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.
G5 PowerMacs MWNY -- Read this...
- Thread starter j763
- Start date
- Sort by reaction score
Informed Discussion!
1. Because my post count is low I don't know **** about computers? This is just plain flame bait.
2. 64-bit is the holy grail of computing? Is it? Why?
3. Because I disagree with you you assume I didn't do research???
4. Complain? Where was I complaining. Exactly the opposite, I was saying how content I would be with a faster 32-bit processor. I never said I specifically wouldn't want a 64-bit processor. If Apple released one tomorrow I would be extremely happy. All I was saying was that I don't think (from my researched knowledge of what 64-bit is) that I or the vast majority of computer users would see a great performance increase just from the fact of it being 64-bit.
5. Vector-based and Matrix math are already taken care of in the Altivec unit. This is very specifically what the Altivec unit is for.
6. If you had read my post, I did mention that film editing could make use of 64-bit. But high-res video would not at this time. All of the video standards DV, DVCPRO, DVCPRO-50, Digital Beta, even the High Definition standard use a color depth of 32 bits. That means each pixels color information can be fully represented by 32 bits. Any transformations that need to be done on those pixels would be 32 bit operations not 64. The only case in which 64 bit would be useful is (as I said before) for editing for film which has a much greater color depth.
7. Third Party hardware capabilities? Give me an example of a piece of hardware that requires the chip to have a certain bitness.
8. Multitasking is a function of numer of pipelines, not bitness. A 10 pipeline 32-bit processor will be a better multi-tasker than a 5 pipeline 64-bit processor any day.
9. Enhanced compilers? Huh? This is a benefit of 64-bit? 64-bitness will not enherently enhance compilers. The compilers will have to be rewritten to take advantage of the 64-bitness. And if history tells us anything about the addition of instructions to an instruction set (MMX, SSE, Altivec) it is that it takes a long time for compilers to come out that efficiently use the new instructions. So when the 64-bit chip first comes out you will have poor compilers not enhanced ones.
10. The internal bitness of a chip has nothing at all to do with its ability to work in a multi-processor set-up. The only time 64-bitness would help a multi-processor set up is in its ability to address more memory. But since multi-processor macs won't hit their addressing max on memory for years going at their current rate I don't think this is a problem.
11. 32-bit has to "fake" high end math. Whoever called Sparklytone on this (I don't remember who that was) sorry to say but he's right on this one. If you need to do a math function that requires 64-bit precision on a 32-bit processor you do have to use a work-around. I won't argue this point. However the number of times that these type of problems arise is very, very small.
I think that just about covers all of Sparklytone's points. I think I have been specific enough that I certainly don't deserve to be called a "newbies pretending to know $h1t about computers ". If you disagree please cite the specific points I was incorrect about and I'd be happy to look into them and entertain the idea that maybe I goofed. (It happens).
Oh, and one last thing. I thought that my strongest argument about why Apple would not release a 64-bit G5 at MWNY was the developer support issue. If anyone can point out the fault in that logic, I would like to hear it. I noticed that Sparklytone didn't mention that at all...
lets take this point by point.
1. Because my post count is low I don't know **** about computers? This is just plain flame bait.
2. 64-bit is the holy grail of computing? Is it? Why?
3. Because I disagree with you you assume I didn't do research???
4. Complain? Where was I complaining. Exactly the opposite, I was saying how content I would be with a faster 32-bit processor. I never said I specifically wouldn't want a 64-bit processor. If Apple released one tomorrow I would be extremely happy. All I was saying was that I don't think (from my researched knowledge of what 64-bit is) that I or the vast majority of computer users would see a great performance increase just from the fact of it being 64-bit.
5. Vector-based and Matrix math are already taken care of in the Altivec unit. This is very specifically what the Altivec unit is for.
6. If you had read my post, I did mention that film editing could make use of 64-bit. But high-res video would not at this time. All of the video standards DV, DVCPRO, DVCPRO-50, Digital Beta, even the High Definition standard use a color depth of 32 bits. That means each pixels color information can be fully represented by 32 bits. Any transformations that need to be done on those pixels would be 32 bit operations not 64. The only case in which 64 bit would be useful is (as I said before) for editing for film which has a much greater color depth.
7. Third Party hardware capabilities? Give me an example of a piece of hardware that requires the chip to have a certain bitness.
8. Multitasking is a function of numer of pipelines, not bitness. A 10 pipeline 32-bit processor will be a better multi-tasker than a 5 pipeline 64-bit processor any day.
9. Enhanced compilers? Huh? This is a benefit of 64-bit? 64-bitness will not enherently enhance compilers. The compilers will have to be rewritten to take advantage of the 64-bitness. And if history tells us anything about the addition of instructions to an instruction set (MMX, SSE, Altivec) it is that it takes a long time for compilers to come out that efficiently use the new instructions. So when the 64-bit chip first comes out you will have poor compilers not enhanced ones.
10. The internal bitness of a chip has nothing at all to do with its ability to work in a multi-processor set-up. The only time 64-bitness would help a multi-processor set up is in its ability to address more memory. But since multi-processor macs won't hit their addressing max on memory for years going at their current rate I don't think this is a problem.
11. 32-bit has to "fake" high end math. Whoever called Sparklytone on this (I don't remember who that was) sorry to say but he's right on this one. If you need to do a math function that requires 64-bit precision on a 32-bit processor you do have to use a work-around. I won't argue this point. However the number of times that these type of problems arise is very, very small.
I think that just about covers all of Sparklytone's points. I think I have been specific enough that I certainly don't deserve to be called a "newbies pretending to know $h1t about computers ". If you disagree please cite the specific points I was incorrect about and I'd be happy to look into them and entertain the idea that maybe I goofed. (It happens).
Oh, and one last thing. I thought that my strongest argument about why Apple would not release a 64-bit G5 at MWNY was the developer support issue. If anyone can point out the fault in that logic, I would like to hear it. I noticed that Sparklytone didn't mention that at all...
macstudent said
I can't ask myself a question directly because I'm not speaking to myself, fortunately I am schizophrenic so I get another me to ask myself a question. The other me then relays the response back to me.
Generally I don't believe the responese I get from myself since the arguement, I tend to lie to myself or do I double bluff myself???
I can't ask myself a question directly because I'm not speaking to myself, fortunately I am schizophrenic so I get another me to ask myself a question. The other me then relays the response back to me.
Generally I don't believe the responese I get from myself since the arguement, I tend to lie to myself or do I double bluff myself???
Re: Informed Discussion!
at first i thought sparkley was referring to gyroscope saying:
but he didn't correct you in thinking he was talking about you so i don't know..
either way, i think that the basis of "newbie" as a reason to disregard someone's information/opinion/insight is unsound. whereas clearly gyroscope said something that was not backed up and he didn't expand on it, just spit it out there, which i could see having a problem with. but to act as though that's a "newbie" thing is kinda dumb. even the oldest veterans of the forums sometimes say things like that. anyhoo
at first i thought sparkley was referring to gyroscope saying:
but he didn't correct you in thinking he was talking about you so i don't know..
either way, i think that the basis of "newbie" as a reason to disregard someone's information/opinion/insight is unsound. whereas clearly gyroscope said something that was not backed up and he didn't expand on it, just spit it out there, which i could see having a problem with. but to act as though that's a "newbie" thing is kinda dumb. even the oldest veterans of the forums sometimes say things like that. anyhoo
thats what i call thinkink different
man or woman i dont know your gender but i must say you really made some analyitic thinking and found a nice evidence on our upcoming, long awaited friend G5 i just wanted to congragulate you for your nice work
This is a Different thinker
continue to do so
apple for life
man or woman i dont know your gender but i must say you really made some analyitic thinking and found a nice evidence on our upcoming, long awaited friend G5 i just wanted to congragulate you for your nice work
This is a Different thinker
continue to do so
apple for life
Re: Informed Discussion!
I am not sure this is true. If you have a longer pipeline and your chip makes a poor branch prediction you end up with much more wasted space in the pipeline. This in turn, would cause poorer multitasking performance unless you had awesome branch prediction (ala the Hammer). The longer pipeline chip is going to be fast though if it hits everything just right or the predictions are dead on.
(at least this is how I understand it)
I am not sure this is true. If you have a longer pipeline and your chip makes a poor branch prediction you end up with much more wasted space in the pipeline. This in turn, would cause poorer multitasking performance unless you had awesome branch prediction (ala the Hammer). The longer pipeline chip is going to be fast though if it hits everything just right or the predictions are dead on.
(at least this is how I understand it)
So what's so hot about the G5?
Humor me here. I'm just an average Joe Mac user who's heard all this hype about the G5.
My question is, How is the G5 really gonna help me? Will it make my 3D games run faster? Will Photoshop benefit from it? iMovie? Web surfing? Or is the G5 mainly useful for high-end scientific and database applications?
I've heard so much, yet no one's really been able to say, in concrete terms, what kind of boost it'll give to my Mac experience.
One thing I've read is that the G5 supports DDR Ram whereas the G4 doesn't. What other new technologies will the G5 incorporate to make things faster, better?
On a separate note, I don't think they're gonna introduce a bluetooth mouse unless bluetooth came standard on a system. So maybe they're introducing a new Pro tower, with bluetooth built in, to go along with the new mouse...
Humor me here. I'm just an average Joe Mac user who's heard all this hype about the G5.
My question is, How is the G5 really gonna help me? Will it make my 3D games run faster? Will Photoshop benefit from it? iMovie? Web surfing? Or is the G5 mainly useful for high-end scientific and database applications?
I've heard so much, yet no one's really been able to say, in concrete terms, what kind of boost it'll give to my Mac experience.
One thing I've read is that the G5 supports DDR Ram whereas the G4 doesn't. What other new technologies will the G5 incorporate to make things faster, better?
On a separate note, I don't think they're gonna introduce a bluetooth mouse unless bluetooth came standard on a system. So maybe they're introducing a new Pro tower, with bluetooth built in, to go along with the new mouse...
thanks phantommaul...
dongmin -- The obvious guess would be that the G5 improved performance. The entire system (OS and Apps) will benifit from this. I would imagine that the general consumer market will be fine with G4 Macs, the G5 will be for the pro market (for now anyway). We'll see the G5 PowerMac and then tibook.
On the subject of bluetooth, Apple will most likely build it into the G5 PowerMacs... They're really pushing the Blueooth Technology (WWDC Sessions etc.) which is good for all users!
Also, it is becoming increasingly likely that Apple will charge for an upgrade to 10.2 from 10.1 ... Why bother releasing 10.1.5 otherwise?? I guess we'll find out soon enough...
Re: So what's so hot about the G5?
Possible G5 performance boosts (+ what it effects):
1) Higher clock frequency (everything, processor heavy stuff most)
2) On chip memory controller (everything, memory heavy stuff most, multi processor systems most)
3) DDR RAM (everything, memory heavy stuff most)
4) 64 bit (mostly scientific stuff, a few other things)
5) Hypertransport (everything, not sure what most)
6) general architecture improvements (varies, probably everything)
7) RapidIO (same as mem controller [not as good for multi proc systems], mutually exclusive with it)
8) PCI-X (PCI cards)
9) ATA100 or greater (many things, hard drive heavy stuff most)
10) AGP 8x (supports faster graphics cards)
11) USB 2 (peripherals)
12) FireWire2 (peripherals)
13) Improved Altivec (Altivec optimized stuff)
note: this is ALL the reasonable G5 rumors I've heard (that I can remember). Some are almost definitely not true.
Possible G5 performance boosts (+ what it effects):
1) Higher clock frequency (everything, processor heavy stuff most)
2) On chip memory controller (everything, memory heavy stuff most, multi processor systems most)
3) DDR RAM (everything, memory heavy stuff most)
4) 64 bit (mostly scientific stuff, a few other things)
5) Hypertransport (everything, not sure what most)
6) general architecture improvements (varies, probably everything)
7) RapidIO (same as mem controller [not as good for multi proc systems], mutually exclusive with it)
8) PCI-X (PCI cards)
9) ATA100 or greater (many things, hard drive heavy stuff most)
10) AGP 8x (supports faster graphics cards)
11) USB 2 (peripherals)
12) FireWire2 (peripherals)
13) Improved Altivec (Altivec optimized stuff)
note: this is ALL the reasonable G5 rumors I've heard (that I can remember). Some are almost definitely not true.
which first?
So should I expect to see a new PowerMac G5 before or after an updated iMac G4? say 1ghz/firewire2/1284x??? monitor bitdepth/GeForce 4.
