people have gotten so much into the attitude of "wait for the G5" that i dont think PM sales will recover until they use G5 or maybe quad-high performance G4 with a much much updated motherboard. At least i know i aint buying anything until that happens. Then again apple will never go under, even if their hardware bites the dust. They have too much going for them in the software house. If os X were to go down, i would go back to snail mail, writing on paper, and board games
<strong>Optimal server I/O architecture and desktop architecture are related but not the same. A server's CPU does not need to see all the data it is shuffling about so memory throughput to the CPU is not as critical as memory throughput on the disk/network bus using DMA.
The current Xserver design is not optimized for use in the renderfarm, but optimized for web/streaming/database operations. It doesn't suck at computation-farm operations, but the memory bus could be a bottleneck for the wrong set of conditions.
And because it is designed as a web/streaming/database server the fact there are no great leaps in desktop number crunching optimizations doesn't tell us much about future desktops.
Great point. They will make killer streaming content servers I bet. Apple is shoring up the industires it already owns before tackling new ones. Media providers used mac to produce content then stream it off non-Apple servers. They will also make lovely file servers for other mac centric businesses. I think Apple would have released spec benchmarks if these servers could compete as computational workhorses.
A big part of the G5 rumor mix came from Architosh. Simply because they were not considered a rumor site, their stories seemed to have more validity. After the recent server stories, however, do they have any credibility left? Oh, by the way, those stories mysteriously disappeared after the real server was introduced. Revisionist history, I guess. Anyway, take away their contribution to the G5 rumors and what do we have left. The Register and MOSR? Both seem to have backed away. Dorsal? Did he every really mention seeing one? Ah, the heck with it. I wanted to believe the G5 was coming any day now for the better part of a year, Finally, however, I am beginning to face reality. I, like so many others, want a revolutionary turn of events for Apple. We want Apple to blow the competition away. The truth, however, is the revolution won?t happen overnight. It will be long and slow, and, while requiring a lot of work from Apple, will probably seem quite boring. The G5 will come someday, but probably not at MWNY. When it finally arrives, the surrounding competition will make it about as exciting to us as the G4 is now. At least I can still dream about Jaguar. Maybe Apple really is a software company.
A message to Architosh:
I hope you?ve learned your lesson. The next time some crack-pot comes around saying he has inside info, please don?t publish it on the Architosh site. It could damage your reputation. Tell him to post it on AI instead.
Amiga? Man I've seen the 3 Stooges run a platform better than that..... <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
TING5</strong><hr></blockquote>
Maybe the 3 stooges platform will be leap frogged by Amiga.
In the Register Amiga article posted May 10th, they note that somebody who posted in ANN.lu wrote that Matrox has a new video GPU that "destroys the Geforce 4 completely."
The specs of this gpu posted on May 14th 4 days after this article was posted by the Register confirm that it is much better than the Geforce 4.
He also mentions in the same ANN.lu post that "just as Matrox is reclaiming the crown now, the G5 will do the same."
The interesting part is that the Register says they checked with the guy about these predictions and he confirmed by email that "the postings were genuine."
Well, the Matrox part seems right. So, is he right about the G5 "reclaiming the crown" from the X86?
Too bad the Register didn't email him back and ask him when he thought the G5 would be ready.
[quote]Maybe the 3 stooges platform will be leap frogged by Amiga.<hr></blockquote>
LOL!
[quote]He also mentions in the same ANN.lu post that "just as Matrox is reclaiming the crown now, the G5 will do the same."<hr></blockquote>
Heh, I remember when the Amiga fanboys were raving about the not-yet-introduced Transmeta Crusoe a couple of years ago(when Amiga were planning on using it on future machines) and how it would blow away Intel.
the amiga crwod is a biased on this whole thing as we are: they rely and have to rely on PPC chips, and their current hardware is even mroe non-existant than ours. No wonder they want something to crush the competition, to prove that their choice of platform wasn't wrong after all.
I doubt they know more about the G5 than we do, in fact they're probably getting their info from sites likes this one here.
