–
By Jeff Field –
With prices for RDRAM (the memory used for Pentium IV-based machines) about twice as much as the competing DDR memory (used primarily in Athlon- and Duron-based machines), Intel had to find a way to compete with Athlon on the low end. Intel has released the PC133-supporting i845 chipset, and today, I review a Pentium IV board based on that chipset, the Soyo P4ISR.
The board
When designing the P4ISR, it would seem Soyo was aiming to please those who want ample expansion on a motherboard. The board has six PCI slots, an audio-modem riser slot, and an AGP slot for internal expansion, as well as two USB ports on the ATX I/O plate and two more pin headers (for four total USB ports) on the board. Other ports include two serial ports, one parallel port, as well as the onboard audio in/out and game port.
Memory expansion on the board is provided by three 168-pin DIMM sockets, with support for up to one-gigabyte DIMMs in each slot for a maximum total of three gigabytes. Also interesting is that this board not only supports PC133, but PC100 memory as well, allowing you to upgrade without the hassle of replacing older memory. The board has four ATA drive connectors for up to eight ATA drives. The first two connectors are controlled by the i845’s south bridge, while the other two are controlled by the additional Promise RAID controller.
The chipset
The i845 chipset from Intel with PC133 support is, to be honest, the result of a business mistake. Intel signed a contract with Rambus, the company that owns RDRAM, or RAMBUS DRAM, the memory used on all Pentium IV systems up until now. Due to this contract, Intel can’t make a chipset using a competing technology, such as DDR, until 2002. In order to hold people over until then, Intel created the i845 chipset using older PC133, which costs significantly less than RDRAM, but does not perform as well as DDR, which costs about the same as PC133. Unfortunately for Intel, this kludge does not help the Pentium IV’s performance at all.
Board layout and design
The P4ISR’s layout is excellent overall. All six PCI slots should be able to accept full-length cards, depending on the size of the IDE cables installed in the third and fourth ATA ports. My only complaint is that because of the locations of the ATX power, floppy, and primary/secondary hard-drive controllers, the top right of the board can get very cluttered, making it somewhat difficult to work around when the board is installed in a case, but this is a minor issue. Otherwise, I had no problems with the design of the board, Soyo seems to have put a good bit of thought into it.
Documentation and setup
The documentation included with the board was the usual Soyo quick guide, which does the job for those who have set up motherboards before, but is lacking some information that may be needed by those who are installing a new motherboard for the first time. This may or may not be a problem, because setting up the board is very easy; it is jumperless, allowing for all configuration options to be set through the BIOS.
Performance
System Specifications
Pentium IV 1.9GHz (Socket 478)
TH7II-RAID Motherboard
256 Megabytes Kingston RDRAM
Western Digital 7200 RPM 10.2 Gig Hard Drive
3Com 3C905TX-C 10/100 NIC (PCI)
400 Watt ATX 2.03 Power Supply
Gigabyte GF3000 GeForce 3 64MB AGP
Slackware 8.0 with Kernel 2.4.14 and XFree 4.1.0
Kernel compiles
In order to test both the board’s stability and speed, I ran three sets of Linux kernel compiles. One is a normal, “uniprocessor” make, or make -j1, which is the default. This uses one process, and does not always maximize system usage. I then did make -j2, which spawns a second process. The last test I run is with make -j3, spawning two extra processes. I do this for several reasons — to find the “sweet spot” for the board/CPU, as well as to stress the system
as much as possible when trying to rate its stability. Also, the kernel is extremely useful as a measure of integer performance. In order to compile the kernel, I untarred kernel 2.4.6, ran make config, and used the default values.
Kernel 2.4.6 Compile Times (Minutes:Seconds)
Lower numbers are better |
| Board |
-j1 |
-j2 |
-j3 |
| Pentium 4 1.9GHz (RDRAM) |
3:30 |
3:24 |
3:23 |
| Pentium 4 1.9GHz (PC133) |
3:56 |
3:49 |
3:48 |
| Pentium 4 1.9GHz (PC100) |
4:18 |
4:14 |
4:00 |
With these kernel compile results, we start to see the negative effect the reduction in memory bandwidth has on the Pentium IV. While these may not seem like huge gaps, for people who do compiling of larger programs on a regular basis, this could mean huge waits compared to RDRAM based Pentiums IVs, or alternatively, Athlons.
