Intel’s Pentium 4 3.2GHz processor

IT’S TIME ONCE AGAIN for the drip-drip-drip progression foretold by Moore’s law to release another drop. After a breathtaking flurry of new chipset releases, including the 875P and 865 family, and after backfilling its processor line to include Hyper-Threading and 800MHz bus support across all its speed grades, Intel is ready to move its Pentium 4’s top clock speed up a notch, to 3.2GHz. Perhaps you’re thinking it’s a little too soon for yet another Pentium 4 upgrade, but in truth, Intel hasn’t ratcheted up the P4’s top clock speed since last November, when the Pentium 4 3.06GHz debuted as the first P4 with Hyper-Threading support.

Since then, as in Dick Gephardt’s campaign headquarters, all the activity has been elsewhere. Intel has upgraded its lineup of Pentium 4 processors and chipsets with an 800MHz bus, dual-channel DDR400 memory, ubiquitous Hyper-Threading, AGP 8X, and Serial ATA—to name just some of the improvements. The Pentium 4 platform practically pulses with bandwidth everywhere, and performance is up as a result.

We found the Pentium 4 3.0GHz chip to be a little bit faster overall than AMD’s latest, the Athlon XP 3200+, in our last round of tests. Still, with a new 400MHz front-side bus and its own dual-DDR400 chipset in the nForce2 Ultra 400, the Athlon XP 3200+ is no slouch. The Athlon turned in the highest scores in many tests, and put up a heck of a fight for the overall crown.

Now we come to the new 3.2GHz version of the Pentium 4. Suppose with me, if you will, what might happen when the manufacturer of the world’s fastest desktop processor turns up the clock speed from 3000MHz to 3200MHz. Lower interest rates? The scent of almond? (What the devil does one say about 200 more megahertz?)

Uhm, sorry about that. As I was saying, we’re expecting the Pentium 4 3.2GHz to take its rightful place at the top of the x86 pecking order. The P4 3.2GHz may be more of the same, but like faithful patrons of the local Luby’s, we’re generally in favor of getting more of a good thing. As always, we’ve loaded up our test bench with a gaggle of the new P4’s competitors and forebears, and the results follow, so read on.


The unassuming Pentium 4 3.2GHz processor

A big, greasy caveat
Some of the numbers you’re about to see below are a little funky. Actually, the numbers are fine, but I’ve kind of forced them into our scheme, so I’d best explain myself. You see, I’ve graphed some of the benchmark results with line graphs, as we have done in our graphics card reviews in the past. The idea in doing so is to show CPU scaling at different clock speeds and with different processor revisions. For the Intel chips, I have two groups, the older Pentium 4 chips with 533MHz front-side busses, and the new ones with the 800MHz bus. I’ve matched up the Athlon XP’s model numbers to the Pentium 4’s clock speeds for the sake of comparison, too, so we have three series on each line graph.

One of those series, though, includes a couple of P4 chips that don’t quite line up. For the “2600” speed/model number, I’ve included the P4 2.53GHz, and for the “3000” speed/model number, I’ve included the P4 3.06GHz. In both cases, these chips are 66MHz off the expected speed—the slower one is under, and the fast over. As a result, the blue line on our CPU scaling graphs, which represents 533MHz-bus Pentium 4 chips, tends to tilt a little more than it should in a perfect comparative scenario. Keep that in mind when you’re looking at how these older P4 chips scale with clock speed.

Our testing methods
As ever, we did our best to deliver clean benchmark numbers. Tests were run at least twice, and the results were averaged.

Our test systems were configured like so:

Athlon XP 3200+ Athlon XP 2500-3000+ Pentium 4 2.53-3.06GHz Pentium 4 2.4-3.0GHz
Processor Athlon XP ‘Barton’ 3200+ 2.2GHz Athlon XP ‘Thoroughbred’ 2600+ 2.083GHz
Athlon XP ‘Barton’ 2500+ 1.83GHz
Athlon XP ‘Barton’ 2800+ 2.083GHz
Athlon XP ‘Barton’ 3000+ 2.166GHz
Pentium 4 2.53GHz
Pentium 4 2.8GHz
Pentium 4 3.06GHz
Pentium 4 ‘C’ 2.4GHz
Pentium 4 ‘C’ 2.6GHz
Pentium 4 ‘C’ 2.8GHz
Pentium 4 3.0GHz
Pentium 4 3.2GHz
Front-side bus 400MHz (200MHz DDR) 333MHz (166MHz DDR) 533MHz (133MHz quad-pumped) 800MHz (200MHz quad-pumped)
Motherboard Asus A7N8X Deluxe v2.0 Asus A7N8X Deluxe v2.0 Aopen AX4R Plus Intel D875PBZ
North bridge nForce2 SPP nForce2 SPP 82845G MCH 82875P MCH
South bridge nForce2 MCP-T nForce2 MCP-T 82801DB ICH4 82801ER ICH5R
Chipset drivers 2.03 2.03 Intel Application Accelerator 2.3 Intel Application Accelerator for RAID 3.0
Memory size 512MB (2 DIMMs) 512MB (2 DIMMs) 512MB (2 DIMMs) 512MB (2 DIMMs)
Memory type Corsair TwinX XMS3200LL DDR SDRAM at 400MHz Corsair TwinX XMS3200LL DDR SDRAM at 333MHz Corsair TwinX XMS3200LL DDR SDRAM at 266MHz Kingmax DDR-400 SDRAM at 400MHz
Hard drive Seagate Barracuda V 120GB ATA/100 Seagate Barracuda V 120GB ATA/100 Seagate Barracuda V 120GB ATA/100 Seagate Barracuda V 120GB SATA 150
Graphics ATI Radeon 9700 Pro 128MB (7.84 drivers)
Sound Creative SoundBlaster Live!
OS Microsoft Windows XP Professional
OS updates Service Pack 1, DirectX 9

