Do not confuse PCI-X and PCI express, they are completely different beasts. What used to be called 3GIO (amongst many other names) is PCI express is a new physical interface, essentially parallel, and very fast, with small connectors. PCI express is not yet available.
PCI-X is an extension to PCI that enables it to run at higher speeds, and much higher bandwidths. PCI-X is widely available already (at least at it's slower speeds).
michael
Sorry, you are right - PCI-express used to be 3GIO but it is a serial technology: "PCI Express is a new serial I/O technology compatible with the current PCI software environment" (from my first link above). I had a pretty quick scan of the press releases for the tech data and missed the difference in the 2 standards.
It looks like either one could be used as a replacement for PCI in different environments but that PCI-X is the one intended for desktop computer use. Sound right?
From the throughput numbers I quoted though it doesn't look like the "200MHz bus" snippet from MB could refer to the card bus. It seems, if true, it must be the memory bus which would be at 200MHz for DDR400 (PC3200).
I honestly doubt that Apple is going to release a serial product based on DDR400 SDRAM, simply because DDR400 SDRAM is not a JEDEC standard specification and is still only widely used by hardware fanatics and so far by no commercial manufacturer, as far as I know.
I consider it much more likely that the machines will be based on DDR-II standard, but I don't know how clockspeeds will look like on that.
I'd be honestly suprised to see Apple ship DDR400 machines.
DDR400 will be certified within the next month or 2. Micron is meant to have certified ram out by midyear and I imagine several other manufacturers are planning the same.
I honestly doubt that Apple is going to release a serial product based on DDR400 SDRAM, simply because DDR400 SDRAM is not a JEDEC standard specification and is still only widely used by hardware fanatics and so far by no commercial manufacturer, as far as I know.
I consider it much more likely that the machines will be based on DDR-II standard, but I don't know how clockspeeds will look like on that.
I'd be honestly suprised to see Apple ship DDR400 machines.
G-News
When JEDEC fails, the computer industry can hardly decide "oh well, let's keel over and die". JEDEC failed by expecting DDR-II to replace DDR333 last year. They were wrong, and the computer industry has gone foward without them. I would have no problem with Apple doing the same, rather than being puritanical.
Sorry, you are right - PCI-express used to be 3GIO but it is a serial technology: "PCI Express is a new serial I/O technology compatible with the current PCI software environment" (from my first link above). I had a pretty quick scan of the press releases for the tech data and missed the difference in the 2 standards.
It looks like either one could be used as a replacement for PCI in different environments but that PCI-X is the one intended for desktop computer use. Sound right?
MM
When I said essentially parallel, I meant essentially serial, I'm too tired and old for this.
PCI express is the probable future of most machines, but just isn't here yet. It has much greater, and more scalable, bandwidth, smaller connectors, easier board layout etc. PCI-X is just a stop gap improvement on PCI, which is reasonably compatible,
but even 80% of the new Intel hardware is going to be a lot better than we've got right now!
Damn right! If the fastest Apple machine can (overall) match (or beat) 80% of the fastest Intel P4 machine, I will be one *ecstatic* camper. I'm hoping (and growing more confident) that IBM has pulled a miracle on a shoe string.
My main worry is that IBM/Apple will have done the near equivalent of coming second in the Marathon while missing one leg and the response from Apple fanatics will be, "Apple's so lame... Intel came in first...".
Of course, being from Canada, maybe I'm just used to gauging progress by whether gap between us and the front runner is shrinking or expanding. The idea of actually closing the gap is too absurd to contemplate and I'm certainly not going to set myself up for disappointment by expecting the impossible.
By the way, I will not be at all suprised if the 970 can actually beat a P4 on a few tests. As I said, the metric of CPU measurement is nowhere near one dimensional.
When JEDEC fails, the computer industry can hardly decide "oh well, let's keel over and die". JEDEC failed by expecting DDR-II to replace DDR333 last year. They were wrong, and the computer industry has gone foward without them. I would have no problem with Apple doing the same, rather than being puritanical.
Actually, standards like JEDEC's are crucial to any commodity market. JEDEC-compliant RAM has a good chance of being compatible between vendors. DDR 400 appeared because DDR II took a bit too long to show up, but as yet it's still uncommon, and since it relies on a 200MHz real clock, it's not ideal. DDR II is quad-pumped, so the real clocks are still 100, 133 and 167. Lower is better from an engineering standpoint, and the presence of an independent standard will make OEMs less nervous about adopting the tech, so I expect DDR II to supplant DDR 400 rather quickly.
Also, once you engineer for DDR II 400, it's a quick and easy path up to 533 and higher.
I would have no problem with Apple doing the same, rather than being puritanical.
Except that going off the beaten track means costs go through the roof, comparatively speaking. [Edit here - DDR400 memory is not quite as rare as I thought. It's only about 50% more than DDR333. Still, it's a pretty big jump in costs]
It's why Apple has to fight so hard for everything. It used to be that if you used the industry standard, your cost might be 'x', and if you did it yourself, your cost might be '2x'. But because of the incredible economies of scale and the vast amount of investment in process, the industry standard now costs '0.1x', while doing it yourself still costs '2x'. Small wonder that Apple has to abandon just about any technology that is unique to it.
It's why innovation is so tough. You don't have to be a factor of 2 better, you have to be a factor of 20. Small wonder that its the Dell's of the hardware world that are still making money.
[I]By the way, I will not be at all suprised if the 970 can actually beat a P4 on a few tests. As I said, the metric of CPU measurement is nowhere near one dimensional.
Heh, you're not kidding -- more likely "infinitely dimensional" is more like it.
Even if a single 970 comes in at 80% the performance of a top-of-the-line P4, Apple will be shipping SMP machines and the 970 + OSX will do better in terms of price/performance in a multiprocessor environment than the P4.
Not that it matters since i am firmly camped in the
"It's not the megahertz Its THE motherboard stupid!"
I haven't checked myself, but if you look at the SPECint of the upcoming P4s (or the Opterons) and the IBM 970 estimates it'll probably be around 80%. Keep in mind that the G4 would be about 20% so the 970 is a huge leap forward.
"Standard" PCI (1.0?) is 33MHz and 32 bit giving 133 MB/s.
There are also 66 MHz and 66MHz/64 bit PCI flavours giving 266MB/s and 533 MB/s respectively.
Going to www.pcisig.org and checking the news page it seems that PCI-X used to be known as 3GIO. It has an "initial bit rate of 2.5 Gigabits per second per lane per direction" and a "16-lane PCI Express interconnect can provide data transfer rates of more than 8 Gigabytes per second". (Here: http://www.pcisig.com/news_room/news...ses/2002_07_23 )
PCI-X 2.0 has 2 flavours: PCI-X 266 and PCI-X 533 which give 2133 MB/s and 4267 MB/s respectively. (These come from here: http://www.pcisig.com/news_room/news...es/2002_07_23b ) I think these must be DDR rates at 64 bit each since they use the phrase "PCI-X 266, runs at speeds up to 266 Mega transfers per second".
MM
It seems that PCI Express is winning "hearts and minds" in the PC development community over PCI-X, but, unfortunately, PCI-Express also appears to have slipped in its intended deployment from fall of '03 to fall of '04.
Is there any actual indication whether Apple is moving away from the existing PCI bus? What about AGP 8X in the mean time (one or two AGP 8X slots are permitted under the spec)?
I also wonder if Apple will adopt serial ATA on the new logic boards. At the present time the performance difference between serial and parallel ATA is not large, but the configuration and cabling is said to be simpler and better. It would also open the way for the improved performance serial ATA drives that will undoubtedly be available during the manufacturing life cycle of the new boards.
This seems to come down to the old argument of which is more important. Total power or the complete package. If the 970 upon initial introduction was only 80% of the fastest Pentium or AMD overall, but provided a "complete" package, that functioned cohesively and reliably, day in and day out. I will take the complete package.
What we will see, is the same thing as is currently seen in the Wintel world. The Power PC will win out at certain tasks, and the P4 3.06 and AMD 3000's will win out at certain tasks. The differentiating factor will be the total package including the OS. It is there that Apple will win hands down. Intel and AMD will bring out faster chips. Apple and IBM will bring out faster chips. Based upon a pragmatic roadmap, they will each periodically leapfrog each other, with each platform having specific areas where it excells. Due to Motorola's mis-managment and shortsightedness, Apple was left in the lurch, falling further behind each year. If they attain overall parity or near parity in performance, then the rest of the package will drive them forward.
I suspect that Apple and Big Blue will now lead the charge, and will become the driving forces for growth and innovation in an industry that was stagnating.
