Intel Pentium 3 1GHz Processor Review
By: Mike Chambers - July 25, 2001
It wasn't too long ago, September of last year to be exact, that I purchased a barebones system from GamePC. I went with an Intel Coppermine processor after reading about the successful overclocking stories of the 700MHz model.
After successfully reaching 840MHz, which was a far cry from the 933MHz other users had achieved, I ran into a problem using an early revision of the GeForce3 reference graphics card. With the GeForce3, my system refused to run at any front side bus (FSB) speed great than 105MHz, which indirectly throttled the CPU speed down to 735MHz.
Realizing that a 700MHz processor wasn't going to push the GeForce3 close to its performance potential, I felt that a processor upgrade was a good idea.
Intel 1GHz Pentium 3
With the arrival of the Pentium 4 processor, it's no secret that Intel is phasing out the Pentium 3 based Coppermine. Unless you're accustomed to staying on the bleeding edge of technology, the end of a product life-cycle is an ideal time to upgrade. Prices are cheaper and in my case, a CPU swap was all I needed as the i815E chipset based Asus CUSL2 gladly accepted the Socket 370 based GHz processor.
FC-PGA Socket 370 Motherboard Layout
Next to changing the CPU multiplier in the BIOS, the hardest part of the upgrade was putting the Thermaltake Golden Orb cooler back on.
Thermaltake Golden Orb CPU Cooler
Within 15 minutes, I was up and running and ready to find out how much the extra 300MHz in processor speed would help.
I concentrated on OpenGL and Direct3D games to measure the effectiveness of the processor upgrade. Benchmarks were run with vsync disabled using a 75Hz refresh rate and unless specified 32-bit color was used and sound was disabled.
The following table lists the specific settings for each of the games that were tested.
||Version 1.17, Demo001, Lightmap Lighting, 32-Bit Textures, Maximum Texture Detail, Texture Compression Enabled, Trilinear Filtering, High Geometry
||Homegrown Demo, 32-Bit Textures, Maximum Texture Detail, Trilinear Filtering, Texture Compression Disabled
||Version 1.003, Standard Demo, Maximum Texture Detail, Trilinear Filtering, Hardware T&L Enabled
||Version 1.02, Karnak Demo, Beyond 3D Presets for Maximum Quality (Texture Compression Disabled, 32-Bit Textures)
||GeForce2 Normal Mode, 16-Bit Textures, 512x512 Texture Size, Trilinear Filtering, 24-Bit Z-Buffer, Bumpmapping Disabled
|Version 436, Thunder Demo, High Texture Detail, High Skin Detail
|Standard Demo, 16-Bit Color, 2/3 Detail Setting, 1/2 Draw Distance, 1/2 Special Effects, High Detail Cars, Mirror Off, Sound Enabled
||Build 200, Default Settings For Game 1 and 3, 32-Bit Color, Double Frame Buffer, Compressed Texture Format, 24-Bit Z-Buffer
|Pixel Shaders Disabled, 24MB Textures
Testing was done using a reference GeForce3 graphics card, which was running at default core and memory speeds, at resolutions of 1024x768 and 1280x1024. Results are reported in average frame rate per second.
Rounding out my system are 256MB of Mushkin PC150 rated memory, two 20GB Maxtor 7200RPM ATA-100 hard drives, and Windows 98 with DirectX 8.1. Beta version 12.90 of the Detonator drivers were used for the GeForce3.
Before checking out high resolution gaming with each processor, I tested the "Fastest" Quake 3 Arena setting where the memory bandwidth of the graphics card wasn't a limiting factor in determining processor performance.
P3-700 vs. P3-1GHz - Quake 3 Fastest
The following results are based on the 3DMark2001 Game 1, 2 and 3 low and high quality tests with software transform and lighting enabled. Software transform and lighting uses the central processor to carry out certain instructions that are normally done via the graphics processor (when hardware transform and lighting is enabled). Tests were run at a resolution of 512x384 in 16-bit color.
P3-700 vs. P3-1GHz - 3DMark2001 Software T&L
Right off the bat, the benefits of an additional 300MHz of processing power are apparent. The 1GHz CPU provided a performance increase of 29% in the Quake 3 Fastest setting and well over a 30% increase in all of the low quality 3DMark2001 Game tests.
The largest gain in performance was in the Game 1 high quality benchmark which increased by 62% (12.5 vs. 7.7 frames per second).
Will we see similar results when games are tested at high resolutions using high quality settings? Read on to find out.
Next Page: OpenGL Game Performance