AMD64 3000+ review

http://www.hardtecs4u.com/reviews/2003/amd_athlon64_3000/

It goes bad for AMD really bad.I do my normal 6 month prediction.Prescott will be clear winner in all the Q1 Q2 Q3 Q4 of 2004.It will see if i can keep my 100% accuracy.
<P ID="edit"><FONT SIZE=-1><EM>Edited by juin on 12/15/03 07:47 AM.</EM></FONT></P>
31 answers Last reply
More about amd64 3000 review
  1. According to the website a 3200+ use 89 watt TPD.

    I dont like french test
  2. <A HREF="http://www.hardtecs4u.com/reviews/2003/amd_athlon64_3000/" target="_new">Clicky</A>

    Quote:
    It goes bad for AMD really bad.

    Can you please tell us what led you to this conclusion? The way I see it, from the benchmark you gave to us, the A64 is a MUCH better choice over the P4 3.0c. It is slower in video encoding but faster in everything else, and it costs right now at Newegg $56 less than its competitor ($215 vs $271).

    Quote:
    Prescott will be clear winner in all the Q1 Q2 Q3 Q4 of 2004.

    Again, it seems to me that you are living in a world of your own. If P4 EE didn't manage to beat FX-51 I doubt that Prescott will manage that, especially since we know that Intel won't be releasing a CPU faster than 3.6GHz, at least before H2 2004. And even IF it does, we also all know that the S939 "San Diego" A64 is coming with 1000MHz HyperTransport bus and lower latency not-registered DDR memory. I am really really curious about how you are being so sure that Intel will dominate in 2004.
  3. Quote:
    It goes bad for AMD really bad

    Try this link on for size:

    <A HREF="http://www.theinquirer.net/?article=13187" target="_new">Inquirer link to A64 3000+ article</A>

    It tells a different story than what you're saying. The only problem is that the original article is in German, but the Inquirer summarizes the results.

    <i>Money talks. Mine always likes to say "goodbye." :smile: </i>
  4. You can read the review yourself if you like! I used the translation tool at Altavista for that website and it works really well. That review shows the Athlon64 3000+ beating 3.0c in everything except video encoding. I am very surprised by this guy's conclusion as well and I can't wait seeing his reply! I'm really curious about this, and about his 2004 prediction too.
  5. I thought the same thing as you I read the post (very early this morning) came back 10 minutes later to respond but you had already said what I was thinking. I figured why double post the same question. I'm also curious on his reasons (conclusion) for the post.

    If I glanced at a spilt box of tooth picks on the floor, could I tell you how many are in the pile. Not a chance, But then again I don't have to buy my underware at Kmart.
  6. Is Intel going to move its critical data centers to the Opteron platform?????

    <A HREF="http://www.theinquirer.net/?article=13172" target="_new"> Click to read </A>


    If I glanced at a spilt box of tooth picks on the floor, could I tell you how many are in the pile. Not a chance, But then again I don't have to buy my underware at Kmart.
  7. This guy basically just spews Intel PR and lives in a world of his own. I wouldn't pay him much attention. :)
  8. Quote:
    This guy basically just spews Intel PR and lives in a world of his own. I wouldn't pay him much attention. :)

    Lol ... He's gotten better, but his English still isn't that great. I'm never really quite sure what he's saying or what his reasoning is. We need a "Juin English to English" translator, but Altavista doesn't have one.

    <i>Money talks. Mine always likes to say "goodbye." :smile: </i>
  9. You know what's more interesting?
    Quote:
    It will see if i can keep my 100% accuracy.

    ... no comment!
  10. juin read this and tell me what you think(please dont start crying :)

    Afterall, a Elvis Prescott @ 2.2 GHz doesnt sound so bad :)
    http://www.eet.com/semi/news/OEG20031215S0017
  11. "Can you please tell us what led you to this conclusion? The way I see it, from the benchmark you gave to us, the A64 is a MUCH better choice over the P4 3.0c. It is slower in video encoding but faster in everything else, and it costs right now at Newegg $56 less than its competitor ($215 vs $271)."-pitsi

    Amd is not going to support the AMd64 3000 or AMD 3200 with an upgrade path,so if you're building one of those now you'll be throwing away the $280 you spend on good single channel ram,which makes the intell cheaper if still inferior.I was going to build a amd64 3200;but now I'm waiting for socket 939 :( or will just build around the xp2500+barton.
  12. Is it confirmed there will be no upgrade path for these chips? The last I heard there was alot of confusion over the latest roadmaps - I find it hard to believe there will be no viable upgrade path.
  13. Where on earth do some people get there information from??

    Yes AMD is certainly going to be supporting the athlon 64 with an upgrade path, namely the the socket 754 platform on which it currently runs. You seem to be getting yourself confused between the the athlon 64 platform and the athlon 64 FX.
    The difference being that the normal athlon 64 only has a single channel ddr memory controller where as the FX range has dual channel, thus its need for the extra pins on the cpu (754 compared to the 940/939) oh yes and alll those extra $$$$$.
    The new socket 939 platform are to support the new athlon 64 fx cpus that will be released early next year and will no longer require the use of registered ECC ram needed by the socket 940 they will not effect the athlon 64 3000+ and 3200+ which run on the socket 754.

    Arguing on the internet is like running in the special olympics. Even if you win your still retarded!
  14. <A HREF="http://www.xbitlabs.com/news/cpu/display/20031030090354.html" target="_new">AMD Confirms Socket 940, Socket 754 AMD Athlon 64 CPUs Next Year</A>

    How is it possible to even think that a company would introduce a new socket, release 2 processors for that socket and then discontinue it? It doesn't make sense! If I would upgrade I would probably wait for S939 as well but if I wanted to upgrade right now and not wait 4 months or so, I would choose S754 over Pentium 4 anytime.
  15. In case your wondering does this mean those who have gone and bought Athlon 64 FX 51 chips will be left with out an upgrade path.
    No it does not
    The socket 940 platform will still be used by Opteron processors.
    So opteron 1XX boards will be compatible with athlon fx chips (for all intents and purposes they are exactly the same chip anyway). These boards however will most probably not be geared towards gaming as the opteron is a workstation/server chip.

    Arguing on the internet is like running in the special olympics. Even if you win your still retarded!
  16. All aplication of all type are faster on a P4.A64 are fast on game but will never improve my frame rate.After FSAA and ASNI are on and quality are high the cpu clock speed or else is unrelevant.

    My prediction is base on new PCI expresse DDR-2 coming future manufacturing process and how the platform evolve.

    I dont like french test
  17. So in other words, you have no idea about what you are talking about.

    Quote:
    All aplication of all type are faster on a P4.

    If you ignore all the synthetic benchmarks (as you should, since they don't play any role in real performance), the P4 won only in video encoding. That's not "all applications of all types" in my book.

    Quote:
    A64 are fast on game but will never improve my frame rate.After FSAA and ASNI are on and quality are high the cpu clock speed or else is unrelevant.

    Then why not choose a Barton 2500+ which is cheaper as well? Answer me this if you can!

    Quote:
    My prediction is base on new PCI expresse DDR-2 coming future manufacturing process and how the platform evolve.

    Yeap, those are great reasons! Way to go!
  18. "All aplication of all type are faster on a P4"

    What are you basing this on?

    <i>The wrath of penguins shall be felt. No windows shall be left unbroken.</i>
  19. I think that for NOW this is unparalleled in price/performance, especially for games. It also has the benefit of running 64-bit. And it seems as Intel is moving towards x-86-64 this is definitely a bonus (http://www.xbitlabs.com/news/cpu/display/20031216070734.html)

    Sure, other platforms will support PCI-Express, DDRII memory in the FUTURE, but for NOW, this is the processor to get. For gamers NOW; this is the chip to get. If you do multimedia/encoding, the P4 might be better

    The only drawback is the current lack of a good chipset and mobo if you are interested in overclocking; the VIA K8T800 lacks a PCI/AGP lock which seriously impairs overclocking :(
  20. Personally for that price I find the 3000+ at that price very tempting, considering all I do with my computer is play games and type papers. The main reason I haven't grabbed on yet is I'm waiting on a chipset that can do everything I want. IIRC (and my memory isn't always that good) the upcoming SIS chipset that was reviewed recently looked like it might be ok.
  21. "I think that for NOW this is unparalleled in price/performance, especially for games. It also has the benefit of running 64-bit. And it seems as Intel is moving towards x-86-64 this is definitely a bonus (http://www.xbitlabs.com/news/cpu/display/20031216070734.html)"

    If you meant this as a reply specificaly to me, all I can say is you're preaching to the choir. :smile:

    [begin anti-fanboy rant]
    This guy keeps claiming that P4s are faster in all applications other than games, and the times I've asked him where he gets his numbers or what he bases his statements on, he doesn't reply.

    In another thread, he told me that 64bit Linux was only available in high end expensive enterprise versions of Linux that cost much more than windows, unless you recompile your kernel. He said this, despite the fact that I run Suse 9 professional for AMD64, which cost 119.00, which is cheaper than any price I've ever seen for Windows XP. He also made some statements I didn't understand regarding the performance benefits of 64bit Linux versus 32bit. I asked where he got the numbers from and he didn't reply.

    I'm only asking him to produce some evidence of his claims. I hate fanboys who spread misinformation. If he knows of some info that can back up his claims regarding the P4, then he ought to share it, because the review he pointed to, contradicted not just the initial post of this thread, but also his later statement that "All aplication of all type are faster on a P4". If it is faster, I want to know so I won't be misleading people by saying otherwise, but I'm not just going to ignore the numerous comparisons I've seen and take his word on it.

    And when Intel releases chips that are faster, I'll be doing the same things to the AMD fanboys. If people want to be loyal to a side, pick a fricking sport. Don't go mucking up tech forums with biased nonsense. I mean, I like Linux, but I'm not about to go lie to someone and tell them that Linux is better than Windows in everyway just because I like Linux. That's ridiculous and dishonest.

    The vast majority of this wasn't targeted towards you waho, I hope juin and every other fanboy reads this. As long as you have to lie, YOU'RE WRONG. As long as you have to spin, YOU'RE WRONG. Crap, I'm a Raiders fan, but I'm not about to go out and say they could clobber every team in the NFL, because the fact is, this season, they can't. And this isn't even a sport for pete's sake...
    [end anti-fanboy rant]

    <i>The wrath of penguins shall be felt. No windows shall be left unbroken.</i>
  22. Changed my mind;but I'm only going to put in an athlon64 3000+.I wish I knew whether its 1.8 or 2ghz-venders are listing it as both;but newegg has it just as 1.8ghz now!

    With the msi k8t neo-fis2r kingston reconmends KVR400x64c25 which only costs $115 a stick or $91 at newegg(via pricegrapper). Anyone have recommendations for coolers for the ram and for twin raptors of the 36gb variety?

    When I stated AMD had no upgrade path I meant next to none as they'll only introduce new ones during the first half of 04 and only up to 3700+ making their upgrade path and length as short as the bartons without the massive overclocking ablity.Who would pay $418-450 for a 3200 only to upgrade it to a 3700 when the better 949 socket will already be out,just like upgrading to a 3200+barton now!
  23. I dled Suse 9.0 off their site for free...free is pretty cheap :smile:

    <font color=blue>If the <font color=yellow>laurel</font color=yellow> is to big for your head, it becomes a hoola-hoop, and you have to keep your butt really busy.</font color=blue>
  24. The Athlon64 3000+ at NewEgg features 2GHz, 512Kbyte L2 cache. Some info at <A HREF="http://hardocp.com/article.html?art=NTYw" target="_new">HardOCP</A>. There's an Athlon 64 3000+ chip for desktop replacement notebooks which features 1Mbyte of L2 cache and runs at 1.8GHz. Good explanation at <A HREF="http://www.xbitlabs.com/news/cpu/display/20031214095239.html" target="_new">Xbit</A>.
  25. <geek power on> ahhhh mmmmdahhhhh i ahh got my 5 jigahurts hard drive that will blow away the amd athalon 64
    <geek power off>

    LOL! so ahh ya i say go with what works :)

    "Bread makes me poop!" - Special Ed

    <A HREF="http://www.anandtech.com/mysystemrig.html?id=9933" target="_new"> My Rig </A>
  26. That is cheap! But, and I could be wrong, the dload version isn't setup for 64bit "out of the box", which seemed to be what he was saying, so I was trying to compare apples to apples.

    Although, you'd probably know better than me if it is setup for 64bit already... I was under the impression that the download one did a 32bit install. If that isn't true, so much the better. :smile:

    <i>The wrath of penguins shall be felt. No windows shall be left unbroken.</i>
  27. Why there are slower and thing will goes worse for AMD.

    The athlon64 3000+ have the same stepping and model number that athlon 64 3200+.There are faulty chip that some bit ligne were not good so AMD disable the 1/2 cache so the chip work.

    Tendency of K7 core, yes a K7 core they have the same core.When barton was introduce many were believe that would lead to the same increase that in NW.The increase in FSB to 200 mghz dont have yield the same result.Those platform improvement yield better result on a P4.On clawhammer thing are the same the disable cache dont even change the performance by only 1% or 2%.There old cache policy is not good any more and the intergrade controleur dont have that much lantency in clock cycle.Also they have to think about a direct feth in the L2 for excution only I2 use right now on FPU data access.

    DDR2 is coming they will be in the market in the same time that Granstdale will be release.

    DDR-2

    DDR-2 400 work only at 100 mghz.SO all timing must be 1-1-1-2 to equal PC3200 as it run twice as fast but DDR-2 will have higher timing for a long time.The good thing when DDR-2 800 they will be as fast as PC3200 and having twice the bandwith.DDR-2 should not have very long life spam in the market and make place fast for DDR-3.

    That will affect the market for sure.AMD must 1 create a MCH that work well.Many will agree that intel make the best controleur best balance between stability/performance.For AMD they might have a small advantage there with memory type support DDR-2 400 DDR-2 533 can be support and be fully use right at the start but they have to support DDR in the same time as none of the big RISC or intel will support DDR-2 until 2H of 2004 in low-end server and in larger systemes only in 2005 should be even longer for mission critical.The opteron target the low-end server so as long they have support for DDR-2 for about Q3 should be OK.

    On intel side they have 2 chose create longer pipeligne in the MCH to have a better ratio or have total sync.Going with the 1 one would yield better performance but dissipate more power and will be harder to put in place (higher price of manufacturing).On the good size of intel they dont have intergrade MCH so there lantency will be higher that AMD.
    On DDR-2 most of the lantency will happen in the ram itself so a intergrade controleur dont give much advantage.There still 1 but in % the gap is closing.


    Now PCI expresse or 3GIO.High speed serial bus that will come in 2 version and have a bandwith range very large with good lantency.PCI the former is still faster on intel controleur having about 126mb/S compare to via sis nvidia amd ali none are close to 120mb/S.Also how via and others will put that in place.When AGP 8X standart was done at intel via was allready place AGP 8X in shipement produce same thing for SIS.That was a big mess most card wee crashing Asus P4S8X some might remember nothing was working.E7205 was also having probleme data prefetch was not working either a bios updata was give after 2 day for solving the probleme.So in short intel will have faster pci exp should be stable from day 1 unlike VIA and the others.We will see how nvidia will act on this one should be interesting.

    Manufactring process

    Intel is king there no argument to do with that.Everything is develop by intel or IBM nothing else.The rest like nothing on isulator from toshiba never have see the light of the day.

    I dont like french test
  28. Quote:
    The athlon64 3000+ have the same stepping and model number that athlon 64 3200+.There are faulty chip that some bit ligne were not good so AMD disable the 1/2 cache so the chip work.

    Yes, and the yield on P4EE chips is so poor, most of them get the L3 cache disabled and the clockspeed reduced, then get sold as M0-stepping regular P4s. For most practical purposes, the P4EE launch hasn't even made it off paper, but at least Athlon64 3200+ supply is still decent.

    Quote:
    Tendency of K7 core, yes a K7 core they have the same core. When barton was introduce many were believe that would lead to the same increase that in NW.The increase in FSB to 200 mghz dont have yield the same result.Those platform improvement yield better result on a P4.

    And they also yield better results on an AMD64 part. AMD64 is not <i>just</i> a K7 with 64-bit extensions; a lot of the rough spots were ironed out, and almost every part of the core was significantly reworked. Memory bandwidth utilization is one of the rough spots that got ironed out.

    Quote:
    On clawhammer thing are the same the disable cache dont even change the performance by only 1% or 2%.There old cache policy is not good any more and the intergrade controleur dont have that much lantency in clock cycle.

    So you're basically saying AMD can safely reduce the Athlon64 cache to 512K, and they'd still be creaming Intel. I would think this is good news for AMD...

    Quote:
    Now PCI expresse or 3GIO.High speed serial bus that will come in 2 version and have a bandwith range very large with good lantency.PCI the former is still faster on intel controleur having about 126mb/S compare to via sis nvidia amd ali none are close to 120mb/S.

    Actually, Intel chipsets have had the same PCI bandwidth limitations as everyone else ever since the i850. It's a design decision based on choosing fair device arbitration over maximum PCI throughput. Most chipset manufacturers choose fair device arbitration for greater stability--not all PCI devices behave properly when they're data-starved.

    <i><Lionel Hutz> I'll be defending...The SCO Group!!!??? Even if I lose, I'll be famous!</i>
  29. I want my Athlon 64 3000+ now! Incredible performance for the price! And you get a platform that will be upgradable (at least for the next year).

    In bonus, AMD64 instruction set for LINUX or Windows with AMD64 support! I want it! I want it right now!

    <A HREF="http://www.anandtech.com/cpu/showdoc.html?i=1937" target="_new">http://www.anandtech.com/cpu/showdoc.html?i=1937</A>

    --
    Would you buy a GPS enabled soap bar?
  30. Yes, and the yield on P4EE chips is so poor, most of them get the L3 cache disabled and the clockspeed reduced, then get sold as M0-stepping regular P4s. For most practical purposes, the P4EE launch hasn't even made it off paper, but at least Athlon64 3200+ supply is still decent.

    They get sold as Xeon gallatin or XEON.The price is still high and gross margin is still very good.

    And they also yield better results on an AMD64 part. AMD64 is not just a K7 with 64-bit extensions; a lot of the rough spots were ironed out, and almost every part of the core was significantly reworked. Memory bandwidth utilization is one of the rough spots that got ironed out.

    Actualie the increase in L2 cache for desktop use have done alomst nothing the 3000+ is so close to the 3200+.Nothing yield good imprvement except lower lantency.The core was not rework there more change in Presscott that in K7-K8 you know i just finish me 1 64 bit register 64 bit add.That is very simple to add more bit across the chip was is hard is to have a high speed as RC delay will rise

    Actually, Intel chipsets have had the same PCI bandwidth limitations as everyone else ever since the i850. It's a design decision based on choosing fair device arbitration over maximum PCI throughput. Most chipset manufacturers choose fair device arbitration for greater stability--not all PCI devices behave properly when they're data-starved.

    Still they are faster and much less issue on intel south bridge and RAID will goes faster.Useful for small server like CSA.


    I dont like french test
  31. Quote:
    They get sold as Xeon gallatin or XEON.The price is still high and gross margin is still very good.

    <A HREF="http://www.hardforum.com/showthread.php?s=&threadid=693498" target="_new">http://www.hardforum.com/showthread.php?s=&threadid=693498</A>

    And just to stack likely rumor on top of confirmed fact...

    <A HREF="http://www.theinquirer.net/?article=13286" target="_new">http://www.theinquirer.net/?article=13286</A>

    Oh well, there go Intel's Xeon/Gallatin margins.

    Quote:
    Actualie the increase in L2 cache for desktop use have done alomst nothing the 3000+ is so close to the 3200+. Nothing yield good imprvement except lower lantency.

    And higher bandwidth.

    Quote:
    The core was not rework there more change in Presscott that in K7-K8 you know i just finish me 1 64 bit register 64 bit add.

    Oh yes, the core was massively reworked. This confirmed by AMD engineers several months back. The overall core layout looks the same from a bird's-eye view, but that's only because the thing was reworked in a piecemeal fashion.

    Quote:
    Still they are faster and much less issue on intel south bridge and RAID will goes faster.Useful for small server like CSA.

    Intel's desktop southbridges are more-or-less stuck at the same performance level as everyone else. 32-bit/33MHz PCI, confined to ~90MB/sec due to fair arbitration and bus contention.

    Meanwhile Intel's server chipsets are busy going head-to-head with AMD kit. All things considered, they're about equal. Intel might have an edge with CSA--except that CSA doesn't necessarily do well for enterprise features like adapter teaming and hot-swap NICs. Add to that, CSA is already outpaced by 64-bit/66MHz PCI.

    <i><Lionel Hutz> I'll be defending...The SCO Group!!!??? Even if I lose, I'll be famous!</i>
Ask a new question

Read More

CPUs Font AMD