Duron or Duroff

Why I initially asked

When I initially asked the question, “There are no entries for Durons and T-Birds. How come?” I just wondering about not seeing any, not implying that you should all have run out and bought one by now. 🙂

I’ve since found out that we manually update the database weekly because there are a lot of garbage entries that have to be weeded out. We’ve also had some server-related problems that have caused us some problems, but we should get them straightened out pretty shortly.

What I Got

Anyhow, intentionally or not, your responses gave us a pretty good cross-section of not only what people think of the AMD chips, but of overclocking strategies in general.

Different people look at overclocking different ways. We have a proportion of the audience that are early adapters; they like to buy the newest thing. More of you would like to see some proof from the early adapters before jumping in.

Some of you like to trade and swap all the time for fairly minor increases in speed. A lot of you have owned two, three, or even more Coppermines already. More of you don’t.

Some people are building new systems from scratch, other are upgrading new systems.

Because this audience is so diverse, it’s hard to please all the people all the time. We shifted heavily to AMD coverage for a while because they looked like interesting chips, we knew at least some early adapters were very interested, and besides, Intel isn’t doing much at the moment.

What We Think So Far

So far, we like the performance of the chips, but we don’t like the power they require, the heat they generate, and the problems that causes.

While we think people building a new, relatively cheap computer system (a back-to-school special) should give the Duron serious consideration (since these problems can be kept to tolerable levels), there
are still a lot of open questions for everybody else.

We are skeptical about T-Birds being significantly better than Intel equivalents right now (especially if air-cooled). We are more than skeptical about the current AMD offerings being worth a CPU+motherboard upgrade, as are many of you.

We saw with the KX133 that buying a motherboard and expecting future upgradability is a bad idea, and probably will continue to be so with DDR boards coming out months after the new KT133s. It’s by no means certain that a TBird you buy today will even work on a 266Mhz DDR system tomorrow.

I’m saying this because the AMD codes distinguish between 200Mhz and 266Mhz chips. If you’re an AMD advocate, don’t send me “of course it will work” unless you’re going to reimburse everybody if they buy a T-bird now and it doesn’t work later.

Don’t write me, write AMD and try to get something definitive from them saying whether or not some or all 200Mhz processors will work on 266Mhz systems. I’ll be delighted to announce that. That would be useful and helpful.
to all of us.

Intel In The Near Future

Intel also looks to be more competitive in the overclocker’s realm than expected a few months ago this fall and winter for three reasons:

  1. The upcoming cC0 stepping
  2. DDR boards for the PIII
  3. .13 micron PIIIs early next year.

First, let’s mention what we did not mention:

We don’t consider Willamette a potential overclockers’ chip until well into 2001. We don’t think you should buy a Willamette until it goes to .13 micron because the .18 micron one will probably be very hot, and heat will limit the overclock. It’s like the Katmais all over again.

The 100Mhz FSB Celerons may be good news to nonoverclockers, but not to overclockers, unless Intel also otherwise improves the performance. The current Celerons are great for a quick cheap pop in performance from a current BX board setup. No worries about O/Cing video cards or your RAM not being good enough for 133Mhz or better. The introduction of 100Mhz Celerons means that
overclocking them brings you to the same problems Coppermine upgraders face. If I just described you above, stick with the 66Mhz.

cC0 stepping

Intel shrunk the die a little, so maximum performance should increase a little. If you absolutely have to have 1Ghz from an Intel chip with air cooling, you probably should wait for this one. I wouldn’t expect too much more than that, though.

DDR boards for the PIII

I don’t think 200Mhz means RDRAM, so I guess these are going to be 200Mhz DDR mobos. We may see these before Athlon DDR boards, and these are going to present an interesting overclocking challenge. Again, like the T-Birds mentioned above, we don’t know yet if Intel has stuck something in that will prevent current chips from running at 200Mhz.

Actually, we know already that most overclocking chips being used today won’t work on such boards simply because you’d better expect divine intervention to go along with expecting a multiplier-locked 100Mhz FSB chip to run at a 200Mhz FSB, a 100% overclock. I don’t doubt one or two will do it, and that’s about it.

What it does do is get the right EB chips into the overclocking ballgame. Prior to the release of .13 micron chips, a cC0 667EB and 733EB look like they’ll become the O/Cing chips of choice. The first should be able with a cC0 stepping to get to 1Ghz, the second may get to 1.1Ghz. A 200Mhz bus speed should give you about 5% more speed than current setups running at 133-150Mhz.

