About the back edge of the board not being a straight-edge. If you look at current G4s, both the IO panel and the PCI slot panels are recessed into the case about 3/4 of an inch. Perhaps Apple desires to have the IO panel more flush with the back of the case while keeping the PCI slots recessed...
I am going to tinker with the image in Photoshop, enlarge it and hopefully get some detail, then label as many things as I can as clearly as I can. Wish me luck.
I have been told by a friend who works closely with Apple that rack mount servers are coming very soon but he is not sure about the exact timing although his boss was assured by Apple they would be out by WWDC.
The interesting thing he said to me when I showed this MB to him is that he has heard that the rackmounts will NOT ship with a hard drive. He also mentioned that there was some type of Fiber communication being implemented but that was all that was said. Take it all with a grain of salt.
Right... but that's the STUDENT'S computer that needs the AirPort card... the AirPort base station is still connected to the server via Ethernet... correct?</strong><hr></blockquote>
Not unless they upgrade to the 2.0 base stations. And why not; every single computer Apple ships today supports airport.
Hats off to Apple, they got onto this one very fast. Just goes to show that if you want to pass comments onto Apple, here is as good a place as any. Someone at Apple will surley read it.
For my two peneyth, It is boviously as a protype and I see now reason why a finished product should bear the same layout. If you are building prototypes you only need to build a board that is easy to workwith. It doesn't matter about location it is performance etc that they will want to test.
If Apple wants a rackmount, it's going to need something more volume efficient than what that board implies. It needs at least 4 CPUs per 1U, or even better, a blade set-up with 16+ CPUs per 3U or something like that. It needs a robust OS and incentive for people to switch too.
Could that mystical 3rd "IDE" slot be in fact some new interface to have multiple of these boards hooked together for a superfast cluster?
The image with the prototpye in the case makes sense, but as you all noticed, power and IDE are on the totally wrong side, and even worse, that third "IDE" slot seems to face right into the top of the case...
Maybe my idea with a case-to-case connector for clustering wasn't all that far off. Maybe you'll be able to stick the boxes together like LEGO with that interface, no cablesalad behind the 19" rack.
[quote]The interesting thing he said to me when I showed this MB to him is that he has heard that the rackmounts will NOT ship with a hard drive. He also mentioned that there was some type of Fiber communication being implemented but that was all that was said. Take it all with a grain of salt.<hr></blockquote>
hmmm...that's interesting. i actually have heard a very similar story.
people were complaining about the lack of renderfarm-like hardware. now imagine this: apple just delivers a 1u rackmount cpu-module connected via firewire2 to another one. no special intelligence needed. brute cpu-power. they could make such a machine quite cheap.
this fits with what kerris said about 10.2 "we're doing some stuff under the hood that's never been done on any other platform."
if this is indeed true, i'll predict they'll sell like hot cakes.
<strong>Ebay pulled it. Great big bunch of wankers, in my opinion. I bet it was a doctored image from someone who knew that Apple test mobos are red.
133 MHz bus? For the NEXT generation? Please. Apple would be foolishto release anyhting lower than 200 MHz.</strong><hr></blockquote>
True, ture. I've heard that it wasn't just EBat pulling it, as anyone would expect (if someones already mentioned this, than I'm sorry, but I felt the need to spread the word )
Apple had something to do with the thing getting pulled.
They might be going for a total make over on there QuickSilver models... No one knows. Maybe new technology has come out that we don't know about... Who knows. Guess we just have to wait to find out...
Taiwan newspaper reported that a Taiwan factory is making "server" for HP and Apple a few months ago. I thought is was a mistake, because before iPod, they reported an Apple "PDA" which play music.
I don't have any hard information from anyone on this, and not really sure if my input is going to help as a few people are noting similar things, but...
I can't see it's practical to have IDE connectors on the open side of a "door", so this board is not flipped. This implies that the PCI slots are at the top of the board when a tower like the El Capitan is used - to me this doesn't sound logical since that would put the conectors to the PCI/AGP at the top and the USB/modem/ethernet/Firewire at the bottom, which seems a bit of an ungainly configuration when adjusting cables etc.
However, this is a bit thin on reasoning, I do know that there have been complaints about Apple's boards having the PCI cards the "wrong way around" in the sense that when put into rack-mount cases the PCI/AGP cards require risers that are "wrong way around".
The risers in this case rotate PCI cards through 90 degrees to be parallel with the motherboard. The point being that a G3/G4 board can be put in a 2U rack case. One supplier of Mac rack-mount cases is supposed to be working on a 1U case. In these third party case I think they can only handle a single PCI card.
It's my opinion that Apple can't afford to have multiple high-end boards, so it may make sense to have a G4 board - for a limited client base - that can easily be put in a rack-mount case while not having access to all PCI slots.
I really don't think Apple would build a specific rack-mount/server board with 4 PCI slots - they are just not necessary. The market seems to want 1/2U rack-mounts for economy of space, having 4 PCI slots perpendicular to the motherboard makes a 4U rack-mount which gives no advantages over current units (except that people don't have to buy third party cases).
