• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

I smell a paper launch on or around July 31st :)

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
the athlon does infact show signs of a new generation... look at its professional apps' benchmarks. Sure it doesnt dominate games and office benchmarks, but it does in OGL, CAD, 3d rendering (at least over p3). When Athlon was initialy released it had a 15-30%+ advantage over katmai, clock for clock... accross the board... that is not "minimal" as you put it. Also the true test of the Athlon core will come when p4 arrives... can it compete w/ anothr 7th gen. cpu? I would argue yes, but p4 isnt out yet and we will have to wait and see. We do know that the athlon core will ramp quite nicely and may even top out at 2-3ghz once its all done w/. If that's not a measure of how much more advanced it is over a p3, i dont know what is. I also have to give credit where it is due, Intel's engineers are some smart mofo's... they've scaled this core since 150MHz and it still is competative... the thing is that you may have to give some credit where it is due... AMD did not dissapoint, hell look how its doing now, better than ever.

alin
 
dreamer

The power of the EV6 bus will become apparent when DDR SMP motherboards become available. You'll see Athlon servers and workstations that will quite literally destroy any Intel setup.
 
heh I don't post for a day, so now I have to address everyone starting from NFS4's comment "So you're saying that the athlon/Thunderbird/Duron is trash???"

Yes, considering how well they can make it right now. Look at it this way, it has 1/3 more potential processing power/clock then the Pentium 3, because it has 3 pipelines, and 128 K L1 cache, coupled with now 256K L2 cache with a bit of a bottleneck..

the L1 cache is fine for the 3 pipes, the L2 cache needs a bigger interface with the CPU, they need to get that core pushed to the max, like Intel has with their Pentium series (I highly doubt they are hiding anything in that CPU). If they went as far with the Athlon as Intel has done with that old core, the Athlon would live a longer life then it will in their roadmap (with the Sledgehammer coming up fast). I don't like how they are handling the Athlon, but also, I don't like how they are going to drop it and run with the Sledgehammer (with the K7 core probably in the low end like the K6-2 was until recently).

you understand my rant NFS4?

pm, "Bear in mind that I work in the IA64 group and so I know a fair amount on IA64 products, but very little about IA32."

Does this mean you can tell me anything about if Motherbaords will be different? I mean using IRQ's and the such?

"In fact, most of the time the Register seems to know more about our future products than I do" heh, those of you who think the Register sucks, think again! this guy's an insider!

Dreamer, "I am glad AMD made a better chip and finally improved their supply problems, but this talk about the Mustang being what the Athlon should have been is lame, the Athlon should have came in and just rocked."

Yes you are right. Then the marketing and stuff occurred, where AMD had to make sure that it would go on a Slot, and therefor thought it would be fine with the limited speed cache, becuase they doubted intel's ability to go .18 and increase the P3 line to over 600mhz.

The company decided to use this strategy; the Mustang will be what the Athlon should be, becuase we need all the money we can get.

So they MILKED it by releasing lower performance, but competitive versions of their Athlon.

The athlon still has potential, it's just that the COMPANY came before the product, and therefor the company survived with more money then if the product came before the company, which leads to better products down the road becuase of better R&D.

AMD is smart in that they are looking in the long run. even in the short run, they are doing mighty fine with these CPU's.

The mustang will be what the Athlon should have been, and will go up against Willamette, in mhz, and in pure speed per clock.

Bobberfett - "The power of the EV6 bus will become apparent when DDR SMP motherboards become available. You'll see Athlon servers and workstations that will quite literally destroy any Intel setup."

True, but the Willamette will have a Quad pumped FSB, and if Via doesn't do anything about their PCI bus at 133 vs. Intel's bus at 266, there will become some problems in speed. With the DDR chipset, Via will hopefully be using a new northbridge which might allow the use of a new Chipset Architecture. we will see.

to AMD's advantage, the EV-6 Bus can supposedly go up to 200mhz, which when using DDR is as fast as Intels planned FSB.
 
I am certainly not speaking of any Willamette, Mustange, Itanium, etc.... With the motherboards we have today, all the technology and architectural changes that the Athlon incorporated just didn't perform as many had thought. Many of you keep on exclaiming that it is hampered, held back, or whatever. This is what is available today: you can buy a PIII700E on a BX platform, and an Athlon on a KX platform, and you get pretty much the same performance. All the while, the Athlon has a next generation core with the EV6 bus. In no way am I questioning the success of AMD, or their motivations, tactics, lifestyle, and choice of breakfast sandwich. I am just saying, for a next fookin generation core, clock for fookin clock, it does not stomp a 6th generation core. You quote higher end benchmarks of course for some salvation, but should it not just handily beat it in everything? Yes it should, and no it does not with what there is today.

I really hope that the Athlon can scale up to more than 2Ghz. But that is more a testament of the improvements in semiconductor physics and process enhancements along with the architectural changes made or added. If Mhz scaling was the main focus of the 7th gen core, then the engineers did a great job.
 