.13 micron PIIIs early next year.

These chips should be able to get up about 1.4Ghz, maybe a little more. Additionally, they’ll probably be cheaper than PIIIs are today. Intel looks like it’s going to shift the PIII inbetween the Willamette and the Celeron/Timna categories.

What we don’t know is what flavors these PIIIs will come in, and how far down the scale they’ll go. If they come in 133Mhz FSB versions, people could have a field day running them at over 200Mhz with a 266Mhz DDR board. If not, we’re looking at a 200-266Mhz FSB overclock, and then we’ll be looking for 300Mhz RAM.

This Doesn’t Mean Stay With Intel

The Athlon chips will benefit more from DDR than the Intel chips simply because the FSB is going to jump more. Overclockers are already doing 150Mhz with Intel chips; Athlon overclockers are nowhere near that. If T-Birds and Durons are holding their own or better despite this FSB handicap, they’ll do better still if the playing field is levelled. Going from around 100Mhz to 266Mhz should get you about 15% more performance out of the same Mhz.

We’ll see chip shrinkages there, too, which hopefully will reduce the power/heat requirements somewhat.

Convenience and Security Over Speed?

If I got anything from your responses, it was the feeling that most of you feel speed isn’t everything. Cost and convenience are pretty important to you, too.

Relatively few of you are dismissing AMD out of hand, the general attitude is somewhat wary watchfulness.

For some reason, overclockers get extraordinarily attached to their motherboards, much more so than their CPUs, and it’s more than just the cost of a new motherboard. Maybe it’s the effort of pulling out a motherboard and in another as opposed to that of a CPU. Whatever the reason, it exists, and that’s a big disadvantage to AMD.

Some of you have expressed considerable doubt about the quality and stability of the Athlon motherboards (though outside of FSB, the same type of problems also existed with the PIII Via mobos). Unless Intel subtly or otherwise tells Rambus to drop dead, those DDR PIII boards aren’t going to be using Intel chipsets either, so we expect it will be six or a half-dozen of the other, with no BX board to fall back on.

What we think may be the killer factor for AMD overclocking is heat. When the realization sinks in that these AMD chips are an entirely different beast than the Intel chips, and that air-cooling isn’t really going to cut it, we suspect that is going to cut the potential AMD overclocking population down considerably, just as Athlon overclocking has been a minority up to now.

What you should also think about is that Intel in its own way is not too far behind. The AMD chips are designed to stand a lot of power and heat. The Intel chips are less hardy, and run into heat problems sooner.

There are at least three segments of the overclocking population:

  1. The hardcore hobbyists, who will do whatever it takes, mostly for the personal challenge.
  2. Those who will put up with a moderate amount of effort and hassle, mostly in choosing the right components, but no significant mechanical effort.
  3. Those who’ll overclock if they can with minimal time and effort.

I would say the first group is a tiny minority, with the rest about evenly split between the other two.

Which Category Are You In?

Water cooling.

If you say, “Water cooling, so what?” you’re in at least the lower level of group one (and then have to pass the “Peltier” and “soldering on components” tests to be hard hardcore).

If you start sweating and saying “What, water cooling??”, you’re in group two.

If you say, “What’s water cooling?” you’re in group three.

Overclocking began blossoming in 1998 simply because it was so painless. Buy the right components, put on a big ol’ fan, change a setting or two, and voila!

If it gets much harder than that, most stay away. For more than a few of you, using a conductive pen is too much. Battling heat gets much harder than that. To overclock these chips heavily, either you move to something like water cooling,
or you deal with a lot more instability when overclocking. We suspect most will say “No” to both.

The Long Term Problem

It may not matter too much to overclocking if the Durons and T-Birds are generally avoided because of heat, but what do you do if the Intel chips get too hot to handle easily? Where do you go then? Transmeta?

People will always overclock, the issue is how many.

A shrink to .13 micron will take care of some of the heat problem for a little while, but if you’re overclocking, you’re pushing it right back to the max again. The difficulty will go up and down, but so long as Intel and AMD continues the
trend of packing more and more power into tinier and tinier spaces, heat is going to become a bigger and bigger problem for overclockers.

