FSB Overclocking: Soltek K8TPro-939

While Soltek is best known for delivering good value, part of that value equation has always been the outstanding overclocking capabilities of Soltek boards. For that reason, we expected a lot from the Soltek overclocking tests. As you will see, we had very mixed results.

Front Side Bus Overclocking Testbed

Front Side Bus Overclocking Testbed
Default Voltage
Processor: Athlon 64 FX53
2.4GHz
CPU Voltage: 1.5V (default)
Cooling: Thermaltake Silent Boost K8 Heatsink/Fan
Power Supply: OCZ Power Stream 520W
Memory: OCZ PC3200 EL Platinum Rev. 2 OR
Geil PC3200 Ultra X
Hard Drive: Seagate 120MB PATA (IDE) 8MB Cache
Maximum OC:
(Standard Ratio)
204x13
220x12
2651MHz (+10.5%)
Maximum FSB:
(Lower Ratio)
228 x 9,10,or 11 (2508MHz)
(1:1 Memory, 2 DIMMs in DC mode)

The Soltek did a good job with Standard Ratio overclocking, reaching our highest previous overclock at 220x12 or 2651MHz. However, no matter what lower ratio we chose, the highest CPU clock setting that we could achieve with stability was 228. SATA was not an issue here, since our test bed was running an IDE Hard Drive. All tests were run with the Asynchronous PCI/AGP set to Enabled to fix the frequency.

To try to find the answer to these low clock frequencies, we decided to try to run the K8TPro-939 with the Asynchronous PCI/AGP disabled. Without a PCI/AGP lock, we reached the same 228 frequency as our top stable setting. We can only conclude that the AGP/PCI lock is not working on this Soltek board. Since we have tested several recent VIA K8T800 Pro motherboards that definitely have working AGP/PCI locks, we suspect that the problem is with the Soltek BIOS or design.

It is very disappointing to find such limited frequency capabilities on the Soltek board. With other Socket 939 motherboards reaching 275 to 290 with the Athlon 64, 228 is a very poor performance. The Soltek is, otherwise, a very capable overclocker, matching the best that we have seen with this CPU at stock frequencies.

Front Side Bus Stress Test Results:

As part of our overclocking tests, a full range of stress tests and benchmarks were run to ensure that the Soltek K8TPro-939 was stable at each overclocked 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. 228MHz was the highest overclock that we were able to achieve with the Soltek while running these tests at default ratio and default voltage. This matches the best stock overclock with this FX53 processor.

Tech Support and RMA: Soltek K8TPro-939 Memory Stress Testing: Soltek K8TPro-939
Comments Locked

21 Comments

View All Comments

  • WT - Monday, April 11, 2005 - link

    A Soltek response in the forums:

    http://forums.anandtech.com/messageview.aspx?catid...
  • horsefly24 - Tuesday, January 18, 2005 - link

    Here is an update to the Soltek SL-KT8 Pro motherboard to the DDR400 bios problem http://www.soltek.de/soltek/download/download_all....
  • bob661 - Tuesday, November 2, 2004 - link

    #16
    I have one of those boards but haven't tried the hothardware.com suggestion. I don't feel comfortable with cranking my HT up to 230 just to see if the damn locks work. Also, I'm going to try lowering the HT multiplier to see if that might enable the lock. I just wish Anand was a bit more precise in explaining how he got his lock to work.
  • Schro - Sunday, October 31, 2004 - link

    >We suspect that this generates enough US volume
    >that Soltek must wonder what the real value of
    >a US distributor would be.

    Soltek HAD an exclusive US distributor, a company doing business as "Soltek USA". They were one of the best distributors that I have ever dealt with (sent plenty of promo packs ;)... ). They called it quits because Soltek Taiwan sold direct to the Egg, who was supposed to be purchasing from the US distributor. This undercut Soltek USA's business, etc etc. Hence, they have been gone for over a year now.
  • Wesley Fink - Thursday, October 28, 2004 - link

    #16 - We also tried 233 on the Soltek which also did not work.

    As for 939 reviews, the Gigabyte GA-K8NS Ultra-939 is a lower cost version of the GA-K8NSNXP-939, which we have already reviewed. Since we have already reviewed the flagship Gigabyte 939, we will not be reviewing the lower members of the same family. The same applies to the lower-featured versions of some MSI boards.