All tests on the Pentium 4 ‘C’, 3.0GHz, 3.06GHz, and 3.2GHz systems were run with Hyper-Threading enabled. The other Pentium 4 chips tested here don’t support Hyper-Threading.

Thanks to Corsair for providing us with memory for our testing. If you’re looking to tweak out your system to the max and maybe overclock it a little, Corsair’s RAM is definitely worth considering.

The test systems’ Windows desktops were set at 1024×768 in 32-bit color at an 85Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests.

We used the following versions of our test applications:

All the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.

Benchmark results

Memory performance
Just to satisfy that geeky urge, we generally kick things off with some synthetic memory bandwidth benchmarks. The release of the P4 3.2GHz gives the Pentium 4 yet another chance to show off one of its greatest strengths, as you’ll see below.

All of the chips you see bunched up at the top there with the really long bars are Pentium 4 chips with 800MHz front-side bus speeds. Those chips are all running on Intel’s 875P chipset with 6.4GB/s of bus and memory bandwidth available to them. (As you can see, the lower speed grades with 800MHz bus support are given a “C” tag to distinguish them from older P4 chips.) The older Pentium 4 chips still have a faster front-side bus than any of the Athlon XP chips, so they come ahead of the Athlons.

By popular demand, fancy-looking Linpack graphs are back in force. This one shows the entire memory hierarchy, save storage devices. As the matrix size grows, Linpack’s floating-point math calculations have to be performed in L1 cache, then L2 cache, then in main memory. You can see how performance drops as we move from cache into main memory. For certain types of FP math calculations with really large datasets, the Pentium 4 is very tough to beat because of its exceptional memory bandwidth.

Cachemem tells a similar story with a little more granularity than Sandra, showing us read speed, write speed, and memory access latency. All of these systems have very low latency memory subsystems, relatively speaking.

Now we’ll bust out the freaky 3D graphs to get a closer look at the output from Cachemem’s latency tests. To keep things interesting, I’ve colored the different block sizes depending on whether they fit into a processor’s cache. Yellow is L1 cache; light orange is L2 cache; and dark orange is main memory.

You can see how the 800MHz bus benefits the Pentium 4 3.0GHz and 3.2GHz chips versus the P4 3.06GHz on a 533MHz bus. Despite its relatively slower front-side bus clock speed, the Athlon XP is also very quick getting to memory—especially the 3200+ with a 400MHz bus.

Business Winstone

Despite all the P4’s advantages in memory bandwidth and clock speed, the Athlon XP still rules this test. The “Barton” chips with 512K of L2 cache are especially fast here. The P4 3.2GHz just barely edges out the Barton-based Athlon XP 2500+. Content Creation Winstone

The results here are complicated by the fact that CCWS2002 performance suffers when Hyper-Threading is enabled. Generally, Hyper-Threading helps performance, and where it doesn’t, it usually doesn’t slow things down. However, Hyper-Threading does involve some overhead (managing two logical processors in the OS) and resource sharing (of L2 cache memory, L2 cache bandwidth, registers, and the like). Sometimes, that means lower performance, which is the case here. You can see the drop in the blue line from the P4 2.8GHz (without HT) to the P4 3.06GHz (with HT). I expect that with Hyper-Threading disabled, the P4 3.2GHz would be beating out the Athlon XP 3200+, but we prefer to test the way we’d want to use the processor, and we’d definitely want to use HT-ready P4s with Hyper-Threading enabled.

Incidentally, we mentioned last time out that we were considering whether to include results for the new 2003 version of Content Creation Winstone, which doesn’t suffer from the same Hyper-Threading performance problems. After mulling it over a while, we decided to start using this test, possibly in conjunction with the 2002 edition for a while. However, time constraints prevented us from including Multimedia Content Creation Winstone 2003 among our results this time around. We’ll try to include it next time out.

LAME MP3 encoding
We used LAME 3.92 to encode a 101MB 16-bit, 44KHz audio file into a very high-quality MP3. The exact command-line options we used were:

lame –alt-preset extreme file.wav file.mp3

Unfortunately, LAME isn’t multithreaded, so Hyper-Threading probably won’t help.