I think, once the G5 does come out, it's not going to blow anyone away, it's just going to be a nice incremental step forward. maybe more significant than the G3 to G4 step, but seriously that didn't blow me away either.
apple has a new G4 coming with a faster bus, but it's not fabbed by moto, so they can't release it 'till the deal with moto ends this summer (does anyone know the exact date?).
Or it IS fabbed by moto, but designed for the new 0.13 micron process, and won't ship 'till those fabs are up and running.
They released the rack early to catch the ed buying season, 'cause it still does a pretty god job with the DDR for I/O only, and cause its a tried and true design, wich makes sense in a server.
<strong>Anyone else notice that the new P4 has a 533Mhz bus, 1066Mhz RDRAM, and is expected to scale to 3Ghz in the 4th quarter.... Why fight this fight when AMD and Intel are clearly winning with no hope in sight from Motorola? Even if we get an upgraded G4/G5 at MWNY, maybe we will see a 400mhz bus.. still one revision behind Intel. Give us parity with Intel on this front and Apple's job of expanding marketshare will be that much easier. Flame away.</strong><hr></blockquote>
I for one wouldn't mind being just one rev behind x86 on mobo tech.
does anyone know how difficult it would be to have a mobo with a FSB speed switch? It could be that the rackmount setup is designed for not-yet exsisting chips, which could be verified by the presence of a FSP speed toggle switch of some kind (or something similar) on the mobo. up for 133mhz for the current G4's get a new G4 that handles DDR, flip the switch, and you have a 266mhz FSB or better.
Anyone have access to one of the servers, and could take a look for how the clock speed jumpers are setup?
[just a stab in the dark, but it would be interesting to know how things look on the inside]
as soon as a mobo has jumpers taht determine the speed of the bus, you cna also have a switch.
I've done it myself and it's easy as that.
All you need is a switch fit an on and an off position, the hook it up to the jumper pins and there you are.
Combining multiple jumper settings with a switch could be done withthe aid of a little circuitry, but still easy enough to do it yourself, if you have some idea of electronics.
But Apple can make as many jumpers and switches as they want, as long as the G4 maxes out at 133Mhz, you can switch till you drop, won't help.
<strong>as soon as a mobo has jumpers taht determine the speed of the bus, you cna also have a switch.
I've done it myself and it's easy as that.
All you need is a switch fit an on and an off position, the hook it up to the jumper pins and there you are.
Combining multiple jumper settings with a switch could be done withthe aid of a little circuitry, but still easy enough to do it yourself, if you have some idea of electronics.
But Apple can make as many jumpers and switches as they want, as long as the G4 maxes out at 133Mhz, you can switch till you drop, won't help.
G-News</strong><hr></blockquote>
I think the point was that Apple's new chipset could support a possible future G4 with FSB DDR support.
Wild and crazily optimistic thought: Apple/Motorola know they are behind in this whole DDR ball game, so rather than just build in double-pumping why not go straight to quad-pumping? There's no rule that says you have to go through all the same technology steps as your competition if you fall behind at some point.
I wonder how high MPX can scale. 200MHz? Then if you double pump it, you should be able to acheive 3.2GBps bandwidth to the northbridge. That would be enough for DDR-II or 2 channels of DDR-1.
Considering the new XServer, it would seem that G5 processors are not even close to being ready. Would Apple introduce a new rackmount server with a G4 lacking a DDR frontside bus, only to announce Powermac G5s at MWNY with 500 MHz system buses?
Not a chance, I say.
<hr></blockquote>
So I´ll guess there is no way X86 server makers would use old PIII running at 1.13 ghz with only 133 mhz FSB in their offerings, when there is a 2.53 ghz PIV with 533 mhz FSB available, that would never happen, right.
Well maybe also consider that for the price range for these xservers, there was only so much they could stick in there. Also keep in mind that most of the Wintel servers use PIII's, not the highest possible chip out there. I don't know, another thought. It certainly would be nice now to have some credible source somewhere seeing some amazing new designs, but it is interesting that all these people and sources have all pretty proven to be bogus. Guess again we'll have to wait and see...
When you work with semiconductors chips, your work area, equipment, and you have to be properly grounded to protect the chips from ESD damage . And the chips themselves usually have some kind of ESD protection built into them.