    The Asrock is based on the Uli chipset, and it's a possible review for the future. However, with PCIe boards coming, an AGP/PCI board with both sockets 754 and 939 probably won't find a slot in our review calendar.
  • thebluesgnr - Thursday, October 28, 2004 - link

    Another great review, thank you Wesley.

    About PCI/AGP lock:
    "However, no matter what lower ratio we chose, the highest CPU clock setting that we could achieve with stability was 228"

    Some VIA K8T800 Pro boards without working locks have PCI/AGP dividers for 233 and 266MHz "FSB". Have you tried going from 228 directly to 233 on this board?

    About that, I remember the MSI K8T Neo2 review. You wrote:
    "We have been assured that shipping K8T Neo2 boards will have working PCI/AGP locks and we are passing this information to you."
    I've seen many people complain that their boards do not have working locks; I've read this in a review (HotHardware.com):
    "From 200MHz to about 212MHz, the K8T seemed to function normally. But at speeds from 212MHz through 230MHz, the board occasionally wouldn't POST. Speeds above 230MHz consistently worked, but we weren't very confident in our particular board's overclocking ability."

    It would be great if you guys could get one of these from retail channel (and not directly from MSI) to check this out and update that review.

    Anyway, with this Soltek review there are only two socket 939 boards listed in newegg not reviewd by AnandTech: ASRock K8 Combo-Z (cheapest s939 mobo, by $31) and Gigabyte GA-K8NS Ultra-939. If you could let us know if you'll review them I'd appreciate it. ;)
  • Gnoad - Thursday, October 28, 2004 - link

    If there eventually is a BIOS update to this board that fixes the AGP lock, anandtech really needs to do a second review, or at least an update. This board looks like my future purchase if there ever is a updated bios.
  • KrazyDawg - Thursday, October 28, 2004 - link

    With the newer Asus motherboards, you won't have a problem with resetting the CMOS. There's an option to load defaults in the BIOS and if you're overclocking with an Asus motherboard and it fails to POST, it will reset the values to the ones previously used before you made the adjustment.
  • Aikouka - Thursday, October 28, 2004 - link

    #3, that floppy location is a pain, because if you ever have to get to the BIOS reset jumper, it makes it a lot harder, especially if you have a large case. I have a Thermaltake Xaser III V2000A, and it's a fairly large case. So a lot of my cables are stretched as far as they can go (especially from the PCI IDE card) and the floppy cable is no exception. At one point, I wanted to reset the bios, because the board was acting funny, and this was made extremely difficult by the location of the floppy connector and the bios reset jumper's location. One of the two should be moved to allow easier access. Also, the front panel connectors (Power switch, etc.) are below that, putting even more tight cables in the way.

    Looking at the picture, the JBAT jumper is located right beside the floppy cable, making it a pain to work with anyway.

    Just like on the SL-K8ANE2-GR the purple doesn't suit it very well.

    Soltek is really good with customer support, though. I was having problems with my board and they even sent me a beta BIOS to try out. Responses came back within 3 hours (usually around 1 hour.)

    Wesley, I don't know chip prices off-hand or anything, but I can't see how VIA's solution would add a great cost to the board when Chaintech sells a VIA ENVY pci sound card, which retails for $25 (retail) on newegg.
  • UnderScore - Thursday, October 28, 2004 - link

    Wesley,
    I like that AT includes content creation & encoding tests and I realize that games are popular repeatable ways to stress test a system. Perhaps these MB reviews could be less focused on game performance and more on the MB's subsystems? Why not prove that the Gigabit LAN will be slower on the PCI bus rather than guessing that it will likely be slower. Am I to assume that USB, Firewire, Onboard digital I/O etc by default are all working perfectly if they are not mentioned? Testing & reporting of such items would make AT's testing methodolgy more balanced since just like #4 & #9 mentioned, not everyone is an overclocker & similarly not everyone plays games regularly (esp FPS).
    In closing, I thank you for all the work that you do & I hope you take this constructively as I had not intended to slam you or AT.
    Thanks,
    James

Log in

Don't have an account? Sign up now