Kuro5hin.org: technology and culture, from the trenches
create account | help/FAQ | contact | links | search | IRC | site news
[ Everything | Diaries | Technology | Science | Culture | Politics | Media | News | Internet | Op-Ed | Fiction | Meta | MLP ]
We need your support: buy an ad | premium membership

[P]
Athlon XP and VIA's long expected KT266a both released

By mtgcollector01 in Technology
Mon Oct 15, 2001 at 02:39:32 PM EST
Tags: Technology (all tags)
Technology

On October 3, AMD released its long awaited Athlon XP chip with the new Palomino core to select retailers. The new chip no longer gives speed in terms of Megahertz (MHz), rather with a new Performance Ratings (PR) system. The current PR ratings are as follows:

  • Athlon XP 1500: 1.33GHz
  • Athlon XP 1600: 1.40GHz
  • Athlon XP 1700: 1.47GHz
  • Athlon XP 1800: 1.53GHz

    The prices for these are around $115, $124, $160, $215 respectively.

    The major new technology implemented in the new Palomino core is "Streaming SIMD Extensions"(SSE) as well as "3DNOW! Professional", which can greatly improve the performance in 3D action games such as Quake III.


  • In order to try to get consumers to understand how AMD AthlonXP works, AMD created technology called Quantispeed Architecture. As big as the name might seem, it simply means that the chip is able to complete more work per clock cycle (IPC) compared to chips such as Intel's Pentium IV. AMD is also stressing to consumers that a processor is not faster simply because of clock speed or "MHz". Benchmarks all over the internet are showing the immense increase in performance of the new AthlonXP over the previous Athlon, the Thunderbird.


    On similar news, VIA's long fabled A revision of the KT266 motherboards has finally hit the market. With the new revision, memory bandwidth has increased by up to 30% over the original version, as well as stability and overall system speed. The full specs are as follows

    Supports AMD DuronTM and AthlonTM Socket A processors
    200/266MHz FSB settings
    Support for AGP 2X/4X
    Supports up to 4.0GB DDR200/266 SDRAM as well as PC100/133 SDRAM and Virtual Channel memory
    266MB/sec high bandwidth North/South Bridge V-Link
    Support for Advanced Communications Riser (ACR) Card Standard
    Integrated 6 channel AC-97 Audio
    Integrated MC-97 Modem
    Integrated 10/100 BaseT Ethernet and 1/10Mb HomePNA controller
    Support for ATA 33/66/100
    6 USB ports, UHCI compliant
    Advanced power management capabilities including ACPI/OnNow and AMD's PowerNowTM
    552-pin BGA VT8366A North Bridge
    376-pin BGA VT8233 South Bridge

    Boards that have already hit the market that include this chipset are the Shuttle AK31 (revision 3.1), EPoX 8KHA+, and the MSI K7T266 PRO2.

    Sponsors

    Voxel dot net
    o Managed Hosting
    o VoxCAST Content Delivery
    o Raw Infrastructure

    Login

    Poll
    Will you buy the new Athlon XP and KT266a motherboard?
    o Yes, I will definitely buy it soon 6%
    o Probably within the next few months 20%
    o Maybe, depending on the new develpments 30%
    o No, I'm happy with my current setup. 42%

    Votes: 63
    Results | Other Polls

    Related Links
    o Athlon XP
    o new
    o chip
    o major new technology
    o finally
    o hit
    o the
    o market
    o Shuttle AK31
    o EPoX 8KHA+
    o MSI K7T266 PRO2
    o Also by mtgcollector01


    Display: Sort:
    Athlon XP and VIA's long expected KT266a both released | 31 comments (21 topical, 10 editorial, 0 hidden)
    Marketing Hype vs. real difference (none / 0) (#5)
    by beleriand on Sat Oct 13, 2001 at 11:57:55 AM EST

    The AthlonXP also has a on-die thermal sensor. With the Thunderbird, you didn't really know how hot the die got, only the Temp. of a sensor on the motherboard.
    Another nice advandage over the TBird is that it uses less power.

    The main difference between the Via KT266 and KT266A is that the "a" has vastly improved memory performance. See this article on tomshardware. Lets just hope that it is bug free.

    Silliness (2.50 / 2) (#7)
    by Neuromancer on Sat Oct 13, 2001 at 03:09:46 PM EST

    * Athlon XP 1500: 1.33GHz
    * Athlon XP 1600: 1.40GHz
    * Athlon XP 1700: 1.47GHz
    * Athlon XP 1800: 1.53GHz


    Look at AMD ho'ing themselves to the uninformed public by changing to series numbers that are higher than the clock speeds. Oh well.

