Round 3 Editors Challenge Sponsored...
Top 10 Challenge Round Sponsored by...
Editors Challenge Sponsored by Inte...
FiringSquad Editors Challenge Round...
||5 entry(ies) in this category
| INTEL P4 1.8A -- A 2007 PERFORMANCE ANALYSIS (18 comments )|
by: kevinSpiess (156) | Posted in cluster Round 3 Editors Challenge Sponsored by Intel
Posted 74 months ago ( edited 74 months ago ) in category DEFAULT
|» MEDIA (7)|
Not a UFO, but instead, an old, post-overclocked P4 1.8A
Company of Heroes and C&C 3: TW Benchmarks
Half-Life 2 CounterStrike Stres-Test
Sandra CPU BenchMark
LAME and WinRAR Benchmarks
Moore’s Law can be a real bitch. For example: one day, say in 2003, you decide you need a new CPU. Your Pentium 4 Willamette 1.4GHz just isn’t cutting it anymore. So, you start researching chips online. You see what’s out there. You read some reviews on FiringSquad. You decide to get a P4 Northwood.
You go to the store, and fork over a bunch of cash, for what is at the time, the latest-and-greatest. You think: “Wow, what great performance! That’s so quick!” You may even brag to your friends: “Hey Billy-Sue, my new Northwood processor is so damn fast it’s like quicker than Slap Jack’s gold Toyota Supra in 2 Fast 2 Furious!” or something.
Then you get home. You bust open the Intel box. Your hands are shaking as you plop the CPU into the motherboard. The chip looks dangerous! You start up your computer. You run Half-Life and it’s like 10,000 times smoother. Damn! You think: What a chip! It’s fast! It’s furious! It can handle 1024x768 no problem!
Then, four years later, you end up with a piece of silicon that can’t even handle flash games anymore. So you try to sell it on craigslist. But makes you terribly annoyed, because someone offers you a bag of Doritos and a screw-driver set for it. And he isn’t joking either.
That’s what Moore’s Law is really about. It isn’t about processor development at all. It’s about not being getting decent frame rates in PC port of Grand Theft Auto: Vice City in 2004, even though you bought your CPU in 2003, for like, 200 bucks, from some shady guy.
This article is a performance review of a Socket 478, Intel Pentium 4 1.8A that I purchased back then, in 2003. Back then it was a middle-of-the-road performer. As you know, computing technology gets old quick. I am interested to see how my old CPU stacks up with today’s applications and games. Especially games. One game in particular -- Command & Conquer 3: Tiberium Wars – has a 2GHz processor listed as a minimum requirement. Let’s see how my old P4 1.8A performs.
To get a better feel for the performance of the P4 1.8A, it will be compared to a later-generation P4 3 GHz HT processor. I’m also going to overclock the P4 1.8A like hell, and see how stable it is at a couple of different clock-speeds.
Let’s step into my time-machine for a second. Let’s go back into time. Back before even 2 Fast 2 Furious came out. Back all the way to 2001. Back then, AMD was leading the processor pack with its Athlon XP series of chips.
The limits of the Pentium III had been reached. Intel’s first iteration of the Pentium 4 – the Willamette – was not living up to some of Intel’s promises, and the chip was not winning over any critics. Intel desperately needed to release something that compared to the offerings of AMD design teams.
The next Pentium 4, codenamed the Northwood, was finally ready for the marketplace in January 2002. A lot was riding on such a small CPU. Intel’s market share was at stake!
At the time, one of the most exciting things about the Northwood’s architecture was the evolutionary step taken in its manufacture: the majority of the processors of the day had transistors the size of, on average, 0.18 microns. So Intel decided to really go the distance to stick it to AMD by developing a 0.13-micro manufacturing process. What this 0.05 micron difference is may not be immediately apparent, so let me lay it out for you: Basically, what this means was that Intel was able to cram more processors onto a smaller chip. This new cool manufacturing technique had a few huge positives for hardcore gamers out there: all those super-teeny-weeny transistors made for higher clock speeds; and production costs were kept low by the superior transistor-cramming ability; and the smaller transistors made for lower power requirements. Lower power requirements means less heat. Less heat means more overclocking, so some gamers everywhere broke into spontaneous convulsion of joy when the Northwood chips came out.
Another really fantastic thing about Northwood chips was, for their time, a healthy amount of cache memory. Intel increased the L2 cache on the Northwood chip to 512 kilobytes. The Northwood also featured an 8 kilobyte data-cache right on the processor. In comparison, the Athlon XP of the time, had only 384 kilobytes of cache – so again, Intel was really sticking it to AMD by doing this, because, as you may know: cache matters. More cache means more memory bandwidth and more on-chip memory. These are very good things.
