Reply 160 of 171, by nd22
- Rank
- l33t
Memory timings are 3-3-3-8 CR1T and the latest BIOS version is used on both boards; all Newcastle come with the CG stepping:
Memory timings are 3-3-3-8 CR1T and the latest BIOS version is used on both boards; all Newcastle come with the CG stepping:
To summarize the hardware and software used:
System 1 - socket 754:
CPU: Athlon 64 3400 Newcastle
Cooler: Arctic freezer 64 pro
Motherboard: Abit NV8 with nforce4 chipset
RAM: corsair 2*1gb at 3-3-3-8 CR 1T timings
GPU: radeon x19590 pro
SSD: Kingston 960gb SATA3 that works perfectly on all nforce4 boards!
HDD: WD 74gb raptor for storage tests
ODD: some SATA DVD-RW that I got from storage
PSU: Antec true power trio 650w - I bought it back in 2008; inspected before testing, still works just fine
System 2 - socket 939:
CPU: Athlon 64 3800 Newcastle
Cooler: Arctic freezer 64 pro
Motherboard: Abit KN8 with nforce4 chipset
RAM: corsair 2*1gb at 3-3-3-8 CR 1T timings
GPU: radeon x19590 pro
SSD: Kingston 960gb SATA3
HDD: WD 74gb raptor for storage tests
ODD: some SATA DVD-RW that I got from storage
PSU: Antec true power trio 650w
Drivers:
- nvidia chipset driver 15.23 - latest for nforce4
- radeon driver 10.2 - latest and greatest for radeon!
- realtek AC97 driver version 4.06
Software:
- Windows XP SP3 with net framework 2.0 installed
- msi afterburner 4.30
- fraps 3.4.7
- irfanview for all screenshots
- everest 5.50
For easier viewing I selected the default Luna theme with the green variation for the Socket 939 with Newcastle 3800 and Royale theme for the Socket 754 with Newcastle 3400. Athlon 64 3800 will always be first, Athlon 64 3400 second.
1. First test is Code creatures pro at default settings - that means no AA, no sound:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754
Socket 939 got a 6% lead thanks to dual channel, still quite far from the 11.76% PR difference!
2. Same as above only with sound this time:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754
Only 3% separates now the 2 Newcastle - the difference in the ratings just doesn't add up!
3. Same resolution but with AA this time as it is the only quality setting - besides the resolution and sound - that can be modified in Code creatures pro:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754
4. The difference shrinks ever more! Time to test with sound:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754
Please note the very low impact of sound over the results despite the fact that nforce4 uses the same old AC97 codec!
5. We increase the resolution to 1280*1024 - standard for 17 and 19 inches monitors that can be had for practically nothing today!
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754
6. We stay at the same 1280*1024 resolution and test with sound and got the following:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754
Just under 2% is the advantage the dual channel socket 939 has over the single channel socket 754!
7. And now the last resolution: 1600*1200
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754
8. Final and most demanding settings: 1600*1200, AA 4x with sound:
A. Athlon 64 3800 - socket 939
B. Athlon 64 3400 - socket 754
Despite using an overpowered radeon x1950 pro in this DirectX8 benchmark the difference between socket 939 and socket 754 is far below the performance rating difference of these 2 processors!
Maybe we can get better scaling in the next benchmark which is Aquamark3!