And contrary to many younger Mac users, I never want to see Apple as #1. I would like to see them grow to about 20-25% market share, and then continue to be the driving and innovative force behind the industry.
I don't want what everyone else has. I don't want to have some ficticious bragging rights. I want to be out on the leading edge. That is where Apple belongs.
...I don't want what everyone else has. I don't want to have some ficticious bragging rights. I want to be out on the leading edge. That is where Apple belongs.
wise words
Quote:
Originally posted by Junkyard Dawg
Right on, brother! I hope Apple never becomes another Microsoft. Power has a insidious way of corrupting humans.
And contrary to many younger Mac users, I never want to see Apple as #1. I would like to see them grow to about 20-25% market share, and then continue to be the driving and innovative force behind the industry.
I don't want what everyone else has. I don't want to have some ficticious bragging rights. I want to be out on the leading edge. That is where Apple belongs.
I dunno. My AAPL holdings would certainly appreciate Monopoly status!
And contrary to many younger Mac users, I never want to see Apple as #1. I would like to see them grow to about 20-25% market share, and then continue to be the driving and innovative force behind the industry.
Agreed. Imagine if Microsft was all that was left invoating. When Apple inovates we get iPods, and other cool software and hardware. When Microsoft innovates we get portable toilets with internet access. I think it speaks voulmes.
course the upcoming 980 and 990 will trounce those numbers.
and they don't include a 2.0ghz , 2.3, or 2.5 ghz 970 so his argument seems shallow.
His argument is simply that at the time of the 970's initial introduction it will probably be at around 80% of the top-of-the-line Intel performance of the same day, based on the facts that we know about the 970. There are no facts saying that there will be a 970 at >1.8 GHz so arguing that a 2.5 GHz 970 will do better is currently just wishful thinking.
Comments
Originally posted by mmicist
NO
Do not confuse PCI-X and PCI express, they are completely different beasts. What used to be called 3GIO (amongst many other names) is PCI express is a new physical interface, essentially parallel, and very fast, with small connectors. PCI express is not yet available.
PCI-X is an extension to PCI that enables it to run at higher speeds, and much higher bandwidths. PCI-X is widely available already (at least at it's slower speeds).
michael
Sorry, you are right - PCI-express used to be 3GIO but it is a serial technology: "PCI Express is a new serial I/O technology compatible with the current PCI software environment" (from my first link above). I had a pretty quick scan of the press releases for the tech data and missed the difference in the 2 standards.
It looks like either one could be used as a replacement for PCI in different environments but that PCI-X is the one intended for desktop computer use. Sound right?
From the throughput numbers I quoted though it doesn't look like the "200MHz bus" snippet from MB could refer to the card bus. It seems, if true, it must be the memory bus which would be at 200MHz for DDR400 (PC3200).
MM
Originally posted by Tom West
IBM really has pulled off a miracle with the 970. The miracle is that it looks like it will have 80% of the performance of a P4
I wouldn't be betting on that number being accurate.
I consider it much more likely that the machines will be based on DDR-II standard, but I don't know how clockspeeds will look like on that.
I'd be honestly suprised to see Apple ship DDR400 machines.
G-News
Originally posted by G-News
I honestly doubt that Apple is going to release a serial product based on DDR400 SDRAM, simply because DDR400 SDRAM is not a JEDEC standard specification and is still only widely used by hardware fanatics and so far by no commercial manufacturer, as far as I know.
I consider it much more likely that the machines will be based on DDR-II standard, but I don't know how clockspeeds will look like on that.
I'd be honestly suprised to see Apple ship DDR400 machines.
G-News
When JEDEC fails, the computer industry can hardly decide "oh well, let's keel over and die". JEDEC failed by expecting DDR-II to replace DDR333 last year. They were wrong, and the computer industry has gone foward without them. I would have no problem with Apple doing the same, rather than being puritanical.
Barto
Originally posted by MartianMatt
Sorry, you are right - PCI-express used to be 3GIO but it is a serial technology: "PCI Express is a new serial I/O technology compatible with the current PCI software environment" (from my first link above). I had a pretty quick scan of the press releases for the tech data and missed the difference in the 2 standards.
It looks like either one could be used as a replacement for PCI in different environments but that PCI-X is the one intended for desktop computer use. Sound right?