Here is the point-form nitty-gritty about the P4 1.8A:
Hyper-pipelined technology w/ rapid execution engine and execution trace cache
8KB L1 Data Cache, 512KB L2 advanced transfer cache
55 million transistors
0.13-micron manufacturing process w/ copper interconnects
533MHz system bus
Socket-478 form factor
1.5V core voltage
Available in clock speeds of 2.0GHz and 2.2GHz
Here is the rest of my benchmarking platform:
Processor #1: Intel P4 1.8A, 533 MT/s FSB
Processor #2: Intel P4 3.0, 800 MT/s FSB, w/ H.T
OS: Windows XP, SP-2
Motherboard: ASUS P4PE (Socket 478)
Memory: 1GB of PC3200 Kingston ValueRAM, @ DDR400
Hard Disk: 60GB 5,400 RPM Maxtor ATA
Video Card: ATI Radeon X800 XT, 256 MB
Push it to the Limit!
Overclocking. Ah, even the word itself sends tingles down my spine and makes me dizzy and somewhat giddy.
Northwood chips are notoriously good overclockers. I consulted Firing Squad’s CPU overclock database, and a few other on the web, to see what kind of numbers people were getting with the P4 1.8A.
Well, I was shocked. Astounded even. With fast RAM and after-market cooling solutions, some people were reporting that they were able to get their 1.8A’s up to a massive 3.2 gigahertz! I could not believe it. I promptly decided to see what limits I could push with mine. I had a few things going for this overclocking project me. First off, my motherboard, my beloved ASUS P4PE, is an excellent piece of hardware to overclock with. I have been quite happy with this motherboard, and it was many, many options that can facilitate extreme overclocking. Secondly, my 2600 RPM stock Intel fan and aluminum heat-sink that came with my P4 3.0 GHz is not too shabby, and would offer relatively good cooling for the P4 1.8A, which originally came with a dinky little fan, that would not be up to the task of extreme overclocking, even if I still had it. The third thing I had going for this overclocking project was that I really wasn’t too worried about blowing my CPU up -- because, well, it’s my old CPU and previous to this experiment, it was just sitting in a big pile of dust on the top of my bookshelf, beside an old waffle-iron.
Emboldened by the results I was finding online, I decided to start off my overclocking at the respectable 400 MHz mark. The P4 1.8A’s multiplier is locked at 18x, so, after doing some quick math on a napkin, I figured to reach the 2.2 GHz mark I needed a FSB setting of somewhere in the low 120’s. (As I’m sure most of you know, your clock speed is determined by multiplying the FSB by the clock multiplier.) Going into the BIOS, I increased the FSB to 122, from the 100 default. I reset the computer, confident that this speed increase would not be a problem. My computer did boot successfully, with no blank screen or mysterious beeps from the get-go, and then Windows started up alright, but there was one problem after that: somehow, I lost sound.
At first, I thought the problem may be attributable to RAM mistiming – I never before had trouble with my onboard SoundMAX sound system. So, I started turning down my RAM timings. That didn’t seem to help. I further underclocked my RAM, but that didn’t solve my problem. Then, from the BIOS, I locked my AGP/PCI frequency ratio to 66.6 / 33.3 – and that did seemed to the trick. My sound never failed after that. Even when I really started to push it to the limit.
I decided the next stop on the overclocking train was going to be 2.2 GHz. I had no problems (beside the sound issue) at 1.8 GHz, so I thought another 400 MHz was a nice, round figure to aim for. I wasn’t sure how stable my system was going to be, but, my mission to destroy my P4 1.8A demanded that I find out. So, I grabbed myself a beverage and set my FSB to 145. I also changed my memory timings from their 2-2-2-6 setting to ‘Auto’, to be a little more forgiving on my recently purchased Kingston ValueRAM, which was not made for overclocking as much as a P4 1.8A is.
My computer started successfully at a clock speed of 2610 MHz. I had no trouble going through the benchmarks. Everything ran fine, including the onboard sound.
So, I decided to get a little crazy. I set my sights at a zany 2.8 GHz – one full gigahertz over the recommended operating speed of the P4 1.8. I knew I was in dangerous territory here – I could easily fry my CPU at these speeds. But I went for it, because the glory days of my P4 1.8A are long gone, and these chips are dirt cheap right now, so I didn’t really care if the CPU burst into a dangerous, silicon-searing, motherboard-decimating little fire on my carpet.
I set my FSB to 156, swallowed, and turned my rig on. And nothing happened. It was terribly unexciting – no 2.8 GHz boot-up, no small fires. I thought I might know what the problem was though: big overclocks require big power. So reset my BIOS, and decided to try something else. This time I set the FSB to 155, and boosted my CPU Vcore to 1.7 volts. A healthy P4 1.8A runs at 1.5 volts, so I knew that increasing the power to the CPU to this extreme level was really, truly, pushing it the limit. As a backup plan to disaster, I opened up my office window – I figured that if my computer started to smoke, I could toss it out into the streets, and it would be somebody else problem. Not a bad plan!
I had my roommate I turn on my power-bar while I watched from the other side room, and I was shocked to see that it booted up at 2.8 GHz! I was somewhat surprised that I reached this high speed. My CPU however, was running at 44 degrees Celsius. This is a little on the hot side. Unsurprisingly, although Windows booted up fine, after about 5 minutes into any benchmark, I would get a B.S.O.D (Blue Screen of Death). After shedding some tears, I had to come to terms with not being able to reach the 3 GHz, like some other overclockers out there.
It looks like 2.8 GHz was pushing the respectable Northwood 1.8A just a little too hard, so I slowed the processor down by increments until it ran stable. The maximum stable speed that I could attain was 2.7 GHz – a very respectable overclock of 50%. This clock speed required that the Vcore was boosted to 1.65, but this is somewhat within the operating parameters of the CPU, and the system was completely stable.
Ready, Set, Bench!
I used nine different programs for benchmarking: three games, one sound encoding program, one compression program, three synthetic benchmarks, and one rendering benchmark program. This should offer a good cross-section of applications to better gauge the performance of the P4 1.8A – both overclocked and not overclocked – against a more recent Socket 478 P4 – a Northwood 3GHz, 800 FSB, Hyper-Threaded CPU.
Here are the benchmarks and their settings:
Half-Life 2, Counterstrike
This perennially popular game is a good benchmark because although it is a bit older now, the game still makes full use of all available processing power, including the multi-threaded nature of the P4 3 GHz HT.
The Counterstrike in-game stress-test was used to generate a final, average FPS. The video settings were as follows: all settings on high, 1024x768, and 4xAA, with 2xAF.
Company of Heroes
This game-of-the-year RTS has an excellent in-game benchmarking program. Company of Heroes is fairly recent, so I expected my P4 1.8A to choke on it. The box minimum requirements for this game are a 3 GHz CPU – and this seems exactly right. The video settings were: everything set to high, AA disabled, and a 1024x768 resolution.
Command and Conquer 3: Tiberium Wars
The minimum box requirements for this game are officially a 2 GHz CPU, so, I was very interested to see what kind of performance a gamer could expect with this much processing power. The benchmark was attained by utilizing the program FRAPS to measure the average framerate through a 90 gaming session. The video settings were: everything set to high, AA disabled, and a 1024x768 resolution.
LAME MP3 Encoder
I used the Razor front-end for the LAME MP3 algorithm. I encoded a 500 megabyte music WAV file to a 256 kps MP3 file. The time it took to encode this MP3 was used as a benchmark.
I compressed a 100 megabyte total, 52 file archive with WinRAR. I thought this test would primarily measure the speed of my hard drive, but faster CPU’s did speed up the compression significantly.
The free version of PCMark04 was used to generate my first synthetic benchmark. PCMark04 is a comprehensive suite of 10 separate tests, all of which contribute to final score in ‘PcMarks’.
Sandra Lite XI
Good ole SiSoftware’s Sandra comes with an excellent selection of benchmark utilities, and from the different tests available, I chose the ‘Processor Arithmetic’ benchmark. The test generates a two scores: one is the Dhrystone rating, which is measured in millions of instructions per second (MIPS), and the other is the Whetstone rating, which is measured in million floating operations per second (MFLOPS).
Cinebench is a rendering benchmark program based on the powerful 3D software Cinema 4D. The benchmark takes about two minutes to complete, and produces a number of figures that measure your CPU and the OpenGL capabilities of your graphics card. The test also makes full use of multiple processors. Of all the figures produced by Cinebench, I chose just to compare the CPU rendering score. I ended up ignoring all the other scores completely.
•Northwoods can be purchased very inexpensively
•Northwoods overclock very well
•Arguably the king of Pentium 4's
•Overclocked to the extreme max, a P4 1.8A @ 2.7 GHz is just barely fast enough for today’s requirements
•Most Socket 478 motherboards only support AGP, and not many AGP cards are being produced anymore
•At higher overclocks, and Vcores over 1.7, Northwoods are known to suffer from Sudden Northwood Death Syndrome
It’s 2007 now, and time is moving on, but the P4 Northwoods are standing still. They had their day in the sun, but that day has passed, and now the P4 Northwoods are left in the shade, beside the Pentium Pros and the Willamettes. Although, for the P4 1.8A, it is mentionable that with an extreme overclock, it is still possible to attain average performance with an inexpensive Northwood CPU – so if you have a aging Socket 478 system, and an older P4 in it, and you can pick up a fast Northwood for a song, you might be looking at good, solid cheap upgrade that can give you some real performance benefits before you really have to put down the serious dough for a totally new computer.
|18 User Comment(s) • 14 root comment(s)|
| Domicron (15) Apr 19, 2007 - 02:24 pm|
|good read. nice work. could use a little editing, but that's just the english major (for a semester!) in me talking.|
» Login to reply to this
| ithinkcowsaresexy (7) Apr 10, 2007 - 03:15 pm|
|Reminds me of my first CPU when I was but a wee lad of 12, the Athlon XP 2000+. Fine performance for 2003, but not so much now. Well, at least for gaming. That little sucker is still running Ubuntu amazingly well!|
» Login to reply to this
» Note: You need to be logged in to write a comment!Login here, or if you don't have an account with FiringSquad, register here, it's FREE!
My Media-Blog categories
No categories created yet.