the losers are indie and specialty record stores like Tower (who subsequently filed for bankruptcy), as Wal-Mart and Best Buy will continue to drive them out of business.
Now that is humor, Tower records as an indie store? Whats next, Wendy's as the indie burger joint?
Amoeba Records is an indie store. A chain with stores all over the us is hardly an indie store.
Oh boy, someone writes at least a page of complete crap and get 5: informative??? Having worked on what he is trying to talk about, I can flatly say he doesn't have a clue. I don't think there is even one thing in his post that is correct.
Well if you'd been following the technology of RAMBUS and actually READING the reviews that you mentioned, you'd know that while RDRAM has wonderful bandwidth, it has horrible latency, heat, power, and timing problems. They've known about all of these problems since RDRAM came out and they're painfully obvious now. It may have oodles of more bandwidth than sdram BUT if the latency is bad or if you can't power it or design a decent chipset/memory controller for it, it's really worthless except possibly at the high end server market where people are willing to pay oodles of $$ for extreme bandwidth with massive cooling solutions.
And you know all this because:
A: you read it on some hardware site?
B: you read it on usenet?
C: you are the urban legend spambot?
First of all the power of DRDRAM is roughly the same as DDR DRAM, in fact, depending of the configurations DRDRAM can generate less heat. Go read data sheet and stop reading usenet and hardware review sites. It is true that an individual chip on a rambus channel can get hot, but the other chips will be generating almost no heat. The heat spreader is there to spread heat and keep all the chips at a roughly even temperature.
As far as latency is concerned, the benchmarks on latency with PC chipsets show that depending on the chipset DRDRAM is better or worth than CL2 SDR SDRAM. So again there is no real fact behind the supposed higher latencies of DRDRAM. The overall differences in latencies are once again well overblown, with the highest variable in latency of a system being the chipset.
Why does this rumor keep going around. If you look at the actual data sheets for DRDRAM AND DDR SDRAM you find that they both have roughly the same power consumption. Just take a look at Samsung's data sheets.
>Am I the only one to notice that no one even >tries to defend Rambus on a technological basis >anymore
Maybe its because the people who actually know about the various memory technologies are tired on explaning over and over the exact same thing to people who probably have no clue what they are talking about, but read somewhere that "insert product here" sucks and therefore is bad.
DRDRAM has significant advantages over SDR SDRAM, DDR SDRAM, DDR-II SDRAM and are roughly the same power. Anyone saying otherwise either hasn't looked at the issues or doesn't know what they are talking about(This included a lot of the hardware review sites, but what do you expect when you get your technical information from a Jr High school student.)
>Separately, if the overall level of taxes is generating surpluses,
What surplus? How can people talk about surpluses when they have somewhere on the order of 5 TRILLION dollars in collective debt. IMNSHO, you don't have a surplus until you don't have a debt. Raise the taxes, cut spending, and pay off the debt.
Put down the crack pipe, please. You look at the freaking spec sheet. For a single RDRAM channel transfering a 64Byte cacheline, you have a transfer latency of 40ns. aka, any given transfer will occupy 40ns. So in order to saturate the link you need to be able to transfer a cas in 40 or less ns. Considering they don't make parts with this high of a CAS latency, you're pretty much wrong.
In fact, a single rambus channel has full CAS bandwidth down to a 16byte transfer size. After that you are limited by the CAS bandwidth.
This is actually a lot better than some of the proposals for DDR-II
These numbers are not cross section bandwidth numbers. The 11.2, 44.8 and 716 numbers are aggregate peak bandwidth numbers. If you want to claim these are cross section, then I'll claim C-plant has >1TB/s of bandwidth.
Um, actually they can't. This would be the same thing as violating the 5th admendment. The can't force you to incriminate yourself, at least in the US, and unencrypting incriminating evidence would certainly apply. Granted, you might have to go all the way to the supreme court.
>As much as I hate the phrase, this is really a >paradigm shift in processor and OS technology.
Um, no. Look up the IBM AS/400 which has been shipping for quite a long time. Its contains an MI(Machine Interface) layer to which all programs are written and then the programs are translated and executed by the MI layer. This is basically the same thing that "code morphing" does.
Compaq bought Digital. Digital = Compaq = Digital = Compaq.
