LinkBack URL
About LinkBacksWe have the Fury Nano soon, so less HBM for the Fury X as well
We have the Fury Nano soon, so less HBM for the Fury X as well
Anand said: (http://www.anandtech.com/show/355/24)Originally Posted by Corky34
Dr Tom (http://www.tomshardware.com/reviews/...essor,121.html) said things likeFinal Words
I don't have to give you a conclusion here, the benchmarks speak for themselves, the Athlon is the fastest desktop x86 processor on the market. It would take at least a 700MHz Pentium III to start to beat AMD's Athlon, but even if Intel does release a Pentium III 700 ahead of schedule, the price of a lower clocked but faster Athlon will still be lower than a 700MHz Pentium III. There is nothing in Intel's current family of processors that can match the performance of the Athlon, even the forthcoming Coppermine will have trouble competing against AMD's latest flagship.People talked about the team that designed this CPU, like the ex DEC employee that lead the the FPU design who managed to make it nearly as fast as the one in the DEC Alpha.I will try and make a comparison between the architecture of Athlon and Pentium III as far as it's possible, so that we can see why Athlon beats Pentium III in pretty much any benchmark.
Overall Intel has always had very mixed bag of product quality. Have you ever used an Itanium? Or one of the original Atom chips? I have a D520 somewhere in a drawer, it really is quite a dog of a chip. How about the original Pentium 60, claimed by Intel before release to be the end of all RISC chips such was its mighty performance, but in fact it was late, hot, missed the clock targets and compared to the Fairchild Clipper and NS32K based machines I was using at the time really sluggish. And don't forget the i432, generally heralded as the worst cpu architecture of all time. Mix this in with their success stories like the 8052, the i960 and the one that their modern desktop chips can trace their lineage to the Pentium Pro.
Oddly, ISTR when the Pentium Pro came out it got quite a critical pasting. That seems rather unfair, if anything made by Intel was their Mona Lisa I would say the 'Pro was it. The i7 is this chip polished so hard you can't see much original architecture any more, but the lineage is there. The P3 was this chip cost reduced and SSE added.
Are you even reading some of what is written?
You have linked to an article whose title is "The New Athlon Processor: AMD Is FINALLY Overtaking Intel"... (my emphasis)
And again you are comparing a single CPU made by AMD that the article you linked to says.."The instruction as well as the data caches are with 64 kB no less than 4 times as big as Pentium III" and then using that to validate that ALL Athlon's beat ALL PIII.
I'll tell you what instead of arguing about if the PIII or Athlon were the better CPU why don't you take up this up with the Wiki article that says...
In terms of overall performance, the Coppermine held a slight advantage over the AMD Athlons it was released against, which was reversed when AMD applied their own die shrink and added an on-die L2 cache to the Athlon. Athlon held the advantage in floating-point intensive code, while the Coppermine could perform better when SSE optimizations were used, but in practical terms there was little difference in how the two chips performed, clock-for-clock.
You're not seriously using wikipedia as a source are you?
When it comes to comparing the overall performance differences between two entire lines what else would you suggest?
DanceswithUnix argues that the Athlon Processor was successfully before Intel screwed up with the P4, if that's the case then why is socket 939 the one that everyone remembers and not Slot A? The fact is the only reason everyone remembers socket 939 and the Athlon days is because Intel's offering during that time sucked.
Anything else. Wikipedia is not a source.
He's right - AMD's 3dNow! instruction set was good and helped the K6-2 become a serious competitor long before S939.DanceswithUnix argues that the Athlon Processor was successfully before Intel screwed up with the P4, if that's the case then why is socket 939 the one that everyone remembers and not Slot A? The fact is the only reason everyone remembers socket 939 and the Athlon days is because Intel's offering during that time sucked.
S939 was great not because of p4, but because it was a genuinely good chip - onboard mem controller, cool n quiet, and decent arch from server tech. Dual core versions even better.
Intel's Core 2 was also great, but that was a response to AMD, not in spite of it.
Please don't mention the amd k6-s or k6-3.
Fab chips and better than intel's offerings but let down by some of the worst motherboards in history.
Like buying a Ferrari and using a block of cheese for tyres.
Like I said what else would YOU suggest.
Like I said, anything else. Even personal anecdotes over wiki because at least then you can put a (psuedo)name to the comment.