POVRay benchmarks
POVRay is a multi-platform raytracing program. It is a floating-point intensive task and serves well to help measure the floating point performance of a CPU. For more information on this benchmark, head to the official POVBENCH homepage. The command to run for this benchmark, once you obtain POVRay, you run povray -i skyvase.pov +v1 +ft -x +mb25 +a0.300 +j1.000 +r3 -q9 -w640 -H480 -S1 -E480 -k0.000 -mv2.0 +b1000 from the command prompt. Results are in seconds.
POVRay (seconds)
Lower numbers are better |
| Board |
Result |
| Pentium 4 1.9GHz (RDRAM) |
20 seconds |
| Pentium 4 1.9GHz (PC133) |
20 seconds |
| Pentium 4 1.9GHz (PC100) |
21 seconds |
Because POVRay is more CPU intensive than memory intensive, it does not suffer the same sort of speed decrease as other tasks when faced with lower memory bandwidth, so its results are basically the same — in all three cases it is the same CPU.
Distributed.net Client Benchmark
Distributed.net is a distributed computing network that works on various distributed computing contests. The contests use primarily integer numbers while performing their tasks, and therefore serve as an excellent benchmark for overall integer performance of properly optimized software.
Distributed.net Client Benchmarks
Higher numbers are better |
| CPU |
RC5 Core 8 |
OGR Core 0 |
| Pentium 4 1.9GHz (RDRAM) |
2,757,055 keys/sec |
8,321,706 nodes/sec |
| Pentium 4 1.9GHz (PC133) |
2,734,080 keys/sec |
8,291,806 nodes/sec |
| Pentium 4 1.9GHz (PC100) |
2,734,080 keys/sec |
8,291,769 nodes/sec |
Again, as with POVRay, we find that the distribted.net test depends on CPU speed rather than memory bandwidth, and so there is no maajor difference.
Quake III Timedemos
Quake 3 Timedemos are perhaps the best way to measure 3D Gaming performance under Linux. Timedemos used the four.dm_66 demo included with the latest version of Quake 3 Arena. To run a timedemo, hit the “~” key, type timedemo 1, followed by demo four.dm_66 – once this completes, hit “~” again to see your results. High quality results were done by turning texture and color depth to 32-bit, filtering to trilinear and texture detail to its highest setting. 640 by 480, 800 by 600, 1024 by 768, 1280 by 1024 and 1600 by 1200 are the screen resolutions at which the tests were run.
Quake 3 Arena Timedemos (Frames Per Second)
Higher numbers are better |
| Board |
640×480 |
800×600 |
1024×768 |
1200×1024 |
1600×1200 |
| Default Quality |
| Pentium IV 1.9GHz (RDRAM) |
221.1 |
218.6 |
186.2 |
125.3 |
90.2 |
| Pentium IV 1.9GHz (PC133) |
168.1 |
166.9 |
161.9 |
122.4 |
89.4 |
| Pentium IV 1.9GHz (PC100) |
141.4 |
139.8 |
138.4 |
116.7 |
87.8 |
| Highest Quality |
| Pentium IV 1.9GHz (RDRAM) |
219.5 |
199.7 |
145.4 |
97.3 |
71.2 |
| Pentium IV 1.9GHz (PC133) |
167.7 |
166.3 |
140.2 |
96.3 |
70.9 |
| Pentium IV 1.9GHz (PC100) |
142.5 |
140.1 |
129.5 |
94.5 |
70.2 |
Where we see the largest, clearest difference in performance between these types of memory is in 3D games, such as Quake 3, which need all the memory bandwidth they can get. In lower resolutions especially, where the CPU and system RAM is more the limiting factor than the graphics processor/memory, we find huge differences — and despite the fact that all of these scores are above the “optimal” 60FPS, future games that tax the system more will suffer from the lower memory bandwidth.
Conclusion
The P4ISR is a well built board, definitely a nice first entry from Soyo into the Pentium 4 motherboard market. Unfortunately, the chipset it uses is somewhat lacking due to the PC133 memory used. While much cheaper than RDRAM, it suffers greatly when it comes to performance, which just compounds the performance problems the Pentium IV has compared to the Athlon. If you are looking for a Pentium IV board and are interested in using your previous memory, then the P4ISR is the solution for you, but if you are upgrading, and have your heart set on buying a Pentium IV, I would suggest you opt for a DDR-based solution once they are widely available, or consider a DDR-based Athlon solution. The Soyo P4ISR is available on Pricewatch for around $140.
