test, a 1.8GHz Centrino, with integrated Intel 855PM graphics was able to score 42 FPS in the Antalus botmatch. The 1.33GHz 17" PowerBook scored 24 FPS, and the 1.25GHz 15" PowerBook scored 23 FPS.
In the Cinebench 2003 CPU Render test, a 1.3GHz Centrino completed the test in 150 seconds, and a 1.6GHz Centrino completed it in 124 seconds. The 1.33GHz 17" PowerBook took 214 seconds, and the 1.25GHz 15" PowerBook took 227 seconds.
Obviously the clock speeds are not comparable in all cases (except in the Cinebench 2003 test, comparing the 1.3GHz Centrino to the 1.33GHz 17" PB, where the Centrino wins anyway), but this also reflects the fact that Centrino systems are available at higher clock speeds than G4 based systems are.
oingoboingo said:
I agree with the majority this time. Pentium M is a sweeeeeeeeeeet chip. Think about the powersavings. The fastest Ultra Low Voltage pentium m chip runs at 1 volt and at full blast it consumes 5 watts. That's not thermal dissipation, that's total processor consumption figure. And, AND, it runs at 1.1Ghz with the current ULV chips. Intel's roadmap shows a release of the 1.2ULV soon, possibly coinciding with their new chipsets.
If I can have a decent desktop, I'd take the 1.2ghz ULV, slot loading mobile CDRW/DVD, mini-itx board, mobile harddrive, this thing will make less noise than me when i sleep, as well as not heat up my room in the summer, and gently reminding me the efficiency of the system is helping to slow the devastation on the environment....the benefit goes on...
Hector said:
lol think,
mac os X panther vs windows xp
widows is really slow when browsing the os interface, plus its so bold
plus pcs are made for games, macs are made for luxury
its sorta like saying which is faster a rolce royce, or a ferrari enzo
abviously the ferrari, but it takes a hell of a lot of time to get to that speed, and i doubt its as comfortable as the smooth, safe, luxury ride of the rolz roice
u just cant compare em
JeffTL said:
Problem is that while I do have an iBook, my desktop is just THAT much faster. And my experience has been just fine (I'm not going to say better/worse b/c there's a distinct parity in processing power) with XP. If my experience was bad, wouldn't you be hearing it? You have to remember that whoever has a good experience and no problems, probably is NOT posting on a forum b/c they have no need to. Remember that when you hear about problems with XP--there may be just as many, or more people who do NOT have problems. The same with Linux, OSX, or apply this to car forums, etc. etc.
Forums like these, you will of course hear about the trouble cases--that is the reason we have the forums in the first place, to communicate solutions to those people.
-Do you guys realize that the Pentium M is really a Pentium III with a larger Cache and power-stepping right?
Note that I said is not esentially is in the above sentence.
Note that I said is not esentially is in the above sentence.
patrick0brien said:
Yes, I did realize that
. The chip's performance matters, though, not the architecture the chip is based on.I have used both the Pentium M and the G4 chip almost simultaneously. My dad has a 'M' chip in his Dull Latitude, which is 1.2 ghz. In real world calculations, my G4 PowerBook (1.33) is FASTER than his 'M' by a long way, even though we have the same amount of Ram, and he has a faster FSB. I have to say, it depends on what the computer is to be used for.
Finally, I get the last laugh anyway, I have a longer battery life (what was that about never un-plugging your Laptop?).
Finally, I get the last laugh anyway, I have a longer battery life (what was that about never un-plugging your Laptop?).
patrick0brien said:
I know that...of course the hilarious part is if the Intel users know this too (especially those tricked by the high clock rates of the P4's). The last Pentium III made, the Tualitin, was actually very competitive (finally) with Athlons. Of course, b/c P4's were being released then, this was hush hush.
Of course that's with Slot A Athlons, not Socket A ones that ran the FSB at Full Speed (poor Intel).
-I discussed this thread with one of my cohorts here who was a chip designer during those days. This was his two bits.
(Get your propellers on fellas)
"Advantage is it runs much, much cooler than the equivalent performance P4 at about ½ the clock rate. Which when you think about it is just sad. Technically, its the final rev of the P6 architecture that they started with in the PPro thus, its a P3 with a slightly larger pipeline microarchitecture is designed for power savings rather than clock, which is pretty novel for Intel.
