MSI GeForce FX 5600-VTDR128

◊ High-Performance, High-Precision, 3D Rendering Engine

◊ High-Performance 2D Rendering Engine

◊ Intellisample Technology

◊ Advanced Display Pipeline with Full nView Capabilities

◊ Digital Vibrance control (DVC) 3.0

◊ Rocket Science For a System-Level Solution

◊ Operation Systems and API support

◊ Performance

Here’s a comparison chart of the MSI’s (non-Ultra) GeForce FX product line:

The Card

The board size is reasonable; it’s not extremely long and it does not take up the adjacent PCI slot. The board is sturdy too. (Don’t try the following “test” at home!)

While helping Bryan test the OCZ ULTRA II thermal paste, I somehow did not properly re-attach the CPU heatsink, which fell off a few days later! The heavy, all-copper cooler fell directly onto the FX5600; luckily the heatsink hit the copper heat spreader on the back of the video card. No damage was done. I was using the system at the time, heard the THUNK, and saw MotherboardMonitor jump to 60° C, when my system turned itself off. Also, as luck would have it, my Biostar M7NCG motherboard has built-in thermal protection, which saved my Athlon 1900+ from burning up.

The MSI FX5600 even has a TWIN BIOS feature to let you recover from a bad Video BIOS flash and includes a Windows BIOS Flash program to make this task easier. (Still, I would hope no one would ever need either of these features, but it’s nice to have the added safety of a back-up BIOS, just in case something awful happens like the power goes out while you’re flashing the card.)

I like the quiet, all-copper cooler.

Notice all the yellow QA (Quality Assurance) stickers on the back of the card.

The Bundle

Look at the size of this bundle!

You can find a list of what’s on the 11 CDs, in addition to a lot of other info, on MSI’s product page

The manual is quite thick, but it covers MSI’s entire line of FX cards and much of the manual is devoted to describing the specifications and package contents for each card. The hardware section is decent, although the remote control is not covered and the paragraph on the video-in/video-out dongle does not document the “Safe / Work” switch. The software section only covers driver installation and use; the software on the other 10 CDs is not covered.

Unfortunately, the only game I’ve had time to play (so far) is the Duke Nukem Manhattan Project game, which crashed on me numerous times. It crashed even on the GeForce4 Ti4200, so it must be a problem with the game (or the Detonator drivers) and not a problem specifically with the MSI FX5600.

Testing

Video-Out Test

The MSI FX5600 has greatly improved video-out, when compared to the output from an MSI nForce1 motherboard with an internal GeForce2 MX. In 640×480, Windows text is not exactly readable, but it was clear enough to be able to navigate to the Start button and run WinDVD.

I did run into an anomaly while hooking up the system to a LCD TV projector. Each time I connected RGB, S-Video, or Video-out to the projector, the computer would reset. I’m assuming that it’s a grounding problem and has nothing to do with this card, but it’s not reassuring. Later I noticed that there is an undocumented switch on the video dongle, labeled Safe / Work T-BIOS, which perhaps should have been in the “safe” position, instead of “work”. Unfortunately, due to publishing time constraints, I am unable to retest with the switch in the other position.

There was a solid black rectangle in WinDVD where the video image should be, when using S-Video or Video-out. (Everything was fine when the LCD projector was hooked up to the VGA connector.) I’ve had the exact same problem with an ATI product in the past (which turned out to be a simple Display Settings issue). I searched the MSI website and found this notice in their FAQ, indicating that I may need to flash the card’s BIOS.

Video-In Test

I’m sorry that the subject matter of the video-in test leaves much to be desired, but every VCR tape I tried to snap a shot of has a big white bar running down the middle of the frame. (It looks perfectly fine on-screen.) However, both video and still frame captures have the white bar. (I think it’s some sort of VCR tape copy-protection scheme.) I could capture TV signals without the bar, but my TV reception is poor (and therefore unsuitable for a quality test), so I continued search for a better solution. I discovered that I actually could capture video from the VCR, as long as the movie has not yet started. Here’s a couple of sample frames:

Click for full-size image

I think the white bar is some sort of VCR copy protection. Click for full-size image.

With the exception of the white bar that blocked half the image, I would rate the video capture as good, since it looks about the same on my TV. If you have higher quality video sources, compared to my ageing VCR, I’m fairly certain that your image quality will be better than my sample above.