Dreamer, the thing is, AMD built their CPU not for absolutely wasting the current CPU clock for clock, but just so they could eek out as much money as possible, by keeping just one step ahead of it's competitors, they are sitting back and relaxing on that part, waiting for Intel to make something faster, all the while preparing their next CPU, which will undoubtably attempt to do the same thing (the company comes before the product), if they can.

"I really hope that the Athlon can scale up to more than 2Ghz. But that is more a testament of the improvements in semiconductor physics and process enhancements along with the architectural changes made or added. If Mhz scaling was the main focus of the 7th gen core, then the engineers did a great job."

ACtually, AMD built the CPU to be a speed demon, in that it can be faster per mhz then the competing CPU. It definetly CAN be, but it's a question of them actually improving the core to what it is capable of (tweaking it to the max), as well as programs being designed with the 3 pipes in mind, rather then the dual pipe the Pentium 2, Cleron, and Pentium 3 have.

So, software has a part in the speed, but I no longer believe programmers and benchmark makers actually build CPU's to perform well on a certain CPU architecture (except when 3DNow! and SSE and SSE 2 come into play).
 
You don't seem to quite understand the EV-6 bus. When we see SMP Athlon setups, each CPU is going to have it's own bandwidth to the north bridge. Unless they're undergoing some radical changes with the next IA32 chip (Willamette) then it will continue the tradition of sharing the bandwidth between all CPUs. So:

Dual Athlon = 532 Mhz Aggregate FSB (2x 266 FSB)
Dual Pentium = 532 Mhz FSB (4x 133 FSB)

Quad Athlon = 1064 Mhz Aggregate FSB (4x 266 FSB)
Quad Pentium = 532 Mhz FSB (4x 133 FSB)
 
yeah Bobber I know that, SMP would be a cool thing with the EV-6. It soon will be available with the 770 chipset (end of this year I hope!).

The way you describe how the Athlon's will have an advantage is kindof confusing to me. Basically, each CPU will be connected to the northbridge of the chipset with it's own FSB pipe. each pipe will have it's own bandwidth available, so say at 133 DDR, each CPU will have 64bits times 266mhz (effective bandwidth) divided by 8 = 2 gig's per pipe.

With Intels current setup for dual processors, both CPU's use the same bus, which still runs at a pidly rate of 100-133mhz.

HOWEVER, Bobber, I'm sure intel will address that issue of each CPU needing more bandwidth eventually, don't sell em short, they already changed their chipset design (i820 etc) so that the bus between the North and Southbridge is capable of 266megs/second instead of the crappy 133 that it used to have (and that the Athlons currently have, because of the Via chipset).

Even if they don't change that design, remember it's still QUAD pumped, so each CPU would get half that bandwidth. meaning each CPU would get the same bandwidth as a double pumped fsb..
 
Soccerman

Short and sweet: Athlon/Thunderbird is MUCH cheaper than an equally clocked PIII. You can actually get Athlon's over 850MHz. 1GHz Athlons can actually be gotten by regular end-users. Athlon's are mirror images of PIII's in performance.

That's all I need and that's all that counts. Who cares if it's a 6th, 5th, or 7th generation?? It's here now and that's what matters most.

Athlon/Spitfire takes the cake, and that's final. The Duron just takes over the whole world 😉
 
1.13 GHZ ??

Obviously Intel is grasping at straws here if they're talking in terms of a third of a HZ. (3)

What's next, the Pentium IV 1.537346 GHZ?

Oh brother..
 
I just was to make a small point. There would be no reason for AMD to release a CPU that could kill everything on the market. Why? Because we, the enthusiasts, are the only people that would know what killed what. The person in Compusa buying a compuke would still think that Intel was the best, because Intel has the better marketing department.
Intel could (back in the day) release something that killed everything, because they had the marketing power. So the idiot in Compusa would still buy Intel and the enthusiasts would also buy Intel (which they wouldn't if the K6 killed it).

The P6 core is at it's max, while the K7 core is at its minimum.

One more thing, this is what I've heard & read and I don't know if it is true. The Willy is going to have very poor Clock for Clock speed, I've heard that the Coppermine will be able to kill it at clock for clock. Intel is aiming it for the stupid person in Compusa that thinks MHz is all that matters.

Okay I am done.😛
 
I'm not saying that it's bad in the price/performance ratio, I'm just saying, it's potential hasn't been noticed.

It's the killer CPU for now, and will be for a while. not arguing that.

AMD's just sitting back watching Intel struggle to reach the ballpark at which the Athlon talks about in terms of mhz, AND mhz-mhz power, because of that the Athlon being a 7th gen CPU doesn't seem to make a difference. AMD hasn't really done anything that big to the core, besides adding cache and changing to .18 copper. ok, that DID speed up MHZ, but I'm talking mhz-mhz speed.

Zorba, yeah the Willy will suck at first, then they will try to force everyone to use their new improved SSE instructions. and yes it's aimed at the basic computer user who says.. oooohh 1.5 ghz! OOOH even better INTEL!

hehe.. anyway.... it's good someone has finally sat up and noticed my point NFS4.
 
It's getting close, do you think that Intel will pull through, or will AMD snub them with a 1.2GHz Thunderbird on the 24th? 😀
 
I think that intel will hire GUTB to go to the AMD Dresden fab and piss all over their wafers this week! Then, they will release a 1.001GHz CPU and claim that they own speed crown again!

