PNY GeForce FX 5600 256MB Review
By Doug Hall
July 24th, 2003
Summary
The testing consistently shows that the older and more familiar Ti chipset is still a viable candidate in the gaming world as far as Nvidia graphics cards. ATI may rule at the moment, but none of the scores here are by any means unusable or substandard in terms of gaming playability or compatibility. Each card has stood out in the area that it was designed for. Older benchmarks written for DX 8 clearly like the Ti-4600 better than the newer FX card and in spite of the power behind it, the Ti-4600 is simply not capable of surpassing the 5600 FX in the newer technology written into DX 9. The only exception to that is with the Splinter Cell test which was written with DX 8 and released with DX 9. The game engine benefits from the power increase but eye candy in game is not as nice as the 5600 produces. So our finally decision is that for those of you who are into the newest games it is definitely worth the upgrade for the graphics alone. If older games are more you style and Deus Ex is still on your current play list (and why wouldn't it be?) the upgrade is futile and in most cases a waste of financial resources. I think that the issue will be solved decisively with the release of the NV40 chipset. For the rest of you still whacking away in Quake 3 it makes little difference which way you go. I only hope that Team Arena is not on the menu.
Full Review
Introduction
So what do you do when the release axe falls and you are left suddenly with yesterday's hardware and tomorrows obsolesce? Any savvy gamester wants the newest and best that the hardware manufacturers put out, but where do you draw the line in regards to what you have to pay divisible by what you want minus what you actually get which in turn equals whether the upgrade is worth the cost? We need to determine the effectiveness of the upgrade in terms of value by finding what is an efficient use of funds and what capabilities are gained by spending them. If you upgrade for the sake of bragging rights and being able to score well in the latest benchmark (or running it at all) then this probably is not your sort of problem and none of this will concern you too much. If you don't have a spare 3 C note lying around, then you might want to read on.
So, another day in the life of the video card world. Somewhere between greatest and the latest lies the decision of “to upgrade or not to upgrade”? Currently we see no reasons to doubt the coming releases of the NV 40 and Radeon series of GPUs. But will they usher in a new era of smoking fast graphics or simply squeeze the older technology a little more and call it an upgrade? How will the hardware and the gaming industry embrace the things to come and what can we forecast for the foreseeable future? There is nothing certain except that technology does not go backwards.
Unless Nvidia can pull it out in the near future then I would expect that ATI will be the leader for quite a while to come. Aside from the obvious differences in the Catalyst drivers vs. the Detonators, ATI seems to be on to something in the GPU architecture that allows for more advanced and flexible features for the near future. Personally we are fine with anything that is not choppy or necessitates turning off the eye candy in order to make whatever favorite game is in the CDROM playable. That is enough for gaming. Certainly the added ability to realize frame rate in a setting that is not what one would call realistic but is closer to the real thing that a home movie on your dad's 8mm movie camera when you were a kid is plenty good enough. Usable as it were. But how cool is simply usable? Not very. Graphics have come so far lately that the reason for spending the money is and should be for the graphic content itself. Game play is up to the game itself.
The purpose of this review is not only to see which is faster or has better quality and capabilities, but also to determine whether the last release of the FX chipset, is a worthwhile upgrade. By most accounts it blew chunks in terms of hype to actuality ratio. For this reason we have chosen the PNY Ti-4600 with 128 megs of ram and the PNY 5600FX with 256 megs of ram. Both have identical bandwidth and are comparable in price. The 4600 is faster and more powerful but the 5600 is newer and AGP 3.0 or 8x compatible. Both have their advantages and disadvantages. We can test with the current set of benchmarks and utility offerings available and see how viable an upgrade might be in the real world. Since both are sold by the same manufacturer then it should give us an honest representation of what the possibilities might be.
The 4600 was retail for $299 about 6 months back and the 5600 was $249 about 6 weeks ago so the costs are some what comparable. PNY was chosen because of the overall availability of them and past experiences with them have been positive.
|
PERFORMANCE FEATURES
|
||||
|
256-bit graphics core
|
256-bit graphics core
|
|||
|
128-bit DDR memory interface
|
128-bit DDR memory interface
|
|||
|
10.4GB/sec. memory bandwidth
|
8.8GB/sec. memory bandwidth
|
|||
|
4.8 billion AA samples/sec. fill rate
|
1.3 billion texels/sec. fill rate
|
|||
|
136 million vertices/sec setup
|
81 million vertices/sec setup
|
|||
|
Dual 350 MHz RAMDACs
|
Dual 400MHz RAMDAC’s
|
|||
From the table, we can see that both cards use the same 256 bit core and 128 DDR memory interface. These are two evenly matched cards in terms of core and the interface used to access the memory on board. So why the difference in Bandwidth? The 4600 has a reported 10.4 GBs/sec. while the 5600 with 128 more megs of ram onboard is down a bit at 8.8 GBs/sec. What’s up with that? Let's find out.
