New Athlon vs. Coppermine

[Ulysses]

I've been thinking:

I've seen many benchmarks and it looks to me like (with the exception of professional/OpenGL apps) the P3 Coppermines and the new Athlons are about even in performance, with perhaps a slight nod to the new Athlon. While the old Athlon really surpassed the Katmai, the performance lead was quickly retaken by Intel with their Coppermine. Now with the new Athlon the score is about even. This has disappointed many, including Anand & Tom Pabst.

My question is this:
When you consider that P3 Coppermines are generally much more overclockable than Athlons, doesn't this tilt the performance benchmarks in favor of Intel?

The forums are filled with people with highly overclocked Coppermines and Celerons, but I don't see too many Athlons like that (in fact I've seen a number of Q's like 'why won't my Athlon o'clock higher'). And I know there are individuals that have had great o'clocking results with their Athlons - but I'm talking about the bigger picture here. I had the impression that this poor overclocking record was due to the EV6 bus already running @ DDR.

If I'm right, then for overclockers the price differential in favor of AMD may be an illusion.

I'm not trying to take a position here in favor of Intel or AMD - I just want to be thinking correctly when I choose a platform.

It just seems to me that you are more likely to get a P3 700/100 Coppermine to 840/120 or higher than an Athlon 700/200/100 to 770/220/110. And the 840 P3 is about equal in performance to an Athlon 800, which costs about the same as the P3 700, per SE @ OEM.

Thus, if the performance of an Athlon 800 and a P3 700 (when overclocked to 840) is about the same, and their prices are about the same, then I think I'd prefer an Intel BX or 815E chipset to the checkeded record of AMD or VIA chipsets, or at least be indifferent.


Just curious - not trying to start a war !

😀

 

[Czar]

I have my 550mhz Athlon at 800mhz (can do 850) and only with 100mhz fsb.

Its true that the EV6 bus overclocks badly, the reason not so many have overclocker their Athlon that much is because they need a GFD.

And actually most athlon mobos are good and very stable, that is the best ones 😉

 

[Muerto]

Athlons can overclock very well. The EV6 bus definetly hold them back but if you can adjust the multiplier then you're only limited by chip quality when overclocking. 🙂

 

[Ulysses]

Yes. I agree the GFD is a sort of equalizer. Let's hope that the multipliers can be adjusted more easily on at least some of the new Socket A mobos.

And I guess VIA has improved quite a bit too. 🙂

 

[OneEng]

I agree that the oc ability of the Intel processors up to around 800Mhz has been quite good. The processors from Intel over these speeds have been less than stellar over clockers though.

Considering the minescule difference that you can actually tell with your own senses (without the aid of some sort of benchmark) either one will do; however, I think there is more to the choice than just current performance.

The BX Socket 370 is at end of life. As of next year, you will see no new stuff for these platforms. The next Intel processor Willamette will debut on a new platform. There has already been rumblings of mb manufacturers refusing to support the Willamette due to Intel's second generation Willamette having yet another form factor with more pins than the original Willamette. This is not an accident IMHO. Intel's business plan is not just to sell processors (like AMD), but to sell main boards and chipsets as well. In Intel's business model, it makes sence to change the platform. This ensures that when a new computer is sold as an upgrade, Intel gets 3 compontents (minimum) sold.

While this is great for Intel sales, it sucks for we upgraders.

AMD on the other hand has standardized on Socket A for the forseeable future. A Duron purchased on a Socket A mother board today can be upgraded to a TBird when the price comes down, then to a Mustang when they arive.

Personally, I think that a 700Mhz Duron is enough power for anything TODAY. As long as I can upgrade to a 1Ghz+ Mustang next year for less than say $250.00.....well, its sort of hard to argue with numbers like that.

 

[Modus]

Ulysses,

You're right that the current Athlon T-Bird is not much faster than the P3E in standard apps and games (professional 3D work is a different ball of string and here AMD owns the day). However, you're mistaken in the impression that Intel CPU's are more successful overclockers. Actually, both Athlons and P3's share roughly the same overclocking potential. Some people had limitted success with their Athlon setup because they didn't use a Goldfinger device to break the multiplier lock, and so were confined to minimal FSB increases. With this $50 device, the Athlon becomes an overclocking monster.

