Lies, Damned Lies, and a Different Perspective - Page 2
Home | Reviews and Features | Special Reports | Forums |

Page 2 of 4 FirstFirst 1234 LastLast
Results 16 to 30 of 48

Thread: Lies, Damned Lies, and a Different Perspective

  1. #16
    Join Date
    Sep 1999
    Location
    Laguna Beach, CA, USA
    Posts
    107
    Why on earth would Intel sell a chipset/mainboard that uses a type of RAM that is expensive to buy? To pay $1,200 for PC800 128MB modules is nuts - the ordinary user cannot be expected to pay that amount.

    Which probably explains their CC820 mainboard, basically the i820 chipset but with the MTH which degrades system performance so PC100/PC133 RAM can be used.

    I don't ever remember SDRAM being as high when it initially came out as RDRAM is just now. This RDRAM is off to a very bad start, technically it is doubted and price wise - it's tripped up before it's even got off the starting line.

  2. #17
    Join Date
    Apr 2000
    Location
    Hawaii
    Posts
    1
    I also was disappointed in the article. I don't even fully understand all the technical discussion posted above, but from the actual benchmarks I've seen, seems that in SOME cases it is faster than SDRAM, and in other cases it is not! People mention DDR or QDR ram. Good point. The article seems to tell me that if I only wait, the price will drop and it will be so much better. I actually would bet you otherwise. I think that DDR or QDR or some other format will prove to be the best (speed for the money if nothing else). I ask you, for 90% of us computer buyers, would we rather say invest a couple hundred bucks for a faster processor, or thousand for [questionably] faster memory? (and I think I know which system would be faster in the real world!)
    brian

  3. #18
    Join Date
    Apr 2000
    Location
    orleans, ontario, canada
    Posts
    2
    A few additional points:

    The article criticizes DDR on the grounds of requiring 6 or 8 layer printed circuit boards. Doesn't the same objection apply to RDRAM?

    I find the latency calculations more than a little suspect (worst case timing for SDRAM?). In the processor bus and in the array of memory cells on the DRAM chips, data is handled in an essentially parallel fashion. Any time you do back to back parallel to serial and serial to parallel conversions, as RDRAM does, you have overheads, even if the serial transfer is quite fast. Since the timing associated with accessing the DRAM memory cells and associated circuitry on the DRAM chips themselves would likely be similar for the two types of memory, I find it difficult to believe that RDRAM could have an advantage. For an alternative viewpoint, I have an article from Hyundai that quotes latency times of 68 to 93 ns and 45 to 50 ns for PC100 SDRAM. The lower of the first two figures matches the one quoted in the article while the second is barely half. The Hyundai figures seem more consistent with the actual performance.

    A further observation is that the i820 and i840 chipsets benefit from a hub architecture that is supposed to eliminate internal bottlenecks and AGP4x. If these "advanced" designs cannot show a significant advantage under favourable circumstances (premium PC800 RDRAM and 800 MHz+ processors), it is difficult to buy arguments in favour of RDRAM. Conversely, the fact that intel is selling motherboards with the i820 crippled by the memory translator hub needed for SDRAM interfacing raises questions as to whether the company is serious in their belief that main memory is a bottleneck that requires such a radical solution. commitment in

    Robert Inkol


  4. #19
    Join Date
    Apr 2000
    Location
    Bronx, NY, USA
    Posts
    10
    The Author writes:

    >But let's be fair here; a couple of years
    >ago we still used EDO RAM and SDRAM was
    >something new and pretty expensive. In
    >hindsight we've seen the benefits of using
    >SDRAM and its impact on overall system
    >performance. But if we look at the
    >performance benefits SDRAM offered in its
    >early days on applications that were then
    >popular, it also didn't seem to offer huge
    >advantages. Still, we've come a long way,
    >and SDRAM performance has indeed improved
    >although the technology at first didn't
    >seem to promise that much of an
    >improvement.

    This is misleading. SDRAM was knocked because the "10ns" rating made it seem much faster than "60ns" EDO SIMMs, when in fact the speed difference was slight. The big advantage of SDRAM at the time was that you didn't need to install them in pairs; memory had caught up with bus width once again.

    Once the early compatibility problems were dealt with, SDRAM became a better choice, and the prices dropped steadily (and they weren't as ridiculously overpriced as RDRAM is now, by comparison).

    Thus the comparison is not valid. SDRAM was a good choice to succeed an aging memory standard that was quickly becoming inadequate to the task. SDRAM still has headroom, especially with the DDR variant, and a clear and steady upgrade path that does not involve the cost penalties that RDRAM does.

