Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
×
Intel

Intel Releases Xeon, Look At Those Kernels Compile 82

Posted by Hemos from the faster-stimpy-faster dept.
Nelly Furtado writes "AnandTech has the scoop on Intel's new Xeon processor that was just released today in dual processor form. The review includes Linux kernel compilation tests as well as database server performance measurements. The article also hints at Jackson Technology (SMT) and AMD's 760MP chipset."
This discussion has been archived. No new comments can be posted.

Intel Releases Xeon, Look At Those Kernels Compile More

Intel Releases Xeon, Look At Those Kernels Compile

Comments Filter:

  • You people don't understand (Score:1)

    by Anonymous Coward
    that Looge is way faster than any xeon or athlon chip. ever.

  • Re:Return of the i860 (Score:1)

    by Anonymous Coward
    Hehe. I think I still have that Hauppauge i486-33 + i860-33 motherboard somewhere. Only think I ever was able to do with that expensive i860 CPU was sample program supplied with motherboard that blinked * at top-tight corner of screen. It was really cool 'cause as long as you didn't press reset or hit powerswitch it kept blinking. Thru BIOS, DOS loading, all programs and so on. :) Actually there was also special display adapter that was supposed to use that i860 but never got any image thru that card.

  • Re:Return of the i860 (Score:1)

    by Anonymous Coward
    There is a parallel super computer based on i860 64-bit uP: the Intel-built Los Alamos Paragon XP/S. Each 50 MHz i860 node contains 16-128 MB RAM. Its no longer in operation, was sold as scrap and than bought back because is "secret". If 50 MHz sound low speed, remenber that in 1989 the i486 is a 25 MHz device...

  • Anand? Server platform? (Score:1)

    by Anonymous Coward
    I must not be reading properly, but the new Xeon DP is slated as a workstation platform, or so says Intel, and supposed to take on the almighty Sun and SGI. I'm not sure where Anand got his info from, but all the Intel sheets and presentations focus solely on the workstation segment!

    The Xeon MP will be released later this year and IS targeted towards the server market. So that'll be the day when we want server benchmarks and other stuff like Linux kernel compilations.

    I found another techy website, www.hardwareanalysis.com [hardwareanalysis.com] that does get into the whole workstation benchmarking and environment. They even have a similarly configged dual Pentium III Xeon system they compare it to, good read!

  • make -j3 'MAKE=make -j3' bzImage (Score:4)

    by Anonymous Coward on Monday May 21, 2001 @06:34AM (#208670)
    This is the proper way to make a Linux kernel on a dual processor box.

    Quoting Linus: "A single "-j3" won't do much. It will only build three directories at a time, and you'll never see much load. But doing it recursively means that you'll build three at a time all the way out to the leaf directories."

    The testers used make -j2 bzImage, which didn't make full use of the machine's capabilities.

  • Um... office benchmarks? (Score:3)

    by Wakko Warner ( 324 ) on Monday May 21, 2001 @08:05AM (#208671) Homepage Journal
    Why does Anand include tests like "office application performance" for a dual pentium 4 Xeon???

    Am I the only one who thought those particular benchmarks extremely worthless? Where's the "10,000 client mail-server" benchmarks, or some other thing people would actually use a dual pentium 4 Xeon system for?

    - A.P.

    --
    Forget Napster. Why not really break the law?

  • Well said. Server-chip need server benchmarks: Database, web server, J2EE stuff. In most cases those boxes won't even have a video card...
    Even the Mindcraft benchmark would be a good test.
    --
    The world is divided in two categories:
    those with a loaded gun and those who dig. You dig.
  • So it doesn't offer much improvement over the PIV eh. That's nothing 5000 layoffs can't fix. It certainly looks exotic, like the ISA expansion panels of the late 80's. Either way I expect they'll have to pull a Cisco to return to profitability. Preferrably on the campus by 101. And keep the suicides off the freeway.
  • What quantum leaps are you talking about, though?

    DVD decoding in software? A DVD FAQ [dvddemystified.com] suggests a 400Mhz PIII if you don't use any hardware acceleration.
    Adamation's personalStudio at least _alleges_ a 600 Mhz machine [adamation.com] as a minimum for real-time video effects previews, etc.
    Last year, someone at The Perl Journal wrote about capturing video and comparing individual frames; IIRC, using a specific optimized module, he captured video without dropping frames with a dual 500 Mhz machine... The Perl Journal site is down (pending transfer of ownership, probably) and I don't have that issue handy.

    I wouldn't be surprised if there are killer apps - I mean, I don't know where to look for info on MPEG4/Divx :), but, well, what kind of application are you thinking about?

  • Re:Athlon (Score:1)

    by gid ( 5195 )
    Anandtech has an intel bias, since when? You do know that they run their site on mostly Athlon based servers [anandtech.com] now, don't you?