With the new Socket A chips from AMD, things have changed. It is now possible to change the clock multiplier through BIOS settings. The new socketed .18u process used by AMD produces startling results -- a normally cooled Duron 700 overclocking to 950 MHz for the price of a P3/550 is almost guaranteed. (Tom's Hardware did this using pre-release engineering samples, not even final shipping silicon.)

If you're comparing the Athlon and Coppermine, you're looking in the wrong place. The Duron is, bar none, the best choice for a new system. Its price/performance ratio, even without overclocking, is staggering. At 90% of an Athlon's performance and 100% of its overclockability for 60% of its price, there is simply no valid reason to consider anything else.

BTW nice handle.

Modus

 

[OneEng]

Red Dawn,
Except for the fact that you can upgrade a SocketA with a Mustang next year. This is not true of the SlotA platform. Good point though.

 

[Quickfingerz]

Ok you said how P3's 700 are getting to 840. They actually get to around 870's on average and some above 900+. All 700's now get to 900+. My 700 gets to 950. Some 700's have gotten to 1 ghz. some 550's have gotten to 1050 MHz. The fact of the matter is: the CPU bus of the p3 overclocks well. not really the processor. let's see how high a EB will overclock.

 

[ragiepew]

red dawn: I think oneeng mistyped his comment because you are right... socketA will be the formfactor for mustang.

On the topic of OC athlons, i wonder how the amd760 chipset will do, hopefully they have ironed out some problems w/ the bus so maybe their nextgen mobo's will do well... we can only hope. As far as P3e v. athlon in straight CPU OCing potential i think most people have it backwards. I think a tbird/duron would kill a p3e/celeron in OCing if the bus wasnt an issue. just my opinion though.

 

[Modus]

Red Dawn,

There's no way the price difference between a Duron 700 and an Athlon 700 for Socket A is going to be only $12. At this point, we can't trust PriceWatch because none of the chips are distributing in volume yet. But AMD would never price their chips the way you describe. It just wouldn't make sense.

Rather than look at the currently unstable prices for products that have just reached the market, and decide that I am a miserly idiot, why not use some common sense and think back to AMD's comment that the Duron's would be priced lower than Celerons? Obviously with that kind of pricing spread, the Duron will be a value juggernaut.

Modus

 

[BoberFett]

Modus

I just got an email from my distributor which lists a Duron 700 for $140 and a TBird 700 for $161.

Edit: That same email lists the Duron 600 at $79. That looks like the one to get, IMO.

 

[Modus]

Those prices for the 700 MHz chips look a little better, but still don't hit the marks AMD is targetting. I'm sure the price delta will increase in the next month. Even now, a Duron 700 is still better value than a T-Bird 700 costing 15% more. And of course, the 600 blows everything away.

Modus

 

[OneEng]

Red Dawn,
OOOOOOPPPPS, I fixed it. Of course I wanted to say that Mustang would run in SocketA....not SlotA. My Bad.

 

[Mday]

fett, those prices are making me rethink getting this c2... oh well.

i was really waiting for dual aths anyway 😉

 

[mschell]

You're theory that Intel changes form factors to bolster chipset and motherboard sales is off base. The slot one form factor has been in use far longer than any form factor used by AMD. It's initial implementation was thought by many as an Intel plot to disadvantage AMD whose processors ran on the socket 7 platform. We know now that there was no plot, it was the only to get 512K L2 cache with the CPU fabing technology of the time. Strangely enough, AMD followed suit with their own slot A version, so much for conspiracies.
Now technology allows the L2 cache to be located on the CPU die hence the move back to a socket form factor by both parties but AMD has sort of screwed many initial adopters of the Athlon chip by making the Duron/Thunderbird electrically incompatible with the Athlon. This means no slot converters for all those people who paid premium prices for Athlon motherboards, who's screwed who?
Modus,
I know the new AMD processors are the best thing that's come you're way for awhile but I can think of a few reasons why one should not buy the CPU just yet. One reason is there's an apparent lack of motherboards to run these processors, another is these will be new motherboards which means bugs and more bugs along with weekly trips to the MB's web site for BIOS upgrades but AMD users love being unpaid beta testers so maybe this is ok.
You might check Tom's site for the update on overclocking the Duron/Thunderbird. Seems AMD is using a high powered laser to modify chip parameters after manufacturing. Unless you have one of these in your garage you might have a tough time with OC'ing. - M.