    Also, the fact that Intel has not yet taken those 1 million shares doesn't mean that it isn't a major factor in their support for RDRAM. Those 1 million shares will be worth hundreds of millions of dollars if RAMBUS becomes the memory technology of the future, to say nothin of the additional monies to come from licensing, and the additional power that comes with control of the DRAM market.

    Funny that the author accuses "other websites" of lacking objectivity...


    ------------------
    Milton Teruel
    www.uncamilty.com
    Milton Teruel
    www.uncamilty.com

  5. #20
    Join Date
    Apr 2000
    Location
    Rochester NY
    Posts
    2
    "However, I tend to disagree, as neither have Intel's track
    record of reliability and performance, its resources, or its customer support,"
    What were you smoking? Intel chips have a DOCUMENTED history of being unreliable and buggy. NOT MINOR BUGS even but Major ones such as the 286 not being able to exit protected mode without a reboot or the Entire?(PPro P2 AND P3) Pentium line having problems ADDING.
    Now Why you at First try to dissaccoiate Intel from Rambus and Then bring in a FALSE statement of Intel's reliability and customer support(they lied about the problem and then lied about the severity)I can not fathom.
    Rambus showed Early promise which has not been shown (in ACTUAL TESTS )and will probably be superceded shortly.

    All Pro Intel/Rambus Tests I have seen pit 800 Mhz(another lie as its ddr 400) RDRAM against PC100 Mhz SDRAM instead of VC SDRAM or PC133 SDRAM or DDR SDRAM(available soon) This is INHERENTLY dishonest. Hey why didn't they just test it against FPM dram?
    As to Why Intel was pushing it even though it does not work as advertised. Well they like propriatery solutions(PCI bus anyone?)

  6. #21
    Join Date
    Apr 2000
    Posts
    1
    How can you simply quote a figure for RDRAM latency? Do you know how this stuff works?

    When you initialise an RDRAM system, each RDRAM must have it's latency programmed by software so that the data arrival time from each RDRAM in the channel is equal. This delay will be determined by how many RDRAMs you have in the channel, and the PCB tracking length.

    Basically, the latency is system dependant, and gets worse the more RDRAMs you add to the channel.

    Jeff

    [QUOTE]Originally posted by KaoShen:
    [B]I quote:

    "The total SDRAM system latency is:

    40+(2x10)+(3x10)=90ns for PC100
    45+(2x7.5)+(3x7.5)=82.5ns for PC133...

    ...the RDRAM system latency is:

    38.75+10+18.75=67.5ns for PC800...


  7. #22
    Join Date
    Apr 2000
    Location
    UK
    Posts
    1
    I am amazed at what most people here have failed to notice. If the almighty powerful Rambus is so wonderfully fast and reliable, then why are Intel not going to use them in there up and coming servers but instead are going to use DDR ram ?? Remember that a company that spends thousands on a super fast server would want the very best, and would'nt worry about an extra 10%-20% cost because they would know that they were going to get extra performance in throughput unless Intel talks with forked tougue and DDR ram is actually going to be faster. Oh and not forgetting cheaper too !

  8. #23
    Join Date
    Apr 2000
    Location
    Nieuwegein,Utrecht,The Netherlands
    Posts
    1
    Dil,what you to say is that if someone has a different opinion than yours he better shut up right?.
    I think that his article showed a different light on rambus , and I found It refreshing.
    After all the negative Hype around rambus.
    My view is simply , may the best system win.
    Dil If you don't appreciate someone's view on things like this , GO To a Benchmark site and stay there !.
    I'am getting sick and tired off all those benchmark junks

  9. #24
    Join Date
    Apr 2000
    Location
    Wallaceburg Ontario Canada
    Posts
    16
    Originally posted by Tholos:
    Dil,what you to say is that if someone has a different opinion than yours he better shut up right?.
    He didnt say that at all.. he said he was displeased with the technically inaccurate information that he felt Hardware Central and Sander was posting. Thats a lot different then saying "we dont want your opinion - shut up"


    I think that his article showed a different light on rambus , and I found It refreshing.
    As a lot of other people have already written on here.. it would have been a lot better article if he'd backed up what he said with facts -- aka benchmarks, rather then mainly theory.

    After all the negative Hype around rambus.My view is simply , may the best system win.
    Agreed.. but remember.. not all the hype has been negative over Rambus.. positive hype is what has sent Rambus's shares rocketing to several hundred dolaars US.. It is when other opinions came forward saying that this technology may not be all its cracked up to be that share prices started to fall.

    Dil If you don't appreciate someone's view on things like this , GO To a Benchmark site and stay there !.
    I'am getting sick and tired off all those benchmark junks
    Its nothing to do with not appreciating the column, but its everything to do with disagreeing with it... Havent you noticed the rest of the posts on here? Should they all stay quiet too?