The Pentium 4’s traditional strength in media encoding tasks continues here. You can see how clock speeds matter more than anything else on the Pentium 4 chips. The Athlon XP, which has only received modest clock speed boosts in the past nine months, doesn’t grow much faster as model numbers ramp up. DivX video encoding

Video encoding tasks are especially well accelerated by the Pentium 4. The DivX encoder loves the P4’s SSE2 instructions, and Hyper-Threading is a big gain here, too. Of course, memory bandwidth plays a big role, as well. As a result, the Athlon XP 3200+ needs 66 seconds more in order to encode the same video clip as the Pentium 4 3.2GHz. Speech recognition
Sphinx is a high-quality speech recognition routine that needs the latest computer hardware to run at speeds close to real-time processing. We use two different versions, built with two different compilers, in an attempt to ensure we’re getting the best possible performance.

There are two goals with Sphinx. The first is to run it faster than real time, so real-time speech recognition is possible. The second, more ambitious goal is to run it at about 0.8 times real time, where additional CPU overhead is available for other sorts of processing, enabling Sphinx-driven real-time applications.

Obviously, the new Pentium 4 chips are all very well suited for Sphinx speech recognition. The Athlon XP 3200+ can’t even touch the Pentium 4 “C” 2.4GHz.

Cinebench 2003 lighting and rendering
Cinebench is based on Maxon’s Cinema 4D modeling, rendering, and animation app. This new revision of Cinebench measures performance in a number of ways, including 3D rendering, software shading, and OpenGL shading with and without hardware acceleration.

Cinebench 2003 is multithreaded, so it takes advantage of Hyper-Threading. For the P4 ‘C’, 3.2GHz, 3.06GHz, and 3.0GHz systems, I’ve reported the multithreaded rendering test result, which was always better than the single-threaded result. For the Athlon XP and the older P4 chips, I’ve reported the single-CPU result.

The Pentium 4 3.2GHz takes the top spot once more. With Hyper-Threading and higher clock speeds, the Intel chips are opening up a widening lead over the AMD chips in this test.

The Pentium 4 3.2GHz takes two out of three of the remaining Cinebench tests. However, the Athlon XP 3200+ puts up more of a fight, staying near the top of the pack in the software shading tests and achieving the highest score in the hardware-accelerated shading test.

Quake III Arena

The P4 3.2GHz shatters the 400 fps barrier in Quake III Arena, and all but one of the Pentium 4 chips beat out the Athlon XP 3200+. 3DMark03

3DMark03 is obviously bound by graphics card performance more than anything else. The P4 3.2GHz is fastest here, but by a mighty narrow margin.

The pack separates nicely in 3DMark03’s CPU tests, as one might hope. When that happens, the Pentium 4 processors come out on top of the AMD chips—but only with the faster 800MHz bus. Serious Sam SE

After months of humiliation at the hands of the Athlon XP in Serious Sam SE, the Pentium 4 3.2GHz finally manages to take the top spot here. This one used to be the Athlon XP’s home turf, but no more.

Comanche 4

Comanche 4 continues to be a tough one, but the P4 3.2GHz’s extra 200MHz helps it eke out another 3 frames per second or so. Once again, the Athlon XP is left in the dust. Unreal Tournament 2003

The Pentium 4 scales up quite linearly with clock speed in UT2003, and the P4’s new 800MHz bus again gives it the edge over the Athlon XP.

SPECviewperf workstation graphics
SPECviewperf simulates the graphics loads generated by various professional design, modeling, and engineering applications.

The viewperf tests are a clean sweep for the P4 3.2GHz. The Athlon XP 3200+ used to rule the DX test, but the P4’s new speed bump just took that away.

Conclusions
Having just written this review in one sitting with several pints of caffeine coursing through my veins, I have a hazy recollection of having just described a series of very bad things happening to an Athlon XP processor at the hands of Intel’s newest Pentium 4. Not pretty.

Clearly, Intel has the upper hand right now in desktop processors, and the situation doesn’t seem likely to change until, perhaps, when AMD delivers its new Athlon 64 processor this fall. I’m also struck—again—by how well the lower speed grades of the Pentium 4 perform now that they have a faster bus and Hyper-Threading support. As always, the top-of-the-line Pentium 4 3.2GHz is going to be very expensive for a little while, and the slower variants will be more affordable. AMD may be able to remain competitive for the next few months by undercutting Intel’s prices.

Right now, the Pentium 4 3.2GHz doesn’t look like much of a value. Ok, that’s a bit of an understatement. Preliminary listings at online vendors currently show the P4 3.2GHz somewhere north of $700. When I first saw that price, something pulmonary went wrong for a split second there. By contrast, the Pentium 4 3.0GHz is hovering around $400, and the Athlon XP 3200+ show up at about $440. I’ll let you decide whether having the absolute fastest chip available is worth paying the premium.

I also reserve the right to mock you for paying $300 for an extra 200MHz.