Your Duron comments a little later today.

Email Ed

Here’s some of your comments about why you are and are not buying the new AMD chips.

A few observations:

  • There is virtually no Intel fanaticism in the responses, which surprised me a little. I had expected a few responses like that, given that this is an audience of a website that has until recently almost
    exclusively covered Intel chips. This doesn’t mean there is huge dissatisfaction with Intel, just that people are open to alternatives, which is good for AMD.

  • However, the alternative has to make a clear case that it is better and worth the extra cost, and that hasn’t happened yet with most people. This is not so good for AMD in the short term.

  • It’s happened with some, the comments somewhat underrepresent the proportion of people who have actually bought already, since the buyers didn’t usually go into why. About
    15% of the respondents have either already bought, or will buy shortly.

  • I think the general attitude is “wait and see.” Given that most of the participants appear to have Intel machines running at 800Mhz or better, it’s understandable they want some more improvement for their buck than they can get from the current offerings.
  • Finally. I think the comments for the most part bely the daredevil image overclockers tend to have. Caution is part of the mindset.

    On to the comments:

    Why I’m Buying One Now

    I’m Not Afraid of a Pen

    Luckily I didn’t have to buy the pen but I would
    have. I might not get a peltier because crushing my CPU is scary. I have no
    problem writing on or even soldering it. I would even chisel the bridges off
    of it if needed.


    A conductive pen does NOT scare me off.


    I will still buy a Duron/T-Bird if I only have to connect the
    L1-bridges. That can’t be that bad. I mean, cracking the old Athlon in pieces and
    doing some soldering on it to change the cache divider wasn’t that bad
    at all, and the Golden Finger Device wasn’t hard to get either.

    If I could push a 600MHz chip to 900 or even further, this would be a
    increase of 50%. It is absolutely worth buying a conductive pen and “drawing” some
    lines with it!!!


    Am I reluctant to buy a Duron because I need a conductive pen to unlock it?
    Hell no! When I first read about the Durons in Tomshardware I was looking
    forward to overclocking it like a kid looks forward to a candy store. I am
    in dire need of an entire new system, so I bought a new case, etc. But I
    wanted great cooling, so I bought all sorts of fans, etc. I didn’t want
    super-loud all the time, so I looked into options….

    To make a long story short, in my pursuit of overclocking a Duron, I’ve not
    only taught myself rudimentary metalshop techniques, I’ve taught myself a
    hell of a lot about soldering. Now that I’ve done all this, there’s no WAY a
    conductive pen is going to stand in my way.


    I think that it’s possible some people are just scared of hardmodding
    something relatively expensive like a CPU. I’m a college student, so even
    US$80 for a 600MHz Duron is nothing to scoff at. It’s going to take a lot to
    screw up my courage to make that first mark with the conductive pen, but
    it’s times like that that make overclocking worth it.


    I Bought Now to Get an Unlocked Chip

    I acted early in hopes that I could buy an Unlocked Duron…I hope it will pay off…I called about 10 stores and e-mailed another 15 in search of the elusive unlocked Duron.


    Why I’m Not Buying One Now

    Have To Make Sure You Get The Right Mobo

    Right now, it’s not really clear which stores/sites have mobos with those switches on them. And no, a conductive
    pen doesn’t scare me at all. I think that most people (ocers at least) want
    to make sure that they get a mobo with the multiplier switches on them, as
    they don’t want to go and try to solder stuff on a $125+ mobo and possibly
    mess up the mobo.


    The people I have talked to are waiting until the Asus and Abit boards are out in force, and a few people have gotten the proper setup’s down. There is no loss in waiting, since prices will only go down right?


    I think we’re all waiting for the Abit motherboard

    The first reason is because the overclocking boards (A7V) seem to be just
    coming out in mass quantities.


    The reason I wont buy a duron as of yet is because I’m waiting for the Asus K7V to come to my local store. Then I’ll get one, so it’s just the avalability of the boards with dip switches. I’m too lazy, and scared, to solder anything on my motherboard..


    I’m Waiting for the Price to Drop

    I’m waiting for the dust to settle and for prices to come down on the
    Mobo. I’m an overclocker for over 5 years now and I’m looking forward to my
    next system, hopefully by late September.


    Creatures of Habit

    Being human, I’m a creature of habit. I’ve always bought Intel, I’m used to it. I’m comfortable with it.