Just my $0.02.
 
woohoo, paper launch 😛

and you know what other paper related issues that bring, that is beneficial to "us."

does it really matter anyway... consumers are stupid and will buy what salesmen tell them. we don't. 😛
 


<< Why does it matter who fabs the fastest CPU if no one can buy them anywhere? >>


AMD makes their chips available at launch (or maybe a few days afterwards). Intel takes months after a &quot;paper launch&quot; to get the chip in consumers' hands.
 
If it is true that Intel has altered the die mask for the 1.13Ghz chip and shrunk the die, they may stand a chance of shipping them in volume. I am doubtful.

The Athlon core was designed to beat out the Katamami core. TBird was squared off against Coppermine. Mustang will compete with PIV Willamette.

I believe you will find that in each and every instance, the AMD chip will best the Intel one clock for clock.

As for a &quot;Next Generation&quot; chip....well that is a grey area. We will never again see the jumps in processor generations like the days of the 286 to 386. Even though AMD could have released an Athlon 800 when Intel had only 600Mhz, it is not a good way to make money by showing all your cards in one round.

3 piplines vs. 2 does not mean 1/3 faster. You have to have a program that continually uses the 3rd pipline efficently to see this.....and that doesn't happen.
 
Intel needs to learn simple physics, paper has a greater surface area than an fcpga chip. Therefore, when paper is launched it doesn't fly as well as the chip!

BTW, yeah the 1.13GHz (really 1.133GHz...aka 1133MHz) P3 is using the cC0 stepping, which has a 5% core shrink over the cB0. Though, I highly doubt that it will launch well. I mean, intel can't eve get out 1GHz cB0, maybe cC0 at 1GHz would make the most sense...heh, seems like very little about intel is making sense lately. 🙂
 
zippy,
The REAL reason that Intel has been having yield problems with their high end processors is that they have secretly been using the top secret material &quot;Unobtanium&quot; in these processors. This rare material makes any processor very fast. The speed increase can be seen for many months from the time of announcement to the actual availability of the part. Once in actual use, the Unobtanium is removed from the die and ordinary silicon is used in order to keep this material from falling into unfriendly hands. MY GOD MAN, THE FUTURE OF THE FREE WORLD WOULD BE IN DANGER SHOULD INTEL RELEASE THIS MATERIAL AT THE TIME OF THE ANNOUNCEMENT!
 
Isn't the via apollo 133a and the kt133 fairly simular as in memory performance? If so if you compare a PIII on a via board vs a k7 on a via board you will see how much better the k7 is.
 
Dreamer, I won't argue with you on the Athlon being a 6th or 7th gen core. It's a 7th gen core in that it's definately a big jump over the K6. Whether it's a 7th gen compared to intel...well you decide.

I'd just like to comment on the DEC EV6 bus protocol. You don't seem to understand the bus properly. I've worked with Alpha workstations so hopefully I can shed a little light on the differences between it and GTL+ (this is not pro-AMD or anti-Intel, I'm just trying to describe the bus).

Before, I start one thing: This EV6 bus is WAY OVER HYPED for the Athlon. (ok back to facts)

It is a 100Mhz DDR Bus. Right now it is being paired with 100/133Mhz SDRam so you have no memory bandwidth advantage over the GTL+. There is very little the extra 66 or 100Mhz can do right now. When DDR SDram comes out it supposedly will have room to use the full speed of the EV6 bus.

Also in SMP systems, the single 66/100/133Mhz GTL+ bus is shared between all processors and the north bridge. This means the more CPUs you have the more bus contention you have. The EV6 bus is a point to point bus that has a dedicated 200Mhz (or whatever speed you have) bus between each processor and the north bridge, so less contention.

Also the EV6 can supposedly scale to higher speeds than the GTL+.

(ok, now this part is my opinion)

Now to be fair, in desktop PCs none of this matters much. In the SMP workstations it will help out some. But in PCs it's not really needed. Also with intel's new 100Mhz QDR bus coming, if they design it well AMD could lose that advantage. And even with this fancy bus technology, a good motherboard like a BX is much better, more mature, more stable, better performing. How long have Via (and even AMD) worked with the EV6 bus compared to the GTL+? If DEC (compaq) made a motherboard for the Athlon it would probably kick a$$, but all the good companies like Asus and Gigabyte and MSi have never worked with these kind of boards before, (besides via's chipsets suck s--t, have you seen an Apollo Pro 133A compared to a BX lately? *ugh*).

*shrugs* The athlon has a more powerful floating point unit. The P3 (cumine) has a better L2 cache design. Sure the bus design for the Athlon is &quot;potentially&quot; better, but in read world analysis for a single CPU it doesn't help much.

I dunno, they are pretty close. The only thing that made me choose the Athlon is the price/performance difference, I would give the nod to intel for performance at a given Mhz number in the stuff I do, but for a given amount of $$$ the Athlon (Tbird) is going to kill the P3.


 
You must log in or register to reply here.

TRENDING THREADS

Back
Top