Jump to content
HWBOT Community Forums

Antinomy

Crew
  • Posts

    1893
  • Joined

  • Last visited

  • Days Won

    25

Everything posted by Antinomy

  1. Well, the Throttling is being monitored under load because it's when the heat is at the max. So it looks like the game is memory intensive indeed. Or the MoBo is buggy, which I believe more - it's an MSI! So get 533 or 667 and be happy. You can run the Everest FPU tests to concern that the CPU is not getting slower, only the RAM.
  2. It's not a wall - I would say FSB fear because it becomes less stable and sometimes you need a specific method of overclocking these. And the FSB wall is really a wall - it won't work if you push +1MHz above. So you say the Palomino is a good one too, I didn't have much experience with them. Did you see an interesting behavior when a CPU is not stress instable, but rather stess-off instable? when you go over a certain frequency, it can go all the way on Superpi 1M (or wPrime32, Pifast) but will freeze when it's complete? I wanna see you doing this with the AN7 and the Mendocino on top of that one http://www.oj0.co.za/cpus/cpus.jpg :-P Looks like all the CPU's in the right stack on that foto are s370. And AMD in the left one.
  3. Yes, high bus is needed but not available on every CPU. The old ones don't like high FSB in means of stable frequency.
  4. Two reasons - 1. CPU overheated. Run Throttlewatch or RMCLock to see if the CPU is throttling 2. Your videocard doesn't have enough videomemory and the game is memory intensive. These are the only reasons I see. What's the MoBo? You can change timings via Memset in Windows and overclock a bit further too.
  5. Well, high FSB is a must only for Thoroughbreds and Bartons. Some CPU's have FSB fear - the begin to crash even far from 200.
  6. Why, maybe the subtimings? And besides, Massman - I hope you know that the DFI board (and Epox 8RDA3+) are a bit slower, than Abit, Soltek, Asus for example? And that the rev2 Nforce2 (the one with official 200FSB) are all slower to get more FSB frequency.
  7. Yes, Winbonds say that 3.3 is for kiddies But I'm going on an experiment that would be more useful on A64 but can make a nice work on A-XP. Don't know other platform that requires high speed DDR.
  8. You'd better increase the Trfc timing. It can give you more memory frequency and let to overclock further without the need to additionally lower the memory frequency. You won't get more speed with 400 memory, the FPS lowered to much. Keep 667 or 533. But you can try 400 to check out the potential of the CPU, the required voltage and run stability tests. When you upper back to 533 and 667, the errors you get will be memory problems almost for sure. So push timings and memory voltage.
  9. Because it's not always that easy when you want to go over 3.3V. The Vdimm is often being powered from 3.3. This can be overriden like on the DFI UT NF250GB Vdimm mod but OCZ booster is easier. But some of my thoughts tell me that maybe a Vdimm is not required at all.
  10. In Rev.3 the Global points will be split by #CPU_core. So the Gulftowns get into 6-core rankings. ES are being published in the same category that their specs belong to. If there are no Gulftowns working on 2.4 / 3.07 as a mass product, then there will be only one category, the 3.33. And as the ES specs do not exceed the production ones - I don't see a problem publish them in the 3.33 category.
  11. Toki, is this: clear enough for you? If you use a ES version of a CPU that is available in retail, it's O.K. The rule is only for the ones, that are not announced officially yet. Nevertheless, you can upload a submission with, for example Gulftown (only one, the rule is working for the moment). You just have to select a checkbox "do not participate in HWBot rankings" - so you don't get boints, but the result will be shown and submitted. So, don't worry about this one There is only one disadvantage in the rule - about the steppings. The C3 stepping of Deneb rips the C2, but they are only going to the market. But the stepping exchange lies in shorter time gap, so it doesn't affect the statistics much. Can somebody tell me what is the difference with AMD and Intel? IMHO, there are no differences - both make CPU's, only the names are a bit different So is your question - yes, all three are competitions and that's all. The organizers are different, the hardware being used is different (though, the vendor competitions are focused on LGA1156 - yeah, it's more promotion and more money from the new boards), the rules are different, the benchmarks are different. I don't see anything common between them except they all are competitions and use HWbot for result submission
  12. Ticket ID: 694 Priority: Medium Please, add a mobile category S3 SuperSavage/IXC\r\nhttp://pic.ipicture.ru/uploads/091120/4x5EmW4g7z.gif - the name is identified correctly according to this link: http://www.s3graphics.com/en/drivers/legacy.aspx\r\nand the Dev_ID.
  13. Seems like wPrime remembers the username after the first submission and locks it like - "Who the hell wants to change a username?". So they didn't take this case in view. Try to search for wPrime entries in the registry. wPrime doesn't leave files after himself, so it should be in registry or a temp directory.
  14. That's absolutely normal. If you update your BIOS, you have big chances that it will be recognised fine.
  15. So you made a new account? Try to logout and login again - this makes things OK often.
  16. IMHO, the IE is the one, who screwed up
  17. I agree, if it's not compatible in software means, then it will be very hard to bench old systems. I don't ever want to see a Vista being installed on 486 So many get the win2K or others to work there. It would be very bad if a category will fall off because of a new version. The same was with Super pi 1.4/1.5 that don't want to launch on Win9x but this can be solved.
  18. Em.. what's the system? First of all, you should use wPrime 1.55, because 2.0 isn't allowed As for the second - switch to Windows Vista, it makes multicore CPUs perform faster in this test (or I would say - in every multithread HWBot CPU test). Don't know about Windows 7 though.
  19. Looks like you don't know about the Time Machine http://web.archive.org/web/20040208023530/http://home.comcast.net/~shrmytoon/NF7D_10b4.ace'>http://web.archive.org/web/20040208023530/http://home.comcast.net/~shrmytoon/NF7D_10b4.ace - here's your file. http://web.archive.org/web/ - here's the Time Machine This is why I asked for the dead link. You can make it undead
  20. New stepping by itself never was a reason for a new category. A question that has been discussed a number of times. As the last example - the Wolfdale E0 a zounds of others.
  21. Post couple links where you were searching. doesn't matter that they are dead.
  22. http://hwbot.org/result.do?resultId=689288 - wrong category. The result is of a socket 478 CPU, but in a socket 423 category pretty common mistake, but was validated, so couldn't use the golden bell
  23. Thanks, that was a great one!
  24. Massman, here's a guy whos result will make you cry: http://hwbot.org/result.do?resultId=910987 Next time, you should beat his score and not less!
  25. The rule is "220 is the limit". This means - if you get 220, you are very lucky. Otherwise you should look for not exceeding this value which means - below 220. Can't get exact - get it lower. Otherwise we will always argue on how much can "not much" be As for me, I know that this difference is much lower than even 1 PCMark point, so it doesn't make sense. But dura lex sed lex. P.S. I think that the limit issue should be written on the benchmark page. In the rules section. As a rule - all rules should be mentioned on the appropriate pages, making it easier for everyone.
×
×
  • Create New...