Here comes Intel again..watch out AMD!

[Toaster]

Well, they are out and they are cheap.
Yes folks, the Cel.II`s are out and the 566 that goes to 850+mhz without a backward glance is available for just over 100 bucks.
Now there is bang for the buck, no approved this or that, no guessing games with compatability, no goofy halfwit drivers and these puppies run on more system boards than you might think.
A full speed cache of 128KB thats 4-way pipelined and an enhanced math unit used on the PIII-E.
The "coppermine" based CPU`s require a system board capable of setting the voltage down to 1.6x volts.
So tell me, what else do you need?

[HAL9000]

Got another correction for you Toaster, I just noticed this one myself. While the Coppermine PIII's run at 1.6V, the new Celeron's have dropped that another notch to 1.5V. Could be why my 533Mhz is running at 800Mhz with standard cooling. At the factroy 533Mhz in my other computer, it's running at 43C instead of the 49C that the PPGA version was running at. I'm probably pushing my luck, but if this sucker runs stable through Winstone overnight, I'm trying 133FSB.

------------------
If it ain't broke, you're not pushing hard enough!

[This message has been edited by HAL90001 (edited 06-14-2000).]

[LawyerRon]

Hey Toaster,
I don't know much about these new Cely IIs. What chipset and slot do they require?

------------------
"To speak ill of others is a dishonest way of praising ourselves"

[HAL9000]

BX, 810, same as before, but for 533Mhz and up, your motherboard must support a multiplier of 8 and higher. They are still on a 66Mhz FSB. The motherboard must also support a voltage down to 1.5V which is a hair less than the PIII.

------------------
If it ain't broke, you're not pushing hard enough!

[LawyerRon]

Hal,
Until a newer bios upgrade comes out, I'm out of luck. By Soyo 6BA is Slot 1, BX chipset, supports up to 133fsb, but has a multiplier of 7.5 max.

------------------
"To speak ill of others is a dishonest way of praising ourselves"

[Mattman]

Don't count out AMD yet. AMD's competitor for the Celeron II, the Duron, is suppoused to be nearly as fast as the Athlon and Coppermine. It's also NOT going to be multiplier locked, since the GFD's will be built into the motherboard. That will make overclocking a lot easier, since you won't have to play with the bus speed (not saying you can't though). The onboard cache will make it more stable for overclocking. And with 600MHz processors coming out for less than $100, I think Intel might be in for a real suprise.

------------------
Life without computers would be like the world without war.

[HAL9000]

It will always be an arguement. I've been to no less than a dozen review sites of the "PIII killer" Thunderbird.... most of them said it just isn't there. But if you're going to factor in cost, of course the AMD is quite a bit cheaper.

------------------
If it ain't broke, you're not pushing hard enough!

[Toaster]

Hey Hal,
I had a long talk with a friend at Intel and he said the following:
The new .18 micron CPU`s dubbed the coppermine are all intended to work with a "nominal" 1.60 volts. He went on to say:
While this voltage is new for the industry, these CPU`s will tolerate voltages from 1.45 to 1.70 and perform "as stated" within temp margins "using an Intel supplied" heatsink.
Further, he is well aware that aftermarket heatsink/fan combinations "grossly" exceed Intel recomendations for the most part and say that the above voltage range is considered "normal".
He also stated that the voltage requirement is not exactly "static" and may vary by as much 2% from die to die and system board "loading".
(he clammed up as soon as I mentioned overclocking )
So hal, you are both right and wrong and thanks for giving me a reason to ask someone "in the know".

[Carlos Kerr]

Hey did I miss something? Are the CeleronII's NOT multiplier locked? My P2B supports an 8x multiplier and runs up to 133MHz. If I put one on an Iwill or Asus sloket does it over ride the board jumpers? Please clue me in!

[glc]

They are multiplier locked but NOT bus speed locked.

[HAL9000]

Cool...nice piece of extra info Toaster.

------------------
If it ain't broke, you're not pushing hard enough!

[lsakhvoruk]

This is for LawyerRon! Hey man your SOYO should be able to run the Cel II with no problem! I have a 6BA+IV which has a maximum multiplier of 8X and I'm running a 566 Cel II on it which has a multiplier of 8.5X. Since the Cel II is multilier locked the processor does not ask the mobo for the multiplier setting but overwrites the mobos multiplier suggestion and only uses the FSB that the mobo provides. So even though you will not get the correct CPU speed in the SoftMenu of the BIOS your motherboard should post the CPU's new speed at start-up. My mobo reports the chip as running at 810Mhz with FSB of 155 and some little multiplier but I in reality I set the FSB at 95Mhz it works great! So there you go! The new Celerons are awesome try one for yourself! Leo.