    Then again, I never believed that the 68k series chips ran that fast *G*

    Uninformed public (5.00 / 2) (#8)
    by slaytanic killer on Sat Oct 13, 2001 at 04:26:59 PM EST

    If the public is uninformed, making bad choices because of their ignorance, isn't it better to correct it?

    Intel has after all shown that it's willing to play off the public's ignorance by playing the clockspeed game. Plus, AMD's move is a double-edged sword because Intel would run ads everywhere if AMD's numbers were abusively misleading.

    Mainly I'm for any move that keeps AMD afloat, because that is better for the consumer than the monopoly of Chipzilla.

    [ Parent ]
    Numbers and Joe Schmoe (none / 0) (#29)
    by matthead on Fri Oct 19, 2001 at 07:44:16 PM EST

    Correct their ignorance? Who's going to sit through a 30-minute infomercial on pipelines, pre-fetching, memory bandwidth, bus speeds, and so on? Those that are interested will learn from other places.

    Anyway, I can see Joe Schmoe watching an AMD ad that explains how their 1.53 GHz processor generally outperforms a Pentium 4 1.8 GHz. He says "OK, so, their argument basically boils down to: 'even though our processor is slower, it's faster than theirs.' Right. Any idiot can see that's a stupid thing to say." He goes and buys from Intel, because their numbers are, after all, bigger, and they don't treat him like he's stupid.

    I thought it was a bad idea when I first heard it, too. Now, I changed my mind. Most people don't care to look behind the number, and for them, it emphasizes the point that AMD processors can perform well without an obscene number of MHz. For those of us that care to know more, it's easy to find out the real story.


    - Matthead
    [ Parent ]
    Yes and no (none / 0) (#15)
    by Elendale on Sun Oct 14, 2001 at 04:43:08 PM EST

    I don't have a problem with them marketing it as an "Athlon 1800"- the problem is when they start trying to hide the clockspeed. Consumers are already being misled by Intel's high clockspeed P4, so this is "merely" a counter-attempt by AMD. As sad as it is, this is a step in the direction toward "truth" as far as consumers go. People who are smart enough to realize "Athlon 1800" isn't "Athlon 1800 MHz" are probably going to be smart enough to realize that clockspeed isn't the only measure of performance. At least, IMHO.

    -Elendale
    ---

    When free speech is outlawed, only criminals will complain.


    [ Parent ]
    VIA (2.66 / 3) (#14)
    by pete on Sun Oct 14, 2001 at 03:47:18 PM EST

    Aren't VIA motherboards notorious for causing filesystem corruption under Linux? I seem to remember threads about this on the kernel mailing list every so often.

    --pete


    1 specific chipset, yes... (4.50 / 2) (#20)
    by Rizzen on Mon Oct 15, 2001 at 02:42:04 PM EST

    But VIA boards in general have been rock solid throughout the years.

    There is a *known* bug, and BIOS fixes for it, with the northbridge in one specific VIA chipset (kt133a or something similar). That *one* chipset caused file corruption under just about every OS, although particularly under Linux as it uses async mounts be default (the least reliable setting, but also the fastest).

    Please check your facts before posting rubbish. :)
    ----- The years of peak mental activity are undoubtedly those between the ages of 4 and 18. At age four, we know all the questions; at eighteen, we have all the answers. -- unknown
    [ Parent ]
    hehe (none / 0) (#30)
    by Spendocrat on Tue Oct 30, 2001 at 07:24:10 PM EST

    Please check your facts before posting crap to USENET!@!!1@!!!1!

    1. Check facts
    2. Post crap to USENET
    3. !@!!1!!@!@

    [ Parent ]

    PR = ?Good? ?Bad? (2.00 / 1) (#16)
    by {ice}blueplazma on Mon Oct 15, 2001 at 11:07:14 AM EST

    If I recall correctly, some little company named Cyrix (I think) used a rating other than MHz. I think that all it did was confuse consumers and drive business down, its like with Apple. They're called MHz for a reason, rate your chips using a standard and people will understand what they're buying instead of multiplying by 2.5737383757 to figure out the actual speed of a G4.

    P.S. I do not like Apples mostly. They have some uses. Don't flame me over incorrect numbers or anything insulting about G4s.

    "Denise, I've been begging you for the kind of love that Donny and Smitty have, but you won't let me do it, not even once!"
    --Jimmy Fallon
    If I recall (none / 0) (#17)
    by Locus27 on Mon Oct 15, 2001 at 12:25:12 PM EST

    correctly, the Cyrix 6x86 chip sucked. I know, I owned one. They still rated them in MHz though. Mine was a 6x86/166, indicative of running at 166Mhz.

    "You're one fucked up cookie."
    -Shawn R. Fitzgerald

    [ Parent ]

    Cyrix (none / 0) (#19)
    by twodot72 on Mon Oct 15, 2001 at 12:46:57 PM EST

    No, they did use PR numbers. I had a Cyrix PR200, it was actually 150MHz. Floating point performance sucked, it was unusable for 3D gaming. You can find a table of ratings vs. core and bus frequencies here.

    [ Parent ]
    Not my memory... (none / 0) (#25)
    by SeaCrazy on Mon Oct 15, 2001 at 07:08:25 PM EST

    As far as I am concerned, the cyrix chips were pretty decent, especially for the price tag they had. So what if a PR200 was not as fast as the Pentium at 200 MHz, the cyrix was still cheaper than a pentium at 133 MHz.

    [ Parent ]
    My Cyrix CPU... (none / 0) (#27)
    by Armored Scrum Object on Thu Oct 18, 2001 at 01:41:02 PM EST

    Mine was a 6x86 P120+, which means that it ran at 100MHz. My experience with it wasn't bad. I could play Quake on it without problems, though Quake II was a bit hard on it (I also didn't have any kind of 3D acceleration). For the price, it was an excellent processor.

    [ Parent ]
    Ratings vs. MHz (5.00 / 2) (#18)
    by twodot72 on Mon Oct 15, 2001 at 12:40:50 PM EST

    The PR scheme Cyrix used was supposed to mean that a chip with, for instance, PR150 had performance equal to a Pentium 150. The problem they had was that this was simply not true. They marked the chips with numbers far higher than they should have. AMD does not seem to make the same mistake.

    What do you mean by "the actual speed"? For instace, in a Pentium 4 2.0 GHz, most of the core works at 2GHz, but some parts (the ALUs) at 4GHz. Some parts have self-clocked elements (i.e. they don't use a central clock at all), the CPU to chipset bus works at 100MHz, but can transfer 4 times per clock cycle. The chipset to memory bus is at yet another frequency. Which is the "actual speed", and why?

    Furthermore, to use the clock frequency to separate models is as arbitrary as many other figures you could use, such as AMDs new model numbers. Worse, when people in general believe that you can compare different CPU families just by looking at the (parts of the) core frequency, using MHz numbers does more harm that good. People won't understand what they're buying, they are being fooled!

    You said "They're called MHz for a reason". Yes, the reason is that the clock completes that many cycles per second. So? What was your point?

    [ Parent ]

    Exactly! (4.00 / 3) (#21)
    by Rizzen on Mon Oct 15, 2001 at 02:51:13 PM EST

    The computer industry needs to come up with a standard "benchmark" that rates the performance of a chip, not its internal clock speed.

    For instance, when you buy a car, do you ask how many RPMs it runs at or the max RPMs, or anything like that? Or, do you ask for the horsepower rating? Very different engines, running at very different RPMs, all compared using standard horsepower ratings. Too bad the CPU industry can't find a nice middle-ground like that.

    The problem with the Cyrix PR scheme was that is was no where near reality. The PR200 was supposed to comparable to a Pentium 200. However, it benchmarked the same as a Pentium 166 at best, and usually only slightly better than a Pentium 150.

    AMD's "PR" scheme is in tune with reality. In fact, it might even be a little on the conservative side (to avoid the kinds of mistakes Cyrix made). In 9 out of 10 benchmarks, the 1800 beats a P4 2 GHz. And yet, the chip only runs at ~1.5 GHz. (Hmmm, maybe the fact that the P4 can handle 5 instructions per clock but the pre-fetch/decoder can only feed it 3 instruction per clock might have something to do with that??) :)

    Intel might be able to say it's reached all the latest MHz milestones, but what's the use of a super-fast CPU if it spends 90% of the time idle? Give me a CPU that burns through mega-multiple instructions in the fastest *AND* most efficient manner any day. So far, AMD is the one with the best technology.

    However, I am going to have to rate this a -1 as there is no real content to this submissions, and it would barely pass muster in MLP let alone Technology.
    ----- The years of peak mental activity are undoubtedly those between the ages of 4 and 18. At age four, we know all the questions; at eighteen, we have all the answers. -- unknown
    [ Parent ]
    We've had them for years (5.00 / 1) (#23)
    by epepke on Mon Oct 15, 2001 at 04:19:14 PM EST

    The Whetstone and satirically-named Dhrystone benchmarks have been around for 15-20 years.

    The trouble is that nobody cares. They don't really want high performance; they want a big number so that they can impress their friends.


    The truth may be out there, but lies are inside your head.--Terry Pratchett


    [ Parent ]
    Performance metrics (4.00 / 1) (#24)
    by sigwinch on Mon Oct 15, 2001 at 05:50:03 PM EST

    Very different engines, running at very different RPMs, all compared using standard horsepower ratings.
    HP isn't necessarily very useful. I'd rather see the power curve (HP vs. rpm).
    Hmmm, maybe the fact that the P4 can handle 5 instructions per clock but the pre-fetch/decoder can only feed it 3 instruction per clock might have something to do with that??
    And it takes a long time to refill the P4's deep pipeline when a mispredicted branch occurs.

    --
    I don't want the world, I just want your half.
    [ Parent ]

    Benchmarks (none / 0) (#31)
    by wiml on Mon Nov 05, 2001 at 08:09:00 PM EST

    Back in the mists of time, benchmarks (whetstone, dhrystone, rhealstone, foostone, foomark, blah blah) were commonly used, almost like MHz ratings are today. Of course, there was still a lot of controversy over benchmarks --- controversy that is very similar to the MHz controversies today. Examples:

    "This benchmark doesn't accurately reflect the way a real application uses the processor!" vs. "Pure MHz ratings don't accurately reflect the real-world performance of the processor!"

    "Vendor X has tuned their product specifically to benchmark Y, to the detriment of actual app Z!" vs "Intel is going for the absolute highest clock speed, ignoring the fact that this requires deep pipelines, poor branch behavior, etc.!"

    "Everyone just looks at the benchmark numbers instead of doing an in-depth analysis of which product is actually faster/better!" vs "Everyone just looks at the MHz rating [etc.]!"

    I think you get my point. :-) Real benchmarks are definitely better than today's fetishistic focus on clock speeds, but they're still not perfect...

    [ Parent ]
    i won one! yay! (4.50 / 2) (#22)
    by sayke on Mon Oct 15, 2001 at 03:18:55 PM EST

    AMD had a promotional giveaway in seattle - i got there at about 5 in the morning, they gave me a half a ticket with a number on it and put the other half in a box, they said fairly typical marketing stuff (explaning their rating system, etc), and then they took some numbers out of the box - one of which was mine =)

    so i am now the proud owner of an athlon xp 1533mhz (1800+), a MSI K7T266 pro-RU mobo (not the best, but not bad), and a decent coolermaster fan. no more k6-2 500 for me! =)


    sayke, v2.3.1 /* i am the middle finger of the invisible hand */

    Could be confusing (none / 0) (#26)
    by Cepper on Tue Oct 16, 2001 at 08:14:33 AM EST

    The new rating scheme could be confusing. As long as they keep emphisising that the rating is not speed of clock but a performance indicator, publish the methodoligy (is it published?), etc. , I see no problem.
    "In the land of Mordor, where shadows lie"
    Improvements? (none / 0) (#28)
    by tc1974 on Fri Oct 19, 2001 at 05:56:27 AM EST

    Seems to me that the new architecture being implemented here will be primarily of interest of the games playing public - I can't imagine the difference between the XP processors and the PentiumIV/Thunderbird clock speed equivalents will be that significant on day to day applications (although I could be wrong).

    So, would it be an idea for an industry standard benchmark for 3D performance (perhaps other than, or a derivation of 3DMark ) looking at the processor itself (assuming, I suppose, a "control" system specification elsewhere, modest enough to be within most people's budget)?

    This way joe public could compare the performance increase across all manner of processors at a glance (ignoring clock speed which seems to be a bit of red herring when it comes to this element of XP processor performance).

    Thoughts please.

    By way of explanation, I'm not a 3DMark Nazi, it's just that (in my experience) 3DMark is aimed more at individual system tweaking by paid up users, rather than providing a publicly accessible source of comparable performance figures.

    Athlon XP and VIA's long expected KT266a both released | 31 comments (21 topical, 10 editorial, 0 hidden)
    Display: Sort:

    kuro5hin.org

    [XML]
    All trademarks and copyrights on this page are owned by their respective companies. The Rest 2000 - Present Kuro5hin.org Inc.
    See our legalese page for copyright policies. Please also read our Privacy Policy.
    Kuro5hin.org is powered by Free Software, including Apache, Perl, and Linux, The Scoop Engine that runs this site is freely available, under the terms of the GPL.
    Need some help? Email help@kuro5hin.org.
    My heart's the long stairs.

    Powered by Scoop create account | help/FAQ | mission | links | search | IRC | YOU choose the stories!