Why is Intel out doing AMD
Why does intel have cpu up to 1.7 ghz, and amd is laging behind with a 1.4. Anyone have any news on the schedule for amd chip speeds, or do i need to buy intel to stay on the cutting edge.
you know... it's this attitude that just reminds me that most people in the world are idiots
Intel's trying the same thing Apple is quite comfortable with... the mhz myth...
the # and efficiency of pipeline stages is just as important (and sometimes more) than sheer mhz...
read this: http://www.amazon.com/exec/obidos/AS...216084-4320003
Thanks... i guess. Just wanted some info.. So i like cutting edge stuff. Thanks for the link to buy a book. Sorry not for me! I guess what i am getting at is if we do not have people pushing the envelope, then we will still be using commador 64! Don't get me wrong c-64 was fun but not that good.
We need to keep having things evolve.
To answer your question is a complicated one! AMD is ramping up their CPU's nearly as fast as intel. The difference is the way they achieve their goal. AMD and Intel both have areas that their CPU's perform better than the other. Currently an Athlon 1.4 blows the doors off a 2.0 P4 and scores nearly the same as a 2.2. Here is a link to a great comparson by the fine folks over at [H]ARD OCP where they compare a 1.4 T-Bird vs. a 2.0 and 2.2 P4. Read that and it should answer any questions you have...steve http://hardocp.com/reviews/cpus/intel/p42ghz/
Oh and Intel is officially at 2ghz now, not 1.7ghz. The 2ghz has a much lower IPC(instructions per clockcycle) so it HAS to run at a higher speed. Intel P4 has to have optimisation to preform. An Athlon runs everything well and just get's better with optimisation. Only now with 2ghz is Intel actually beating the 1.4ghz T-bird in about half the popular benchmarks. Not to mention the tremendous price premium you have to pay for a P4 system. I850 + RDRAM + the P4 itself is quite an expensive package.
true, so true, hellmund...
gotta just love it how one company charges x5 for the same performance, yet still manages to make headway in the workstation market thanks to the oblivious nature of the normal consumer...
if people only took 5 minutes to investigate what they were about to purchase instead of taking the word of some idiot at staples, circuit city, best buy, etc... (no offense if anyone works there... a good buddy of mine works at CC, but he also knows where to get his procs....)
it's like Dell... i swear, nothing pisses me off than a company that is sworn to uphold a less effective price point.. what a "wonderful way to run a business"... jeeez.. what clods...
i'm glad we in the enthusiast crowd have a bit of reasoning going on
Nice handle, Just Having Fun :-).
Remember guys and gals, if it weren't for Apple we would still be typing c:/ to get anything started and if it weren't for AMD we would be squeezing 300 out of a 266 at most. And if it weren't for Intel sharing their goodies we would be paying 4k for that!
It's called competition, folks, and that, and opposable thumbs, have got humanity where it is today.
Maybe Paul, but I think we would still have a GUI, just not as advanced and as easy to use. Actaully I think we'd still be at around 800mhz, intel would be keeping ahead of cyrix's Samuel III still. Of course we certainly wouldn't have P4's around. Intel would probaly just be sitting at 1ghz with the PIII and be charging a HELL of a lot for it. We'd also have only RDRAM.........with I820 ('shudder').
Well .. the reason Intel is pushing so hard to bring out those 298Ghz cpus because they know what AMD cpus can do.
Look at it this way, the main reason the Athlon 1.4Ghz outperforms the P4 1.8Ghz cpu is because intel was rushing as fast as they can to finish the P4 and get them on the shelf since their P3 sucked compared to Duron and Athlon cpus. The price they paid for brining out so fast is performance, they had to drop off A LOT of features, that if were kept then it would kill the Athlon cpu.
Also right now Intel is going up in hertz so fast because they want people who know nothing about pcs to buy them. If you knew nothing about pcs, and i offered you a Athlon 1.4Ghz cpu (From AMD who you have never heard of.. btw has anybody here seen comercials??) or a P4 1.6Ghz (From Intel who you have heard of sicne they rerun their comercials ever few minutes.). You would choose the 1.6ghz right? Of course the people who know even the slightest about athlon processers will of course choose the athlon anyday, unless of course they wanna brag and say "haha i have a p4 and u dont haha u suxxorz".
And also from the benchmarks at anandtech.com, it looks like the Athlon cpus' still stomp the 2Ghz P4 cpu by a score of 8-5.
And imo.. Intel has gotta drop those P4 prices by a lot if they hope to sell them to people who cant dish out $600 for a cpu.
Athlon 1.4Ghz = $130, and performs better than 1.8ghz, and just as well as 2ghz.
P4 2Ghz = $600.. and is really slacking off in performance.
You also gotta consider that the reason AMD is taking so long because they are probably trying to actually make the cpu better overall, rather than just incrase the clock speed.
lol... so basically, this has turned into a circle-jerk of those of us "in the know".... because we all know that the athlon is better and cheaper
oh well, persecution of the better, yet less "tied in" product is the norm in our society
this is definately one of those times that if i were rich i would run an AMD Athlon4 commercial (because AMD seems to not give a flying f*ck about their adverts) so that at least some people knew
sorry for the pointless rant :-/
Run a real-time non-linear video editor on a P4 with Rambus, then on an Athlon with DDR. You'll spend the extra money for Intel's 1500+ MB/s of PC-800 memory bandwidth.
Otherwise, for most day-to-day tasks, AMD does have a superior product. I am using a 1.2 GHz Thunderbird in my primary box.
hate to tell you, but DDR / AMD systems have gotten over 1500mb/sec mem bandwidth (beathen p4/rdram sys in sandra)... it's not quite the gap it's cracked up to be
if anyone to the time to tweak their systems, maybe this would be commonplace, but the age of complacent overclocking has struck us...
Rambus Bandwidth Myths 101:
(I'm back in school, lame pun...hehehehe)
The P-4 uses a 128-byte sectored cache (across 8k, that is 64 sectors), leading to (obviously) 128-byte external bursts. Occasionally 64-byte bursts are seen for write-backs, code-fetches, etc (by the way, this leads to about 20% of the L1 cache being unused, empty, wasted, but that's a delve into microarchitectural theory we'll save for another day, okay?).
The PIII uses 32-byte cache line (not sectored), all (ALL) external accesses are 32-byte (allowing excellent L1 cache hit rates). When a burst from main memory comes in, the information demanded by the processor is put in first, then the neighboring data is brough in (up to the cache line size for that burst). As you can easily predict, the 4-times-larger cache lines of the P4 artificially inflate the Bandwidth to, you guessed it, 4-times the amount (100MB/s PIII vs. 400MB/s P4).
In the end, this results in a latency delta of 19% over the P4's own predecessor, as well as a 33% increase in burst time.
SPECint Vortex Benchmark scores show the P4@1.5GHz running only 3.5% faster than a PIII@1GHz, and the Athlon T-Bird@1.33GHz with DDR running 13.7% Higher than the PIII (scores are 710, 735, and 807 - respectively).
worth twice as much money to you? and when my 1.2GHz AXIA gets paired up with nForce's dual memory controllers, it's going to 1.5GHz (150x10, 300MHz DDRx2!!!!)
I would like to see those benchmarks, and I am not referring to results from a nForce 128-bit dual channel DDR board that isn't even available outside of nVidia R&D. Then again, I have internal specs for that one as well, and they are definitely not near 1.5 GB/s.
The SIS735 chipset is the lowest latency, highest bandwidth EV6 bus compatbile DDR solution available for the K7 architecture. Looking at around 1.0 GB/s with a Athlon MP at 1.5 GHz here, so this doesn't add up here either.
The is only one current Athlon chipset capable of higher memory bandwith than the P4 RAMBUS architecture, and it will never be released for public usage. Micron's MAMBA architecture uses a traditional EV6 design, but a chipset level L3 cache is added to offset memory latency. The internal numbers look impressive, especially for branch-predictive tasks. However, Micron can not release the chipset due to high production costs and potential liscencing issues.
I do hope you're not talking about nVidia's attempt at an integrated low-cost product. The 128-bit nF chipset can not produce significant performance gains in terms of memory bandwidth or latency. The Athlon only allows for a 64-bit memory bus which limits bandwidth without effective ramps in processor to chipset to memory bus MHz speeds. Forget trying to improve latency without a chipset level cache mechanism (nForce does not have!), as the Athlon has relatively poor scheduling efficiency due to its severly limited g-share branch predicition routine combined with absolutely no capabilities for L2 cache instruction re-sequencing without completely stalling and flushing the execution pipeline. Definitely not optimal for ideal memory op efficiency.
While I'm on Athlon chipsets, here is one tidbit that you'll never see in a nVidia press release: The nForce audio controller doesn't even work correctly. And I'm not talking about driver probs either......
While discussing benchmarks, new Spec date was published earlier this week:
P4 2.0 GHz: 640 peak
Athlon 1.4 GHz: 554 peak
AthlonMP 1.2 GHz: 522 peak
P4 2.0 GHz: 704 peak
Athlon 1.4 GHz: 458 peak
AthlonMP 1.2 GHz: 481 peak
People are starting to realize the potential of Pentium IV optimizations. The addition of code vectorization for SSE2 with additional hardware code hint conventions had afforded significant gains. Also, programmers seem to have finally learned how to properly code for datagram alignment. Many write of the P4 FPU as being weak, but this is not the reality. Programs featuring poor data alignment in terms of cache operations will suffer with the P4. The above SPEC results were obtained through optimizated alignment ops. Notice the serious increase in performance.
Why is it hard to believe that the P4 platform is a superior architectural design in certain situations as compared to current Athlon offerings? Intel can't help that none seem to ever bother to read the P4 programmers guide, even though the documentation is freely available. The tricks that were possible with the Athlon or P6 core (PPro through P3) are no longer viable for the P4, but programmers seem to still hold on to these legacy code conventions for some reason.
[This message has been edited by RobRich (edited 08-28-2001).]