Here are the sys specs for our test system. All the speeds and voltages are default except for the FSB which is set at 166. It will default to 100 upon the intial install of the CPU so it was set manually 166. No OC on this machine means that the results are viable for a real wold system and not optimized for either graphics adapter. The same drivers and DX version 9.0a is used for both cards and reinstalled with each hardware installation. Also, Windows XP professional was used with all patches, hotfixes, and Service Packs available applied at the time of testing.
The 4600 is pushing 86 million triangles a sec. and the 5600 FX is pushing 81 million vertices a second. A vertex is the point of a triangle, so each triangle is composed of 3 vertices each. This would lead us to surmise that the value that the 5600 can truly produce is 54 million triangles/sec roughly, which seems realistic, and maps more to the Ti-4400 for the bandwidth available.
Vertex shaders are the calculations applied to each individual point where they meet (the X,Y and Z coordinates) in order to produce the illusion of reality such as skin tones and animation to movement. A vertex is a special point of a mathematical object, and is usually a location where two or more lines or edges meet. Vertices are most commonly encountered in angles, polygons, polyhedra, and graphs. Graph vertices are also known as nodes.
This is what gives an image the 3D quality based on what angle the vertices are connected. There is no easy calculation to determine the amount triangles as it varies based on the shape of the object and how it is put together. A 6 vertices corner has much more dimension than a 2 virtices corner. Counting them is your only option for accuracy.
 |
 |
 |
|
34834 vertices
104288 edges 69451 triangles 1435697 bytes |
22 vertices
60 edges 40 triangles 402 bytes |
8 vertices
18 edges 12 triangles 156 bytes |
The Ti-4600 has dual RAMDACs (random access memory digital to analog converter) @ 350 MHz and the 5600FX has Dual @ 400 MHz. a difference of 11.4 % overall which should even out the scales some what in the performance arena.
The FX chipset which will process a lot more data than the Ti-4600 because it will process DX9 extensions, and at a faster rates because of the 8x capabilities, so it should be a pretty even match in most areas. Scoring will be purely benchmark based with the same drivers (Detonator 44.03 which are WHQL certified for the FX) and Direct X 9.0a with the exact same settings and tests run on the same machine. I feel as though the advantages of either will balance out for the weaknesses of either.
Performance
It would appear at this point that the newer 3Dmark benchmark is a little on the rough side in scoring the Nvidia card in general. This is simply not the case as it was written to test DX9 and AGP 3.0 features from the very start. We simply wish to show here that the differences in scoring from two different generations of video cards can be seemingly different on the same test. As in the 3Dmark2001se it appears that the 4600 is much more powerful overall and the 5600FX is better equipped to handle newer technologies.
The difference in the 3Dmark2001se score would lead you to believe that the Ti-4600 is quite a bit more powerful than the 5600 FX card. I would have thought that the 5600 FX would have shown better results then the Ti-4600. Since that is not the case, and consistent results have been achieved time and time again, that I would have to surmise that this test runs routines that the 5600 FX is not built to run and it focuses more on newer and more advanced technologies ands is optimized for them.
The Ti-4600 smokes the 5600 FX in the UT2K3 benchmark proving its more powerful but why the difference in the 3Dmark scores. Seems odd doesn’t it. Keep in mind that the UT2K3 scores are designed for the Geforce4 at 4x as DX9 and FX chipset did not exist at the time the game was introduced.
All test are run with all available options in the Benchemall! API and framerate scores are either stated as to resolution used for completed tests, or an average of all available modes completed. Interesting to note in this particular set of tests is that though both cards are miles apart in terms of their respective technologies, Quake 3 arena scores are almost identical in this situation. This would indicate that the FX series of cards has poor OpenGL support or nVIDIA is not doing much to improve or develop the technology for the FX line. This is further supported in the Vulpine GL_Mark as it is another OpenGL test where we would have expected to see a great improvement over the Ti-4600 if this was the case.