The one thing that they really borrowed from the P4 is the branch predict in fact, its a bit better than the stock P4 branch predict in that its got loop detect and indirect branch predict, which is getting into, say, a powerPC chip circa 1996. The BP history table is actually quite a bit larger than the P4 version, Northwood rev., anyway. The branch predict in the northwood cores is really hamstrung by the relatively small size of the branch history tablethey made that a lot larger in the M series. Thats one of the ways you get the low power, since youre not spending time spinning in the speculative execution pipeline, which is expensive to run and very expensive to flush, which by its nature you do a lot of.
A couple other neat things in the microoperations side, but thats window dressing, the rest is the bigass cache. Power wise, its a decent story....at same process (0.13um), its a bit more than half the wattage of a comparable PowerPC 970-series for the same core voltage, but with nearly another 20 million gates, which is a heck of a story, especially for intel. Now, that PPC970 will run rings around it, but theres not a lot they can do about that with the freakishly low register count and other baggage that comes along with the x86 instruction set. In the industry, we regarded it as a neat hack, rather than anything evolutionary...the pipeline architecture is vastly inferior to IBM and AMD.
"
(Get your propellers on fellas)
"Advantage is it runs much, much cooler than the equivalent performance P4 at about ½ the clock rate. Which when you think about it is just sad. Technically, its the final rev of the P6 architecture that they started with in the PPro thus, its a P3 with a slightly larger pipeline microarchitecture is designed for power savings rather than clock, which is pretty novel for Intel.
The one thing that they really borrowed from the P4 is the branch predict in fact, its a bit better than the stock P4 branch predict in that its got loop detect and indirect branch predict, which is getting into, say, a powerPC chip circa 1996. The BP history table is actually quite a bit larger than the P4 version, Northwood rev., anyway. The branch predict in the northwood cores is really hamstrung by the relatively small size of the branch history tablethey made that a lot larger in the M series. Thats one of the ways you get the low power, since youre not spending time spinning in the speculative execution pipeline, which is expensive to run and very expensive to flush, which by its nature you do a lot of.
A couple other neat things in the microoperations side, but thats window dressing, the rest is the bigass cache. Power wise, its a decent story....at same process (0.13um), its a bit more than half the wattage of a comparable PowerPC 970-series for the same core voltage, but with nearly another 20 million gates, which is a heck of a story, especially for intel. Now, that PPC970 will run rings around it, but theres not a lot they can do about that with the freakishly low register count and other baggage that comes along with the x86 instruction set. In the industry, we regarded it as a neat hack, rather than anything evolutionary...the pipeline architecture is vastly inferior to IBM and AMD.
"
Maxx Power said:
I achieve the same beneficial effects for the environment, right now with my giant dual-processor tower, by simply putting my system to sleep or turning it off when I'm not using it.
I agree with your overall point: efficient computers are cool (heh - didn't intend that one) and quiet computers are very nice. I'm glad to see that the industry seems to be taking steps away from 120W CPU chips and towards engineering quieter hard disks.
Crikey
patrick0brien said:
Sure, if you consider that the latest Pentium-Ms have 4x the L2 Cache of the last Pentium 3s, 2X the L1 Cache (this itself is enough for most companies to rebrand their cpus), and 3-4x the FSB speed (depending on the variant of Pentium 3). In terms of other changes (all of them quite major you'll find), the Pentium M employs far more advanced branch prediction (Pentium 4 based?), a slightly longer pipeline (read the technical paper on Ars and you'll find that this implementation is primarily for scaling and micro-op fusion), the P4's SSE2 enhancements+FSB, and Micro-op fusion which is also implemented in the PPC970 (the typical performance increase from this feature alone is 5% on integer code and 9% on FP code according to Ars Technica).
I'd recommend looking up some reviews comparing the Pentium M and Pentium 3. You'll find the Pentium M to be quite a bit faster clock-to-clock.
Crikey said:
First, both these things exist. My Athlon XP-M is ALOT of power, for little voltage. (another reason it overclocks so well using stock DESKTOP voltages vs. Laptop voltages).