[HAL9000]

Hey Toaster....we might have to agree to disagree . My boss talked to his Intel friends and while the coppermines are labled 1.6V (like you mentioned an optimal 1.45 to 1.7V), the new Celerons are actually labelled 1.5V. His informant claims that this is due to the reduction in cache (256K is actually on the chip but disabled).

------------------
If it ain't broke, you're not pushing hard enough!

[LawyerRon]

Hey Leo,
thanks for the info, I'll keep that in mind. My Cely 500 is running rock solid at 572 and that's plenty fast for me right now. It doesn't seem to like a higer fsb setting though.

I think your 6BA+IV runs the VIA chipset, is this right? I've heard it's pretty good for overclocking too.

My BX board officially only supports 100mhz fsb, but when I flashed the bios to get the Cely 500 to run, it now has the 133 fsb setting.

------------------
"To speak ill of others is a dishonest way of praising ourselves"

[lsakhvoruk]

No Ron, my motherboard is also a 440BX! If you only have a 100 Mhz setting on your mobo and nothing lower you should still be able to oc the new Celery just add a little voltage Leo.

[LawyerRon]

Hey Leo,
I stand corrected on your Soyo board. Mine has settings from 66 to 133 mhz. In fact, it even has some settings lower than 66 but I don't recall how low it goes.

------------------
"To speak ill of others is a dishonest way of praising ourselves"

[Toaster]

Hal,
Well thats very strange. I have before me a Cel.II-600 that states "1.60v". Then, being the curious person I am, called a few retailers I do business with. From the 12 boxed Cel.II CPU`s read (skt370) 4 read 1.60v, 3 read 1.6v, 3 read 1.65v, and 2 read 1.55 volts.
The differing numbers differed when the "country of origin" differed in "almost" all cases. This holds somewhat true for the "variable" of 2%. I`m surmizing manufacturing differences?

[HAL9000]

Learn something new every day I guess, mine have 1.5V. Interesting.

------------------
If it ain't broke, you're not pushing hard enough!

[Alfie]

Duron released!
Kicking crap out of Celerons,matching Athlons and beating Pentium III's
:c



[This message has been edited by Alfie (edited 06-19-2000).]

[Toaster]

Hey Alfie,
Can Duron do this for 100 bucks on a "typical" system board?
What about "approved this or that"?
Did AMD finally find out that the "cache" was actually useful and decide to let it run at full system clock? (thank Intel!)
Whats the price range for the Duron? The price for the system board? Memory?
Etc...etc...so on and so forth....

[figarowa]

Now now toastie, i'll admit that the Intel cpu's do overclock like a bandit, while AMD has a better FPU, sos its pretty much anyones personal choice..(AMD fo mee) of course..now where would intel be without any competition...a direct monopoly on personal computer cpu's, and what would be to stop chipzilla from charging an arm and a leg...
jus my 2 cents

[Charliey]

heres a link to the duron review at anandtech AMD Duron Review

------------------
-Charlie

[Alfie]

Hey Toaster,
Duron's cheaper than Celeron,memory costs the same,mobos cost more,but as more are released we'll see a change there.
Bottom line they'll cost the same for now.
But Duron rules.
As it turns out they're not clock locked,so can be overclocked by bus speed or by multipliers,or both!
Celeron's still stuck on a 66mhz.bus speed.
So look at the positive,perhaps this will force Intel's hand in upping that!

[LawyerRon]

Hey Toaster,
Want me to jump in on your side?!

------------------
"To speak ill of others is a dishonest way of praising ourselves"

[HAL9000]

I don't really understand the statement of the Duron kicking the crap out of the Celeron. Looking at the review, comparing the Duron 700 to a Celeron 566. I would hope it would beat it. It would be a sad day for AMD if it didn't!. But let's flip the coin here. Lets overclock that 566 to 850 on the good old trusty BX chipset and what happens. The Celeron beats the Duron in all of the benchmarks listed except one. Sure, it's comparing apples to oranges, but isn't comparing a 700 to a 566 a bad comparison as well?

------------------
If it ain't broke, you're not pushing hard enough!

[Alfie]

I was looking at the overclocked Celeron and comparing the two of them!
In addition to that set of benchmarks,there have been others,I'm still trying to find one where it goes mhz.for mhz.
It,however,appears the Duron is the winner here.
As their overclockability becomes better known and more mobos are produced I believe the Duron will come to it's full potential,and as stated the heat will be on Intel to up the fsb for the Celeron!
Here are the benchmarks from Tom's hardware,which does go mhz for mhz. http://www.tomshardware.com/cpu/00q2.../duron-08.html



[This message has been edited by Alfie (edited 06-20-2000).]

[HAL9000]

But again...is that a fair comparison. The Celeron 700 does not yet exist so you are comparing overclocked to a factory clocked processor. I have my 266 Celeron at work overclocked to 533, but the actual performance gain seems to indicate that it is performing at about 450. So would it be fair to compare a Celeron 266@533 to a Celeron 533, no, the Celeron 533 has the advantage of not contending with heat buildup as well as running at spec which the 266@533 is not.

------------------
If it ain't broke, you're not pushing hard enough!

[Toaster]

You folks keep saying "chipzilla" as if Intel had no compitition. You and I BOTH know this is as far from the truth as one could get.
Sure if Duron is released at the same pricing/speed as Intel this will help the end user, this I dont contest to any degree.
But remember..If there is a "chipzilla" there is a "GUIzilla" (aka winderz) But you folks seen to duck that one all together.
Remember, you folks gave Micky$oft the power they now have, you alone. So next time you say "hurray for the underdog" sayso knowing that you mean : "when it`s convienient".
So you dont think Intel learned a bit about the marketplace and such?...think again.
Sure, AMD may be a fine, fast chip, but if you reverse the tables the story remains unchainged...BIG AMD and poor lil underdog Intel. I fear some of you can`t see the forest because of the trees.
For competitors, Intel has:
Cyrix
AMD
and a few "no names"
Plus, intel and AMD as well as the former IBM owned Cyrix (still to some degree) each have "other" microelectronic device subsidiaries.
Chipzilla? no...guess again..but they were the FIRST to give you what you use today...the home computer.
So next time you scream "chipzilla" look about you and your system(s) and see how many DIFFERENT O/S`s you have...then we can discuss the term "monopoly".
I still attest that AMD makes a great product for many folks, I just don`t happen to be one.

Linux...be free to choose...choose to be free...freedom of choice.

Micky$oft..a monopoly..in the "top ten" with AT&T.

[Alfie]

Hal,
Celeron 700mhz are available.
The reveiw pitted mhz for mhz.
Here's an article pitting an overclocked Celeron against a Duron.
Interesting read.
Toater,I'm not down on Intel,nor on Celerons,
Intel makes quality parts(excluding there last batches of chips after BX!)
As I stated,hopefully this will force Intel to change some things with the Celeron(100mhz bus speed?,larger cache,etc.
As far as Windows,it is the OS of choice and until Linux becomes more mainstream,will probably remain so.

Ace's Hardware
AMD's Duron: Salvation For Tight Budgets?


By Johan De Gelas
Monday, June 19, 2000

Overclocking, Memory Performance, and Conclusion

I can imagine that quite a few people might raise the question whether or not the Duron can compete with an overclocked Celeron. Indeed, there is a fair chance that you can get a Celeron 566, which is clocked at 8.5 times 66 MHz, up to 850 (8.5 x 100 MHz). I had to use 1.75v to get the Celeron 566 stable at this high clock-speed. My time was quite limited with this chip, and it remains to be seen if this speed really is rock stable, but there is no denying that the Celeron has a huge overclocking potential.

Our K7T Pro was a beta board, and would not overclock past 110 MHz. It is likely that the MSI K7T Pro will reach at least the same speeds as its predecessor, the MSI K7Pro, which had no problem reaching 115 MHz FSB. The Duron has quite a bit of overclocking potential as its core voltage is not higher than 1.5V, but until we have boards that "unlock" that potential with a multiplier setting, our Duron 650 was limited to 715 MHz.


Benchmark Duron 715 Celeron II 850
Quake 3 Normal 97.6 92.7
UT 61.5 61
Content Creation 28.9 29.9
Povray 33 s 34 s
MP3 encoding 34.4 s 39.2 s

Well, the Celeron 850 shows its clock-speed muscle in Content Creation, but that is it. The Duron 715, with its better memory subsystem (FSB at 220 Mhz vs 100 MHz, Memory at 145 Mhz vs 100 MHz) outperforms the 115 MHz faster celeron II in all other tests.