Conclusion
The testing consistently shows that the older and more familiar Ti chipset is still a viable candidate in the gaming world as far as Nvidia graphics cards. ATI may rule at the moment, but none of the scores here are by any means unusable or substandard in terms of gaming playability or compatibility. Each card has stood out in the area that it was designed for. Older benchmarks written for DX 8 clearly like the Ti-4600 better than the newer FX card and in spite of the power behind it, the Ti-4600 is simply not capable of surpassing the 5600 FX in the newer technology written into DX 9. The only exception to that is with the Splinter Cell test which was written with DX 8 and released with DX 9. The game engine benefits from the power increase but eye candy in game is not as nice as the 5600 produces. So our finally decision is that for those of you who are into the newest games it is definitely worth the upgrade for the graphics alone. If older games are more you style and Deus Ex is still on your current play list (and why wouldn’t it be?) the upgrade is futile and in most cases a waste of financial resources. I think that the issue will be solved decisively with the release of the NV40 chipset, but we will wait for the Nvidia to provide one before we can comment officially on that one. For the rest of you still whacking away in Quake 3 it makes little difference which way you go. I only hope that Team Arena is not on the menu.
Performance
Benchmark session result
Operating System: Windows XP Professional (5.1, Build 2600) Service Pack 1 (2600.xpsp1.020828-1920)
Processor: AMD Athlon(tm) XP 3000+, MMX, 3DNow, ~2.2GHz
Memory: 1024MB RAM
DirectX Version: DirectX 9.0 (4.09.0000.0900)
Card name: NVIDIA GeForce FX 5600
Driver Version: 6.14.0010.4403 (English)
Detailed info dir:
C:\Program Files\BenchEmAll\result\2003.07.16.01.05.36.\
-----------------------------------------------------------
Splinter Cell 1.2b
800x600 Average= 23.617 fps, min 9.879 fps, max 77.945 fps. Effects quality= VERYHIGH
1024x768 Average= 20.390 fps, min 9.146 fps, max 75.238 fps. Effects quality= VERYHIGH
1280x1024 Average= 16.619 fps, min 8.102 fps, max 59.675 fps. Effects quality= VERYHIGH
-----------------------------------------------------------
Unreal II: The Awakening
800x600 Average: 56.38 Unreal 829, 31.41 Min, 124.49 Max,
1024x768 Average: 43.61 Unreal 829, 24.03 Min, 74.38 Max,
1280x1024 Average: 31.47 Unreal 829, 17.21 Min, 50.46 Max,
1600x1200 Average: 23.62 Unreal 829, 12.37 Min, 38.85 Max,
-----------------------------------------------------------
3DMark03
800 x 600 2528 3DMarks
1024 x 768 1938 3DMarks
1280 x 1024 1395 3DMarks
1600 x 1200 1061 3DMarks
-----------------------------------------------------------
ChameleonMark
800x600 glass= 82.8484 fps, real= 115.266 fps, shiny= 83.9416 fps, Average= 94.019 fps.
1024x768 glass= 63.5764 fps, real= 93.7892 fps, shiny= 63.7935 fps, Average= 73.720 fps.
1280x1024 glass= 44.6069 fps, real= 69.5996 fps, shiny= 44.7069 fps, Average= 52.971 fps.
1600x1200 glass= 33.9889 fps, real= 54.5222 fps, shiny= 33.9889 fps, Average= 40.833 fps.
-----------------------------------------------------------
3DMark2001 SE Pro
3DMark Score 9849
-----------------------------------------------------------
Unreal Tournament 2003
800x600 flyby - 157.899776 botmatch - 75.851051
1024x768 flyby - 117.381157 botmatch - 67.614164
1280x960 flyby - 78.706215 botmatch - 53.237398
1280x1024 flyby - 74.273758 botmatch - 49.149206
1600x1200 flyby - 51.358450 botmatch - 36.862101
-----------------------------------------------------------
Quake III Arena
800x600 20.8 seconds: 60.5 fps
1024x768 21.1 seconds: 59.7 fps
1280x1024 21.0 seconds: 60.0 fps
1600x1200 21.6 seconds: 58.3 fps
-----------------------------------------------------------
Vulpine GLmark by I-D Media AG
Command line:
-benchmark -fullscreen -advanced -s3tc -highdetail -res 1024 768 -colorbits 32 -texturebits 32
Average fps : 54.3
-----------------------------------------------------------
Gun Metal Benchmark Results
Created: 14/07/2003 00:04:08
Benchmark 1
Validity
Results valid? Yes
Benchmark version V1.20S
Statistics
Minimum frames per second 5.41 fps
Average frames per second 11.77 fps
Maximum frames per second 38.42 fps
Benchmark duration 318.47 seconds
Total frames rendered 3748
Computer information
Processor AMD Athlon(tm) XP 3000+
Processor speed 2205 Mhz
Operating System Windows XP Service Pack 1
Graphics card NVIDIA GeForce FX 5600
Game settings
Screen resolution 1024 X 768
Screen colors True (32-bit)
Aspect ratio Normal
Antialiasing 2x
Terrain detail High
Draw distance High
Cloud shadows On
Lens flare On
Sun Glare On
--------------------------------------------------------------------------------
Gun Metal Benchmark Results
Created: 14/07/2003 00:11:01
Benchmark 2
Validity
Results valid? Yes
Benchmark version V1.20S
Statistics
Minimum frames per second 4.21 fps
Average frames per second 14.55 fps
Maximum frames per second 55.12 fps
Benchmark duration 340.64 seconds
Total frames rendered 4957
Computer information
Processor AMD Athlon(tm) XP 3000+
Processor speed 2205 Mhz
Operating System Windows XP Service Pack 1
Graphics card NVIDIA GeForce FX 5600
Game settings
Screen resolution 1024 X 768
Screen colors True (32-bit)
Aspect ratio Normal
Antialiasing 2x
Terrain detail High
Draw distance High
Cloud shadows On
Lens flare On
Sun Glare On
Processor
Model : AMD Athlon(tm) XP 3000+
Speed : 2.16GHz
Model Number : 3000 (estimated)
Performance Rating : PR3136 (estimated)
Type : Standard
L2 On-board Cache : 512kB ECC Synchronous Write-Back
Mainboard
Bus(es) : AGP PCI USB FireWire/1394
MP Support : No
System BIOS : Phoenix Technologies, LTD ASUS A7N8X2.0 Deluxe ACPI BIOS Rev 1003
Mainboard : ASUSTeK Computer INC. A7N8X2.0
Chipset : ASUSTeK Computer Inc nForce2 AGP Controller
Front Side Bus Speed : 2x 166MHz (332MHz data rate)
Total Memory : 512MB
Video System
Monitor/Panel : ViewSonic G810-2
Adapter : NVIDIA GeForce FX 5600
Physical Storage Devices
Removable Drive : Floppy disk drive
Hard Disk : WDC WD800JB-00CRA1
CD-ROM/DVD : E-IDE CD-ROM 52X L
Logical Storage Devices
1.44MB 3.5" (A:) : N/A
Hard Disk (C:) : 74.5GB (71.6GB, 96% Free) (NTFS)
Nvidia nforce2 (D:) : 608MB (0MB, 0% Free) (CDFS)
Peripherals
Serial/Parallel Port(s) : 0 COM / 1 LPT
USB Controller/Hub : Standard OpenHCD USB Host Controller
USB Controller/Hub : Standard OpenHCD USB Host Controller
USB Controller/Hub : Standard Enhanced PCI to USB Host Controller
USB Controller/Hub : USB Root Hub
USB Controller/Hub : USB Root Hub
USB Controller/Hub : USB 2.0 Root Hub
FireWire/1394 Controller/Hub : OHCI Compliant IEEE 1394 Host Controller
Keyboard : Standard 101/102-Key or Microsoft Natural PS/2 Keyboard
Mouse : Microsoft USB IntelliMouse Explorer
Human Interface : Microsoft USB IntelliMouse Explorer
MultiMedia Device(s)
Device : MPU-401 Compatible MIDI Device
Device : NVIDIA(R) nForce(TM) Audio Codec Interface
Device : NVIDIA(R) nForce(TM) MCP Audio Processing Unit (Dolby(R) Digital)
Operating System(s)
Windows System : Microsoft Windows 2000 Professional Version 5.00.2195
Network Adapter(s)
Network Drivers Enabled : Yes
Adapter : NVIDIA nForce MCP Networking Adapter
Specs
| |||
 | |||
| ||||||||||||||||||||||||||||||||||||||||||||||||||