Those prices indicate a bit of trouble for AMD. The Pentium 4 2.8GHz generally matches or outperforms the Athlon XP 3200+, and the P4 2.8GHz is selling for around $275, or over $160 less than the 3200+. Unless AMD recognizes that its ratings system has been blown up by advances in Pentium 4 clock-for-clock performance and lowers prices accordingly, the question for many PC enthusiasts may no longer be whether to buy an Intel or AMD, but simply which Pentium 4 chip to buy.

Comments closed
    • Anonymous
    • 16 years ago

    The only diffrents Prescott from Northwood C —-> is larger L1 + L2 K memory, 13 more instructions and NetBrust Tech.

    So, I think Prescott 3.4GHz will match 3.8GHz Northwood C if existed.

    • Anonymous
    • 16 years ago

    We are reviewing the fastest CPU from Intel and AMD. If you are interested in PPV, you can write another article. I’ll be the first one to review it.

    • atidriverssuck
    • 16 years ago

    nice for the stinking rich. Athlon XP 2500+ for the real world.

    Honestly, I’m still amazed they can command such prices for such a little difference. You’d think people have gotten over having the fastest chip on the block. I guess there always is that small percentage that buys as soon as a new chip is out, regardless…

      • Krogoth
      • 16 years ago

      Well, Intel used to charge that much or even more for a small increase in speed for their from mid-range to top-end processors back in their day with their near-monopoly position on x86 platform during (l990-1998) of course before the RAMBUST/P3 fisco dealing with AMD’s Althon K7/Thunderbirds

      • Yahoolian
      • 16 years ago

      Dimishing returns.

    • Anonymous
    • 16 years ago

    You forgot one very important benchmark. The PPV benchmark. Y’know the Price+Performance+Value one. Ok, so it’s not put out by either a software OR a hardware company, but it’s the one that matters the most to my mind. It’s rather simple actually, in that it’s based on the (Closest) $100 per component rule. (MB/CPU/RAM) (Avoid direct pricewatch pricing, as it can be deceptive…) Just take $300 ($100 * 3 components) and then compare the performance between the setups. (IF you really need stuff like RAID on your MB, just normalize the price between each setup and ignore the extra cost…)

    So using this system (And a price sheet from my closest reputable, local Computer parts dealer….) I find that I will be comparing the performance of: (Assuming Non-Rambus memory for all the P4’s)

    [AMD]
    Athlon XP 2500+(Barton) $115
    MSI KT4-Ultra $105
    512 MB DDR-333 PC2700 $80
    Total cost ($300)

    [Intel]
    Celeron 2.4Ghz $110
    MSI 845PE Max $105
    512 MB DDR-333 PC2700 $80
    Total Cost ($295)

    Now given the Intel ‘competition’…do I really need to even do *any* benchmarks in order to decide where to spend my money? (Next pricier Intel is the Pentium 4 1.8A (400MhzFSB) @ $170!!! While the next pricier AMD is a XP 2600+(333FSB-Non-Barton) @ $125…although the XP2700+ comes in at the P4’s $170 price point.)

    Now extend that up comparison up through all the CPU prices, and tell me: *How* much more do I have to spend until the Intel performance is greater than AMD’s.

    (Although I can say that point is probably way above my *desire* to spend the needed amount of $…)

      • Anonymous
      • 16 years ago

      Well, nobody was reviewing a 2500+ or a celeron here IIRC. There is little (if any) doubt about the good price/performance ratio of AMD in the low end, but when talking about high end, you don’t expect to pay $300 for a machine 😉 .
      PS. I don’t condone Intel policy of high prices. I only happen to realise that they have never been the best bang for the buck.

        • Anonymous
        • 16 years ago

        Well no, I don’t expect to pay that much for a top end machine, although I would *like* to…

        BUT…How can you call the 2500+ the LOW END?!?!?!

      • Anonymous
      • 16 years ago

      Maybe, but since the P4 2.4C lists for $178 and is $176 at NewEgg, buying a slower P4 is silly. Flipping over to the AMD side, it looks like the price-comprable Athlon is the 2800+. Looking at the numbers in the P4 3.2 review, it seems reasonable for those two CPUs to have a similar price, as they seem to have similar performance.

    • Dposcorp
    • 16 years ago

    For all this talk of Intel being so great, just remember how many people they screwed since the original P4 line came out.

    First, Socket 423, 400Mhz FSB.
    Then, socket 478, 400Mhz FSB.
    Then, socket 478, 533Mhz FSB.
    Then, socket 478, 800Mhz FSB.
    I wont even mention the RAMBUS / DDR fiasco.
    (Has anyone else EVER bundled ram with a cpu?)

    It seems that everytime a person invested in Intel, they were left in the cold like 6 months later.

    I know that AMD has done some of the same things, but, at least the socket stayed the same, and the multiplier could be adjusted to compensate for the FSB change, so that even some of the newest CPUs could be made to work in older boards.

