Unveiling the Data Expert EXP 4349: Stability Issues and VLB Graphics Performance
AD
Table of Contents
- Introduction
- Issues with the Data Expert EXP 4349 Mainboard
- 2.1 Recommendation for AMD 486 CPU
- 2.2 Layout Issue with the Power Connector
- 2.3 Unstable Mainboard and Clock Settings
- Clock Generation on the Mainboard
- 3.1 Clock Generation IC: Ali M1429 G
- 3.2 Clock Frequencies for 386 and 486 CPUs
- VLB Performance and testing
- 4.1 Testing with Tank Labs ET4000 Graphics Card
- 4.2 Comparison of ISA and VLB Performance
- 4.3 Overclocking the VLB Clock Speed
- Results and Conclusion
- 5.1 Performance Boost with VLB Graphics Card
- 5.2 The Impact of Math Co-Processor on Doom Performance
- 5.3 ISA Doom 25 FPS Challenge
- 5.4 Impressive Results with Ali Chipset Mainboard
Introduction
In this Second part of the video series, we will continue our exploration of the Data Expert EXP 4349 mainboard. In the previous part, we addressed several issues and made necessary fixes to enable the use of a 386 CPU on this 486 mainboard. Now, we will dive deeper into the experience and discuss the issues we encountered during testing. We will also explore the performance of the mainboard with VESA Local Bus (VLB) graphics cards and compare it to the performance with ISA graphics cards.
Issues with the Data Expert EXP 4349 Mainboard
2.1 Recommendation for AMD 486 CPU
One of the first issues we encountered was related to the manufacturer's recommendation to use an AMD 486 CPU for the best experience. This recommendation seemed odd and required immediate attention. While it is essential to follow manufacturers' recommendations, it is equally vital to ensure compatibility and optimal performance with other CPU options.
2.2 Layout Issue with the Power Connector
Another issue we discovered was the poor layout of the 80 power connector. The connector lacked a gap between itself and the memory slots, which is a standard feature where clips hold it in place. As a result, we were unable to push the plug completely into the connector, leading to an insufficient power connection. To resolve this issue, we had to modify the side clips on the plug, allowing a tighter and more secure fit.
2.3 Unstable Mainboard and Clock Settings
During our initial tests, we encountered significant instability with the mainboard. This prompted us to investigate further and identify the underlying causes. One of the key factors contributing to this instability was the clock settings. Despite our efforts to adjust the clock settings and experiment with different configurations, we could only achieve clock speeds of 20, 25, and 33 megahertz. The documentation explicitly Mentioned the ability to test with a 40 megahertz CPU, but the mainboard did not support this clock speed.
Clock Generation on the Mainboard
3.1 Clock Generation IC: Ali M1429 G
To understand the limitations and possibilities of clock generation on this mainboard, we examined the clock generation integrated circuit (IC) in detail. The mainboard utilized an Ali M1429 G chipset, which differed from other mainboards in the database that featured dedicated clock generators. We hypothesized that the "G" in the model number might stand for an integrated clock generator. However, we encountered challenges in finding the datasheet for this specific variant.
3.2 Clock Frequencies for 386 and 486 CPUs
Our investigations led us to realize that the clock requirements for 386 and 486 CPUs differ significantly. While 486 CPUs are supplied with an external clock that determines the front-side bus frequency, 386 CPUs require a doubled external clock to function correctly. This distinction in clock requirements posed a challenge for mainboards like the Data Expert EXP 4349, which needed to supply different clocks for 386 and 486 CPUs. Understanding and addressing these clock requirements were crucial for achieving optimal performance.
VLB Performance and Testing
4.1 Testing with Tank Labs ET4000 Graphics Card
To assess the performance of the mainboard with VLB graphics cards, we conducted multiple tests using the Tank Labs ET4000 graphics card. This card, which was generously donated by Chris, allowed us to compare its performance with that of an ISA graphics card. We also used a Cyrus Logic VLB graphics card, a commonly used card known for its compatibility and performance. The benchmarks and gaming tests were performed to analyze the impact of VLB on overall performance.
4.2 Comparison of ISA and VLB Performance
In our previous video, we discussed the performance of a 386 DX33 CPU and an ISA graphics card, which yielded 26.9 points in the Season 4 benchmark. With the introduction of the Data Expert EXP 4349 mainboard and the Tank Labs ET4000 graphics card, we aimed to determine if VLB could provide a significant performance boost. Our initial results showed a deviation from the expected 35.9 reference points. However, after adjusting the memory cache read timing, we managed to achieve a score of 36.1 points, surpassing the reference value.