Regarding Motorla's fab's, I met someone once who worked for a Motorola fab as a process engineer and he told me that their fabs are (and I'm putting this mildy) not as well run as they should be.</strong><hr></blockquote>
Thanks. I know about electrostatic discharge and it's threat to fabs, but I'm not familiar with it enough for the acronym ESD to click.
Look's like I'm the monkey's butt here....I'm shocked to hear about Moto's fabs, it sounds so incompetent, as if they don't care about making competitive chips. Go figure. Very funny though.
I've seen a lot of talk about the DDR RAM in the new XServer and how it's not completely useful. I understand the thinking, but I'm curious. What's most common in the Intel server world? Are there PIII servers with DDR Ram? Or is it just uncommon?
:Considering the new XServer, it would seem that G5 processors are not even close to being ready. Would Apple introduce a new rackmount server with a G4 lacking a DDR frontside bus
Do you even know what you're talking about? There's no such thing as a 'ddr frontside bus' or whatever. 'ddr frontside buses' run at 133 mhz [so far].
Comments
<strong>Optimal server I/O architecture and desktop architecture are related but not the same. A server's CPU does not need to see all the data it is shuffling about so memory throughput to the CPU is not as critical as memory throughput on the disk/network bus using DMA.
The current Xserver design is not optimized for use in the renderfarm, but optimized for web/streaming/database operations. It doesn't suck at computation-farm operations, but the memory bus could be a bottleneck for the wrong set of conditions.
And because it is designed as a web/streaming/database server the fact there are no great leaps in desktop number crunching optimizations doesn't tell us much about future desktops.
[ 05-15-2002: Message edited by: AirSluf ]</strong><hr></blockquote>
Great point. They will make killer streaming content servers I bet. Apple is shoring up the industires it already owns before tackling new ones. Media providers used mac to produce content then stream it off non-Apple servers. They will also make lovely file servers for other mac centric businesses. I think Apple would have released spec benchmarks if these servers could compete as computational workhorses.
A message to Architosh:
I hope you?ve learned your lesson. The next time some crack-pot comes around saying he has inside info, please don?t publish it on the Architosh site. It could damage your reputation. Tell him to post it on AI instead.
<strong>
Amiga? Man I've seen the 3 Stooges run a platform better than that..... <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
TING5</strong><hr></blockquote>
Maybe the 3 stooges platform will be leap frogged by Amiga.
In the Register Amiga article posted May 10th, they note that somebody who posted in ANN.lu wrote that Matrox has a new video GPU that "destroys the Geforce 4 completely."
The specs of this gpu posted on May 14th 4 days after this article was posted by the Register confirm that it is much better than the Geforce 4.
He also mentions in the same ANN.lu post that "just as Matrox is reclaiming the crown now, the G5 will do the same."
The interesting part is that the Register says they checked with the guy about these predictions and he confirmed by email that "the postings were genuine."
Well, the Matrox part seems right. So, is he right about the G5 "reclaiming the crown" from the X86?
Too bad the Register didn't email him back and ask him when he thought the G5 would be ready.
Here is the link again...
<a href="http://www.theregister.co.uk/content/54/25238.html"; target="_blank">http://www.theregister.co.uk/content/54/25238.html</a>;
LOL!
[quote]He also mentions in the same ANN.lu post that "just as Matrox is reclaiming the crown now, the G5 will do the same."<hr></blockquote>
Heh, I remember when the Amiga fanboys were raving about the not-yet-introduced Transmeta Crusoe a couple of years ago(when Amiga were planning on using it on future machines) and how it would blow away Intel.
I doubt they know more about the G5 than we do, in fact they're probably getting their info from sites likes this one here.
I think, once the G5 does come out, it's not going to blow anyone away, it's just going to be a nice incremental step forward. maybe more significant than the G3 to G4 step, but seriously that didn't blow me away either.
G_News
apple has a new G4 coming with a faster bus, but it's not fabbed by moto, so they can't release it 'till the deal with moto ends this summer (does anyone know the exact date?).