    Hell, you could even buy boards that let you use your old PC133, then put in DDR at a later date.

    I hesitant of investing in Intel since the P4 first came out, and I will keep my eye out for Athlon64.

      • Anonymous
      • 16 years ago

      Remember that the majority of PC users never upgrade anything on their system, and the overwhelming majority never upgrade their CPU. If you go more than three years between upgrades, this is reasonable anyway, because you probably want to upgrade all of your components.

      • Steel
      • 16 years ago

      q[

      • indeego
      • 16 years ago

      AMD low-mid range PC’s and motherboards are still cheaper than Intel low-mid-range.

      I know it’s not hip to ever buy less than the very top performance anymore, but when you can build entire mid-range systems, *[

      • Anonymous
      • 16 years ago

      The 423-478 was definately a fiasco. But if you bought a i850 or i850e motherboard your had the fastest performing chipset which too DDR a couple of years to catch up.

      Like anyone here gives a damn about socket changes. How many AMD enthusiasts are using a KT133 or chipset? Okay how bout 133A, 266, 760, 266A, 333, Nfarce1?

      Wow they are all on the same socket but who actually puts a XP 2400 or a 266A motherboard here? It’s a lame duck argument. Get over it, sockets will change usually for the better. Prescott will initially be produced for socket 478 and we wont see a socket change until next year.

        • Forge
        • 16 years ago

        Actually I know many folks with KT266A and KT333 mobos in use, with recent CPUs, and I can think of a few folks on AMD-760 (non-MP) as well..

        Even seen a Barton on KT333? I have. It ran nicely.

          • spike
          • 16 years ago

          So Forge are you saying that I would be able to use a 2400+ on my brothers Abit KR7A-Raid (it’s a 266A) motherboard, because he is stuck with Tbird 1400 and is dying to make an upgrade.

            • Forge
            • 16 years ago

            Yeah, no sweat. Unless Abit specifically locked it out in the BIOS, it’ll run fine. No 333 FSB support, so that limits your options a bit, but it should run the 2400+ and the 2600+ 266 version.

            • Anonymous
            • 16 years ago

            So, to expose my ignorance… I’ve got a KR7A as well. A 2400+ runs at 2.0 Ghz, which gives a multiplier of 15. Can this board do that? The Abit site only lists up to TBred 2000+…

            -Rick

            • Forge
            • 16 years ago

            Shouldn’t be any problem. You won’t get multiplier control, but it should boot and run fine.

          • Xylker
          • 16 years ago

          I actually have quite a bit of love for the 760 chipset. I am running an A7M266, a GA7DXR+ and a S2466-4M (OK, so that is not exactly 760…) And a couple NF2 systems. AND, I can swap CPU’s at any time so if there is ever a failed component, I can use any of the parts that I have lying about to troubleshoot.

      • Forge
      • 16 years ago

      You think that’s bad? ALL 478 is about to be put out to the curb. We’ll get one or two token Prescott releases as a ‘quit whining, look, we give you CPUs’, and then Intel will be releasing the exact same CPU on the exact same chipset, with a new socket, followed by turning around and cranking out Grantsdale/ICH6. Intel has no love for those with ‘old’ (>6 months) motherboards.

      On the other mitt, I booted and ran a Barton on my KT133A, just for kicks.

      • absinthexl
      • 16 years ago

      The 533 -> 800 jump isn’t necessarily screwing me over. I’m running on a P4Pe, which is perfectly capable of running at 800MHz with decent RAM. Only problem is single vs. dual-channel DDR.

      My old AMD board, on the other hand, is running with a 1.2GHz proc and isn’t getting an AthlonXP to work in it anytime soon. Thankfully, I didn’t get stuck with one of the awful Slot A boards.

      • Anonymous
      • 16 years ago

      Can you install an Athlon64 onto an AthlonXP board?

        • Krogoth
        • 16 years ago

        no, A64s will feature a completely diffierent socket design due A64 different architech design i.e (Intergrated Memory Controller) and power requirements so, simply put that it’s physically impossible to install a A64 to AXP Motherboard

      • Anonymous
      • 16 years ago

      Over the last 10 years, I’ve owned motherboards which had some upgrade headroom on the processor end of things. However, by the time it came for me to want to change my processor, I either found that the maximum processor the motherboard could take was less than what I wanted, or there was some sort of socket design change which made upgrading impossible.

      There is some light at the end of the tunnel this time around. I bought a 2.26GHz P4 with an Asus P4B533 motherboard a year ago right now. Lucky for me, I can still easily get 533MHz FSB processors for this motherboard, and at prices which aren’t too shabby (the rules of short supply have not come into effect yet). Although it’s not a priority at the moment, I would not mind getting rid of my 2.26GHz processor and replacing it with a 2.8GHz version. I think, however, this good fortune is because processors have been stuck at about 3GHz for quite some time.

      I still consider it to be a bonus if you can buy a motherboard and change the processor on it a year later.

