Tulatin Spanks Athlon!

[lhgpoobaa]

What the -Peeparooni- is Intel doing?

i suppose its all becauze they dont want to make the pentium4 look bad.

a. discontinuing the p3/tully line.
b. rereleasing it as the 100mhz fsb cellery
c. not letting get above 1.26Ghz
d. very limited distribution

intel has a great processor that can spank the athlon in a significant number of benchmarks...and produce it far cheaper than the p4... yet for the general public its basically stillborn.
*sighs*



In memory of the 90+ Auzzies missing in the WTC disaster. An attack that has changed the world.

 

[AMD_cErTiFiEd]

Granted it did win but we need to look at the simple aspect.... It beat a cpu that is essentially 1+ yrs old with only clock adjustments

 

[Crashman]

Yes, but the PIII Coppermine is over 2 years old, and the Tualatin is nothing more than a minor revision of it. Plus, it's based on the Pentium Pro which is over 6 years old!

Back to you Tom...

 

[Matisaro]

I work in a fab, and I found this statement tom made to be slightly in error.

The miniaturization to 0.13 microns even enables a higher yield rate.

While the number of good die will increase simply due to the smaller die size(unless your fab is totally out of control) the %yielding will decrease, smaller die features are INHERANTLY more defect prone than larger ones.

~Matisaro~
"The Cash Left In My Pocket,The BEST Benchmark"
~Tbird1.3@1.5~

 

[Crashman]

What size die process on the original P-Pro?

Back to you Tom...

 

[Guest]

Tualatin also spanks P4....

But I must note that in this test the Athlon on SIS735 scores less than it did in the original SIS735 test...

So you should not focus on the small margins.

Regards,
lurqa

 

[Matisaro]

I dont work for intel, so I may be wrong, however I remember reading the origional ppro was either a .5 or .35 (not sure which) micron process. The killer defect size scales with the feature shrink.
All chips have defects even working ones, particle contamination, debris etc. The smaller the critical dimentions(die features) are the smaller a particle has to be to kill the die. On a .35 micron gate array a .05 micron particle may not (and most likely would not) affect the functionality of the gate, however on a .13 micron process the size ratio between the defect and the feature is much smaller, a .05 micron particle would have a MUCH higher chance of killing the gate, thereby killing the processor. Since even clean rooms have many particles floating around which are of a small size, the smaller feature size will take a larger yield hit due to particulate contamination, not to mention process variations.
A thickness differential of say 10 nanometers may not affect the focus of a .18micron photo stepper, but .13 micron steppers may be unable to function with that kind of variation. Flatness on a wafer is important for focus, I think crash man will get it but if anyone else dosent follow I can do a folowup post explaining what I mean. The small line widths require extremely sharp photo imaging, without it you get bad exposures and faulty lines.(smudged, shorting etc) this with the increased damage from contamination results in lower yield. This lower yield however is generally more than made up for by increased die per wafer.(especially on 12 inch wafers). I have a picture of an actual defect I took at work on a scanning electron microscope, (and emailed to myself at home). I may get some webspace and link to it sometime this weekend if anyone is interested.



~Matisaro~
"The Cash Left In My Pocket,The BEST Benchmark"
~Tbird1.3@1.5~

 

[Guest]

Just noticed the subject heading.

Keep in mind the following. the tests were done with sis735, the revised kt266 offers a lot more performance. (it goes with out saying the nforce will do more so but theres always a better thing 'about to come out' from both camps).
and the tualitin adds some nice stuff like hardware prefetch and isnt really comparable to the thunderbird. the corresponding athlon would be the palomino. as the desktop palamino is about to launch and the MP is already out im surprised it was not tested or mentioned.

take care

 

[Guest]

yep
an athlon 1,4GHz on an kt266 get a framerate on quake 3 of 220 thats al lot different then the 194 he
an athlon would have reached 15-20fps in flask mpg4 this is why this bench is strange because it gets only 11. ive seen in flask even higher scores of an athlon that reached the 24fps
so i wouldn't pay much attention on this benchmarks because it uses the best intelchip but sis735 is a nice ship but an kt266A is much much better so i think this comparison is usseless. and remeber this is done with an tb not with a XP that will come out soon and for i understand is an athlonXP at 1200MHz faster then an P3 @ 1200MHz.

computer

 

[Guest]

I think that should be the REAL heading.

I did read the article and i did see the PIII scores, and you know what? I'm glad... the PIII was and STILL IS a very good system. The fact that it beats out the PIV prooves that the p4 is just a stripped down overclocked p3.

PIV SUCKS!!! (Had to get that out of my system).

Now as for the PIII vs. the Athlon, i hate to point out the fact that the KT266A gives the athlon the boost it needs to beat out the more expensive PIII. And i didnt see and Athlon MP scores in those benchmarks, i'd like to see a rough comparison of the Tual Vs. the MP.