Infrared Remote Test

There’s a nifty infrared remote that comes with this (and many other MSI cards). Unfortunately, I was unable to get it to work properly. I think it’s because I’m running WinMe. I got an error while installing the MSI Media Center Deluxe II program, which (even though the manual fails to mention it!) is required for using the infrared remote control. The Media Center software installation failed, since the WINDEx.dll failed to “properly self-register” and then I got another error message, which indicated that the version of Microsoft Media Player 9.0 included was only compatible with WindowsXP and Windows2K.

Since I run WindowsMe, I guess I’m out of luck. I tried it anyway. Whenever I press a button on the remote, I get the following error message:

That’s exactly what happens when I try to run the MSI Media Center Deluxe II program as well. I’m going to test this (in the future) on a WindowsXP machine. Check back later for an update (here in this article).

Benchmarks

Test System Configuration

• AMD AthlonXP 1900+
• Corsair 512 MB 3200 CAS2 DDR
• FX5600-VTDR128 – 128 MB
• NVIDIA reference GeForce4 Ti4200 – 64 MB
• Drivers: 44.03 (DetFX drivers)

I ran a few benchmarks on the GeForce4 Ti4200 to make sure that the DetFX (ver. 44.03) drivers were not slowing down older architecture boards. In some cases, the 44.03 drivers were nearly twice as fast (on the Ti4200) as the older 41.09 drivers were.

There’s been a lot of discussion about “cheating” in the DetFX (version 44.03) drivers, especially when it comes to the 3DMark2003 benchmark, so while I’ve done limited testing of this board using that benchmark, I’ve chosen not to publish the results. I also ran a lot of other benchmarks, like Quake3 and ChameleonMark; I only tested at one or two resolutions (just to make sure they ran correctly on this board). However, I chose to omit those results for brevity, since the results are quite similar to the other scores I’m presenting here. (I’m getting tired of endless benchmarks in reviews, aren’t you?)

One thing I did notice, is that the 44.03 drivers fixed a shadow bug in UT2003 Demo. On the outdoor level, sometimes the bots had shadows projected upwards into the trees instead of downward onto the ground. I consistently saw the bug with the version 41.09 drivers, but it’s now been fixed.

If you’re constrained by your CPU, like I am, antialiasing and anisotropic filtering can be practically free with the MSI FX5600, since the video card is waiting around for the CPU to “catch up” anyway. That means you might as well turn on the high quality settings, and perhaps even lock to V-Sync to remove “tearing” artifacts. One older game I still play (Parsec) looks great at 1024×768 8xAA 2xAF. I get 100 fps, with V-sync ON. I still like to test with Parsec because it makes tearing obvious and therefore looks MUCH better with V-Sync enabled. It also benefits greatly from antialiasing.

3DMark2001SE – Build 330
Setting: Default – NoAA, NoAF, Quality Mode

GFX Card/GPU	Score
	1024×768
MSI FX5600	8869
GeForce4 Ti4200	10023
	1280×1024
MSI FX5600	7298
GeForce4 Ti4200	8131
	1600×1200
MSI FX5600	5856
GeForce4 Ti4200	6629

When running in lower quality modes (no antialiasing or anisotropic filtering), the GeForce4 Ti4200 is clearly faster, but that’s about to change.

3DMark2001SE – Build 330
Setting: 4xAA, 8xAF, Quality Mode

GFX Card/GPU	Score
	1024×768
MSI FX5600	5687
GeForce4 Ti4200	4118
	1280×1024
MSI FX5600	3791
GeForce4 Ti4200	2423
	1600×1200
MSI FX5600	2498
GeForce4 Ti4200	1603

As soon as we turn on AA and AF, the MSI FX5600 becomes the obvious winner. If you like to run your games in higher quality antialiased modes, who cares if NoAA is faster? Let me make it clear: Games are quite playable on my Athlon 1900+ and the FX5600 when antialiasing is turned on, although for some games, I turn the AA down to 2x.

Unreal Tournament 2003
Setting: NoAA, NoAF, Quality Mode

GFX Card/GPU	Score
	1024×768
MSI FX5600 – Flyby	118.94
MSI FX5600 – Bots	47.30
GeForce4 Ti4200 – Flyby	131.78
GeForce4 Ti4200 – Bots	49.25
	1280×960
MSI FX5600 – Flyby	85.25
MSI FX5600 – Bots	44.87
GeForce4 Ti4200 – Flyby	100.19
GeForce4 Ti4200 – Bots	48.01
	1600×1200
MSI FX5600 – Flyby	56.88
MSI FX5600 – Bots	35.91
GeForce4 Ti4200 – Flyby	67.80
GeForce4 Ti4200 – Bots	41.00

I used the UPT Benchmark application to test UT2003. I see pretty much the same story as before, except that my system is definitely CPU limited when running the Bots demo. The FX5600 and Ti4200 are running neck-and-neck even with AA turned off, due to my underpowered 1900+.