    Overclocking Intel Is Easier

    It just seems easier to buy a cbo stepping p3 than to mess with the amd scene right now. (imho)
    I am leaning toward a 700e — should get me at 933. Seems worth the little extra more than the Duron mess.

    I’m holding off on the tbirds/durons because a celeron 566 or 600
    system is still pretty hard to beat:

  • Celeron 566 $100
  • Super mature BX board $110 (asus, abit)
  • Standard PC100 ram
  • Semi-cheap 3D AGP video card

    The celeron at 900Mhz hardly generates heat compared to the AMD chips,
    and the motherboards are much more mature.

    The only thing holding me back is likely that the AMD mobo’s are VERY
    new, VERY expensive – plus the multiplier is set with jumpers and not
    in a slick softmenu.


    I’m Not Made of Money

    When I upgrade, I can usually only afford to replace one thing at a time (cpu, mobo, video, etc.) Changing to AMD would cost me up to $300, a goodly chunk.


    Switching to AMD requires
    both a new chip and a new (much more expensive motherboard $150 vs. $100).


    The second reason is people who have classic Athlons and Pentium/Celeron
    don’t want to upgrade their motherboard again. Not because they don’t want
    to, but cash flow isn’t always what we want it to be. . . . . I don’t think the conductive pen is the limiting factor. It’s the other
    parts which cost 6 or 7 times as much (motherboard) that keep the masses
    from filling their hearts’ desire.


    I own a BX board (BH6 rev 1.0x), I really didn’t want to buy a new
    mb to accomadate a new proccesor. So it all comes down to MY budget (or any
    BX board owner who wants to o/c with less $$$), I need a cost-efficient way
    to upgrade and still o/c.


    The only thing keeping me personally from buying a T-Bird or Duron is the fact that I have an Intel Mobo, right now the cost is too much for me.


    Since most of us are using Intel chips the move to AMD is not going to be a
    cheap one. Also, from what I understand about these new AMD chips, you have
    to buy a new motherboard to support the Socket-A format even if you are a
    previous AMD owner. Buying a new motherboard and CPU is asking alot of
    somebody, especially since most overclockers are doing it to save money, not
    spend more.


    Price/performance Doesn’t Justify the Shift

    Factoring in that I would have to buy a $150 Asus motherboard vs. a $95
    Abit motherboard, even if I were starting from scratch, the
    price/performance advantage isn’t that great.


    The only benchmarks I care about are the applications benchmarks (content
    creation, Quake, & 3DMarks) at comparable FSB’s and CPU speeds. There, the
    Duron had less than a 9% edge. Why should I give up my Intel infrastructure
    and proven BX stability for a 9% edge?


    Most people are probably contented with their coppermines and their comfy BX
    boards that they don’t want to give up for a chance to get a few more fps.
    They would need a new mobo, processor, most likely new cooling. and the
    price factor must come in with the T-birds somewhere.


    I’m Waiting Until They Get It Right

    I personally am NOT buying a T’bird or Duron until they have been available for a few months.
    I bought one of the first CuMine P3s, and was sorely dissapointed by the Intel’s stability, lack of SMP (Yes I bought 2), and initial lack of Mobo support.


    I will wait. It’s always a safe bet to wait.


    Give it about 6 – 12 months, and we will be looking at some really aweome chipsets for T-Bird / Duron. May have to get one at that stage.


    Jeez, gimme a break, these processors require yet another chipset to be stable. That’s
    what, 3 different chipsets in one year? (750, KX and KT?). I started with a
    750 board and just bought a KX board about 2 months ago. I’m not real
    anxious to go out and buy a KT based board and new processor just yet,
    especially with DDR RAM just around the corner. Is the KT chipset gonna be
    100% compatible? I’m not bettin’ on it.


    Waiting for the whole thing to shake out a little more- I have no problem
    with connecting the dots, but am very curious about the Abit Socket A board
    and want prices to stabilize a little as well.


    I think its more of a wait and see. I believe most of us on the site are pretty happy with what we have and
    a lot of us spent a lot of money to get here I think its still a little early to go
    spend 160+ for a mother board and 80+ for a processor + ? for pens and dip
    switches (unless you can get a good board) all for what? A little better
    benchmark scores?