Leave some room for the front panel and speaker, and this is what you get:
All this information is just me making educated guesses. What do you think??
-Mike
All I will say is that from what I held, is that you had better leave a LOT more room for the heat sink than you have in your illustration. Let's just say 3 times more space for it.
"All I will say is that from what I held, is that you had better leave a LOT more room for the heat sink than you have in your illustration.
Let's just say 3 times more space for it. "
Undoubtedly I made several errors in the illustration. But I should say that it does no justice to the 3-dimensional world we live in. Remember that the heat sink could easily extend 4 inches up from the board itself since there appears to be nothing blocking it from entering the depths of the case.
It occured to me just now that if Apple was half as smart as we might expect, that they would intentionally place this picture on eBay for the sake of tracking our speculations on their design, and from that make their final revisions. Another crazy speculation in itself i know, but it seems like already there have been several comments made that could potentially improve the design from a user experience stand point.
I'd like to see what anybody else has come up with for a possible case design/layout...
It occured to me just now that if Apple was half as smart as we might expect, that they would intentionally place this picture on eBay for the sake of tracking our speculations on their design, and from that make their final revisions. </strong><hr></blockquote>
What if those aren't really IDE connectors? They could be SCSI, would make sense in a server.
And this coulod be someone making a hoax. Take a mobo from a el-capitan Mac and color it red. Then add some DDR slots, move some ports around trim the board in a few places so you can just super impose one on the other and see it's a fake as easily. Then put it up on eBay for a while, then tell eBay that Apple wants the auction removed. Sure it's a bit far fetched and overly complicated but it's possible.
I think it's real though. Apple never seems to do what I think they should. I didn't think they would make a fp iMac or that an iMac would have a G4 before the tower got a G5. I didn't think there would be a larger iPod. I didn't think the towers would get a speed boost as soon as they did, and I thought the middle MHz one would be the dp.
And this could be a G5 ready motherboard. Apple releases the mobo now, then the G5 daughter cards a couple months later. Wasn't the yosemite mobo G4 ready (heard that somewhere)?
I think when the G4 first came out Apple went the other way around. They slap a G4 on an essentially B&W PCI MoBo, and hen a fewq months later they gave the G4 it's very own AGP MoBo.
<strong>I think when the G4 first came out Apple went the other way around. They slap a G4 on an essentially B&W PCI MoBo, and hen a fewq months later they gave the G4 it's very own AGP MoBo.</strong><hr></blockquote>
As far as I remeber it the first round of G4s were new boards, except for the entry level model which used the old B&W G3 board. The entry level being 400MHz to start with, than downgraded to a 350. :-)
A little later AGP was introduced across the three machines... but this may have been a year to 18 months after the initial release.
The easist way to tell these first generation G4s apart is that the entry level model has the speaker/mic jacks side by side, and the modem port is up and to the right, whereas the mid and high end machines have the speaker/mic jacks in vertical alignment, and the modem is directly to the right of them (not above and right).
Hmmm. Has anyone else noticed that with the new layout it might be easy to remove some or all of the PCI slot end of the motherboard and only lose non-server related components?
If you cut along the AGP slot and battery, you get rid of PCI slots, airport, and modem. Things that probably aren't needed in a rackmount render-server. Replace the AGP slot with onboard video, and things get really compact.
In fact, I wonder if you could fit two of the resulting motherboards in a nice 1U case...
Comments
The interesting thing he said to me when I showed this MB to him is that he has heard that the rackmounts will NOT ship with a hard drive. He also mentioned that there was some type of Fiber communication being implemented but that was all that was said. Take it all with a grain of salt.
<strong>
Right... but that's the STUDENT'S computer that needs the AirPort card... the AirPort base station is still connected to the server via Ethernet... correct?</strong><hr></blockquote>
Not unless they upgrade to the 2.0 base stations. And why not; every single computer Apple ships today supports airport.
For my two peneyth, It is boviously as a protype and I see now reason why a finished product should bear the same layout. If you are building prototypes you only need to build a board that is easy to workwith. It doesn't matter about location it is performance etc that they will want to test.
[ 05-04-2002: Message edited by: Eugene ]</p>
The image with the prototpye in the case makes sense, but as you all noticed, power and IDE are on the totally wrong side, and even worse, that third "IDE" slot seems to face right into the top of the case...
Maybe my idea with a case-to-case connector for clustering wasn't all that far off. Maybe you'll be able to stick the boxes together like LEGO with that interface, no cablesalad behind the 19" rack.
G-News
hmmm...that's interesting. i actually have heard a very similar story.
people were complaining about the lack of renderfarm-like hardware. now imagine this: apple just delivers a 1u rackmount cpu-module connected via firewire2 to another one. no special intelligence needed. brute cpu-power. they could make such a machine quite cheap.
this fits with what kerris said about 10.2 "we're doing some stuff under the hood that's never been done on any other platform."
if this is indeed true, i'll predict they'll sell like hot cakes.
<strong>Ebay pulled it. Great big bunch of wankers, in my opinion. I bet it was a doctored image from someone who knew that Apple test mobos are red.
133 MHz bus? For the NEXT generation? Please. Apple would be foolishto release anyhting lower than 200 MHz.</strong><hr></blockquote>
True, ture. I've heard that it wasn't just EBat pulling it, as anyone would expect (if someones already mentioned this, than I'm sorry, but I felt the need to spread the word
Apple had something to do with the thing getting pulled.
They might be going for a total make over on there QuickSilver models... No one knows. Maybe new technology has come out that we don't know about... Who knows. Guess we just have to wait to find out...
I can't see it's practical to have IDE connectors on the open side of a "door", so this board is not flipped. This implies that the PCI slots are at the top of the board when a tower like the El Capitan is used - to me this doesn't sound logical since that would put the conectors to the PCI/AGP at the top and the USB/modem/ethernet/Firewire at the bottom, which seems a bit of an ungainly configuration when adjusting cables etc.
However, this is a bit thin on reasoning, I do know that there have been complaints about Apple's boards having the PCI cards the "wrong way around" in the sense that when put into rack-mount cases the PCI/AGP cards require risers that are "wrong way around".
The risers in this case rotate PCI cards through 90 degrees to be parallel with the motherboard. The point being that a G3/G4 board can be put in a 2U rack case. One supplier of Mac rack-mount cases is supposed to be working on a 1U case. In these third party case I think they can only handle a single PCI card.
It's my opinion that Apple can't afford to have multiple high-end boards, so it may make sense to have a G4 board - for a limited client base - that can easily be put in a rack-mount case while not having access to all PCI slots.
I really don't think Apple would build a specific rack-mount/server board with 4 PCI slots - they are just not necessary. The market seems to want 1/2U rack-mounts for economy of space, having 4 PCI slots perpendicular to the motherboard makes a 4U rack-mount which gives no advantages over current units (except that people don't have to buy third party cases).
All this information is just me making educated guesses. What do you think??
-Mike
All I will say is that from what I held, is that you had better leave a LOT more room for the heat sink than you have in your illustration. Let's just say 3 times more space for it.
Of course, this is a prototype.
Let's just say 3 times more space for it. "
Undoubtedly I made several errors in the illustration. But I should say that it does no justice to the 3-dimensional world we live in. Remember that the heat sink could easily extend 4 inches up from the board itself since there appears to be nothing blocking it from entering the depths of the case.
It occured to me just now that if Apple was half as smart as we might expect, that they would intentionally place this picture on eBay for the sake of tracking our speculations on their design, and from that make their final revisions. Another crazy speculation in itself i know, but it seems like already there have been several comments made that could potentially improve the design from a user experience stand point.
I'd like to see what anybody else has come up with for a possible case design/layout...
<strong>
It occured to me just now that if Apple was half as smart as we might expect, that they would intentionally place this picture on eBay for the sake of tracking our speculations on their design, and from that make their final revisions. </strong><hr></blockquote>
No, Apple is not that stupid.
And this coulod be someone making a hoax. Take a mobo from a el-capitan Mac and color it red. Then add some DDR slots, move some ports around trim the board in a few places so you can just super impose one on the other and see it's a fake as easily. Then put it up on eBay for a while, then tell eBay that Apple wants the auction removed. Sure it's a bit far fetched and overly complicated but it's possible.
I think it's real though. Apple never seems to do what I think they should. I didn't think they would make a fp iMac or that an iMac would have a G4 before the tower got a G5. I didn't think there would be a larger iPod. I didn't think the towers would get a speed boost as soon as they did, and I thought the middle MHz one would be the dp.
And this could be a G5 ready motherboard. Apple releases the mobo now, then the G5 daughter cards a couple months later. Wasn't the yosemite mobo G4 ready (heard that somewhere)?
<strong>I think when the G4 first came out Apple went the other way around. They slap a G4 on an essentially B&W PCI MoBo, and hen a fewq months later they gave the G4 it's very own AGP MoBo.</strong><hr></blockquote>
As far as I remeber it the first round of G4s were new boards, except for the entry level model which used the old B&W G3 board. The entry level being 400MHz to start with, than downgraded to a 350. :-)
A little later AGP was introduced across the three machines... but this may have been a year to 18 months after the initial release.
The easist way to tell these first generation G4s apart is that the entry level model has the speaker/mic jacks side by side, and the modem port is up and to the right, whereas the mid and high end machines have the speaker/mic jacks in vertical alignment, and the modem is directly to the right of them (not above and right).
Any board that came later, inlcuding the downgraded ones, had AGP 2x, codename Sawtooth.
the audio input wasn't abandoned before the digital audio came in, which could be called Sawtooth 1.5.
Atm were at quicksilver without audio in and AGP 4x.
G-News
If you cut along the AGP slot and battery, you get rid of PCI slots, airport, and modem. Things that probably aren't needed in a rackmount render-server. Replace the AGP slot with onboard video, and things get really compact.
In fact, I wonder if you could fit two of the resulting motherboards in a nice 1U case...
=The Cynic=