The big advantage is, because you can run at laptop voltages, you can get away with much, much quieter cooling (i.e., almost none or 5-volting your fans and still getting load temps lower than 45C).
Quieter HDs? They already exist in laptop drives. Compare laptop drives to desktop drives--there is a CLEAR difference. Of course, the big hurdle is the cost. Count how many people have 200-300GB of desktop HD storage (not uncommon for media storage). Compare that to laptops...um, yeah, those laptop models don't exist.
An 80GB Seagate PATA ATA/100 8MB Cache 7200RPM? $67.
80GB Toshiba ATA/100 8MB Cache 4200RPM (Mobile)? $149.
Not only is cost a factor, but the RPM's are significantly lower.
The only 7200RPM Mobile HD with even close to that capacity (80GB is not much in the desktop world, but its huge in laptops):
>> the Hitachi 60GB (w/ 8MB Cache) >> $166.
http://www.zipzoomfly.com/jsp/ProductDetail.jsp?ProductCode=100519-1
and just in case any of you fellas are wondering how a Pentium-M based laptop stacks up against desktop P4 and Athlon64 systems, heres a review:
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2129&p=7
Please note the Pentium M is actually very competitive clock-clock with the Athlon64 and even significantly better in some programs (Business Winstone, UT 2004, etc). The same certainly can't be said about the Pentium 3 to the original Athlon and the Athlon64 is a hell of alot better clock-clock than the latter processor (which was beaten fairly believably by the Pentium 4 in the end, even after significant enhancements). The Pentium M certainly beat the 3.2 GHz Pentium 4 in almost every applications, sometimes by 10+%.
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2129&p=7
Please note the Pentium M is actually very competitive clock-clock with the Athlon64 and even significantly better in some programs (Business Winstone, UT 2004, etc). The same certainly can't be said about the Pentium 3 to the original Athlon and the Athlon64 is a hell of alot better clock-clock than the latter processor (which was beaten fairly believably by the Pentium 4 in the end, even after significant enhancements). The Pentium M certainly beat the 3.2 GHz Pentium 4 in almost every applications, sometimes by 10+%.
hjhhjh said:
wtf dose that have to do with anything?, i want to see a proper benchmark and you blathering on about how great OS X is dose not mean sh*t as i use OS X every day and not even if a pentium m is five times faster than a powerbook would i buy one purely because i'm to lazy to put linux on it windows pisses me off and i'm much happier with OS X, what i would like to know is how the speed compares in a fair test so i dont just chat bull sh*t to my friends when there buying a laptop and need advice.
don't patronize someone that has over 1k posts at a forum you are a newb in.
More than MHz
Most of you, are talking about the Megahertz, yes; it might be a contributing factor to peformance, but what about the Pipeline Stages
(which apparently... which kicks bum) , and the Cache isn't that important as well, and most of all the R.A.M, if have more, the more stable its gonna be. Windows or Mac. G4 might not be the geeks dream (but G5 might ) , but it certainly gets the job done without hassle, and has a good processor architecture.
I hope you think beyond the MHz.
Most of you, are talking about the Megahertz, yes; it might be a contributing factor to peformance, but what about the Pipeline Stages
(which apparently... which kicks bum) , and the Cache isn't that important as well, and most of all the R.A.M, if have more, the more stable its gonna be. Windows or Mac. G4 might not be the geeks dream (but G5 might
) , but it certainly gets the job done without hassle, and has a good processor architecture.I hope you think beyond the MHz.
I'm not sure if that barefeats test really says much. The first test is a game, which is OK if you want to test game performance, but we all know that all game s are optimized for PC/Win but for Mac they are just ported so that they barely work. Games aren't good benchmarks for the performance potential of processors.
The cinebench test is more disturbing, but still: "The tool is set to deliver accurate benchmarks by testing not only a computer's raw processing speed but also all other areas that affect system performance such as OpenGL, multithreading, multiprocessors and Intel's new HT Technology."
Which means that it doesn't just test the cpu speed. I also find it very strange that the four PC laptops that were used in the first test are not the same as the three that were used in the second test Overall a very little documented test that really doesn't say much.