    • eitje
    • 16 years ago

    q[

      • absinthexl
      • 16 years ago

      Stock speeds are for wusses.

    • meanfriend
    • 16 years ago

    /[

      • eitje
      • 16 years ago

      a luby’s is like a furr’s or a wyatt’s.
      §[<http://www.lubys.com/index.asp<]§

        • meanfriend
        • 16 years ago

        Well I’m in Canada so I have no idea what a furr’s or a wyatt’s is either…

        However, from your link, they are evidently a chain of restaurants 🙂

          • EasyRhino
          • 16 years ago

          NOT MERE RESTAURANTS!

          Lubys is the greatest chain cafeteria either.

          It’s mainly a South thing.

    • eitje
    • 16 years ago

    isn’t it odd that, in the Sphinx tests, the Intel compiler runs slower than the Microsoft compiler on the P4s, and the converse is true for the AMD CPUs?

    i think we need to have an expose on Intel optimizing their compiler for Athlons. 😉

      • Anonymous
      • 16 years ago

      These tests weren’t run on Unix; they were run on Windows. That MS’s compiler should be able to take advantage of Windows’ undisclosed optimizations, not to mention pre-loaded code, is a given. That Intel’s compiler can even manage to make a race out of it is to be commended, at the least.

      As Gates would say, “It’s good to be the king.”

      • Anonymous
      • 16 years ago

      Cough! [b]WHAT?![/b]
      Back in the day everyone was accusing MS of using undocumented calls in their applications to make faster. So you mean all those other application vendors would’ve had faster apps if they just compiled with the MS compiler? Wait a minute, they [b]were[/b] using the MS compiler. Let’s keep our conspiracy theories straight, shall we?

      Look, I’ve disassembled code emited by the MS compiler, and I haven’t seen any undocumented calls. I’ve seen some pretty fuky op combinations when optimizations were turned on, but that was doing non-inutitive things to make it go faster, not calling undocumented entrypoints. I could be wrong… but Intel has better engineers than me, and I’m sure they could find and exploit any undocumented calls for their Windows compiler.

      Then again MS has some pretty smart engineers, maybe they can hide those calls from me and the Intel engineers. Or maybe they spent that effort on making a better optimizing compiler. Hey, there’s a thought.

      • eitje
      • 16 years ago

      note that “converse” is a type of shoe, but in this context also means “reverse”.

    • wiper
    • 16 years ago

    I would just like to point out that AMD has consistantly said that the new pr system is not a comparison of their processors to Intel’s, but a comparison of performance to the tbird so I hardly think it’s fair to call this an apple’s to apple’s comparison in the sense that a 3200+ should perform on the level of a 3.2ghz p4 since that’s not what AMD is saying. Anyone who think’s that a p4 clocked 1ghz faster than an athlon should perform on the same level as eachother might have caught a case of RDF syndrom from a mac fan or somthing.
    I think it would have been more fair to compair ghz to ghz since AMD in no way implies thier pr system relates to p4’s and you can see this by looking at the 3200+ (2.2ghz) scores compaired to the 2.4ghz p4, their scores are much closer together.

      • Anonymous
      • 16 years ago

      Intel also claims that having a 20 stage pipeline increases performance (not frequency, /[

      • Anonymous
      • 16 years ago

      AMD can “claim” whatever they want. When Joe “I just want a computer so my kids can play games and the wife can work on her novel” walks into the computer store, how do you think he views the PR ratings? How do you think the computer salesman explains them? And don’t tell me it’s not AMD’s fault. They knew that this is exactly what would happen — they get to claim better performance than they actually have, without getting called on it because they have “explained” it to the experts.

      • Anonymous
      • 16 years ago

      Who cares about comparing Ghz to Ghz when you can’t get a Athlon that runs as high as a P4? Only AMD fanboys trying to make stupid points on tech boards..

        • wiper
        • 16 years ago

        q[

      • Dissonance
      • 16 years ago

      Comparing GHz-for-GHz would really only be an academic exercise. In the real world, AMD’s fastest (and most expensive) chip is the 3200+. Intel’s fastest is the 3.2GHz. Damage just compared the fastest processors for each platform, and results are also there for the performance of price-competitive processors.

      • Anonymous
      • 16 years ago

      Then please ask AMD to provide the author a 3.2GHz CPU for comparison. I’d like to review the result.

    • Hockster
    • 16 years ago

    Nice processor from Intel, and as usual, nice review from TR.

    I was actually having a quick play around on a 3.2GHz system yesterday funny enough. I was in a PC World store fairly near to where I live and I couldn’t believe that they actually had the 3.2GHz in a computer they were selling so soon! I was really shocked! The computer was an Advent computer I believe; although it might have been HP I can’t remember. Anyway the system was actually very well priced at first glance – £1300. In that price you got: 3.2GHz, the Sprindale board (not Canterwood), 512MB Dual Channel 333 Memory, and a GeForceFX 5600. Well, I’m a bit disgusted at the GPU myself, so I guess if you were to have a better GPU that would higher the price, but still: the price did look pretty good. Of course the system was obviously missing other things such as good speakers and was using on-board sound, but the average consumer doesn’t really care for those features.

