Socket 754 Roundup: Comparing Generation 2
by Wesley Fink on May 28, 2004 5:57 PM EST- Posted in
- Motherboards
Epox 8KDA3+: Overclocking and Stress Testing
FSB Overclocking Results
| Front Side Bus Overclocking Testbed | |
| Default Voltage | |
| Processor: | Athlon 64 3200+ 2.0GHz |
| CPU Voltage: | 1.5V (default) |
| Cooling: | AMD Stock Athlon 64 Heatsink/Fan |
| Power Supply: | Antec TruePower 430W |
| Maximum OC: (Standard Ratio) |
248FSB x10 2480MHz (+24%) |
| Maximum FSB: (Lower Ratio) |
283FSB x 8 at 1:1 Memory |
The working AGP/PCI lock on the nForce3-250 is allowing all our test boards to reach the 246-248 range with this late 3200+. A 24% overclock at stock multiplier is the kind of overclocking performance that many have been looking for since the Athlon 64 was first introduced. Frankly, tests of earlier chips were lucky to reach a 10% overclock, so our results here are no guarantee that your Athlon 64 can achieve this same overclock level.
Dropping the multiplier, and testing 1:1 with one DDR550 DIMM, we achieved the highest 1:1 FSB overclock that we have ever accomplished on an Athlon 64 board at 283FSB or DDR566. This result is as good as we have been able to reach with this same memory on our Intel memory test bed; that speaks well for the stability of the Epox 8KDA3+ design. Epox has done an outstanding job with the BIOS updates for the 8KDA3+. The Epox was not very impressive when we first saw it, but the Epox BIOS magic has turned the 8KDA3+ into a remarkable overclocker.
HyperTransport could be maintained at the 4X (800 setting) up to a 266 FSB setting, which is excellent performance. Above this point to the 1:1 maximum of 283 FSB, we needed a 3X HT setting. We're a little surprised that Epox did not provide a 5X HT setting in BIOS, but even without that option, the Epox 8KDA3+ was an exceptional performer.
Memory Stress Test Results:
The memory stress test is very basic, as it simply tests the ability of the Epox 8KDA3+ to operate at its officially supported memory frequency (400MHz DDR), at the best performing 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: | 10T (10T for Best Performance)* |
| Precharge Delay: | 2T |
| Command Rate: | N/A |
*Several memory tests have shown that memory performs fastest on the nVidia nForce and VIA K8T800 chipsets at a TRas (RAS Precharge) setting in the 9 to 13 range. We ran our own Memory Bandwidth tests with memtest86 with TRas settings from 5 to 15 at a wide range of different memory speeds. The best bandwidth was consistently at 9 to 11 at every speed, with TRas 10 always in the best range at every speed. The memory bandwidth improvement at TRas 10 was only 2% to 4% over TRas 5 and 6 depending on the speed, but the performance advantage was consistent across all tests. Since best performance was achieved at 2-2-2-10 timings, all Athlon 64 benchmarks were run at a TRas setting of 10.
The original BIOS of the Epox 8KDA3+ was rock solid with one DIMM, but not very stable with two DIMMs - requiring slower timings. The latest BIOS greatly improves the Epox memory performance, allowing the same aggressive timings with 2 DIMMs that we could use with one DIMM.
Filling all three available memory slots is more strenuous on the memory subsystem than testing 2 DIMMs on a motherboard. Epox does not recommend using 3 double-sided DIMMs, and in this case, it is solid advice. No matter what we tried, or how slow the timings we set, 3 double-sided DIMMs would not operate in the 8KDA3+. We do not consider the inability to run 3 DS DIMMs a real handicap on a Single-Channel memory motherboard, but if this is an important feature to you, the Epox is not the motherboard to run 3 DIMMs. Despite the fact that 3 DS DIMMs are not officially supported on the nForce3-250, the other nF3-250 boards did manage to run 3 DIMMs, some at the same aggressive timings used for 2 DIMMs.
Facebook
Twitter
RSS
30 Comments
View All Comments
Odeen - Monday, May 31, 2004 - link
There is a difference between SATA native to chipset, and SATA native to the OS.SATA native to chipset means the chipset runs the SATA internally, off the Hypertransport or V-Link bandwidth, which is at least a gigabyte/second.
Intel's implementation of SATA is cool because to the OS it emulates a standard IDE controller. (Thus it's "native" to the OS.) The downside of this approach is that every OS other than XP gets horribly confused seeing two primary and two secondary controllers. (i.e. your PATA1 is Primary, your PATA2 is Secondary, your SATA1 is.. again Primary) Without "compatibility" options in the BIOS, which limits you to four drives total (i.e. SATA channels become masters on IDE1 and IDE2, with PATA masters becoming slaves, and PATA slaves dropping off the map, or, as an alternative, PATA2 disappearing, and SATA1 and SATA2 becoming PATA1 Master and PATA1 Slave) Win2K and DOS-based utilities (such as bootable Antivirus or Partitioning program CD's and utilities like the drive test disks that you get with a hard drive,) fail on startup.
Running SATA as a SCSI-over-IDE, requiring drivers, is a more flexible approach, but requires the use of driver floppies. Still, there's something neat about having four drives all hooked up as masters (2 SATA / 2 PATA) and installing XP without driver floppies.
I'm not sure how it can be remotely possible with a 4 drive SATA controller, though.
sprockkets - Monday, May 31, 2004 - link
Does anybody know if the NF3 chipset has any functionality similar to Intel's SATA, like is SATA done natively without needing any special drivers or programs for the os to use or understand?rms - Monday, May 31, 2004 - link
I also would have preferred to see feature benchmarking instead of cpu/memory benchmarking.rms
Zak - Sunday, May 30, 2004 - link
2 RAM slots on the Abit mobo??? They call THAT an improvement??? Why can't there be at least 4? With 1GB chips' prices being still very high that would be a major selling point for many. I'd upgrade my mobo instantly if I could stick 4 512MB DDR400 chips and not have them run at 333...Zak
Odeen - Sunday, May 30, 2004 - link
I'm very surprised that none of the motherboards except for MSI actually implemented all the features of their chipsets. Both the NF3-250GB and the K8T800 Pro support 4 chipset-level SATA ports, but only MSI has all 4. If it wasn't for that Corecell silliness, I'd be taking a long, hard look at the MSI board.Crassus - Sunday, May 30, 2004 - link
Whats the point of showing benchmarks when all the boards perform within margin of error? When the memory controller is part of the CPU there's IMHO little point in benchmarking it.Why not go after the components that make a bigger difference, esp. HDD, Ethernet and stuff in terms of throughput, CPU utilisation and so?
JustAnAverageGuy - Saturday, May 29, 2004 - link
#11In RAM, generally speed increases are more noticible in real world performance than timings.
Obviously if you have 400 cas 3-3-3-8, versus 400 cas 2-2-2-11, 2-2-2-11 would win. Generally though, speed is more important than timings after a certain point.
bigtoe33 - Saturday, May 29, 2004 - link
#9I think you may have one of these supposed 3000 boards that have non-pro chipsets that Abit says are pro chipsets but really appear to be not..
I would take your issue to Abit.
qquizz - Saturday, May 29, 2004 - link
Concerning the overclock. I can overclock the crap out of my XP2100+, but I keep it at levels where it's stable using Prime95 and Memtest. I wonder if these overclocks can meet my standards?gplracer - Saturday, May 29, 2004 - link
All of the ram in this comparision was CAS3. I wonder how the CAS3 at 270mhz compares to CAS@ at 250mhz. I run my corsair at that speed now.