Unreal Tournament 2003
Setting: 4xAA, 8xAF, Quality Mode

GFX Card/GPU	Score
	1024×768
MSI FX5600 – Flyby	66.25
MSI FX5600 – Bots	37.04
GeForce4 Ti4200 – Flyby	40.68
GeForce4 Ti4200 – Bots	25.83
	1280×960
MSI FX5600 – Flyby	42.28
MSI FX5600 – Bots	25.43
GeForce4 Ti4200 – Flyby	24.69
GeForce4 Ti4200 – Bots	14.56
	1600×1200
MSI FX5600 – Flyby	24.37
MSI FX5600 – Bots	14.91
GeForce4 Ti4200 – Flyby	17.54
GeForce4 Ti4200 – Bots	11.07

All of these scores are a little low for my taste. I like 1024×768 2xAA 4xFF on the MSI FX5600; that plays great. If you really want 4xAA, you’d probably need to drop the resolution down a notch, to 800×600, for reasonable performance.

Return To Castle Wolfenstein – version 1.31, “Checkpoint” demo
Setting: NoAA, NoAF, Quality Mode

GFX Card/GPU	Score
	1024×768
MSI FX5600	90.5
GeForce4 Ti4200	103.6
	1280×1024
MSI FX5600	67.0
GeForce4 Ti4200	100.0
	1600×1200
MSI FX5600	47.5
GeForce4 Ti4200	89.9

The GeForce4 Ti4200 is one heck of a card! Too bad it doesn’t do DirectX 9.

Return To Castle Wolfenstein – version 1.31, “Checkpoint” demo
Setting: 4xAA, 8xAF, Quality Mode

GFX Card/GPU	Score
	1024×768
MSI FX5600	65.2
GeForce4 Ti4200	70.8
	1280×1024
MSI FX5600	42.3
GeForce4 Ti4200	42.3
	1600×1200
MSI FX5600	28.7
GeForce4 Ti4200	N / A

The Ti4200 did not actually fail to run the demo at 1600×1200. What happens is that it runs at 2xAA instead of 4xAA, so I felt obligated to drop its score.

Comanche 4 Demo
Setting: NoAA, NoAF, Quality Mode

GFX Card/GPU	Score
	1024×768
MSI FX5600	31.18
GeForce4 Ti4200	32.57
	1280×1024
MSI FX5600	29.67
GeForce4 Ti4200	32.28
	1600×1200
MSI FX5600	26.19
GeForce4 Ti4200	29.95

Comanche 4 is primarily CPU limited, even on far more powerful CPUs (than my Athlon 1900+)

Comanche 4 Demo
Setting: 4xAA, 8xAF, Quality Mode

GFX Card/GPU	Score
	1024×768
MSI FX5600	23.34
GeForce4 Ti4200	17.53
	1280×1024
MSI FX5600	17.22
GeForce4 Ti4200	8.26
	1600×1200
MSI FX5600	12.35
GeForce4 Ti4200	6.84

Again, the FX5600 shows its stuff once antialiasing is turned on.

Overclocking

One of the best features of NVIDIA’s newest drivers is the Auto Detect button, which will run through some tests to determine the maximum safe overclocking speeds (core and memory). In my case, it suggested 362 / 627 (stock is 325 / 550).

When I overclocked at the suggested speeds, 3Dmark2001SE rebooted (even when I cranked it down to 360 / 625), so all of my overclocking scores use 350 / 625, which was completely stable for the benchmarks. However, when I switched to 8xAA, 8xAF, Quality, I saw a few “sparkly” texture artifacts, so I may need to turn the frequency settings down a little more.

Overclocking – 3DMark2001SE
Setting: Default – NoAA, NoAF, Quality Mode

GFX Card/GPU	Score
	1024×768
MSI FX5600 @ 350 / 625	9195 (+3.7%)
MSI FX5600	8869
GeForce4 Ti4200	10023
	1280×1024
MSI FX5600 @ 350 / 625	7701 (+5.5%)
MSI FX5600	7298
GeForce4 Ti4200	8131
	1600×1200
MSI FX5600 @ 350 / 625	6362 (+3.7%)
MSI FX5600	5856
GeForce4 Ti4200	6629

When overclocked, the FX5600 can almost catch the Ti4200.

Overclocking – 3DMark2001SE
Setting: 4xAA, 8xAF, Quality Mode

GFX Card/GPU	Score
	1024×768
MSI FX5600 @ 350 / 625	6246 (+9.9%)
MSI FX5600	5687
GeForce4 Ti4200	4118
	1280×1024
MSI FX5600 @ 350 / 625	4259 (+12.3%)
MSI FX5600	3791
GeForce4 Ti4200	2423
	1600×1200
MSI FX5600 @ 350 / 625	2841 (+13.7%)
MSI FX5600	2498
GeForce4 Ti4200	1603

A 10% increase (or more) is nothing to sneeze at.

Overclocking – Unreal Tournament 2003
Setting: NoAA, NoAF, Quality Mode

GFX Card/GPU	Score
	1024×768
FX5600 @ 350 / 625 – Flyby	125.42 (+5.4%)
FX5600 @ 350 / 625 – Bots	46.91 (-0.8%)
MSI FX5600 – Flyby	118.94
MSI FX5600 – Bots	47.30
GeForce4 Ti4200 – Flyby	131.78
GeForce4 Ti4200 – Bots	49.25
	1280×960
FX5600 @ 350 / 625 – Flyby	94.10 (+10.4%)
FX5600 @ 350 / 625 – Bots	45.71 (+1.9%)
MSI FX5600 – Flyby	85.25
MSI FX5600 – Bots	44.87
GeForce4 Ti4200 – Flyby	100.19
GeForce4 Ti4200 – Bots	48.01
	1600×1200
FX5600 @ 350 / 625 – Flyby	63.41 (+11.5%)
FX5600 @ 350 / 625 – Bots	38.90 (+8.3%)
MSI FX5600 – Flyby	56.88
MSI FX5600 – Bots	35.91
GeForce4 Ti4200 – Flyby	67.80
GeForce4 Ti4200 – Bots	41.00

The slight drop (-0.8%) in performance is odd, but nothing you’d ever notice.

Overclocking – Unreal Tournament 2003
Setting: 4xAA, 8xAF, Quality Mode

GFX Card/GPU	Score
	1024×768
FX5600 @ 350 / 625 – Flyby	74.01 (+11.7%)
FX5600 @ 350 / 625 – Bots	45.71 (+23.4%)
MSI FX5600 – Flyby	66.25
MSI FX5600 – Bots	37.04
GeForce4 Ti4200 – Flyby	40.68
GeForce4 Ti4200 – Bots	25.83
	1280×960
FX5600 @ 350 / 625 – Flyby	47.52 (+12.4%)
FX5600 @ 350 / 625 – Bots	28.34 (+10.3%)
MSI FX5600 – Flyby	42.28
MSI FX5600 – Bots	25.43
GeForce4 Ti4200 – Flyby	24.69
GeForce4 Ti4200 – Bots	14.56
	1600×1200
FX5600 @ 350 / 625 – Flyby	27.62 (+13.3%)
FX5600 @ 350 / 625 – Bots	16.74 (+12.3%)
MSI FX5600 – Flyby	24.37
MSI FX5600 – Bots	14.91
GeForce4 Ti4200 – Flyby	17.54
GeForce4 Ti4200 – Bots	11.07

A 23% increase is quite good! (And 45-49 FPS is about as fast as my CPU can play this game.)

Conclusion

For those of you who are looking for an NVIDIA-FX-based graphics card with both video-in and video-out, the MSI FX5600-VTDR128 may be exactly what you’re looking for. It’s a well-built card that comes with a lot of extras. It’s an excellent mainstream graphics card, well suited for all but the most extreme gamers, and it’s especially useful for those who need video-in as well as video-out. It might also be appropriate in business situations, like content creation.

When gaming, I’ve settled on 1024×768 4xAA 4xAF in “Performance” mode; if a game plays too slowly, I either crank the AA down to 2x or try 800×600. This card plays demanding games well, even with antialiasing and anisotropic filtering enabled.

I really like the 8xAA with 4x texture sampling mode too: It looks great (especially at high resolutions), but it’s only playable at 960×720 and below, at least on my computer. I get roughly 40 fps in Quake3 (at that res – when overclocked to 350 / 265), so it’s a little jerky at times, but it still plays fine.

Final Score