MM
When I said essentially parallel, I meant essentially serial, I'm too tired and old for this.
PCI express is the probable future of most machines, but just isn't here yet. It has much greater, and more scalable, bandwidth, smaller connectors, easier board layout etc. PCI-X is just a stop gap improvement on PCI, which is reasonably compatible,
michael
Damn right! If the fastest Apple machine can (overall) match (or beat) 80% of the fastest Intel P4 machine, I will be one *ecstatic* camper. I'm hoping (and growing more confident) that IBM has pulled a miracle on a shoe string.
My main worry is that IBM/Apple will have done the near equivalent of coming second in the Marathon while missing one leg and the response from Apple fanatics will be, "Apple's so lame... Intel came in first...".
Of course, being from Canada, maybe I'm just used to gauging progress by whether gap between us and the front runner is shrinking or expanding. The idea of actually closing the gap is too absurd to contemplate
By the way, I will not be at all suprised if the 970 can actually beat a P4 on a few tests. As I said, the metric of CPU measurement is nowhere near one dimensional.
Originally posted by Barto
When JEDEC fails, the computer industry can hardly decide "oh well, let's keel over and die". JEDEC failed by expecting DDR-II to replace DDR333 last year. They were wrong, and the computer industry has gone foward without them. I would have no problem with Apple doing the same, rather than being puritanical.
Actually, standards like JEDEC's are crucial to any commodity market. JEDEC-compliant RAM has a good chance of being compatible between vendors. DDR 400 appeared because DDR II took a bit too long to show up, but as yet it's still uncommon, and since it relies on a 200MHz real clock, it's not ideal. DDR II is quad-pumped, so the real clocks are still 100, 133 and 167. Lower is better from an engineering standpoint, and the presence of an independent standard will make OEMs less nervous about adopting the tech, so I expect DDR II to supplant DDR 400 rather quickly.
Also, once you engineer for DDR II 400, it's a quick and easy path up to 533 and higher.
Originally posted by Barto
I would have no problem with Apple doing the same, rather than being puritanical.
Except that going off the beaten track means costs go through the roof, comparatively speaking. [Edit here - DDR400 memory is not quite as rare as I thought. It's only about 50% more than DDR333. Still, it's a pretty big jump in costs]
It's why Apple has to fight so hard for everything. It used to be that if you used the industry standard, your cost might be 'x', and if you did it yourself, your cost might be '2x'. But because of the incredible economies of scale and the vast amount of investment in process, the industry standard now costs '0.1x', while doing it yourself still costs '2x'. Small wonder that Apple has to abandon just about any technology that is unique to it.
It's why innovation is so tough. You don't have to be a factor of 2 better, you have to be a factor of 20. Small wonder that its the Dell's of the hardware world that are still making money.
Originally posted by Tom West
[I]By the way, I will not be at all suprised if the 970 can actually beat a P4 on a few tests. As I said, the metric of CPU measurement is nowhere near one dimensional.
Heh, you're not kidding -- more likely "infinitely dimensional" is more like it.
Even if a single 970 comes in at 80% the performance of a top-of-the-line P4, Apple will be shipping SMP machines and the 970 + OSX will do better in terms of price/performance in a multiprocessor environment than the P4.
the oct 970 presentation?
I thought the numbers were MUCH closer than that.
Not that it matters since i am firmly camped in the
"It's not the megahertz Its THE motherboard stupid!"
camp.
Originally posted by keyboardf12
where is he getting this 80% number?
the oct 970 presentation?
I thought the numbers were MUCH closer than that.
Not that it matters since i am firmly camped in the
"It's not the megahertz Its THE motherboard stupid!"
I haven't checked myself, but if you look at the SPECint of the upcoming P4s (or the Opterons) and the IBM 970 estimates it'll probably be around 80%. Keep in mind that the G4 would be about 20% so the 970 is a huge leap forward.
course the upcoming 980 and 990 will trounce those numbers.
and they don't include a 2.0ghz , 2.3, or 2.5 ghz 970 so his argument seems shallow.
Originally posted by MartianMatt
"Standard" PCI (1.0?) is 33MHz and 32 bit giving 133 MB/s.
There are also 66 MHz and 66MHz/64 bit PCI flavours giving 266MB/s and 533 MB/s respectively.
Going to www.pcisig.org and checking the news page it seems that PCI-X used to be known as 3GIO. It has an "initial bit rate of 2.5 Gigabits per second per lane per direction" and a "16-lane PCI Express interconnect can provide data transfer rates of more than 8 Gigabytes per second". (Here: http://www.pcisig.com/news_room/news...ses/2002_07_23 )
PCI-X 2.0 has 2 flavours: PCI-X 266 and PCI-X 533 which give 2133 MB/s and 4267 MB/s respectively. (These come from here: http://www.pcisig.com/news_room/news...es/2002_07_23b ) I think these must be DDR rates at 64 bit each since they use the phrase "PCI-X 266, runs at speeds up to 266 Mega transfers per second".
MM
It seems that PCI Express is winning "hearts and minds" in the PC development community over PCI-X, but, unfortunately, PCI-Express also appears to have slipped in its intended deployment from fall of '03 to fall of '04.
Is there any actual indication whether Apple is moving away from the existing PCI bus? What about AGP 8X in the mean time (one or two AGP 8X slots are permitted under the spec)?
I also wonder if Apple will adopt serial ATA on the new logic boards. At the present time the performance difference between serial and parallel ATA is not large, but the configuration and cabling is said to be simpler and better. It would also open the way for the improved performance serial ATA drives that will undoubtedly be available during the manufacturing life cycle of the new boards.
What we will see, is the same thing as is currently seen in the Wintel world. The Power PC will win out at certain tasks, and the P4 3.06 and AMD 3000's will win out at certain tasks. The differentiating factor will be the total package including the OS. It is there that Apple will win hands down. Intel and AMD will bring out faster chips. Apple and IBM will bring out faster chips. Based upon a pragmatic roadmap, they will each periodically leapfrog each other, with each platform having specific areas where it excells. Due to Motorola's mis-managment and shortsightedness, Apple was left in the lurch, falling further behind each year. If they attain overall parity or near parity in performance, then the rest of the package will drive them forward.
I suspect that Apple and Big Blue will now lead the charge, and will become the driving forces for growth and innovation in an industry that was stagnating.
And contrary to many younger Mac users, I never want to see Apple as #1. I would like to see them grow to about 20-25% market share, and then continue to be the driving and innovative force behind the industry.
I don't want what everyone else has. I don't want to have some ficticious bragging rights. I want to be out on the leading edge. That is where Apple belongs.
Originally posted by Shaktai
...I don't want what everyone else has. I don't want to have some ficticious bragging rights. I want to be out on the leading edge. That is where Apple belongs.
wise words
Originally posted by Junkyard Dawg
Right on, brother! I hope Apple never becomes another Microsoft. Power has a insidious way of corrupting humans.
power corrupts everything
coöperations
country's
religions
to use their power to corrupt people
(being the corruptor or the corrupted)
Originally posted by Shaktai
And contrary to many younger Mac users, I never want to see Apple as #1. I would like to see them grow to about 20-25% market share, and then continue to be the driving and innovative force behind the industry.
I don't want what everyone else has. I don't want to have some ficticious bragging rights. I want to be out on the leading edge. That is where Apple belongs.
I dunno. My AAPL holdings would certainly appreciate Monopoly status!
Originally posted by Shaktai
And contrary to many younger Mac users, I never want to see Apple as #1. I would like to see them grow to about 20-25% market share, and then continue to be the driving and innovative force behind the industry.
Agreed. Imagine if Microsft was all that was left invoating. When Apple inovates we get iPods, and other cool software and hardware. When Microsoft innovates we get portable toilets with internet access. I think it speaks voulmes.
And if you think I am making that up about Microsoft, read here: http://www.inq7.net/brk/2003/may/07/brkinf_1-1.htm or http://wireless.ziffdavis.com/articl...059589,00.asp. It has also been mentioned on Letterman and Leno as well (both of course makign jokes about the move).
Originally posted by keyboardf12
oh ok. Upcming machines. prob. true.
course the upcoming 980 and 990 will trounce those numbers.
and they don't include a 2.0ghz , 2.3, or 2.5 ghz 970 so his argument seems shallow.
His argument is simply that at the time of the 970's initial introduction it will probably be at around 80% of the top-of-the-line Intel performance of the same day, based on the facts that we know about the 970. There are no facts saying that there will be a 970 at >1.8 GHz so arguing that a 2.5 GHz 970 will do better is currently just wishful thinking.