I know it might seem strange but that is the way it is. Think of it this way, there are more Digital "Heritage" people at Compaq than non Digital "Heritage" people. So, I wouldn't call it taking credit for Dec's work.
Slashdot Mirror
Now that is humor, Tower records as an indie store? Whats next, Wendy's as the indie burger joint?
Amoeba Records is an indie store. A chain with stores all over the us is hardly an indie store.
Oh boy, someone writes at least a page of complete crap and get 5: informative??? Having worked on what he is trying to talk about, I can flatly say he doesn't have a clue. I don't think there is even one thing in his post that is correct.
When I was last in the UK, a couple years ago, it cost more to call cross town than it cost in the US to call cross country.
And you know all this because:
A: you read it on some hardware site?
B: you read it on usenet?
C: you are the urban legend spambot?
First of all the power of DRDRAM is roughly the same as DDR DRAM, in fact, depending of the configurations DRDRAM can generate less heat. Go read data sheet and stop reading usenet and hardware review sites. It is true that an individual chip on a rambus channel can get hot, but the other chips will be generating almost no heat. The heat spreader is there to spread heat and keep all the chips at a roughly even temperature.
As far as latency is concerned, the benchmarks on latency with PC chipsets show that depending on the chipset DRDRAM is better or worth than CL2 SDR SDRAM. So again there is no real fact behind the supposed higher latencies of DRDRAM. The overall differences in latencies are once again well overblown, with the highest variable in latency of a system being the chipset.
Why does this rumor keep going around. If you look at the actual data sheets for DRDRAM AND DDR SDRAM you find that they both have roughly the same power consumption. Just take a look at Samsung's data sheets.
Think before you spew.
>Am I the only one to notice that no one even >tries to defend Rambus on a technological basis >anymore
Maybe its because the people who actually know about the various memory technologies are tired on explaning over and over the exact same thing to people who probably have no clue what they are talking about, but read somewhere that "insert product here" sucks and therefore is bad.
DRDRAM has significant advantages over SDR SDRAM, DDR SDRAM, DDR-II SDRAM and are roughly the same power. Anyone saying otherwise either hasn't looked at the issues or doesn't know what they are talking about(This included a lot of the hardware review sites, but what do you expect when you get your technical information from a Jr High school student.)
The AlphaServer SC machines don't use myrinet, they use quadrics interconnect and routers. You can find more about quadrics at www.quadrics.com
>Separately, if the overall level of taxes is generating surpluses,
What surplus? How can people talk about surpluses when they have somewhere on the order of 5 TRILLION dollars in collective debt. IMNSHO, you don't have a surplus until you don't have a debt. Raise the taxes, cut spending, and pay off the debt.
Put down the crack pipe, please. You look at the freaking spec sheet. For a single RDRAM channel transfering a 64Byte cacheline, you have a transfer latency of 40ns. aka, any given transfer will occupy 40ns. So in order to saturate the link you need to be able to transfer a cas in 40 or less ns. Considering they don't make parts with this high of a CAS latency, you're pretty much wrong.
In fact, a single rambus channel has full CAS bandwidth down to a 16byte transfer size. After that you are limited by the CAS bandwidth.
This is actually a lot better than some of the proposals for DDR-II
even better, would it be illegal to link to a site with an auto redirect to the "illegal" information?
These numbers are not cross section bandwidth numbers. The 11.2, 44.8 and 716 numbers are aggregate peak bandwidth numbers. If you want to claim these are cross section, then I'll claim C-plant has >1TB/s of bandwidth.
Um, actually they can't. This would be the same thing as violating the 5th admendment. The can't force you to incriminate yourself, at least in the US, and unencrypting incriminating evidence would certainly apply. Granted, you might have to go all the way to the supreme court.
>As much as I hate the phrase, this is really a
>paradigm shift in processor and OS technology.
Um, no. Look up the IBM AS/400 which has been shipping for quite a long time. Its contains an MI(Machine Interface) layer to which all programs are written and then the programs are translated and executed by the MI layer. This is basically the same thing that "code morphing" does.
Compaq bought Digital. Digital = Compaq = Digital = Compaq.
I know it might seem strange but that is the way it is. Think of it this way, there are more Digital "Heritage" people at Compaq than non Digital "Heritage" people. So, I wouldn't call it taking credit for Dec's work.
The best place I found were the talk page and products pages off of www.bitmover.com
There isn't a whole lot there right now.