    ---
  • dual-processor server platforms based on the Intel Xeon processor to be available in the second half of 2001 [yahoo.com]

    In other words, the official announcement of dual Athlon availability is imminent, and Intel is making a pathetic attempt to steal some of AMDs thunder. Amusingly, today the price-per-share gap between Intel and AMD grew to 4-1/2 points.

  • Memory is cheap now I think most people would be better served maxing out their memory rather than seeking faster CPU's

    I beg to differ -- at least once you get into the 256MB territory. I have 128MB in my machine and my fiancee has 256MB in hers. We spend very little time actually hitting swap. (I spend slightly more, since I have less RAM and I master CD-Rs. I also don't close Moz very often.) But still, I don't spend much time at all in disk-wait even if there are pages on disk, since what gets paged out is idle anyway. And on my fiancee's machine, lessee.... 8K swap used out of 512MB... :-)

    Basically, once you get to 256MB or above, unless you're doing some crazy video and image editing, you quickly get into the domain of diminishing returns. Once your working set fits in RAM, you're CPU bottlenecked again.

    Now on a different topic of when to upgrade -- I seem to wait for a factor-of-five to factor-of-ten difference myself for major upgrades. I went from a TI-99/4A to a 4.77MHz Tandy 1000 to a 386SX25. During college, I had a bunch of minor mobo upgrades spanning a 486DX33 through a 5x86-133, but that's not my preferred way of doing things anymore. I sat with the 5x86-133 for awhile, jumped to a Pentium II-300MHz, and there I stayed until now. My next system will likely be a dual 1.5GHz Athlon early next year (factor of ten). My present strategy is to buy a computer to keep for ~4 years between upgrades, so I have bragging rights at both ends of the spectrum (start with "ha ha, my computer kicks your computer's ass" and end with "ha ha, I don't run bloated software that needs all those MIPS and all that RAM...."). :-)

    --Joe
    --
  • The original poster was talking about CGI performance under Apache, which in some cases is actually CPU-bound. This is not that uncommon when doing some types of intensive processing in a scripting language.

    Admittedly, finding a good test case of a benchmark is not easy. Serving static files is largely irrelevant, but easy to test.

    -
  • Don't be fooled, it is not double-buffering that uses more resources but that distracting transparency.

    Yeah, transparency takes resource but so does double buffering. The difference is that double buffering reduces flicker and headaches while transparency is just eye candy.

    Btw, compiling a single file java app with Project Builder is slower than compiling a 10Klines java project with Microsoft Visual J++ on a Pentium 120...

    That's because you're using two diffefent compilers. One's written in Java and one in C. Try using Jikes [ibm.com] (open source java compiler from IBM) for speed.

  • I would like to see 2 (or 3) new benchmarks added:

    Apache repsonse time. Have some scripts and or servlets which are more processor intensive than bandwidth intensive. They do exists.

    MySQL (or appropriate) - Are the processors available so fast as compared to disk I/O that we could see improvements over different chips in equivalent systems?

    A combination of the first 2


  • That makes it highly likely that a dual Athlon will significantly outperform the dual Xeon, does it not?

    Well, according to www.spec.org the fastest processor at the market is the Alpha 21264 at 833 Mhz..
    It runs rounds around any PIII or Athlon on the market today.
    Second, there are already logicboard with an AMD 760 and a 21264 processor.
    Yes, it uses DDram without any problems.
    Guess what happens if you have logicboards which can support two 21264 processors and DDram..
    Who needs an Intanium anymore?
    And this processor is available for at least a year now.
    And 64 bits.
    I like to see a kernelcompile on a dual 21264 machine.....
  • Would I rather run Blender, 3DSMax 4, Adobe Premiere 6, and Adobe After Effects 5 on a P90 or a 1.7GHz Pentium 4 (or a Gig+ Athlon)?

    Hmmm, let me thing about that for a moment...

    Don't discount multimedia work as "not for your average user", either. With most common desktop OSes (Mac OS9/OSX and Windows ME) shipping with video production software and photo-editing tools graphic manipulation speed is squarely in the realm of the average user's needs now.

  • It's only bloat when you can't turn it off ;-)
  • ... wasn't made by the engineering team, but by their marketing department. They managed to convince the world that the following foolishness was actually the truth:

    CPU clock speed == CPU speed

    And, judging simply by the title of the article, they are now attempting to convince the journalists of the world that:

    CPU clock speed == Workstation Power

    Hmm... I wonder if there's a better CPU for workstations and servers out there.... Moreover, I wonder if anyone whose chip design knowledge goes beyond the buzzwords has given any awards for the best Workstation/Server CPU architecture [sun.com]. Are we ever going to get back to reality in terms of computer processing power? Or, have the marketing people succeeded in snuffing out another truth?

    --Mid

  • ISTR that recent make version should do the right thing with 'make -j3' as well, but now that I search for it I can't find it anywhere. Judging from the >40% performance boost, they pretty much maxed the dual boxes as well.

    However, I'd be more conserned with the 2.4.4ac9/2.4.4 difference in the benchmark. I think the Athlon benchmark should've been rerun on 2.4.4ac9. The -ac kernels (and recent -2.4.5pre's I think) incorporate such optimizations and VM changes that I think the benchmark numbers can vary just because of that. Particularly on these kind of SMP tests that stress the scheduler, VM, fs and cache system -- and not just the CPU. Of course, there might not be a significant difference -- propably isn't -- but now we'll never know.
  • I have two and a half words for you: Gulp N Blow

    No wait, one word and a number. Doom 3. That's why we need every teency incremental increase in processor speed we can get. Humph! Kernel compiles! Who needs 'em?


    --Gfunk
  • a Sparc microprocessor might be considered a pure RISC chip but the UltraSparc II isn't and the UltraSparc III most especially isn't. As a matter of fact, the underlying architecture bears a lot of resemblance to x86 implementations. Go read this article [arstechnica.com] for an explanation of the fact.
  • You dont upgrade when theres a 10% or 50% speed increase unless you love wasting money on nothing. You upgrade when you cant stand your old machine anymore and the new one you buy will be at least 4 times faster. Seems to be around a 3 year cycle.

    My 300MHz pII is still working fine, but its slowly reaching the point where a new one would be useful. But it's not like I care if the new one would be 4.1 or 4.2 times faster.

    Then again, I work with bleeding edge million dollar crap machines every day and Im not very easily impressed by gee-whiz-gotta-have-that hardware anymore.

    But of course, what really drives the speed market is the unbelievable crap that the a lot of so-called 'programmers' in the commercial software buisness of today deliver.
  • If you read his CPU reviews he seems pretty consistent dismissing AMD's benefits and talking up Intel, even in cases such as Duron vs Celeron where it's totally one sided. At least this was try up to about a year ago when I stopped following things so closely.
  • Remember the source of this "information" - Anandtech. *ahem* Intel bias *ahem*

    Of course seperate northbridge - CPU buses can't hide the fact that there's only limited bandwidth to any given resource, but it's unlikely that multiple CPUs are contending for the same resource at any time.

    The only trouble I have with thr 760 MP (same as for any GHz SMP system) is that the power requirements are so high that you need an expensive power supply too.

  • Re:Time for more benchmarks? (Score:4)

    by SpinyNorman ( 33776 ) on Monday May 21, 2001 @07:10AM (#208691)
    Apache benchmarks don't really make much sense - a fairly low end processor can easily saturate a high bandwidth pipe. The Microsoft vs Linux web server benchmark wars are just a marketeering pissing context - they have essentially zero real-world relevance.
  • Some of the benchmarks there are silly is speed of kernel compile really that important? All that matters to me is that it compiles properly and in a relatively timely manner.

    The importance of the kernel compile time is that it serves as a measure of compile times in general, which is quite a reasonable benchmark for programmers.

    I would say the office and gaming benchmarks are also reasonable, since probably most of the readers (at least many of them) will never run a database on their computer, but may be tempted to buy the latest and greatest chip, only to find out that it's slower than an athlon at 2/3 the clock speed.

    I personally am glad they ran enough tests to show that the pentium is still behind the athlon on FP performance, which is the only thing I care about (except perhaps compile time), as I run floating point stuff at work (physics stuff).

  • *shrug* as far as i saw the AMD chip kicked the P4 Xeons arse with room to spare on a 1 to 1 match.
    the dual CPU benchmarks showed negligible performance gains for every test except multimedia. the P4 also got kicked hard on Floating point (huh? 2 intel CPUs cant keep up with 1 AMD running at half the speed with 1/4 the cache? wtf?).
  • I can't seem to remember where i read it, (it may have been on [H]ardOCP [hardocp.com].) but whatever it was, it stated that the Pentium 4 is messed up because it is configured for pure MHz. However, the way that it uses those cycles is very inefficient. it is very common to see 1.33Ghz athlons, or even 1Ghz P!!!'s outperform the 1.4Ghz p4's as far as throughput and other factors besides clock speed.
    ----------------------
  • My first (intel) PC was a 286

    then I went to a 486-50

    then a P90

    then a P2-333

    now I'm running a P3-1GHZ

    Each has lasted 3 to 4 years, however I have upgraded other components in the boxes. Added hard drives and memory about every 1.5 years, replaced video cards and modems every year. But the latest system will probably only get one upgrade in its life when a reasonably priced DVD burner (whose DVDs work on my DVD player) comes out. My CD burner is Good Enough, even after 3 years. I have 30 GB of hard drive, on two drives, and use about 10 of that. The 256 MB of ram is plenty. The ATI All-In-Wonder Radeon does MP2 recording in real time.

  • > You upgrade when you cant stand your old machine anymore and the new one you buy will be at least 4 times faster.

    Four times? I tend to get a cpu twice as fast as my old one (on Mhz and Benchmarks.)

    > Seems to be around a 3 year cycle.
    Yeah, that's about right.

    My upgrade schedule was:
    Apple, 1 Mhz
    XT, 4 Mhz
    286, 12 Mhz
    386sx, 16 Mhz
    Pentium, 100 Mhz
    Pentium Pro, (180 o/c to 200 :)
    Dual Cel 550s (366 o/c to 550 :)
    Athlon 1.2 Ghz (not o/c as I prefer 100% stable)

    My upgrade schedule has been around every 2 to 3 years as well.

    Currently, I'm not upgrading for another 2 years when 2+ Ghz machines are out (Just upgraded back in Feb. Probably will upgrade the GeForce 2 first though next year.)

    > and Im not very easily impressed by gee-whiz-gotta-have-that hardware anymore.

    I hear ya. As you get older, computers just don't have the same "magic" or "pizzaz". (I grew up with an 8-bit 1 Mhz Apple ][ w/ 64k. Now we have video cards with 64 Megs of ram. Times sure change :)

    i.e. P4 1.5 Ghz. Yeah, so. It's not THAT MUCH faster then an Athlon 1.2. :)

    Don't get me wrong, I still want faster hardware, but it just doesn't phase me the same way when I got my 386sx-16.

    e.g.
    Cant' wait for 3D graphics to look just as good as 2D. I want a real 3D MMPRG to make Quake 3 look like Donkey Kong.

    * Can't waiting for the GeForce 3 to become "bottom end" / ubiqiutous :) *

    Cheers
  • I'm sorry, but if you need 30Gs of hardware for a MySQL database you've designed something seriously wrong. And yeah, 2 Xeon 1.7Ghz, 4GB RDram, and 4 good 10k drives for quick raid will be around that. Surely you've got some more interesting information to serve than that which is flat? For that matter, apache across 20 boxes you found in a pentium junkyard with a cheap loadbalancer in front are going to run circles around the dual Xeons, especially when you factor in things like Uptime and concurrent load -- 2 things at once is nice. 20 things at once, even at 1/4 the speed is much better. Less time spent task switching, more time spent serving files. Odds are it'll appear faster to client as response time is reduced even if overall throughput is *potentially* increased.
  • Point-to-point can only give the processors dedicated bus up to the Memory Repeater Hub at which point the processors will still compete for the same memory bandwidth. However point-to-point is still better than FSB.
  • So let me get this straight. You should buy a system that performs the best under conditions that will occure .001% of the time and not one that performs better the rest of the time? That makes lots of sense. I love the way you can compare a specific chip (xeon) to an entire class of chip (risc) and make a statement like this. In this day and age there are no pure RISC or CISC architectures. It's' like China, not even they are 100% communist in practice.

    Hmmm. Were you confused when I said RISC chips like the Sparc? Ok, obviously the Pentium is a RISC core with a CISC emmulations, and the lines everywhere have been blurred. Let me rephrase RISC with "high-end server chips". Are you still confused? You should never buy a chip that handles 1/1000 situations. However, high load is a fairly predictable and common situation for many of the more demanding server roles these days. I get the feeling /.'s database (quad Xeons) probably runs under fairly continous load during the day. Do you think they screwed up and they should downgrade to regular pentiums? From the performance I've seen today on /. they should be considering a 64bit chip with better throughput on all of their servers.

  • throughput throughput throughput (Score:5)

    by selectspec ( 74651 ) on Monday May 21, 2001 @06:29AM (#208700)
    I remember working at a place where this guy was complaining that some timing tests he ran on these E250's running 350Mhz Sparc II's vs. 450Mhz PIIIs clearly favored the PIII. I told him that his test was meaningless because he was not running the systems under full load. When I demonstrated by essentially DOSing the two systems with SSL requests, he saw how the throughput tends to smoke clockspeed in the end. High end chips come down to saturation performance. Of course compare a Xeon to pure RISC chip like the Sparc under these high load conditions and you'll see similar results. Clock speed loses to throughput under load.
  • Can you do real-time editing of the digital video clips you downloaded from your digital video camera? Do you even have enough storage for that kind of content? Didn't think so.

    The need for speed does not ramp up continuously. I'll agree with that. But certain classes of applications require quantum leaps of processing power. We've gone from blinking LED's to text to still images (with hints of motion thrown in for effect). But today's computers cannot reasonably be expected to handle high-bandwidth streaming multimedia except in fits and starts. They will though, and this will require more horsepower. Much more than even this processor improvement provides.

    So don't say "No one with a modern computer will ever need to upgrade". Amend that to "You might consider waiting until you'll really be able to notice the difference." Because depending on what you're doing, you really may notice the difference.

  • I dunno, my 1GHz Athlon gained almost a minute when it went I compiled $KERNEL on a default redhat kernel, and then again compiling the same kernel while running 2.4.3-ac14. Now, going from one ac- to another may not make a difference, but leaving _default redhat kernel_ did (isn't that what they used on the k7 compiles?)
    --

  • What is Jackson Technology? (Score:3)

    by mclearn ( 86140 ) on Monday May 21, 2001 @06:05AM (#208703) Homepage
    In case you don't know what Jackson Technology is, here's an article over at Tech-report. However, is appears that Intel has sold their soul to Lucifer. Ah well, let 'em join the ranks of other companies to have done the same. :-) http://www.tech-report.com/onearticle.x/1947 [tech-report.com]

  • Re:What is Jackson Technology? (Score:4)

    by mclearn ( 86140 ) on Monday May 21, 2001 @06:27AM (#208704) Homepage
    Crap. I meant to point to this [theregister.co.uk] article over at The Register.
  • with default options obtained by running 'make menuconfig' and exiting without changing any values. You could also type 'make config'


    Of course, make oldconfig would do the job just as well.


    Incidentally, Linux 2.5 will feature CML2 [tuxedo.org], developed by ESR and friends, to make the configure and compile process more dynamic. The number of threads to be used will no longer have to be specified by the user as the system will discover this automatically.

  • The vapours from the dual-Athlon hardware will clearly overwhelm this real operating hardware.

    To quote the last paragraph of the article:

    "The real question on everyone's mind is how does the i860 and the Intel Xeon compare to the upcoming 760MP and the Athlon 4? We have been benchmarking that very combination for weeks now and soon enough we will be able to provide you with the definitive answer in many more test scenarios than those we just presented to you."

    Yup, just vapor. You moron, dual Athlons motherboards have been out for months, and everyone who has one is under NDA. Do you think Anandtech went out to Best Buy and got a P4 Xeon? P4 Xeons will probably not be available from anyone for at least a few weeks commercially: Intel gave them one, they benchmarked it, and when they got the OK from Intel they released the results.

  • Will someone explain to me exactly what Nelly Furtado [nellyfurtado.com] has to do with all this (see the related links)

    --jon

  • What about more users on one workstation at the same time? Two users on dual processor workstation will utilize more power than one user on one CPU.

    You can put two or more video cards in a PC. Or use dual head card. With USB, its easy to plug in few keyboards, mice, ... The only problem is that operating systems do not support such setups at the moment.

  • Maybe I'm being naive here, but...

    In order to make room for all of these memory slots they are located on a riser card that sticks up out of the board. This unfortunately means that there is no hope of getting [the MB] to fit in anything smaller than a 4U or 5U rackmountable case.

    So what's stopping someone from using a little adapter and some standoffs to have the riser board sit above and parallel to the MB? (of course, you'd need a different CPU fan, but for a 2U or 3U box, you've got to completely rethink airflow anyway.)

    Inquiring Dremel-wielding hardware h4XXors want to know.

  • but then again I am using plan9
    .oO0Oo.
  • Speed always matter. It just depends on the application. For editing text, a 4.77 MHz 8088 is ample, as was a 1 MHz 6502 before that.

    But forget office applications, no one is buying a new computer to edit text. How about games. Take the "frames per second" indicator in 3D simulations, it still very far from the ideal. The computer of my dreams would be able to do 60 frames per second when ray-tracing a scene with 100 million objects. And at the same time, it should also have the capacity to solve partial differential equations fast enough to simulate waves hitting a beach, accurately calculating the position of all the bubbles in the surf, and all the grains of sand. And how about strectching and deformation of solid materials: when will Lara Croft have a natural-looking smile in her face?

  • Return of the i860 (Score:3)

    by ozbird ( 127571 ) on Monday May 21, 2001 @05:43AM (#208712)
    Flashback: 1989 - Intel releases the i860 [si.edu] 64-bit microprocessor, dubbed a "Cray on a chip" (okay, so it was only a Cray 1, and I think it was only a third as quick as that, but hey - it's a "Cray"!) I even remember some of the RMIT [rmit.edu.au] ubergeeks designing a kick-ass computer based upon the i860... Unfortunately, I don't think it was ever built.

    2001 - Intel releases the "i860" chipset to support the latest of its flagging 32-bit microprocessor range. Intel's 64-bit microprocessor, the "Itanium", is due for release real soon now...

    </irony>
  • I'm starting to think that we need a system where *every* slashdot reader has the opportunity to moderate any message as "incorrectly moderated as 'funny'" and likewise for the other options. The amont of scoring a moderator gets would effect how often he or she gets moderation again. This also favors new blood since first time moderators would have the highest possible score (0 bad points). On the other side, some type of aging would have to be present so that active moderators aren't disproportionately hurt. In any event, I think the current system has to change.
  • I don't think you read what I wrote.
    But then again, you're just another AC.

  • Simply put, the cost of DUAL XEON 1.7Ghz cpus makes them un unrealistic purchase for small businesses and power users.

    AMD's cpu's are less expensive and give a MUCH better performance for their price than an Intel PIII or PIV.

  • It does make a lot of sense under some circumstances. If a system is just ticking along not doing much, it doesn't really matter how well it performs (assuming some reasonable level of average performance). Think of a low-end Toyota and the best BMW available. Both of them perform superbly at 50kph. There's virutally no difference between the two systems. You buy the BMW because it's peak performance is superior, even though you only use that peak performance occasionally. In the server world, if you're only asking for a few X per second, performance is frequently meaningless - buy whatever is cheapest, taking into account things like maintenence. It's only when you're looking at high load that performance matters.
  • June 15 [tombraidermovie.com]


    Enigma
  • >when will Lara Croft have a natural-looking >smile in her face?

    And natural-looking breasts, instead of a hexapod bra.
  • Alright, it is of course no problem to come up with thousands of applications where unlimited CPU would be nice. But still:

    Graphics can easily be handled in the Graphics card. The framerate in Q3 or whatever is not really there thanks to the CPU, its there thanks to the graphics card (I dont play games).

    ...and when it comes to most real applications (scientific work) memory is usually the showstopper. If the program does not fit in the cache it often wouldn't help even if you could do floating point operations in zero clockcycles - the cpu would just be waiting for memory all the time. And RAMBUS/Intel, memory latency is the main problem, not memory bandwidth.

    If Intel put 32Mb full-speed L2 cache on their Xeon, then its high clockfrequency would be _really_ useful.

    ...and of course you can find lots of more or less useful applications for fast cpus with just 128kb of cache...
  • It would be more fair to compare UltraSparcs, Alphas and Itaniums when they start rolling out.

    The Itanium will be at least a generation ahead of Alphas and UltraSparcs when they are released so I dont really agree it would be more fair, but it would look better for Intel. And it will be very interesting to see what the Itanium can perform.

  • Is there any sane reason why I should put a CPU optimized for multimedia and with a lousy integer part in a file/mail/print/backup-server or a W2k Domain Controller? I guess not. How soon will Dell put 2 of these CPUs in there general purpose servers? I guess quite soon...

  • Relevant to compare with Athlon? (Score:4)

    by Zo0ok ( 209803 ) on Monday May 21, 2001 @05:59AM (#208722) Homepage
    It is of course impressive that Intel made it run at 1.7GHz, but given this, the benchmarks are quite what could be expected. The Xeon is intended for High-End workstations and servers. I dont think many people will actually choose a system, just becaues it is Xeon-based (the OEM has made the choice).

    When buying servers the CPUs are not really the most important thing (at least now where I work - we focus on disk/RAID/memory and takes whatever CPU comes with it, and maybe double it). If CPU is really important (massive database, scientific applications etc) 64-bit CPUs or vector machines might be more adequate even though the operate at lower clock freq.

    Most Xeons will be in Windows (and of course, Linux/BSD) servers. On these servers PIII/Thunderbird is not really an option anyway.

    So, what I have wanted to say all the time is that I'd rather see a benchmark/comparision between UltraSparcs, Alphas and Xeons.
  • Totally agree. I'm going for a dual Athlon system, when I start at my new job. The point-to-point bus in AMD's 760MP sounds interesting. One thing i didn't understand in the article though was the statement that a point-to-point bus doesn't keep the CPU's from competing over bandwidth. What's the point with a P2P bus then?
    ---- "Ok! Now that I have an extra brain, the only thing missing is an extra pair of arms."
  • And I'm going to build a beowulf cluster of these.

    No foolin'.

    My company is working on a huge database project, with the need to look at the data in every way conceivable. That, and they want to rent out computer time for data analysis to other companies in our industry.

    The boss told me to budget for a 16-node cluster, and I talked him into waiting for the Athlon 4 to come out, later this summer. Dual-processor boards, a couple of gigs of DDR ram per box, and we are talking greased lighting :)

    I figure it'll set us back about $50k, assuming we can build each box for $3k or so.

  • I wonder if it's one of these...

    Dear ValueWeb Customer,

    ValueWeb is performing a system-wide upgrade for all users with domains residing on servers running the BSDi operating system.

    If you are receiving this e-mail , your domain is currently on one of these servers and your site is scheduled to be moved to a dual processor VA Linux server.

    Some of the benefits of this upgrade are that you will soon have the ability to use Chili!Soft Active Server Pages, the latest version of Miva engine, plus the ability to run Linux supported software. Upgraded processors and increased memory will also result in improved performance.

    --
    All your .sig are belong to us!

  • Imagine a Beowulf cluster of THESE babies!

    Sorry, it was gonna happen eventually.


    --

  • So, I have a 1.2GHZ athelon. Do I really need a Xeon. No. There compes a point where the marginal benefit of speed increases becomes negligable, and the cost effectiveness of upgrading to get that additional marginal performance, disappears. I believe that occurred when we entered the GHZ range of measurement for clock speeds. Aside from compiling kernels (which an earlier poster correctly pointed out, should only happen infrequently) What does your adverage user need this kind of speed for?

    Is it so we can tell our friends "My Computer is bigger than yours" ?

    Do we really need to continue to upgrade at this point? I grant you the marginal benefit of moving from 9 Mhz to 16Mhz is extremely large, as is the marginal benefit of moving to 33 and 66 mhz. As soon as we hit 90 though, that benefit began to dwindle. It became a crutch for bad code, an excuse for Microsoft to write bloated operating systems.

    Then AMD began competing heavily and forced a shortening of the product cycle... Was this truly good for consumers? I ended up upgrading my system three times in two years. It gets expensive after a while, and with OS vendors all too happy to force consumers onto new hardware, through distribution of poorly written overdone and over-typed software, the consumer doesn't truly benefit at all.

    Certainly hardware vendors benefit, but when it comes down to it you have to ask yourself, Do we really need all this speed?

    OK, to be perfectly fair, some of us do; yah, those of us who are processing data from the Genome Project [ornl.gov] in our spare time...

    --CTH

    --
  • That's the story submitter, silly. Nothing new there, the submitter is always listed in the related links area if they have a homepage or an email address registered.

  • Athlon (Score:5)

    by OblongPlatypus ( 233746 ) on Monday May 21, 2001 @05:50AM (#208729)
    For those of you who can't be bothered to read the entire article, the most interesting part by far is the real-world database benchmark towards the end. The 1.2GHz Athlon places right smack in the middle between the single Xeon and the dual Xeon (both 1.7GHz).

    That makes it highly likely that a dual Athlon will significantly outperform the dual Xeon, does it not?

    Even though dual Athlon systems aren't available yet, I'm willing to bet that when they are, the price tag of one such system will quickly drop far below the price tag of one of those dual Xeon beasties.
  • The Intel Xeon processor shares the exact same core as the desktop Pentium 4, meaning that the same features the Pentium 4 can boast, the Xeon can do the same. This also unfortunately means that the same shortcomings which affected the Pentium 4 will also affect the Xeon.
  • Maybe they just wanted to show people that getting big, nasty, high-end chips for their office applications is a stupid idea? That's what I got from it - the right tool for the job. Gimme an Athlon for my desktop, and a dual P4 Xeon system for my database server.
  • The person who uses all that io for quake and photoshop needs to be hanged by their toenails. I suppose anand has to put those type stats in so the 'gamer' community has something to relate to but in this instance it's not really relavent is it? This isn't a 'pc' processor. Some of the benchmarks there are silly is speed of kernel compile really that important? All that matters to me is that it compiles properly and in a relatively timely manner. Office suite benchmarks just get filed with the rest of the fluff, in short this review should have been more in context with the processor (the database stuff was good).
  • Your right speed doesn't matter anymore, so why did you upgrade 3 times in 2 years?

    Then AMD began competing heavily and forced a shortening of the product cycle... Was this truly good for consumers? I ended up upgrading my system three times in two years. It gets expensive after a while, and with OS vendors all too happy to force consumers onto new hardware

    Sure its good for consumers, Nobody is forceing you to upgrade 3 times in two years. Its all about choice, if AMD wasn't digging at Intel we wouldn't be blessed like we are now. If you want to live out on the edge, you pay the price. Me I've only upgraded 4 times in twelve years. Each time I get more for less money than the previous system cost. Lets see I went from 286 to 486-25 to pentium 166 to celeron 566. Maybe I'm weird but i tend to buy the bottom of the barrel as opposed to the bleeding edge, When your 1.2 GHZ Athlon is the entry level CPU I might get one of those machine or hold out for the next level of entry, (64 bit systems ?).

    Memory is cheap now I think most people would be better served maxing out their memory rather than seeking faster CPU's.

  • Certainly hardware vendors benefit, but when it comes down to it you have to ask yourself, Do we really need all this speed?

    No. If you do not want the functionality of your computers to increase, then you don't really need all this speed.

    Every time Intel or AMD raises the bar, there is always a crowd asking this question. There were actually people who believed that anything more than a 386 was a waste.

    Forget the fact that, historically, any excess in average CPU speed is almost instantly consumed by more feature rich software. The obstinate call this bloat, but most people recognize it for what it is - improvement.

    So, although your 1.2 Ghz Athlon is certainly powerful enough for you now, I know that deep down you don't really believe it could last you for the next 10 years. Surely you're aware that by then every last cycle of your 100 Ghz CPU will be saturated with software more advanced and intelligent that you can even imagine.

  • I am a bioinformatics graduate student who IS processing data from the genome project in his spare time.

    I regularly run jobs on supercomputers.

    Often, I bring work home with me, because it's more comfy there. Last fall, I bought a Pentium 1 Ghz w / 384 Mb RAM.

    I frequently kick myself for not buying more RAM (I'll probably upgrade to >=512 Mb soon), and yes, the difference between 1 and 2 Ghz matters to me.

    I might represent a small proportion of users, but to us, it makes all the difference to have home computing systems pushed to the limits of performance and affordability.
  • This is all well and great if you have the money to afford something like this. IN a lot of our cases, we have 1.2 ghz athlons, and we like them just fine. Having more speed is great. The only thing being that if you spent so much on a cpu and mobo that you can't afford the power it runs on, then you have a problem.

    Not to put it down, but I'm going to wait for it until it comes down to like, oh, under 400 dollars a unit. Then it will be 'affordable'.

  • We used an i860 based system here for a while. It was an Alliant FX2800 with (IIRC) 26 i860s for processing power and a further 2 dedicated purely to I/O. It was pretty efficient and could handle multi-processor vector optimisations pretty well.

    All well and good, except that the OS (concentrix) was hideously buggy and prone to crashes. In the end it was junked in favour of a 4-processor Alpha box which was faster, had more memory, and cost the same amount as the yearly hardware support for the FX2800 :)

  • Actually I got burned on several of Intel's chips they were all O/Ced of course :)

  • Programmers will make it matter (Score:3)

    by melquiades ( 314628 ) on Monday May 21, 2001 @06:49AM (#208739) Homepage
    Sure you don't need more speed now, but remember someone-or-other's law: Software will expand to consume all available resources.

    My Apple ][+ had 48k of RAM, and somehow still managed to do all sorts of cool stuff. And although it was probably about four orders of magnitude slower than my current machine, it didn't feel that much slower. Somehow, modern software manages to use resources at rates that we would hardly have dreamed of in 1980.

    This isn't necessarily a bad thing. It's easy to blame programmers for being lazy, but they're actually making intelligent decisions that lead to ever-increasing resource use. Every new increase in speed or memory opens up options that just hadn't been there before -- scripting languages, multitasking, multiple users, OOP, application frameworks, garbage collection...all made possible because there was a little power to spare. OS X uses gobs and gobs of memory double-buffering everything on the screen, which is just fine with me -- memory is so cheap they gave me 256 megs free when I got my machine, and the UI looks really cool for all the extra RAM it uses.

    So hang in there. I'm sure somebody will find a way to burn up all the power your hardware can muster. And oddly, it will probably be worth it.
  • Still have my K6 300Mhz with 128Meg and petite 4 Gig HD . It gets the job done, for internet, for office apps, and most games. Only add-on was a DVD-Rom and REALMAGIC PCI board which gets the job done for DVD movies.

    Do I want the fastest out there? Yes, I do. But I tell you. I saw 600mhz and 850 mhz Athlons running and except for games, increased clock speed increased nothing much else. Windoes still takes forever to load. Office apps didn't jump to attention, and I wanted to hook up a hand crank to jack up the internet download speed. So obviously, there is something else needed to get PCs up to the next speed level, and clock speed isn't it. How about get front/side bus speeds up and harddrive data transfer up and how about that 133 -SDRAM? Shouldn't it be equal or almost equal to CPU speed? And finally the 56k modem. Sorry to say most of us still use it and will continue to until DSL or cable internet prices come down AND ACTUALLY BECOMES AVAILABLE!!! So although some people got the fancy 1 or 1+ gigahertz clockers, my question is; really?? or is it just a marketing scheme. I haven't seen the difference. My 1966 Mustang far surpassed the car Ford calls Mustang now.

  • The point is that Athlon outperforms Xeon when measured in comparable enviroment ( not to mention significantly lower price.)
    These "Athlon fanboys" are into something here, don't you think ?

  • I already got this story [kuro5hin.org] on the other site.

    This post is only meant to be read by authorized readers

    Authorized readers includes but is not limited to people rating this post up

    UnAuthorized readers includes but is not limited to people rating down

    Author of post is not liable for any spelling mistakes, or idiot comments made in this post

    That would be because of an recently discovered Slashdot feature ironicly called a "lameness filter"

    This post will self-destruct in 10 secs.

  • Everyone keeps talking about getting a dual proc machine. I tend to leave those for the server types. I know linux has SMP support but most consumers use Win95/98/Me Those opperating systems will only use one processor. Sure, once you put Win2K on your bok you get to use that second proc, but the ral advantagees come when your software is designed to use multiple procs. At home the typical things that need horse power are games, and you probably won't be doing something in the background then. I would save he money and spend it on the 3d card in that situation.

    Forgive the typos, I have two broken wrists.
  • Yeah! but the Intel chip looks much nicer in dimed light.
  • The point isn't 'it is only 30% faster than my Athlon'.. This is not a dektop product - you're preaching money saving techniques to businesses who could care less if a system costs $500 more or performs 5% worse. They want reliability and speed, this offers both. Will the 762? Maybe, that has yet to be seen - I think it will be very fast, but will it be reliable, and will the business community accept it? I don't know. Neither AMD nor Intel CARE what us home user desktop people think, this is about OEM's and the business sector.
  • The XEONS will be shipping in two weeks, the Athlon NDA doesn't even end till then. They will ship around the same time - and Intel had planned this release for May 8th, I don't think they're that interested in stealing thunder. If anyone is trying to outdo the other, it is AMD and their benchmark leaking funniness. I just know that in the business community 'no one ever got burned going with Intel'...
  • Everyone seems to be combining two fdifferent markets - desktop and workstaton/server. The XEONS aren't meant for desktops, if they were, they would suck. They are meant for workstations and when the MP shows, servers - the 762/dual athlon will compete with the XEONS, not the 1P Athlon.

Slashdot Top Deals

Only God can make random selections.

Close