Understanding the Duron

So, why does the Duron perform like it does now, and can DDR SDRAM or VCSDRAM boost its performance? Let's find out. The Linpack graph tells more than a thousand words.

Yes, the Duron outperforms the classic Athlon 650 until the matrix size gets larger than 107 KB. You can also see the results from our AMD Duron 700 test unit, which we will go into more detail over in the future.

To understand why the Duron performs like this, you got to have a basic understanding of the exclusive cache. Each time a needed piece of data can not be found in L1-cache, the data that is found in the L2-cache (or worse, in main memory) will be written into the L1-cache. The least recently used (LRU) cacheline will not simply be replaced like in a normal cache, but it will be written to the L2-cache. You can understand that all this writing demands extra bandwidth between the L2-cache and the L1-cache. Both the Athlon 650 and the Duron have a 64-bit pipe from the L2-cache to the L1-cache, but the Duron's cache is clocked twice as high. Therefore, the Duron's cache has twice the bandwidth of the Athlon 650, 5.2 GB/s (8 bytes x 650 MHz) versus 2.6 GB/s (8 bytes x 325 MHz). As we have learned, the exclusive nature of the Duron's cache will eat quite a bit of that bandwidth advantage. The maximum amount of data or instructions that the Duron can store on the die is less than 128 KB. Because the data cache is 64 kb and it is very unlikely that the whole L2-cache will be used for data or instructions. That is why you see the Athlon 650 outperform the Duron once the matrix size gets past 107 KB.

Most popular applications have a memory footprint (amount of repetitively used data and instructions) which is more or less 256 KB to 512 KB in size, as we can clearly see that the Coppermine 600 outperforms the Katmai and the Thunderbird outperforms the Athlon in almost every test. This is why the Duron cannot outperform its older brother in most benches.

The small cache also means that the Duron is very sensitive to the memory subsytem. Watch this:


Benchmark Duron 650 with 100 MHZ SDRAM Duron 650 with 133 MHZ SDRAM Duron 650 133MHZ VCC SDRAM
Quake 3 Normal 82.1 86.7 89.3
Quake 3 HQ 81.3 82.9 85.1
Expendable 64.6 69.8 69.9
MCAD 150 17.63 20.68 20.53
CPUMark 55.2 57.9 57.9

Oh! This is really surprising! The VIA KT133 is better optimized for the VCSDRAM! VCSDRAM (Virtual Channel SDRAM) provides no less than a 3% boost to the Quake 3 scores. The 100 MHz scores reveal the achilles heel of the impressive Duron. You can use your old PC66 or PC100 SDRAM, but you will lose a lot of performance.

Final Evaluation

Without any exaggeration, the Duron offers an earth shattering price/performance ratio! Its Arch-enemy, the Intel Celeron, is too crippled by its 66 MHz bus, and must throw in the towel at standard clock speeds. The Celeron has a better overclocking potential for now and comes closer to the Duron at overclocked speeds, but even at lower clock-speeds, the Duron leads in most benchmarks. No other chip benefits so much from better memory technology. With DDR SDRAM and other faster memory technologies around corner, the Duron has a bright future ahead. And let us not forget that the Duron has more than enough frequency headroom, so future Socket A motherboards might offer an greater overclocking potential.

If you have super 7 system and you like to play games, then the Duron offers remarkable performance (up to 250%!) gains, even though you have to invest in a new motherboard.

We appreciate that AMD offers a very strong CPU to the low budget market, but nevertheless we hope that socket A may live longer than Slot A. While the Slot A processors still offer excellent performance, we feel that AMD should offer a better upgrade path to all the owners of the Slot A motherboards.

Special thanks to Drew Prairie and Patrick Brouckaert of AMD for their quick support in this review and in the past. Mina Hsu and especially Angelique Berden of MSI also rank among our heros for helping to supply Ace's Hardware with elusive K7T Pro engineering samples. Finally, thanks the PC Magazine Belgium crew (Thomas, David, and Kevin

[HAL9000]

Well put Toaster. The heated competition benefits both AMD and Intel users with lower pricing. Hopefully quality won't suffer though.

------------------
If it ain't broke, you're not pushing hard enough!

[This message has been edited by HAL90001 (edited 06-22-2000).]