    Anyway, I started to have a little play around with it. Of course, there weren’t any games or anything on the system: basically just what comes with Windows XP, lol. But still, after just clicking around the desktop and silly little stuff like roaming Windows, it did actually seem faster than my 2.53GHz Pentium 4. I didn’t expect to be able to see a difference simply by doing simple things around Windows – but I did. The biggest difference in speed that I noticed was launching the Pinball game that comes with XP – it loaded it a matter of seconds! Now I know the Pinball games is nothing amazing, but for some reason it takes a while to load on my 2.53GHz, and also on all the other computers I played with at the store. The 3.2GHz was definitely faster.

    Anyway, I just though I’d share that…I’m still surprised I actually saw the 3.2GHz in the store, even before places like TR got hold of a review unit.

      • Anonymous
      • 16 years ago

      Hard drive speed — what you saw with Pinball may have more to do with the speed of the hard drive than anything else. That would speed up a lot of other general Windows operations too. Hard to know without comparing specs.

        • Hockster
        • 16 years ago

        I guess so, although I didn’t think the hard drive was any faster than mine or any of the other ones that I tested. Then again, it was probably just that bit more optimised or something, but I still think the extra CPU speed must have somewhat helped it.

      • indeego
      • 16 years ago

      Are you kidding? Faster pinball? working around in the OS? muhahahahahag{

        • Forge
        • 16 years ago

        Funny, I’d think pairing a 200 FSB CPU with 166 MHz RAM would be a lame idea…..

        I also find it unusual that ANY OEM would incite Intel’s wrath by putting a CPU in a shipping box before Intel announced it.

          • Hockster
          • 16 years ago

          I found it unusual too, but I guess Intel ships their processors out a day or two before the official date for obvious reasons.

    • AmishRakeFight
    • 16 years ago

    the cool thing about that review and the graphs is I got to see how my Barton 2500 and my Thoroughbred 2600 actually compare to /[

    • Hattig
    • 16 years ago

    AMD is back to being a provider of excellent value budget processors.

    Sure, a 2000+ (or whatever) might only provide ~60% of the performance of the 3.2GHz P4, but it costs 1/10th of the price.

    Now if only there were dual-Athlon motherboards with modern features and faster bus speed support, available at a good price… no PCI-X or server features required. A couple of $100 Athlon processors and you’d have a great system with more power and for less money than a 3.2GHz P4.

    Because AMD are not going to be able to compete at the high-end with a single CPU solution. I doubt that the 3400+ Athlon 64 will compete well with a 3.2GHz P4, nevermind a 3.4GHz P4. let’s hope that AMD surprises us.

      • eitje
      • 16 years ago

      agreed.

    • house
    • 16 years ago

    These PR ratings are getting more and more ridiculous and if this continues, consumer mistrust is going to start to build. AMD needs to go back to it’s position it had in k-6 times and secure the budget end of the spectrum with the best valued chips out there. Stay low for now AMD and keep working away to put out another original Athlon for us ( I dont think the 64 will be the savior they are looking for.)

      • Anonymous
      • 16 years ago

      If they do that they are doomed. They aren’t making enough profit on the low-end chips to stay in business. They have to get an entry into the server / workstation market. “Another XP” is not going to do it. 64bits has to succeed for them — at reasonably [i]high prices[/i] — or AMD is not long for this world.

    • Anonymous
    • 16 years ago

    §[<http://biz.yahoo.com/rc/030622/tech_intel_1.html<]§ Intel said that its latest, and fastest, Pentium 4 will run at 3.2 gigahertz, or 3.2 billion cycles per second. With the hyper-threading technology, a PC with that chip can convert one minute of digital video to the MPEG 4 standard while also converting 26 minutes worth of music to the MP3 format more than four times faster than a Pentium III chip running at 1 gigahertz, Intel said. ----- Sounds like Intel is moving towards a TPI -like stance also. 4 Times more powerful than a chip that came out almost 4 years ago. A CPU that you can find on pricewatch for 90 bucks. Wow - doesn't technology move through your wallet fast?

    • droopy1592
    • 16 years ago

    I don’t think Hammer can stand up to this type of ass kicking if it doesn’t have the memory bandwidth use-ability like the P4’s do. They just keep getting better the more you feed them. I guess they keep guessing until they get it right… the chips, I mean.

    I’m not down for socket changes though..

    • Anonymous
    • 16 years ago

    *yawn*
    Wake me when Newegg is selling these for $160

      • Anonymous
      • 16 years ago

      They are not AMD you know… 😆

    • Zenith
    • 16 years ago

    AG#11 – Your being sarcastic…right?

    • Anonymous
    • 16 years ago

    Let’s not forget the other benefits of a the P4 chips, they run *much* cooler, thus allowing you to pick a much quieter case design. Paired with a Seagate drive and an ATI gpu you can have a virtually silent box and less power hungry box that perform as well as its louder and equivalently priced competitors.

    Meet the new boss, same as the old boss.

      • Forge
      • 16 years ago

      The overall dissipation of current P4s and current Athlon XPs are not that far apart… The Athlons are actually lower, most times. The reason the P4s run cool is the Intel HSF retention mechanism. It’s designed for a HSF roughly 70 by 90mm, while the Socket A clearances only require a 60x60mm square over the socket… Very few allow 80mm or larger HSFs.

      With similar Alpha coolers (8045 and 8942, same, but 8942 is perhaps 20mm longer), I have a 2400+ and a 2.4C that get idle and load temps within 1 or 2 degrees C of each other…. Unless I enable HT, in which case the P4 picks up a few more degrees C on it’s load temp.

      Basically, P4s do not run cooler on their own, they just have better HSFs by default.

        • gordon
        • 16 years ago

        The reason P4 processors sometimes seem cooler isn’t really some mystical belief P4s by nature are cooler, but that Intel includes some pretty damn good stock coolers with their retail processors. AMD uses smaller and seemingly less effective stock coolers on the retail Athlons. That’s why you’ll see huge drops going from a stock to retail cooler on Athlons, but sometimes negligible results on a P4.

      • eitje
      • 16 years ago

      current P4s with HT consume between 70-100 W, iirc, which is on par with current Athlons. now that P4s are using their functional units more heavily (iow – keeping the pipeline more full), the “power consumption/heat production” discussion is no longer worthwhile. 🙂

      • Buub
      • 16 years ago

      P4 chips don’t run “much cooler” than anything. High-end P4’s generate as much heat as any Athlon.

        • Anonymous
        • 16 years ago

        or even more.
        so let’s forget the ‘benefits’ of a cooler p4.

      • absinthexl
      • 16 years ago

      Not that it’s a fair comparison, but my 2.53GHz P4, overclocked to 2.9GHz, still runs faster than my old 1.2GHz Athlon with close to the same HSF combo (Swiftech copper/aluminum HS same fan)

        • Forge
        • 16 years ago

        You’re right, that’s not a fair comparison.

        My 2.4B@3.0 put out much more heat than my 2500+@3400+, too, but neither of our examples are very relevant.

          • absinthexl
          • 16 years ago

          I just wanted to feel useful *sniff*

      • eitje
      • 16 years ago

      §[<http://fiehog.de.s18.evanzo-server.de/PCDOC/tt-h-p.htm<]§ old charts, but i think they're still fairly relevant. found them looking for something else, and thought of this thread.

        • Forge
        • 16 years ago

        They’re too far out of date to really be helpful, unfortunately. They don’t reflect the higher clocked NWs and the impact of HT, for one thing.

        A good spot, though.

    • Anonymous
    • 16 years ago

    Congratulations!

    Nice review… very nice conclusions…

    • AnimeEd
    • 16 years ago

    hmm.. AMD haven’t released much lately
    they better do it quick or they’ll be left behind (further behind)

    • JustAnEngineer
    • 16 years ago

    q[

      • wesley96
      • 16 years ago

      What he needs is a connected scattergram. Then you don’t have to be fixated to a very coarse quantum value like a regular line graph would.

        • JustAnEngineer
        • 16 years ago

        Yes, a “connected scattergram” is usually referred to by the rest of us as an X-Y plot. I’d guess that Damage is using MS-Excel to generate his charts. It does a fine job with X-Y…if you tell it to.

      • Anonymous
      • 16 years ago

      Bar charts rule!

    • Anonymous
    • 16 years ago

    q[

    • yokem55
    • 16 years ago

    q[

    • gordon
    • 16 years ago

    Intel just loves rubbing it in :(. Kinda like ATI releasing a new R3xx product. So mean!!

    Edit: On second thought it’s not really that big of a deal. In most areas that matter the AXP is neither far behind or outperforms the processor. Some people like that extra 5% performance that comes with the extra 100% in price so I guess it’s all a matter of choice really =/.

    • Bomber
    • 16 years ago

    No overclocking results??

    Anyway, I think that unless the Athlon64 is all that and a bag of potato chips, then AMD might as well give up and stop competing directly with Intel for a while and just do what they do well – make a damn good cheap and powerful CPU. Intel had their downtime but they are definately back and with a vengence…400fps in Q3! Damn!

    • Hellsbellboy
    • 16 years ago

    Why on Tom’s Hardware site, they calling the Prescott the Pentium 5.. is that it’s name? I haven’t seen this anywhere else..

    Great review.. wish i could afford one..

      • Entroper
      • 16 years ago

      Hopefully their True Performance Initiative will amount to something. I have my doubts, though, as to AMD’s ability to convince anyone that their performance rating scheme is less slanted than any other scheme out there.

        • Glorious729
        • 16 years ago

        Don’t hold your breathe. Nothing else has been said about it since they announced it, like what?, more than a year ago?

Pin It on Pinterest

Share This