MSI K8N Neo Platinum: First nForce3-250Gb
by Wesley Fink on April 26, 2004 12:05 AM EST- Posted in
- Motherboards
Overclocking and Stress Testing: MSI K8N Neo Platinum
FSB Overclocking Results
With all the excellent options for overclocking, the MSI K8N Neo should be capable of excellent overclocking. Given the great overclocking results that we found in the review of the nForce3-250 Reference Board, we expected the K8N Neo would do even better with the expanded ranges on the K8N Neo. We weren't disappointed.| 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) |
246FSB x10 2460MHz (+23%) |
| Maximum FSB: (Lower Ratio) |
300FSB x 8 (BIOS Limited, Asynch Memory) 270FSB x 9 at 1:1 Memory |
This late 3200+ achieved a new overclocking record on the K8N Neo Platinum at 246FSB, at the stock 10X multiplier and default voltage. This 23% overclock at stock multiplier is more like those that we are accustomed to seeing on Pentium 4. Dropping the multiplier, the highest FSB overclock that we could achieve was the BIOS limit of 300. We fully expect that the K8N can go much higher on FSB, but Clock Gen for the nForce3-150 from www.cpuid.com only partially works on the 250 chipset. The FSB portion of Clock Gen does not function correctly yet on the updated 250, so the BIOS limit is the current FSB limit.
We were also able to run OCZ DDR4400EL at a 1:1 ratio at a FSB of 270x9, or DDR540, which is near the limit of this CPU on this board. We tried to lower the ratio to 8, but we still could not achieve 1:1 memory performance at any speed higher than DDR540, as the board would automatically reset to stock speed if we selected a 1:1 setting above 270. Perhaps the overclock protection on this board needs to be a little less aggressive, because this memory is capable of performance at even higher speeds.
HyperTransport could be maintained at the 4X (800) setting up to a 260 to 265 FSB. Above this point to the BIOS maximum of 300FSB, we needed a 3X HT setting. There is clearly additional headroom with the higher HT of the 250 chipset, and it is likely that we can reach much higher FSB settings with a greater range of BIOS settings or a Clock Gen that fully supports the nF3-250 chipset.
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 K8N Neo 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. 246MHz was the highest overclock that we were able to achieve with the MSI while running these tests at default voltage, an astonishing performance with the Athlon 64.Memory Stress Test Results:
The memory stress test is very basic, as it simply tests the ability of the K8N Neo Platinum 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: | 5T (10T for Best Performance)* |
| Precharge Delay: | 2T |
| Command Rate: | N/A |
*Several memory tests have shown that memory performs fastest on the nVidia nForce chipsets at a TRas (RAS Precharge) settings 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 performance 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. This will be shown in more detail in an upcoming Memory review. While benchmarks here were run at 2-2-2-5 for consistency with past results, future nForce benchmarks will be run at a TRas setting of 10.
The MSI K8N Neo was completely stable with 2 DIMMs at the most aggressive settings of 2-2-2-5 at default speed. Higher overclocks could be achieved with 1 DIMM compared to 2 DIMMs, but at default speed, 1 or 2 DIMMs were reliable at the same aggressive 2-2-2-5 timings.
Filling all three available memory slots is more strenuous on the memory subsystem than testing 2 DIMMs on a motherboard. We were very pleased to find that 3 DS DIMMs (1.5GB) of memory worked fine at timings almost the same as the aggressive timings we used for 2 DIMMs. This is the first Athlon 64 board we have tested that was completely stable with 3 DIMMs at DDR400.
| Stable DDR400 Timings - 3 DIMMs (3/3 DIMMs populated) |
|
| Clock Speed: | 200MHz |
| Timing Mode: | N/A |
| CAS Latency: | 2.0 |
| Bank Interleave: | N/A |
| RAS to CAS Delay: | 3T |
| RAS Precharge: | 5T |
| Precharge Delay: | 2T |
| Command Rate: | N/A |
Three DIMMs required 2-2-3-5 timings for stable operation, which is remarkably close to the most aggressive 2-2-2-5 timings available on the K8N Neo. You do need to be aware of one quirk in the MSI memory setup. If 3 DIMMs are installed with the memory speed set to AUTO, the board will down-clock the memory to DDR333 speed. To achieve DDR400 with 3 DIMMs, we had to move off Auto and specify DDR400 speed. However, DDR400 with 3 DIMMs worked very well in our tests, and was very stable, which is certainly a first for the Athlon 64.
Tech Support and RMA: MSI K8N Neo Platinum
Standard Performance Test Configuration: 3200+ with ATI 9800 PRO
Facebook
Twitter
RSS
26 Comments
View All Comments
kmmatney - Tuesday, April 27, 2004 - link
Ahhhhhh! A long spiel about how great the board is at overclocking....and then no overlocked benchmarks!! I want to see how much gain you get! How about a single game and encoding benchmark?Pumpkinierre - Monday, April 26, 2004 - link
I agree with #10, Wesley, you should include at least one synthetic with your overclocking tests because you have'nt done any good overclocking/high FSB tests on a64s as you havent had a mobo with good PCI/AGP lock or >250 FSB til now. It makes me wonder why because you are thorough with the p4/i875 and by your own admission the gigabyte and shuttle nF3-150 have PCI/AGP locks and ClockGen adjusts the FSB. The location of the a64 memory controller ondie means the memory sync/async tests results from intel systems cant be translated to the a64 ones. So, in my opinion, for conservative a64 overclockers, benchmark results at the same cpu speed but different FSB (clockspeed) sync or async and HT alteration would be of good use. I know you are just seeing the limits etc. but you must be running stability tests so it costs you nothing in time to throw in the sandra or Memtest86 bandwidths or even 3DMark2001/2003 results.And I agree there is merit in running cool and quiet with overclocked systems(if possible) as the windows software overclockers are still not perfect and only available on some brands.