I'm selling my Pismo500/60GB/640MB/DVD/airport/warranty for $1600 so I can get a new G4 system (or better if it comes) to do video/DVD burning.
I'm wondering if I should hang around a bit to see if Apple speedbumps it's iMacG4 and to see what the new PowerMac's look like.... well, actually, I will be waiting, but I want to know what to expect first:
revised iMac or new PowerMac
REMEMBER: the original bondi blue iMac (I've got one, also for sale) was updated to a new graphics card in just 3 Months!
So should I expect to see a new PowerMac G5 before or after an updated iMac G4? say 1ghz/firewire2/1284x??? monitor bitdepth/GeForce 4.
I'm selling my Pismo500/60GB/640MB/DVD/airport/warranty for $1600 so I can get a new G4 system (or better if it comes) to do video/DVD burning.
I'm wondering if I should hang around a bit to see if Apple speedbumps it's iMacG4 and to see what the new PowerMac's look like.... well, actually, I will be waiting, but I want to know what to expect first:
revised iMac or new PowerMac
REMEMBER: the original bondi blue iMac (I've got one, also for sale) was updated to a new graphics card in just 3 Months!
Nipsy said,
I said
Note the use of the word originally. Have you seen the modern linux distros mandrake,suse, redhat they are definately aimed at the average joe.
I said
Note the use of the word originally. Have you seen the modern linux distros mandrake,suse, redhat they are definately aimed at the average joe.
But then you get distro's like slackware and debian which definitely aren't aimed at the average joe..
So? That's saying, "The distros NOT aimed at consumers are hard for consumers, therefore, Linux is not easy for the consumer!" Those distros are NOT meant for the average joe. Those are for experienced Linux users. Not all Linux distros are meant for the beginning users. Some distros are (RedHat, Mandrake, SuSE) and some distros are not (SlackWare, Debian, LFS).
(Whoever implied that all distros are aimed at consumers is incorrect. However, there are many ones that are aimed at consumers, and they do a good job of it, IMO.)
Those distros are NOT meant for the average joe. Those are for experienced Linux users. Not all Linux distros are meant for the beginning users. Some distros are (RedHat, Mandrake, SuSE) and some distros are not (SlackWare, Debian, LFS).(Whoever implied that all distros are aimed at consumers is incorrect. However, there are many ones that are aimed at consumers, and they do a good job of it, IMO.)
Re: Re: Informed Discussion!
You are absolutely right. A longer pipeline does cause problems with branch prediction. However reread my original post. I was not talking about a 10 STAGE pipeline. I was talking about 10 discrete pipelines. Most (if not all) chips today use multiple discrete pipelines. My point was that if you want a processor to be good at multitasking it is better to spend your transistor count on more 32-bit pipelines than fewer 64-bit pipelines.
You are absolutely right. A longer pipeline does cause problems with branch prediction. However reread my original post. I was not talking about a 10 STAGE pipeline. I was talking about 10 discrete pipelines. Most (if not all) chips today use multiple discrete pipelines. My point was that if you want a processor to be good at multitasking it is better to spend your transistor count on more 32-bit pipelines than fewer 64-bit pipelines.
Re: Re: So what's so hot about the G5?
All of these things would be great, and I'm not arguing that we may see these in the next Towers, but 8-12 are not actually done in the processor. The processor communicates with another bridge chip which communicates with the various periferals. And if Hypertransport is implemented in the G5 then it will be the standard the processor uses to communicate with the various other bridge chips.
I also have a question if anyone out there knows... Is rapidIO solely a memory IO subsytem, or is it similar to hypertransport and used to connect to bridge chips for PCI, etc? Maybe I need to do some more reading.
All of these things would be great, and I'm not arguing that we may see these in the next Towers, but 8-12 are not actually done in the processor. The processor communicates with another bridge chip which communicates with the various periferals. And if Hypertransport is implemented in the G5 then it will be the standard the processor uses to communicate with the various other bridge chips.
I also have a question if anyone out there knows... Is rapidIO solely a memory IO subsytem, or is it similar to hypertransport and used to connect to bridge chips for PCI, etc? Maybe I need to do some more reading.
.posting to move this up to the top of the forums so that people who haven't seen this thread can have a look. and no, i'm not that desperate for my 'tar.