    As for being "sick" of benchmarks, thats the way to find out whether statements such as Sanders and others make are correct or not. Unfortunately, his article (and your argument) sound a lot like Rambus's investment firm pumping up Rambus after they had their first stock dip a month ago - partially due to Van Smith's original report in March (on "Tom's site") with benchmarks claiming the technology was over rated. These guys come out and claim the company is rated at an overly high value of 500$, and when asked about the report that Van Smith wrote, dismissed it as saying it had inaccurate info, but then didnt bother to elaborate. Well.. I am sorry.. but that doesnt cut it.. why was it inaccurate? What didnt you like? Show the public proof (ie BENCHMARKS!) that they werent on the ball. As it was, that site and Van Smith published their own rebuttal 2 weeks ago and put out even more facts to state their case.
    Its one thing to attack someone elses opinion.. but at least show an alternative set of facts that can back you up. The investment firm was wrong to omit this, and I believe the same about Hardware Central and Sanders. I hope he and Hardware Central will address all the conflicting evidence and disagreement that has shown up here (assuming he/they careswhat is said or actually reads the discussion group) and clarify his post - or else make a retraction stating he omitted some facts or was wrong.

    One good thing about his post and the controversy - it'll make the site more popular - I see "The Register" has posted a link to here about the article and all the flak that readers are giving it.

  10. #25
    Join Date
    Sep 1999
    Location
    Netherlands, Europe
    Posts
    170
    Hello there,

    Let me start off by thanking you all for your honest and upfront comments. I'm in the midst of my research for the follow-up as I'm not about to make all of you find out the truth of the matter on your own account. Also I'll be sure to provide answers to the questions you've raised and any issues that I may have left out in the initial article.

    Many of you also suggested that I'm 'bought' by Rambus or Intel; let me just assure you that that's not the case. I'm just as eager to find out what this RDRAM vs. SDRAM discussion is all about as you are. Also many of you state that I'm ill-educated and should leave this kind of stuff to the 'professionals'. Just to satisfy your curiousity, I have a masters degree in computer science and a bachelors in micro-system technology; you can rest assured that I'm capable and on top of things.

    Regarding benchmarks; there are many ways to conduct a benchmark, and there're even more ways to interpret its results. Choosing the wrong benchmark can spoil the outcome and vice versa. Benchmarks are usually written to demonstrate a product's performance, and when issued by the product's manufacturer, they are mostly tailored to the product's features. These kinds of benchmarks generally give an estimate how the product performs in relation to others, but more importantly they show the performance that that product is capable of delivering.

    And that's exactly the problem, because 'capable of delivering' and 'actually delivering' are two entirely different things and must not be interpreted as being the same thing. Real world performance is an entirely different story than a controlled benchmark environment. Even the fastest CPU/system can be crippled by either an ill configured system or running software which is not tailored to make use of the CPU's/system's features.

    Furthermore, if you know how to 'benchmark' a product it is very easy to make one come out on top whereas the other comes in second place by focussing on the actual performance enhancing features of one, and neglecting those of the other. Just to give you an example, Apple's G4 CPU was boasted to offer 'supercomputer performance on the desktop', but after close examination of the benchmarks and SIMD optimizations used, it is very clear that it only does so under certain conditions. If another benchmark had been used which did not include the SIMD optimizations the outcome would have been entirely different.

    So nothing new here, as the saying goes 'Lies, damn lies and benchmarks'. To do away with some of the obvious pitfalls, I've decided not to include benchmarks in the initial article as that would take the focus off of the theoretical discussion. I will however include those in the follow-up, after I have found a way to properly and more important objectively benchmark the different memory types.

    I'm currently talking to Rambus, Intel, AMD, Samsung and Micron about how to best go about this and provide results that are accurate and can be easily reproduced by anyone willing to do so.

    As much as I would like to participate in this discussion thread and answer everyone's concerns individually I cannot due to time constraints, I hope you all understand.

    Best regards and thanks,

    Sander Sassen
    Email : ssassen@hardwarecentral.com
    Siteleader at HardwareCentral
    Visit us at : http://www.hardwarecentral.com



  11. #26
    Join Date
    Apr 2000
    Posts
    4
    Thanks for posting a reply, Sander. I am impressed that you would take the time to respond to your critics. That takes character. Sorry if I got a little overzealous with the "leave it to the pro's" comment, I did not mean to personally offend.