4.3 Overclocking the VLB Clock Speed
To further explore the performance possibilities, we decided to overclock the VLB clock speed. The mainboard offered various clock settings in the BIOS, allowing us to experiment with faster clock speeds. As we overclocked the VLB clock speed, we conducted tests with the aim of achieving higher framerates in games. The performance improvements with VLB were promising, especially in Windows, showing a noticeable boost in performance compared to ISA graphics cards.
Results and Conclusion
5.1 Performance Boost with VLB Graphics Card
Our tests conclusively demonstrated that using a VLB graphics card on the Data Expert EXP 4349 mainboard provided a considerable performance boost. While the boost varied across different applications, the overall improvement exceeded 300 percent in synthetic benchmarks. This increase in performance showcases the potential of VLB and its impact on enhancing the visual experience and overall system performance.
5.2 The Impact of Math Co-Processor on Doom Performance
As we explored the performance of the mainboard with various graphics cards, we also examined the role of the math co-processor on Doom's performance. Contrary to popular belief, adding a math co-processor did not significantly influence Doom's performance. Doom primarily relies on integer calculations, rendering the math co-processor ineffective for this particular Game.
5.3 ISA Doom 25 FPS Challenge
Our findings and discussions with other enthusiasts, such as CPU Galaxy, led us to participate in the ISA Doom 25 FPS Challenge. This challenge aims to achieve stable framerates of over 25 FPS on a system with an ISA graphics card. We were impressed by the results, which demonstrated that with appropriate tweaks, overclocking, and a well-configured bias, stable framerates of 26.2 FPS were achievable.
5.4 Impressive Results with Ali Chipset Mainboard
Throughout our testing and experimentation, we discovered that mainboards featuring Ali chipsets, despite their reputation for slower performance compared to their competitors, could deliver impressive results. The stability and performance achieved on this Data Expert EXP 4349 mainboard showcased the potential of Ali chipsets when combined with proper optimization and tweaking.
In conclusion, our exploration of the Data Expert EXP 4349 mainboard revealed several issues and opportunities for performance improvements. We successfully addressed the initial issues, tested the performance with VLB graphics cards, and achieved impressive framerates in Doom with proper overclocking and bias configurations. The Ali chipset mainboard, despite preconceived notions, delivered exceptional stability and performance. We look forward to further exploring its capabilities and pushing its performance limits in future endeavors.
Highlights:
- The Data Expert EXP 4349 mainboard presents several issues, including a layout problem with the power connector and instability with clock settings.
- The clock generation on the mainboard, using the Ali M1429 G chipset, requires different clock frequencies for 386 and 486 CPUs.
- Testing with VLB graphics cards, such as the Tank Labs ET4000, showcases a significant performance boost compared to ISA graphics cards.
- Overclocking the VLB clock speed further enhances performance, particularly in applications like Windows.
- The Ali chipset mainboard surprises with impressive stability and performance, debunking the Perception of slower performance.
FAQ
Q: Can I use an AMD 486 CPU on the Data Expert EXP 4349 mainboard?
A: Although the manufacturer recommends using an AMD 486 CPU, it is possible to explore compatibility and performance with other CPU options.
Q: Are there any layout issues with the power connector on the mainboard?
A: Yes, the power connector lacks a gap between itself and the memory slots, requiring modifications to ensure a secure fit.
Q: Can the clock generation on the mainboard be adjusted for optimal performance?
A: Clock generation on the mainboard relies on the Ali M1429 G chipset. Understanding the different clock requirements for 386 and 486 CPUs is crucial for achieving optimal performance.
Q: Does using a VLB graphics card provide a significant performance boost?
A: Testing with VLB graphics cards, such as the Tank Labs ET4000, demonstrates a noteworthy performance boost compared to ISA graphics cards.
Q: Can I overclock the VLB clock speed for enhanced performance?
A: Yes, the mainboard's BIOS offers various clock settings, allowing for experimentation with faster VLB clock speeds.
Q: What are the performance implications of adding a math co-processor?
A: Contrary to popular belief, adding a math co-processor does not significantly impact Doom's performance, as the game primarily relies on integer calculations.
Q: Can stable framerates of over 25 FPS be achieved with an ISA graphics card?
A: Yes, through appropriate tweaks, overclocking, and a well-configured bias, stable framerates of 26.2 FPS can be achieved in Doom.
Q: Is the Ali chipset mainboard capable of impressive stability and performance?
A: Despite preconceived notions, the Data Expert EXP 4349, featuring an Ali chipset, delivers exceptional stability and performance, debunking the perception of slower performance.