AOpen AK86-L UPDATE: Uncommonly Good
by Wesley Fink on March 30, 2004 1:00 PM EST- Posted in
- Motherboards
AOpen AK86-L: Overclocking and Stress Testing
FSB Overclocking Results
BIOS options don't mean much unless they can translate into improved performance. The AOpen AK86-L delivers overclocking as good as we have seen with this early Athlon 64 3200+.| Front Side Bus Overclocking Testbed | ||
| Default Voltage | Maximum Overclock | |
| Processor: | Athlon64 2.0GHz |
Athlon64 2.0GHz |
| CPU Voltage: | 1.5V (default) | 1.55V (Maximum vCore) |
| Cooling: | AMD Stock Athlon64 Heatsink/Fan | AMD Stock Athlon64 Heatsink/Fan |
| Power Supply: | Antec TruePower 430W | Antec TruePower 430W |
| Maximum OC: | 2223MHz (+11%) 222FSB |
2233MHz (+11.5%) 223FSB |
The above overclocking setup at default voltage allowed us to reach a stable FSB of 222. As you can see, increasing voltage did not help very much, topping out at 223. These are the highest overclocks that we have seen with this Athlon 64 and we suspect 222 is the limit on air-cooling with this CPU.
By dropping multipliers in the BIOS, we were able to reach FSB settings as high as 252. It appears that the AGP/PCI is ratio controlled and that it again drops to 33/66 at a setting of 234. This means the 252 is a PCI/AGP overclock of about 19 which is in-line with the maximum overclock our picky ATI 9800 can handle. This is not a working PCI lock as we have seen on Intel chipsets; it is a ratio-driven PCI/AGP frequency similar to the one we first saw on the Abit K8T800 motherboard.
Front Side Bus Stress Test Results:
As part of normal overclocking tests, a full range of stress tests and benchmarks were run to ensure the AK86-L was stable at each overclocked FSB speed. This included Prime95 torture tests, and the addition of other tasks - data compression, various DX8 and DX9 games, and apps like Word and Excel - while Prime95 was running in the background. Finally, we ran our benchmark suite, which includes Veritest Winstone 2004 suite, Unreal Tournament 2003, SPECviewperf 7.1, and Aquamark 3. 222MHz was the highest overclock that we were able to achieve with the AOpen while running these tests at default voltage.Memory Stress Test Results:
The memory stress test is very basic, as it simply tests the ability of the AK86-L to operate at its officially supported memory frequency (400MHz DDR) at the lowest supported memory timings that our Mushkin PC3500 Level 2 or OCZ PC3500 Platinum Ltd Modules will support. Memory stress testing was conducted by running RAM at 400MHz with 2 DIMM slots filled.| Stable DDR400 Timings - 2 DIMMs (2/3 DIMMs populated) |
|
| Clock Speed: | 200MHz |
| Timing Mode: | N/A |
| CAS Latency: | 2.0 |
| Bank Interleave: | N/A |
| RAS to CAS Delay: | 2T |
| RAS Precharge: | 6T |
| Precharge Delay: | 2T |
| Command Rate: | N/A |
We had no problem running 2 DS 512MB DIMMs of our standard Mushkin PC3500 Level2 or OCZ PC3500 Platinum Ltd in the AOpen AK86-L. Both of these memories were completely stable at the default memory voltage of 2.5V with the fastest 2-2-2-6 timings at DDR400.
Filling all available memory banks is more strenuous on the memory subsystem than testing 2 DIMMs. AOpen states clearly that only 2 DIMMs are supported at DDR400 speed, but we decided to try 3 DIMMs at that speed. At first, we thought 3 DIMMs were working fine in this board, but the AOpen BIOS had automatically lowered memory speed when 3 DIMMs were detected. We then forced DDR400, but could not get 3 DIMMs to work at this speed. By lowering the memory speed to DDR333, or setting to Auto and allowing the AOpen BIOS to manage the speed, we were able to run 3 DIMMs at fastest timings at DDR333. We did not have 1GB DIMMs for testing to check the AOpen claim of 2GB support for DDR400 and 3GB at DDR333.
| Stable DDR400 Timings - 3 DIMMs (3/3 DIMMs populated) |
|
| Clock Speed: | 166MHz |
| Timing Mode: | N/A |
| CAS Latency: | 2 |
| Bank Interleave: | N/A |
| RAS to CAS Delay: | 2T |
| RAS Precharge: | 6T |
| Precharge Delay: | 2T |
| Command Rate: | N/A |
We tested the memory timings with 2 banks filled using several stress tests and general applications to guarantee stability. Prime95 torture tests were successfully run at the timings listed in the above charts. We also ran Sciencemark (memory tests only) and Super Pi. None of the three stress tests created any stability problems for the AK86-L at these memory timings. These memory performance results are much better than we normally see with K8T800 chipset motherboards.
Facebook
Twitter
RSS
22 Comments
View All Comments
howminn - Thursday, May 20, 2004 - link
I've got that board and am having hell with it. No matter what agp card, no matter what memory and in which memory slot, the board won't switch the monitor back on after restart. Which means it is not possible to install an OS onto it.I've returned the original board and bought another one, both boards have displayed the same trouble, and have both displayed cmos checksum error, default values loaded. Batteries have been changed, still no solution.
The vendor to whom I RMA'd the board have not given me any feedback as to what is wrong.
Has anybody else seen this before? Please!
cowdog - Sunday, April 4, 2004 - link
Good to see the update with information that AOpen "opened" up the bios options! Good job AOpen!Ditto what #14 said about the AK89 Max. Come on AOpen, don't overlook your nForce3 150 board!! DDR to 3.0v, cpu to 1.8v, fsb to 300Mhz. That would make my day, esp if AOpen also put that hyper-active watchdog on a shorter leash.
Ronnie - Wednesday, March 31, 2004 - link
Hey TrogdorJW, I also have the MSI K8T Neo and found that OCZ pc3200 works great for my setup, I tried some kingston hyperx pc-4000 with no luck.elixia - Wednesday, March 31, 2004 - link
Hey AMD4ME2, I had that problem on an old FIC motherboard. I simply pinned the power cable to the power supply and plugged it in. It will not hang in front of the CPU and block airflow this wayResh - Wednesday, March 31, 2004 - link
There was mention of a new board revision (as opposed to BIOS version). Could AT post the number codes for the new and old rev. to help us in purchasing?Thanks!
AMD4ME2 - Tuesday, March 30, 2004 - link
I'm a bit confused by the wording of the location of the ATX power connectors. they look to me like they are behind the processor blocking air flow, at least in my case they would be.Chuckles - Tuesday, March 30, 2004 - link
There is an error on page one (Index)."Adjustments for memory voltage have now been extended from 2.5V to 3.0v in 0.5v increments. This is an extremely wide and useful range for users trying to get the most from their memory."
Should be 0.05V instead of 0.5.
Pumpkinierre - Tuesday, March 30, 2004 - link
I posted my question on Feb.16(#12) and got my answer on March 30th (#13). Its never too late. Thanks for the follow-up.Venomous - Tuesday, March 30, 2004 - link
Westley, re: the AK89MAX.. Have them do the same kind of BIOS mods they did on the AK86. If they can bring those voltages up to AK86 specs, im sure hitting 280 fsb wont be hard. The watchdog thing IS annoying.Wesley Fink - Tuesday, March 30, 2004 - link
Pumkinierre -1.06Q was never a BIOS published by AOpen, so 1.08c is the first published BIOS that has multipliers. Ratios are the same as 1.06q, which is whole numbers - no option of 0.5 multipliers.
As I discussed in the PCI lock article and comments, the AK86-L seems to float the PCI/AGP up to 233, then at 234 it drops back to 33.3. This is a ratio arrangement very similar to what we saw in the Abit K8T800 motherboard we reviewed, and is much less flexible than a true adjustable PCI/AGP lock. If you look at the new screen captures you will see PCI/AGP now reports the rising frequency in the BIOS.
As mentioned in the update, the nForce3-150 based Gigabyte K8NNXP and the Shuttle AN50R, both nF3-150 based, are the only 2 Athlon 64 boards that I am aware of that are reported by respected writers to have AGP/PCI lock. I have not tested either with the PCI Geiger, so I can not personally confirm this. If you want fine adjustments for OC, then either of these would appear to be a better choice, at a higher price.
The AOpen AK89 Max, which is a very late nF3-150 board, also appears to have a working PCI/AGP lock, but there are some issues with high settings prematurely invoking the watchdog feature and resetting the frequency - requiring a CMOS clear. We have asked AOpen for help in resolving this issue with the AK89 Max, and if it is fixed we will post a review on Anandtech.