    AMD Could Pull The Plug

    I won’t buy either processor. I was eagerly anticipating the release of
    these parts by AMD and when they were released there was so much question
    about weather or not they would be multiplier locked that it seemed only
    reasonable that AMD would make a statement clarifying their stance on the
    issue. Unfortunately, AMD has failed to make any kind of useful statement
    at all. What this means, AMD can change the Duron or T-Bird at any time.
    They could choose to simply stop connecting the necessary pins of the CPU
    package, which allow the motherboard to overclock the processor.


    I Don’t Know This Stuff

    While I have read many of the informative articles on overclocking the AMD. I’ve never had my hands on one. I’d hate to spend all my money just to fry the thing because of an error, whereas my chances of making an error and frying my intel is greatly redued.


    No, I Don’t Want to Use a Conductive Pen

    I’d rather search around for an unlocked processor, even at the
    slower 650mhz(soon to be changed :), than buy one and have to modify it
    myself… some people just aren’t comfortable with doing things like that to
    their new cpu.


    You have to play connect the dots with a conductive pen…(I am nowhere near steady in the wrist for small work like that, and wouldnt want to risk damage to the CPU)


    Yeah. I find it a real hassle to have to jump through these hoops in
    order to overclock the sucker. I am going to build a new PC for my
    friend, and it’s really a shame that it’s more difficult than it should
    be to choose the Duron.


    I was planning on hopping on the Duron train to speed when I read that they were “multiplier-locking” their chips, so I went Celeron II instead… now I have a 1.133 ghz computer, which is cool.
    I slap a 52W peltier on my Celeron II 566 and run at the 133 mhz bus, she’s doing 1133 mhz. I don’t want the extra time and effort of connecting those rediculous little bridges that needen’t be there.


    I thought to myself, yeah, the conductive pen would be too much trouble. Then I glanced at my watercooling setup and had to rethink my situation.

    I Don’t Want to Break It

    The real reason why overclockers don’t buy a duron or thunderbird is


    AMD Hasn’t Proven Itself Yet

    I certainly havent discounted AMD, and am watching very closely, but at this stage, they are really too new in the performance market…there just isnt the same sort of chipset support out there for them, yet…
    The fact that AMD’s CPUs are kicking serious ass technically hasnt really translated to the real world, yet…given time, it will, but until then, I will be sticking with what I have…


    Why bother with all the trouble when I can easily overclock a Coppermine
    and get performance that is on par with TBird? My last two Athlon
    setups were nightmares – special power supplies, video cards, driver
    updates, and lousy stability. Beats me why this hasn’t caught AMD by
    the balls yet. Sure my Athlons were a couple percentage points ahead of
    a similarly clocked Pentium – when my system was actually running
    correctly, and that wasn’t often.


    I Don’t Think It’s Good for Gaming

    I look at gaming benchmarks and I see the BX
    pound on AMD. People blame it on the 83 or 89MHz AGP 2x, but that’s not
    right because the AGP 4x is running 66×4 and the BX is running 83×2 or
    89×2…… I don’t wanna get my calculator out to check what the AGP
    overclock for 133 is. All I know is that in Quake 3 and Unreal
    Tournament (two major real-world benchmarks), the equivalently clocked
    BX133-based Coppermine has pounded on the Athlon as well as the T-bird
    and the Durons.


    Waiting for DDR

    I think that the AMD scene will be much better when we can use the bus speed to overclock. It will also help the Duron out a lot more to have the option of 266 bus and DDR memory because it has such a small L2 cache. The AMD platform just need time to mature like Intel has had over the last 2 years. We have 3 different official bus speed at intel (66, 100, 133 — these give you a lot of room to play with) and only 1 at AMD. Wait until we have at least 2 official bus speed at AMD and then you will have more Duron sightings in your database. (And I will have to get one myself).


    When I can afford it and DDR is finally mainstream along with a good dose of AMD LDT, I may look into getting a socket A Thunderbird as my main system.


    I’m not migrating to that platform until chipsets are tidied up a bit more
    and include SMP and DDR RAM. Personally ive got no problems with busting out
    the soldering iron etc..


    I’ll buy them in a few months time, when DDR boards and
    memory are out and hopefully by then things have
    stabalized. But untill then, my C566@900 is doing me
    just fine 🙂


    Email Ed

  • Be the first to comment

    Leave a Reply