Jeff7181 - Monday, April 26, 2004 - link
Wesley,It may not really fit in with a motherboard review, but seeing as how it seems overclocks on the nF3 150 and KT800 chipsets have been limited by the motherboard, it makes sense to test the overclocked performance to see if it's worth paying for a motherboard with lower performance at stock speeds that might provide a lot better performance once overclocked. For example, if the nF3 250 can only match KT800 performance after overclocking, it makes the choice much more clear.
cnq - Monday, April 26, 2004 - link
Wesley,Thanks for the quick response. Glad to hear you are considering mixing manual overclocking with cool'n'quiet in future tests.
You mention that "I am not convinced that it matters to most overclockers". I think that's just because c'n'q' is fairly new. If you tell them that c'n'q is just the thing to help prolong the life of their overworked system, they'll start paying attention. Only those who run 100% cpu utilization 24/7/365 won't care.
For the rest of us, mixing Cool'n'Quiet with overclocking makes perfect sense (throttle back the cpu when just web browsing; crank it up to overclocked when you need a burst of bower, which in my case is compiling. Note that I'm not counting MSI's CoreCell auto-overclocking because it's not powerful enough...)
I'd also claim that Cool'n'Quiet is *doubly* useful in a heavily overclocked system. Such systems are often on the ragged edge of overheating, and thus could benefit more than anyone from throttling at times of light or moderate load.
Naturally, when the load jumps back up, we'd want the fully overclocked manual settings to automatically kick back in. And that's the rub. Will that part work?
Has anyone tried running Cool'n'Quiet on a manually overclocked A64? Does it (a) work perfectly, (b) crash, (c) not crash but "forgets" your overclock settings, (d) other? I'm guessing it will crash, esp. if manual overclock settings included changing the CPU voltage or raising the HTT while lowering the CPU multiplier. Hoping to hear I'm wrong (for at least one A64 board) before making purchase...
Schnieds - Monday, April 26, 2004 - link
Do you have any idea when this motherboard will be available for purchase in retail outlets? From the review it sounds like this board is a release version and ready for retail, but no one seems to know when it will actually be available for purchase... Thanks for the great review.Wesley Fink - Monday, April 26, 2004 - link
#7 - nForce3 chipsets automatically control Bank Interleave and Command Rate - there are no manual options. Some VIA chipsets do allow interleave and command rate settings. We always specify the setting we used in the review - if those settings are available. If they are not an available option we state NA for Not Available in our Memory Stress Testing tables. We also always enable interleave if it is an option.#8 - I added Cool'n'Quiet to the Feature list at your suggestion. Most overclockers turn off auto features for overclocking, because frankly I can always out-tweak Auto settings on OC. I will consider doing more with Cool'n'Quiet in overclocking, but I am not convinced it matters much to most overclockers.
#9 - Corrected. Not to be defensive, but if you want to see how much AnandTech found with this board that other sites missed, read through K8N Neo reviews that others have posted.
#10 - What would it take to make you happy on OC results in board reviews? I have explored 1:1 vs aynch in memory reviews. Performance results at a full-range of OC and timings are always a part of memory reviews. What would you like to see in motherboard reviews?
Jeff7181 - Monday, April 26, 2004 - link
No overclocked comparisons? Awwww man... questions still unanswered...KillaKilla - Monday, April 26, 2004 - link
How well did the Albatron GeForce FX5950U work with those ATI Catalyst 4.4 drivers?Not to be sarcastic, but simply alerting those who can correct this to the typo.
cnq - Monday, April 26, 2004 - link
Wesley,You mentioned that you disabled Cool'n'Quiet to prevent possible interference with overclocking.
I wonder if it's even worse than "interference": is it even *possible* to run Cool'n'Quiet on a heavily overclocked system? (especially one with aggressive overclocking: significantly raised HTT with lowered CPU multiplier).
This is one of those things that should be a FAQ, yet no one seems to have tried. Cool'n'Quiet is the perfect complement to an aggressively overclocked system: prolonging its lifespan by giving it a "breather" when just (e.g.) web browsing. But are Cool'n'Quiet and overclocking compatible?
(Pls refer to my same question in your Aopen AK89 Max review for specific technical reasons why I think c'n'q and aggressive overclocking may not be compatible. I hope I'm wrong.)
Myrandex - Monday, April 26, 2004 - link
My K8T Neo supports cool and quiet completely, and toms hardware confirms this too with an older A64 mobo roundup. Looks like a hell of a board, but I am wondering why so many reviews at Atech are run w/ Bank interleave set to disabled. I remember back in the K6/2 days, this was a major bios tweak for memory. Does it not matter, or negatively effect performance now?Jason