JNav89GT's Test Setup |
CPU | Athlon FX53 939 Retail-Cooled with retail cooler |
Mainboard | Asus A8V rev 2.0, bios 1006(shipping bios) |
Memory | OCZ PC4000 Gold Rev 2.0--2x512mb or 1GB. using timings of 2.5:3:3:8/1T at 2.8v(bios max) |
Video | PNY 6800GT@410/1100 61.77 driver |
PSU & HDD | * Enermax EG465P-VE * Western Digital WD800JB 80GB 7200rpm 8MB cache |
Software | Windows XP Professional, DX 8.1 for all benches except Doom3 for which DX9b was added |
Testing:
Due to limited time, and basically a plethora of benches available around the internet running the Asus A8V at stock speeds I will jump straight to the "sweet spots" of this board.
I chose OCZ PC4000 Gold Rev 2.0 memory to attempt to show how high the board can overclock running the memory at 1:1 with the HTT bus. After hitting a wall with memory at 1:1 I then set memory at the 5:3 divider and continued to run the HTT bus to the highest stable speed I so I could find the limit of my particular board. All testing was able to be completed using the 4X HT multiplier in bios. Lowering this to 3x did not improve HTT overclock over the stable limit I found. CPU multipliers were altered to remove CPU frequency as much as possible confounder in stability.
That said, I could hit 250HTT bus at 1:1 on the memory. I will say that this memory has been tested in other boards up to 260mhz at Cas 2.5, 3:3:8 timings stable, so I am going to assume the limited bios voltage of 2.8v is somewhat holding me back from going higher, or possibly it's a bios issue. Memory cooling was provided by a homemade 60mm fan mounted on a bracket I made.
Of note, I could bench higher but stability was not good and Prime 95 would error almost instantly. At 250HTT and memory at 250 MHz the system was rock solid during my testing. By moving the memory to a 5:3 or PC2700 divider, I was able to scale the HTT to a very respectable 283HTT bus. I chose to back off HTT to 278 and bring CPU multiplier back to 9x to bring MHz back to 2.5ghz to closer correspond with testing done at 250HTT.
What does difference does the HTT speed make in benchmarks? Find out on the next page ->