I'm not really disputing the claim that the Pentium M is a very good chip, my understanding as also that it is good, but we haven't really seen any benchmarks that compare the Pentium M and the G4 in meaningful ways.
The cinebench test is more disturbing, but still: "The tool is set to deliver accurate benchmarks by testing not only a computer's raw processing speed but also all other areas that affect system performance such as OpenGL, multithreading, multiprocessors and Intel's new HT Technology."
Which means that it doesn't just test the cpu speed. I also find it very strange that the four PC laptops that were used in the first test are not the same as the three that were used in the second test
Overall a very little documented test that really doesn't say much.I'm not really disputing the claim that the Pentium M is a very good chip, my understanding as also that it is good, but we haven't really seen any benchmarks that compare the Pentium M and the G4 in meaningful ways.
gekko513 said:I'm not sure if that barefeats test really says much. The first test is a game, which is OK if you want to test game performance, but we all know that all game s are optimized for PC/Win but for Mac they are just ported so that they barely work. Games aren't good benchmarks for the performance potential of processors.
The cinebench test is more disturbing, but still: "The tool is set to deliver accurate benchmarks by testing not only a computer's raw processing speed but also all other areas that affect system performance such as OpenGL, multithreading, multiprocessors and Intel's new HT Technology."
Which means that it doesn't just test the cpu speed. I also find it very strange that the four PC laptops that were used in the first test are not the same as the three that were used in the second test
I'm not really disputing the claim that the Pentium M is a very good chip, my understanding as also that it is good, but we haven't really seen any benchmarks that compare the Pentium M and the G4 in meaningful ways.
i wasnt insulting ne one ... jeez, plus what would a 1.33 ghz g4 compare to pentium m wise?
I found the machines that have the Pentium-M respond more slower than my Powerbook G4 1.5Ghz. The Pentium-M may be faster but it's poorly implemented with Windows. I haven't tried it with Linux yet.
Also my G5 spanks the Pentium-M.
Also my G5 spanks the Pentium-M.
gee whiz
jeez folks.
I get tired of reading internet forums sometimes.
The pentium M is a far superior architecture to the G4.
Honestly, the PPC 970 is not so good in comparison to anything that intel/AMD have to offer, either.
A DUAL G5, runs close to what a SINGLE Athlon64/Pentium 4/Pentium-M as far as raw number crunching goes.
Hopefully, the new PPC970GX with 1MB of L2 cache will narrow the gap.. Even better, if Apple decides to go with the PPC 970MP. drool. dual dual-core chips. that would be very nice, and quite probably enough to take the processing crown away from the AMD64.
All of the specifications for the above chips are available on spec's site. The integer/floating point capabilities of these chips. I'm not trying to troll or spread FUD.
The operating system that these chips run... that is something that you have control over. And the overall snappiness of your interface has a lot to do with how much system overhead there is.
OS X is plenty use-able on a 1GHz G4. Windows XP is not.
Just chill out. It's a computer. Usability is what matters to the average user. If you are crunching numbers, or making movies, or compiling code... That is when all of the above stuff really matters.
Moo
jeez folks.
I get tired of reading internet forums sometimes.
The pentium M is a far superior architecture to the G4.
Honestly, the PPC 970 is not so good in comparison to anything that intel/AMD have to offer, either.
A DUAL G5, runs close to what a SINGLE Athlon64/Pentium 4/Pentium-M as far as raw number crunching goes.
Hopefully, the new PPC970GX with 1MB of L2 cache will narrow the gap.. Even better, if Apple decides to go with the PPC 970MP. drool. dual dual-core chips. that would be very nice, and quite probably enough to take the processing crown away from the AMD64.
All of the specifications for the above chips are available on spec's site. The integer/floating point capabilities of these chips. I'm not trying to troll or spread FUD.
The operating system that these chips run... that is something that you have control over. And the overall snappiness of your interface has a lot to do with how much system overhead there is.
OS X is plenty use-able on a 1GHz G4. Windows XP is not.
Just chill out. It's a computer. Usability is what matters to the average user. If you are crunching numbers, or making movies, or compiling code... That is when all of the above stuff really matters.
Moo