Then please provide it, if anything else would be better than a wiki article on comparing the overall performance differences between two entire lines of CPUs then please provide something that's more substantial, or are you saying that, in fact the P4 was better than the PIII, because that's not what you seem to implying here.
Smaller L1 cache was Intel's design choice, you can't blame AMD for that.
As for Coppermine, yes when it was released it was faster than the Slot A processors it was competing against. That is a process advantage, nothing to do with design talent. When the Athlon went on die cache shortly after, it was faster, because the cores were better.
As for Wikipedia, if you referred the Athlon page rather than the P3 page you would see a very different story, because Wikipedia is like that. That's why I referred to Dr Tom and Anand from articles at the time.
And what is it with this "comparing a single CPU" malarky? Slot A vs Slot 1 was a win for AMD across the board. Coppermine vs Thunderbird was a win for AMD across the board. Only if you compare Coppermine to Slot A do Intel come out with a win, which is what the Wikipedia article is doing. Well it is a win, if you were buying during those months it counted, but comparing a 250nm to a 180nm device isn't a good indicator of architecture is it.
You have to remember the history here: the DEC Alpha team at Compaq had just been handed over to Intel as part of the Compaq/HP merger thanks to HP not wanting to take on a CPU design team (possibly due to a contractual issue thanks to the Itanium we will probably never know). It seems that the design team of what is probably still the best CPU architecture ever didn't want to work for Intel and a bunch of them went to AMD. That was a big injection of talent into AMD, who were already doing some good things. At this point AMD had the architects of the 29K RISC chip who had gone on to design the K5 (AMD's first out of order chip, beating Intel by years) as well as the acquired NexGen team that had designed the K6, and now some choice DEC Alpha engineers. Really, if AMD didn't come out with something special at that point then something was seriously wrong. Specially as they picked up floating point talent from DEC, as that is what the K5 and K6 had been really missing on for gaming use.
You should be able to find my entire PC CPU history + drop in upgrades here somewhere*, but I lack the ability to compare competing products exhaustively, but from my impressions comparing research at the time and friends and families systems, the PIII was great, certainly. The p4 was not anything like as good clock for clock, but later in the series it was certainly better than what was available years ago in P3. However the AMD chips by this time were better than both.
*It goes something like: Intel286+287, Intel486dx33->dx66?, AMD K6 200?, IntelP3 800/133->1ghz, AMD X2 3600?->4400+?, Intel i7 950.
I remember being very impressed by the K6-2 systems, but couldn't afford one at the time (everything up to then was second hand - the P3 800 was my first new computer).
Average age of your sample group?
I've built one s939 PC in my life. I've lost count of the number of socket A builds I did. Back in my day it was a no brainer: you built Socket A. Even before Netburst, Athlon was competitive with or better than the Northwood P4s. Athlon 64 on s939 was simply a further refinement of the already-better tech in Athlon XP. And at the low end of the market Duron ate Celeron for breakfast, even before AMD rebranded the lower end Athlon XPs to form the Sempron line. AMD was climbing all over Intel long before s939 made netburst look silly. s939 is not the one that everyone remembers.
OK, so let me get this right, you're saying that in fact the P4 and NetBurst wasn't a micro-architecturally dead end, that even before the P4 and NetBurst, that AMD ruled the roost because their 7th generation x86 CPU microarchitecture was so good that it beat all 6th generation x86 CPU on a clock by clock basis that Intel had at the time, that we don't in fact think so fondly of the AMD Athlon era because Intel's 7th generation x86 CPU microarchitecture was a stinker, that people in fact look back at AMD's 7th generation x86 CPU because they were good enough to stand on their own.
Maybe someone need to do a re-write of the history books and tell everyone that in fact the P4 and NetBurst wasn't a micro-architecturally dead end.
I don't think anyone's saying netburst wasn't a mess - they're saying every generation of Athlon was competitive with the contemporary Intel processors. You've even been shown graphs clearly showing Athlon @ 500MHz hammering PIII @ 500 MHz, which is going back WAY before Netburst. So I'm not quite sure what your point is, to be honest: there's clear evidence in this thread that Athlons were competing with, and often beating, contemporary Intel processors for years before netburst, yet you appear to have some hang up that Athlons only looked good because netburst was so bad. That's patently not true.
?? Are you replying to the right thread? Sounds like you're talking about another discussion somewhere, which might be adding to the confusion!
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
Reply With Quote