Or it IS fabbed by moto, but designed for the new 0.13 micron process, and won't ship 'till those fabs are up and running.
They released the rack early to catch the ed buying season, 'cause it still does a pretty god job with the DDR for I/O only, and cause its a tried and true design, wich makes sense in a server.
As for PM G5's. Not till 2003.
Just speculating
[ 05-16-2002: Message edited by: LowB-ing ]</p>
<strong>Anyone else notice that the new P4 has a 533Mhz bus, 1066Mhz RDRAM, and is expected to scale to 3Ghz in the 4th quarter.... Why fight this fight when AMD and Intel are clearly winning with no hope in sight from Motorola? Even if we get an upgraded G4/G5 at MWNY, maybe we will see a 400mhz bus.. still one revision behind Intel. Give us parity with Intel on this front and Apple's job of expanding marketshare will be that much easier. Flame away.</strong><hr></blockquote>
I for one wouldn't mind being just one rev behind x86 on mobo tech.
Anyone have access to one of the servers, and could take a look for how the clock speed jumpers are setup?
[just a stab in the dark, but it would be interesting to know how things look on the inside]
I've done it myself and it's easy as that.
All you need is a switch fit an on and an off position, the hook it up to the jumper pins and there you are.
Combining multiple jumper settings with a switch could be done withthe aid of a little circuitry, but still easy enough to do it yourself, if you have some idea of electronics.
But Apple can make as many jumpers and switches as they want, as long as the G4 maxes out at 133Mhz, you can switch till you drop, won't help.
G-News
<strong>as soon as a mobo has jumpers taht determine the speed of the bus, you cna also have a switch.
I've done it myself and it's easy as that.
All you need is a switch fit an on and an off position, the hook it up to the jumper pins and there you are.
Combining multiple jumper settings with a switch could be done withthe aid of a little circuitry, but still easy enough to do it yourself, if you have some idea of electronics.
But Apple can make as many jumpers and switches as they want, as long as the G4 maxes out at 133Mhz, you can switch till you drop, won't help.
G-News</strong><hr></blockquote>
I think the point was that Apple's new chipset could support a possible future G4 with FSB DDR support.
Wild and crazily optimistic thought: Apple/Motorola know they are behind in this whole DDR ball game, so rather than just build in double-pumping why not go straight to quad-pumping? There's no rule that says you have to go through all the same technology steps as your competition if you fall behind at some point.
Nah, too optimistic...
That's it. Give in to the dark side...
lemon bon bon
IBM touts 200MHz busses on the 750FX. (SDR)
However their tradtition for feeding us every smallest crumb of performance increase and new tech is BIG.
Powermac 9500: 120/132/150/180/200/225/233
talk about small steps.
G-News
[QB]Nudge------>>
Considering the new XServer, it would seem that G5 processors are not even close to being ready. Would Apple introduce a new rackmount server with a G4 lacking a DDR frontside bus, only to announce Powermac G5s at MWNY with 500 MHz system buses?
Not a chance, I say.
<hr></blockquote>
So I´ll guess there is no way X86 server makers would use old PIII running at 1.13 ghz with only 133 mhz FSB in their offerings, when there is a 2.53 ghz PIV with 533 mhz FSB available, that would never happen, right.
[ 05-16-2002: Message edited by: Decurion ]</p>
<strong>
ESD stands for electrostatic discharge.
When you work with semiconductors chips, your work area, equipment, and you have to be properly grounded to protect the chips from ESD damage . And the chips themselves usually have some kind of ESD protection built into them.
Regarding Motorla's fab's, I met someone once who worked for a Motorola fab as a process engineer and he told me that their fabs are (and I'm putting this mildy) not as well run as they should be.</strong><hr></blockquote>
Thanks. I know about electrostatic discharge and it's threat to fabs, but I'm not familiar with it enough for the acronym ESD to click.
Look's like I'm the monkey's butt here....I'm shocked to hear about Moto's fabs, it sounds so incompetent, as if they don't care about making competitive chips. Go figure. Very funny though.
Do you even know what you're talking about? There's no such thing as a 'ddr frontside bus' or whatever. 'ddr frontside buses' run at 133 mhz [so far].