In my opinion the AthlonMP on KT266A will hold out against the new (and more powerful) PIII Tual at the same clock speed and completely frag the PIV at any speed.

If you like the PIII, kudos.

<font color=blue>I for one run Quake 3 on a P133(No MMX)</font color=blue>I have no affiliatioin w/ Intel

 

[Makaveli]

it doesn't really matter if the revised P3 did win some benchmarks. It will never see the light of day past 1.26 Ghz! Also have u seen the price of them!!!!

It's to little to late, Intel has seen this.
R.I.P P3

 

[Crashman]

I think I remember a mention of it being .35 micron come to think of it. Or was that the PII? At any rate, it's transistors were so large and it's die so spread out that it required 5V(!) to operate and had major heat production!

Back to you Tom...

 

[Crashman]

I wouldn't call it a spanking like the original poster did. It does spank the Willy P4 though.

Back to you Tom...

 

[rcf84]

I wanta see some Northwood Benches. Can't wait maybe the FPU will work right.

Nice Nvidia and ATi users get a Cookie.... Yummy

 

[Guest]

Man, I think most here are missing the REAL important underlying message. Intel now has 0.13 micron fabs functional.

Even if PIV is a hunk of cr@p, and I'm not saying it is, by bumping the clock speed a few gigs they will tear some serious @ss.

 

[Guest]

oops, sorry RC, that was not aimed directly at you.

 

[zengeos]

Actually, Knew...try to actually find Tuallies in stock... VERY scarce. So, while Intel may have them in production they seem to be coming out at a trickle. Is this due to problems with production, high demand or a combination of the two?

I suspect it may be more due to demand....mostly from notebook makers. This is a large reason why I think the Palomino Athlon has taken so long to come out in desktop form...AMD is simply selling all the Palominos it can make for notebook and MP demand.

Mark-

Mark-

When all else fails, throw your computer out the window!!!

 

[Guest]

zengeos,
I'm sure you are correct about there being production problems. Weren't physisists claiming, just a little while ago, that 0.13 micron was impossible (something about the wavelength of light in the visible spectrum)? If they have already done the impossible, I am confident they will be able to improve yields as well. In my opinion it is only a matter of time before they work out the production kinks.

I hope AMD for their own sake has something similar up there sleeves, as I'm sure they do.

I would hate to see this intense competition between AMD and Intel stop until my machine can simulate 3D weather patterns in real time in the background while I'm playing Quake 10 at 100fps.

Kevin

 

[Matisaro]

KNewton, .13 requires exomer steppers and high uv steppers to image such small feature sizes, regular light dosent get that small, thats what they were saying.

~Matisaro~
"The Cash Left In My Pocket,The BEST Benchmark"
~Tbird1.3@1.5~

 

[lhgpoobaa]

hehe
and that quake3 benchmark @ 640x480 always has me in stitches.
i.e. the athlon can do 194 and the p4 200+
ummm who cares? lol
find me a monitor and a pair of eyeballs that can tell the difference then i might be interested.

on a serious note, anyone seen the damn price of the tully? especially the 512k model? you could buy TWO athlon 1.4's for that!

and for intels sake i really do hope that the new pentium4 with supposed improved core is better, not just faster mhz wise.

Religious wars are 2 groups of people fighting over who has the best imaginary friend.

 

[Guest]

Too true about the Q3 benching. The analogy I use to convince myself to keep dumping my money is: because a yugo will effectivly break all speed laws doesn't mean I wouldn't rather have a mustang that can wheelie while breaking all the speed laws.

 

[Guest]

Hey, if it's Ghz that counts for the home user, then why don't all the OEM's ship Via C3's?

A rig with a C3 and a CryoCooler would probably cost less than a comprable PIV system. And you could OC the son of a b17ch way past 2Ghz.

Not a sermon just a thought!!!!!


<font color=blue>I for one run Quake 3 on a P133(No MMX)</font color=blue>I have no affiliatioin w/ Intel

 

[silverpig]

You're missing the point of the Q3 benches. It's to show which processor is faster, and by how much. It's not really to show which processor can play q3 faster because, as you said, who cares at those framerates?

Think of it more as an OpenGL bench rather than a q3 bench.

Lyrics. Wasted time between solos.

 

[flamethrower205]

Personally, I have nothing against P3. I even like P3. I am happy to see that it's succeeding, but sdaly will be taken away. The reson I hate Intel is b/c P4. Otherwise, they're a great company, and the P3 is a kickass processor. However, I doubt anything from Intel will be able to stand up to the all mighty Hammer when it comes out. As one sig goes, *News flash* Intel suddenly hit by a hammer....suspect is AMD. I love that Sig.
That's mere specualtion though, I pray not, but teh Hammer may turn out to be junk (I had high hopes for the P4, but when it came out.....)

Your brain: PC
You brain on drugs: Mac