Just under 4% advantage for the dual channel platform! Small, very small difference! I should add that all the screenshots with the socket 754 NV8 can be seen in the old topic where I tested only the first platform for the Athlon 64: Re: Battle of the platforms: socket 754!

16. 3dmark 2000 - 1280*1024
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754

17. 3dmark 2000 - 1600*1200:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754

And the difference shrinks to less than 1%! Until now Athlon 64 3800 does not deserve its rating which should be far lower than what AMD went for! We shall see if the situation changes in the most popular benchmark on VOGONS: 3dmark 2001.

Why i love S754 more than S939, it has AMD Turion 64, Mobile Sempron support, low power options.
Why i love S939 more than S754, it has Dual Core, it has Dual Channel and its faster than Socket 754.

18. 3dmark 2001 - the yardstick after which all retro systems are measured! Default settings first:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754

Less than 4% advantage for the dual channel platform!

19. 1280*1024 - max details:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754

The difference shrinks to just 2.5%!
20. 1600*1200 - max details:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754

Again the PR difference between these 2 processors is not justified! Dual channel does not bring an extraordinary jump in 3d performance. Socket 939 Athlon 64 3800 is far closer to socket 754 Athlon 64 3400 than AMD would have us believed!

21. The last 3dmark version is 2003! Already radeon x1950 pro is beginning to limit performance in this benchmark at higher resolutions so no point in testing 3dmark 2005!
1024*768 default settings first!
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754

Dual channel?? What dual channel? The socket 939 Newcastle suffers a humiliation as it provides practically the same performance as the socket 754 counterpart!

22. 1280*1024 - max details:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754

23. 1600*1200 - max details:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754

24 &25. The only version of 3dmark to feature a sound test, the NO SOUND and 24 SOUNDS can be done even with the basic AC97 codec included in the nforce4 while for the 60 SOUNDS you need a hardware accelerated solution.
You can see the enormous difference - 18% and 16% for the no sound, respectively 24 sounds - between socket 754 and the newer socket 939. So dual channel does indeed help a lot with the sound processing - this should be seen later in the games.

26 & 27 & 28. We reached benchmarks that test overall system performance and the first is PCMARK 2002.
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754

CPU scores are practically indistinguishably, after all we are talking about the same 2,4 GHz/512 KB L2 cache processor; memory shows a 24% better score in the case of dual channel KN8 and raptor score improves on the socket 939 board despite using the same chipset!

29 & 30 & 31 & 32. PCMARK 2004 is a completely overhaul suite with a graphics score added into the mix!
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754

For some mysterious reason I forgot to select memory tests the first time around so I retested! CPU scores are again equal as they should be; so are the hard drive score; socket 939 flexes its muscles with a 23% better memory score and also better performance in the graphics department!

33 & 34 & 35 & 36. Last PCMARK to be tested is the 2005 one which is an enhanced version of the previous one.
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754

Again CPU and HDD scores are really close; dual channel leads to 17% better memory score and also 10% better graphics performance! So dual channel is working as it should be.
It is time to see if games take advantage of this feature or not!

37. Finally we reached the games; the first one is the Directx8 Unreal tournament 2004. Umark benchmark utility refuses to run with UT 2k4 GOG edition that I have so I used fraps to test the Death match Antalus map in spectate mode. After loading the map I waited 1 min before pressing F11 for benchmark. After 1 minute of testing I changed the resolution and waited again 1 minute before benchmarking and so on. I repeated 3 times and kept the final 3 runs at 1024*768, 1280*1024 and 1600*1200. Please note that I used maximum details and also sound! The radeon card is bottlenecked heavily by the CPU in this game!

1024*768:
ATHLON 64 3800 NEWCASTLE - SOCKET 939
2026-05-19 20:40:16 - UT2004
Frames: 7820 - Time: 60000ms - Avg: 130.333 - Min: 77 - Max: 219

ATHLON 64 3400 NEWCASTLE - SOCKET 754
2025-09-22 04:45:51 - UT2004
Frames: 7714 - Time: 60000ms - Avg: 128.567 - Min: 89 - Max: 185

38. And almost no advantage for a CPU with a launch price of 720USD versus the older socket model with a cost of 417USD!

UT2K4 - 1280*1024
ATHLON 64 3800 NEWCASTLE - SOCKET 939
2026-05-19 20:46:37 - UT2004
Frames: 7795 - Time: 60000ms - Avg: 129.917 - Min: 86 - Max: 186

ATHLON 64 3400 NEWCASTLE - SOCKET 754
2025-09-22 04:44:25 - UT2004
Frames: 7496 - Time: 60000ms - Avg: 124.933 - Min: 75 - Max: 195