Suchergebnisse

  1. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    Small contribution to the list. This is the 306MHz stock board. Notice both NB and SB are Korean, not sure if that is relevant or just the production date. NF7 v2.0, PCB version 0.52, plastic socket lever, blue DRAM slots Northbridge: 0402 A1 SPP Ultra400 KOREA Southbridge: 0346 A4 MCP KOREA...
  2. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    Got a second try on the stock NF7 v2.0. I'm waiting for some capacitors for a full recap and then will do voltmods. Mostly interested in highest stable FSB 1:1 with BH-5, but nevertheless tried to see how how I can go with 6:4 divider. Last time it stopped at 295, now - 306.83 and I saw 308 as...
  3. I.nfraR.ed

    CPU-Benchmark: Maxon veröffentlicht den Cinebench 2024

    Wraith Prism (box cooler from 3900X) 7950X@PBO -26/-22, FCLK 2200 2x16GB DDR5@8000MT/s 36-46-46-58-94 1T CPU (Multi Core): 2242 pts Chilled water 7950X@5700/5550MHz, FCLK 2200 2x16GB DDR5@8000MT/s 34-45-40-44-84 1T CPU (Multi Core): 2464 pts Thought 2500 wouldn't be that hard to reach...
  4. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    If SYSDCIN is 4 bits, then it should go to 0xF, but it's another thing if the chipset/cpu actually allows it. I guess original one was 3 bit and the rest reserved for future use, then they extended it to 4 bits. Similarly, WRDATA might be extended too, but that means the X in 15x8 could be as...
  5. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    That dropdown was not functional anyway. My initial idea was to have separate "romsip" profiles as a subset of a bigger profile and list them in that dropdown. I've implemented the profiles functionality in a different way - you can choose which section to save/load, so it is still possible to...
  6. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    Und mein Deutsch ist auch nicht gut. I've learnt it in high school and then when I was working for a German software company (more than 12 years ago). But I've never really used it since then. Attached is the latest version of the tweaker. There was a stupid mistake which prevented writing main...
  7. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    I don't have many 32M scores with socket A and most of them are old and just "click and run". The closest one I have is this screen from 2012 (with a Barton), but I can do better for sure. I don't think his score is something exceptional, just tweaked OS + copy wazza. The OS is Taiwanese, not...
  8. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    These buttons use "Webdings" font. 3 and 4 are the codes for left and right triangle. I guess you don't have that font. It's probably a good idea to replace them with images or with simple text, - and +.
  9. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    Profiles are now implemented and ready for test. I've included 2 sample profiles in "profiles" folder. One of them contains all possible settings, the other one (in romsips subfolder) includes just ROMSIP values. You can chose which sections to load or save. Upon load, the app tries to be smart...
  10. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    Haven't tested CAS2, but CAS2.5 with BH-5 and CAS3.0 with TCCD didn't seem to make a difference with this 6:4 divider. It's all about romsip and alpha timings.
  11. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    No performance in mind, just exploring nforce2 limits, trying to figure out what helps for high FSB. This would eventually help running higher stable FSB 1:1 without big performance degradation. It's probably a good way to bin chipsets as well. @The Sandman Maybe ~300MHz is the internal PLL...
  12. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    Tried another board, this one is stock NF7 v2.0, no mods, except EBED bios from @digitalbath 297.1MHz is max I see, 295.83 is max I have as a validation file. Plus we can take every single nforce2 reference frequency record on the bot :D Seems to be a nice way to bin motherboards :d...
  13. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    What about tweaker doing the validation for you :d It was just a fun experiment, but I will leave it there (Tools -> Auto Validation Bot). All these steps have validation files in the CPU-Z folder, the tweaker increases FSB automatically and sends "F7" to the cpuz window to validate...
  14. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    New beta 11 is up and has a fully working AGP/PCI frequency control. You can download from the github repo, source code is also up-to date. Let me know if something doesn't work as expected. https://github.com/irusanov/nForce2-xtreme-tweaker-vcl/releases From the S2K options, only first 3...
  15. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    Sorry, I meant 68 and 6A. I've left both at 55 in the end.
  16. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    Also, I've found that romsip regs 6A and 66 from the tweaker are important for 3D stability, at least on my system. Adjusting TREF had some positive effect to eliminate the stutter in car tests with 3850, although scores are still low for the clocks. Run AGP freq as high as possible. I'm running...
  17. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    Why do you blur the fps? 😛 3850 is in the 430-450 range at that point and no other AGP card can beat it at this test. Generally, run the High tests first and maybe try Nature as last. It was not better for me, but I'm just an average 01 bencher and it might be different on nvidia. Formula is (CH...
  18. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    You have to run the tests in a specific order (needs testing for every card/platform/driver) and one by one. Fresh boot is always the best, you also don't rerun tests, except Car Low at the end. You also need to find the best LOD for each test and ideally the CPU and VGA clocks - some tests...
  19. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    Well, it seems to be normal then, sorry for that stupid idea I gave. There seem to be a white layer under the black one and the markings are "engraved" byre moving the black part. At least on a genuine Barton, so I wonder how they did that on the fakes. Seems to be the same?
  20. I.nfraR.ed

    [Sammelthread] AMD K7 - Sockel A (462)

    Yep, the font on your sticker doesn't match. So they've just removed the original markings from the sticker and added a "random" one, thus the AQYHA with that date code. From fab51 PS: Wait, on that photo it looks like there's a new slightly larger label pressed over the original one - you...
Hardwareluxx setzt keine externen Werbe- und Tracking-Cookies ein. Auf unserer Webseite finden Sie nur noch Cookies nach berechtigtem Interesse (Art. 6 Abs. 1 Satz 1 lit. f DSGVO) oder eigene funktionelle Cookies. Durch die Nutzung unserer Webseite erklären Sie sich damit einverstanden, dass wir diese Cookies setzen. Mehr Informationen und Möglichkeiten zur Einstellung unserer Cookies finden Sie in unserer Datenschutzerklärung.


Zurück
Oben Unten refresh