    I am anxious to see the results of your unbiased benchmarks. I'm sure you will be careful to make the test a fair one, since a lot of us are sure to go over your results with a fine-toothed comb. I'm not sure there is any way, currently, to measure Rambus vs. SDRAM without the chipset getting in the way, unfortunately. Clearly, the i8XX series of motherboards from Intel have so far been much less efficient than the 440BX and may hurt Rambus's numbers. However, testing an i820 with SDRAM through a Memory Translator is unfair to SDRAM. Ditto for using a Via chipset, which have historically been slower than Intel's. Here is my idea: i810e/SDRAM vs. i820/Rambus with a basic Intel video card similar to the integrated video in the i810e. Not an ideal setup, I agree, but possibly the closest we can get to similarity if we are testing the difference in the memory performance, not the rest of the system.

    What do you think? I'm probably overlooking an obvious problem with calling this a similarity. If that is the case, Via Apollo Pro/SDRAM (PC133 of course) vs. i820/Rambus is probably the best test. We eagerly await the results!

    (this will all be a moot point anyway when a chipset is released that uses DDR, and then its successor QDR)

  12. #27
    Join Date
    Apr 2000
    Location
    Bronx, NY, USA
    Posts
    10
    Sander,

    I don't see where anyone implied that you were a shill for Intel/Rambus. Nor do I see a post which questioned your knowledge without providing a basis for their doubts.

    Many of us pointed out that Intel does indeed have a vested interest in seeing Rambus succeed, and thus their support of Rambus may be influenced by concerns other than performance.

    It's good to see that you'll provide benchmarks. But keep in mind that many other sites and magazines have run more than just synthetic benchmarks. They have run actual applications and application specific benchmarks, which you agree is an important part of any test suite. And those benchmarks show that Rambus does not distinguish itself in the way a new and superior technology should. In fact, Tom's review pointed out that Intel's benchmarks unfairly favored Rambus by comparing it to a system with PC100 SDRAM which was poorly optimized for performance.

    If Rambus is to become an industry leader on its merits alone, it will need to be refined and improved. A future article discussing ways in which this may happen sounds like a worthwhile pursuit...


    ------------------
    Milton Teruel
    www.uncamilty.com

    [This message has been edited by UncaMilty (edited 04-07-2000).]

    [This message has been edited by UncaMilty (edited 04-07-2000).]
    Milton Teruel
    www.uncamilty.com

  13. #28
    Join Date
    Apr 2000
    Location
    Berkeley Heights, NJ USA
    Posts
    5
    If anyone wants to see REAL WORLD performance between DDR-SDRAM and Intel i840 dual-channel RDRAM, please reference this article at Tom's Hardware. They looked at an engineer's sample motherboard from Micron that used Micron's Samurai DDR-SDRAM chipset...the technology of which VIA has subsequently licensed.

    Also relevant are the real world performance results from InQuest research here.

    You can spin techno-babble all day long, attempt to confuse people with circuitous logic, and wax loquacious about theory, but what matters in the end are hard numbers. "Rambus by Intel"(TM) loses, DDR wins.

    P.S. And no where mentioned in the article was SyncLink DRAM (SLDRAM) technology, an OPEN standard that is a prime competitor to RDRAM, especially since it utilizes a low pin count necessary for future memory technologies.

    Be careful who you trust.

  14. #29
    Join Date
    Apr 2000
    Posts
    2
    bdwilcox1 beat me to a reply, but he/she is absolutely correct. We listened to the propaganda from Intel and set up 5 trial systems as part of our normal planning testing. ALL performed, overall, less effective than then the BX systems we currently use and no better then the VIA based test systems. Initial Micron board results (DDR) we've seen from friendly competitors mirrors the Inquest data..

    I would repectfully suggest that you do your benchwork and get the facts before you repeat the Intel "party line"..

    Very disapointed in your effort.....

  15. #30
    Join Date
    Jul 1999
    Location
    Petaluma, CA USA
    Posts
    15
    The mere fact that so much controversy has arisen concerning RDRAMs performance should tell us a lot about the real state of affairs. OEMs are always anxious to incorporate new and better technologies that enhance system performance simply because it gives them a competitive edge when selling their systems. Had RDRAM and the 8XX chipsets been a clear performance leap in computer system performance, the new platform would have been embraced without hesitancy.
    Intel destroyed public confidence in these new technologies itself with it's flaw in the bus design that required junking huge amounts of motherboards before they even made it to market. Intel's own junking of a couple of it's own 820 boards later didn't help either. Then the flaw with the 840 chipset and ECC memory further destroyed public confidence.
    Adding to this, the reviews with benchmarks done by sites that the public looks to for honest appraisals of new technology, certainly did not see any huge performance leaps with the new RDRAM and 8xx chipsets. Motherboard manufacturers have had very disappointing sales based on this new technology. This doesn't even touch upon the price\performance issue, which is paramount in the computer industry. VIA has had record sales of it's chipsets because of their ability to fill the gap in the price\performance requirement. All in all I think there is only one conclusion: Intel blew it big time and they probably know it better than anyone else.

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •