Page 4 of 5 FirstFirst 1 2 3 4 5 LastLast
Results 46 to 60 of 67

Thread: Intel PR berating AMD PR rating

  1. #46
    Member wallie_x's Avatar
    Join Date
    Nov 2001
    Location
    Central Calif. USA
    Posts
    444
    Ah, come on P4, you know I'm waiting, as everyone, to see the performance the P4 can do once it's coupled with the architecture it was designed for. And AMD's secretive "Hammer" ploy makes me even wonder if AMD can compete with Intel. Time will tell. But at least in the mean time they got people like us speculating (and equivocating in terms of MHZ) who’s best.

  2. #47
    Gone
    Join Date
    Dec 2001
    Posts
    5,713
    wallie_x

    True. But I think the Hammer is gonna be pretty good. If it is half what they claim I want one. But then maybe they are doing a bit of hype themselves? As you said time will tell.

  3. #48
    Junior Member
    Join Date
    Mar 2002
    Posts
    8

    Intel MGZ scaling

    P4 Gamer..

    Wuh?!?

    Dude, stop doing crack. I have no idea what you just wrote (referring to the previous post about Intel P4), but I think you are basically saying that Intel's scalability is going to be far superior to that of AMD? If that's true, then AMD is still using the old architechture, and since by your own admission, AMD's new Clawhammer/Thunderhammer will be better than Intel, that would surmise that since AMD still has yet to unveil their new technology, how do you know its not going to blow the socks off Intel? AMD is managing to beat Intel at their own game, that's what this boils down to. They are more efficient, and I don't care if you do own your own oil well, making an engine perform, and at a much better efficiency, is better than just wasting fuel any day. Ask any engineer.. or race driver.. An engine that manages to eek out about 400+ hp and get about 2 miles to the gallon, isn't as good as getting an engine to perform at 350 hp more efficiently.

    IT isn't always pure hp, if the engine is doing it at the top of its RPM range. There is a little thing called torque. So just rating an engine or cpu purely by its hp rating, is a farse to be sure. Its still about efficiency. AMD is able to utilize more work, at a given RPM because its more efficient (less is more). Intel does 2 instructions per cycle. AMD does 3 instructions per cycle. So go ahead, waste that fuel, bad boy.. meanwhile, after the race is on for about an hour, where are you? Not only are you eating my dust, but you are stuck out in the desert somewhere, because you are out of gas!

    I like your comparison to Ford vs Toyota.. That's like comparing a vintage wine with some cheap Publix brand. A Ford mustang is a classic. Its in a class all by itself.. Toyota (key part of that being T-O-Y) that's the definitive rice burner, if I ever saw one... hahaha..


    There is another saying, that "racing is 10% car, and 90% driver." In this case, the car being the CPU and the driver being the OS. Speed isn't worth a flip, if you are all over the road and burning your tires. Just like golf, I don't care if you can hit the ball harder than Tiger Woods, if it aint straight, what good is it?

    Anyway, like you said, you are entitled to your own opinion
    Last edited by Hotlanta; 04-02-2002 at 07:12 AM.

  4. #49
    Ultimate Member Rugor's Avatar
    Join Date
    Dec 2001
    Location
    Pacific Northwest, Earth
    Posts
    2,694
    As to scalability, yes the P4 scales better than the Athlon, the question is how much better.

    On the .18micron process the P4 topped out at 2.0GHz, the Athlon XP is currently at 1.73GHz. That means AMD is currently able to scale to 85% of Intel on the same process. Intel processors above 2.0GHz are all on the .13micron process. When the Athlon migrates to .13micron it too will scale higher than it currently does. If it continues to scale to 85% of Intel's speeds it will remain very competitive, at least until it runs into bandwidth limitations.

    The only real advantage P4 has is its 3.2GB/sec FSB, which is soon to scale to 4.2GB/sec. It may waste that bandwidth, but it does have it. AMD should be able to keep hot on Intel's heels for the near future at least. However, by continuing to raise IPC more than clockspeed, which seems to be the case with Hammer, AMD is going to be able to sidestep some of the bandwidth limitations through keeping clockspeed down.

    Remember a 1.5GHz P4 on the current quad-pumped 100MHz FSB has the same bandwidth per clock tick as the 2.0GHz will on the upcoming quad-pumped 133MHz bus. So the faster the chips scale the more they will starve for bandwidth, since the FSB is a hard limit for CPU bandwidth. What's going to happen when AMD has more bandwidth per clock tick, given the P4's insatiable appetite for bandwidth. Admittedly AMD would have to raise IPC significantly for that to happen, but if the P4 were to scale even to 4GHz on the current or projected platforms it would have huge problems.

    Take a look at this table:

    P4 1.5GHz, 3.2GB/sec bandwidth, 2.1B/tick
    P4 1.8Ghz, 3.2GB/sec bandwidth, 1.8B/tick
    P4 2.0GHz, 3.2GB/sec bandwidth, 1.6B/tick
    P4 2.2GHz, 3.2GB/sec bandwidth, 1.5B/tick
    P4 2.4GHz, 3.2GB/sec bandwidth, 1.3B/tick

    XP 1.53GHz, 2.1GB/sec bandwidth, 1.4B/tick
    XP 1.60GHz, 2.1GB/sec bandwidth, 1.3B/tick
    XP 1.67GHz, 2.1GB/sec bandwidth, 1.3B/tick
    XP 1.73GHz, 2.1GB/sec bandwidth, 1.2B/tick

    As you can see as clock speeds increase the differential in bandwidth decreases, so Intel really needs the faster FSB.

    Hope this gives people something to think about.

    PS: and yes I rounded my numbers off, but it should still illustrate the point.

  5. #50
    Ultimate Member araaraara's Avatar
    Join Date
    Oct 2000
    Location
    Canada
    Posts
    1,007
    The Athlon may not appear to scale as well as the P4, but that's because it is a much older chip that was originally a P3 class chip. If we were to say that the max speed for the Athlon is 2.0ghz, which is probably is, and we divide it by its original speed of 500mhz, the result is that the Athlon has scaled to 4X its original speed. Now, if we multiply the P4s original speed by 4, we get 5.2ghz, which is what speed the P4 will have to reach before it has scaled to the same extent as the Athlon has. Currently, the P4 has only scaled to less than half of what the Athlon has.

    I think that Intel has done a pretty bad job of implementing the P4's scalability. I mean, it required a new socket with extra voltage pins in order to get it past 2.0ghz. So, to go up from 1.3ghz to 2.0ghz (700mhz), it required an a change to its platform. That doesn't seem right for a chip designed with the future in mind and speeds upto 10ghz. I mean, is Intel going to come out with a new socket with another volt or two every 1ghz just to get the P4 upto 10ghz? I'm pretty sure that AMD could get the Athlon up to 4ghz if they came out with a new socket that gave out 4.0v or something. Also, the Athlon has managed it's large increase in clock speed using the same socket as it originally had. AMD hasn't made any changes whatsoever to the Athlon's socket462, and yet it has managed to make almost a 4X increase in clock speed (1.5ghz). To me, that sure sounds like the Athlon has much better scalability that he P4.


    About AMD's PR system. Well, I think that it is actually comparing the AthlonXP to an Athlon TBird and the P4 at the same time. Why? Because the Athlon TB was even with the P4 when it came out. So, an AthlonXP 1.4ghz (1600) = AthlonTB 1.6ghz (=P4 1.6ghz), so then an AthlonXP 1.4ghz (1600) = P4 1.6hz. Right?
    I know that the above formula is changing now with the P4 Northwood and its 512kb of L2 cache. This leads into my next topic.


    I think that Intel has really messed up on trying to increase the P4s lackluster performance. Instead of really trying to work on the issues which are holding the P4's performance back, they went ahead and did the easy thing: double the cache and hope that it doubles the performance. Well, unfortunatly it didn't. The performance increases between the Williamete? and Northwood cores is much less than would be expected from doubling the L2 cache. Obviously Intel knows this, and is hoping that its "more Mhz and more cache is better" marketing scheme will continue to get "uneducated" PC buyers to choose the P4 over the Athlon. Yet, the Athlon still manages to keep up with the P4, despite its seemingly "infererior" specs. Read this article to get an understanding of the problems with the P4 which were present from the day it was made.
    Last edited by araaraara; 04-03-2002 at 04:20 AM.

  6. #51
    Gone
    Join Date
    Dec 2001
    Posts
    5,713
    All very good points araaraara & Rugor.

    As far as Hotlanta, I think you need to chill man. Don't get so hostile. And what does tourqe have to do with it? It was a analogy dont tear it apart for flaws. I was just makeing a point. As far as the Hammer yes I said it was sounding like a good chip! I like AMD, that is nothing new why make it sound like I'm anti-amd or something. Chill.

  7. #52
    Gone
    Join Date
    Dec 2001
    Posts
    5,713

    Rugor

    Remember talking about the whole ram deal? You corrected me saying that the P4 wasnt a 64 bit proc. You said they were all 32 since the 486. Well I just started takeing my A+ course and they say different.It says the Pentium was the first 64bit chip and the first 32 bit was the 386!

    Here is from the book

    The Pentium represents yet another major advancement, containing more than 3 million transistors and a 64-bit data bus. The initial version was designed to have an operating speed of 66 MHz. The standard package is a 273- pin PGA. Microcomputers based on this chip will, for the first time, be placed in the mainframe category of performance.

    This throws me off though(mind you the book is a little out dated)

    " The newest microprocessor chip in the Intel family is the Merced chip. This 64-bit microprocessor is the first of Intel’s IA-64 series and promises to make computing even faster and better than the latest Pentium chip."

    Was that the code name for the P3? And what does IA-64 mean dif from being a 64 bit cpu? Help I'm confused?

    Sorry i know this is off subject but I saw you post here. I'll start a new thread if you want!

    If anybody else knows the dif. help as I'm getting confused on this whole 64bit vs. IA 64????

  8. #53
    Member rlbogue's Avatar
    Join Date
    Mar 2002
    Location
    Indianapolis, IN
    Posts
    54

    Re: Rugor

    Originally posted by $1500-P4 gamer
    If anybody else knows the dif. help as I'm getting confused on this whole 64bit vs. IA 64????
    Let me ask the million dollar question. Why do you care?

    There are so many ways to measure the "width" of a processor. There's the obvious ... ruler. Or you can look at the internal registers of the processor, or the way it addresses memory (the size of the memory WORD), etc.

    There are some implications of wider registers to some people who write in assembly and write compilers but most of the rest of the sane universe doesn't truly care.

    The whole "64-bit" processor discussion is a great one for the people at Microsoft who write operating systems and the people at Intel who want to say that they've done it. For most consumers the difference is indistinguishable.

    Rob

  9. #54
    Member
    Join Date
    Apr 2001
    Posts
    396
    The IA-64 is not the same as the P3 - it is the Itanium, and is currently Intel's server-sider CPU, like the Xeon was in the past. Also, while it is true that the Pentium has a 64-bit data bus, just like the Pentium 2 and 3, all of the Pentium processors only have 32-bit registers, which limit the CPU's processing power. It's like having 4 doors to let people in at Ticketmaster, but only 2 cashiers to help the people coming in.

    Here's a breakdown of some CPU stats regarding how many bits a CPU can handle:

    CPU Registers Data Bus
    8088/80188/
    8086/80186/ 16-bit 8-bit
    80286

    80386/80486 32-bit 32-bit

    Pentium family 32-bit 64-bit

    Sorry I don't have any info for AMD chips, but when I took this class, nobody at my college cared about AMD But I suspect that the Athlon family has the same registers and data bus widths as the Pentiums

  10. #55
    Ultimate Member Rugor's Avatar
    Join Date
    Dec 2001
    Location
    Pacific Northwest, Earth
    Posts
    2,694

    Re: P4 Gamer, thanking those who already answered

    I guess the easiest way to explain it is to say that the bus is not the processor.

    The Pentium sits on a 64bit data bus, allowing for 64 bit data transfers to and from the northbridge and main memory. However, the processor itself is limited to operating in 32bit mode which limits the data size on the internal registers and the amount of virtual or physical memory the chip can address. If you can't describe it with 32 bits of data the chip can't utilize it.

    The internal registers are a limit to the x86 ISA which basically describes how a processor handles data internally, how many registers there are, what the fundamental intstructions are, etc. It is a CISC rather than an RISC processor design, Complex rather than Reduced Instruction Set. The x86 instruction set is what the processor works with, and both Intel and AMD processors are designed around it. If you made a processor in any other fashion everything would have to be recompiled to run on it. Windows is designed to run on x86, the classic Mac OS is not.

    The current x86 ISA is 32bit, but it was originally designed for 8-bit computing and has been extended twice, first to 16 and now to 32 bits. This extension promotes backwards compatibility, and explains how a P4 can still run 16 bit DOS games. However there are much more efficient ways to do things than x86, but if those were to be implemented we would lose backwards compatibility for legacy apps and have to run them in emulation.

    Itanium is Intel's 64 bit processor, it uses 64-bit internal as well as external datapaths. It has a completely new architecture called EPIC, designed for massive parallelism and with none of the baggage of the x86 ISA. But it runs legacy code very slowly, and in order to get the most use of it all software would have to be written specially for it.

    AMD is working on a 64-bit processor too, Hammer. The Hammer uses a 64-bit extension to the current x86 ISA, called IA-64. Now Hammer may not have the theoretical performance of Itanium with native 64 bit apps, but it has much better performance with legacy software which it can run natively. It still has the baggage of the x86 ISA, which EPIC sidesteps, but it allows an evolutionary upgrade path. Intel is rumored to be working on a project called "Yamhill" which is its own processor using AMD's IA-64 extension to the x86 ISA.

    Confused now?

  11. #56
    Gone
    Join Date
    Dec 2001
    Posts
    5,713

    Getting the picture!

    No I'm not confused now. Thanx rugor,gfunkmartin and rlbogue.

    I see what you mean now. Basically in short explanation it comes down to this. Ever since the Pentium,the cpu has had a 64bit bus bandwidth. But internally the cpu is running at 32bit unless it is the Hammer or Itanium which is 64bit>64bit. Thanx guys. I was getting a headache. I kept reading it over and over but nothing came of it. God I love sysopt! Thanx guys once again for the replys.

  12. #57
    Gone
    Join Date
    Dec 2001
    Posts
    5,713

    gfunkmartin

    "Here's a breakdown of some CPU stats regarding how many bits a CPU can handle:

    CPU Registers Data Bus
    8088/80188/
    8086/80186/ 16-bit 8-bit
    80286

    80386/80486 32-bit 32-bit

    Pentium family 32-bit 64-bit "


    So the pentium family can handle 4 words of data in cpu simultaniously. But has 8 words ready on the bus. The IA 64 is 8 words of data on bus and in cpu. I gotcha! There is a big difference between the two. Thanx!

  13. #58
    Gone
    Join Date
    Dec 2001
    Posts
    5,713

    Re: Re: Rugor

    Originally posted by rlbogue


    Let me ask the million dollar question. Why do you care?

    There are so many ways to measure the "width" of a processor. There's the obvious ... ruler. Or you can look at the internal registers of the processor, or the way it addresses memory (the size of the memory WORD), etc.

    There are some implications of wider registers to some people who write in assembly and write compilers but most of the rest of the sane universe doesn't truly care.

    The whole "64-bit" processor discussion is a great one for the people at Microsoft who write operating systems and the people at Intel who want to say that they've done it. For most consumers the difference is indistinguishable.

    Rob
    "the size of the memory WORD"

    I'm confused? Size of mem word. All WORDs are 8 bit data I didnt know that can differ. Do you mean the size of mem cell. I know a mem cell has 1 bit data in it. Millions of mem cells in one mem chip. Can you further explain what you mean?

    As far as why I care. I tend to over study. I love pc's like I love electronics. So I tend to pick the info apart and analyze everything. This threw me off(64bit) so I wanted to understand before it snowballed on me makeing me further miss informed. I got a 98% on the test but I felt I needed to understand this. I know,why do you need to know that to fix a PC. I dont but I like to pass with flying colors on my A+ certification. They got me on a trick question those ------. They asked How many bits are in a Kilobyte. A.1 B.8 C.1024 D.8196 E. none of the above. Dummy me got suckered and choose c.1024. Wrong it is bits not bytes they asked about. So it was supposed to be E. I fudged oh well.

  14. #59
    Member rlbogue's Avatar
    Join Date
    Mar 2002
    Location
    Indianapolis, IN
    Posts
    54

    Re: Re: Re: Rugor

    Originally posted by $1500-P4 gamer

    I'm confused? Size of mem word. All WORDs are 8 bit data I didnt know that can differ. Do you mean the size of mem cell. I know a mem cell has 1 bit data in it. Millions of mem cells in one mem chip. Can you further explain what you mean?
    A byte is 8 bits. that's standard. However a word, or more technically data word, is the length of bits that the processor operates on simultaneously. However, because programmers started using this when a word was 16 bits, we've fallen into the incorrect usage of referring to a word as 16 bits.

    Generally we assume a double word is 32 bits and that a quad word is 64 bits. Confused yet?

    However, if we get away from the mislabeling that programmers had to get into to abstract the lower level hardware from their programs a word is how many bits the processor operates on at one time. This is typically the size of the registers built into the processor -- although theoretically it doesn't have to be.

    Did I just muddy the water or did I make it better?

    Rob

  15. #60
    Member
    Join Date
    Apr 2001
    Posts
    396
    I'm following you rl...(Thank GAWD for that assembly class I took )

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •