knopflerbruce

In some ways I agree, GPU benchmarks shouldn't be about the CPU - just as CPU benchmarks are independant of GPUs. Unfortunately, we don't have a "GPU only" benchmark yet, but folding@home doesn't benefit much from the CPU, so there's hope:) Current 3d benches are more system benches. They're fun to play with, even if the GPU dependancy isn't always the best:p

Pretty clear who started this silly flaming:rolleyes: Now, stfu - PLEASE. If you want to spam down some forum with your crap posts, please do it elsewhere...

I wonder wthat that mod actually does... But cool to see a manufactuerer giving away mods:D

Np, lots of things to learn when you're a mod. Everyone makes mistakes:p

If the validation was currupt, canardpc would've rejected it while uploading. Score OK'ed, and by looking at the pifast and superpi runs there's no reason to think this is a bug. Hopefully it'll stay OK'ed this time.

I've never heard that regular ES chips are not allowed when there are retails available, perhaps we should turn the question around - where does it say that such chips are NOT allowed?

...and M3eh7:p

I think it's completely ridiculous that you can actually own an unbenchable piece of HW, if that's what you're saying. It should be possible to create a set of rules that doesn't exclude any piece of exotic hardware there is - after all, this is the biggest database of overclocking results.

DFI UT 790FXB-M3eH7 - this board is unreleased, but it's not under NDA (I think...), does that matter? IMO we need to find a set of rules that doesn't make certain boards unusable for boints forever (which is the case if a board is never released, and both HW and globals are removed)... hardware boints only for not officially available parts, that's what I call a fair solution.

I got an open box DFI Blood Iron P45, and I could reach 633mhz superpi 1m on air on nb and water on CPU. Pretty good bang for buck IMO, but I think efficiency isa little bit worse than on the DFI DK P45 Plus. Stay clear of the UT p45, btw... 600+ is not common. You could try to PM massman, maybe he knows a little bit about some of these DFI P45 boards.

If Intel chips were benched more, it will often be more difficult to grab the first spots there than for other ranks without the same competition. How can this be wrong? It's impossible for the engine to actually know how much skill it took to achieve some result, and IMO the popularity/rank method is pretty good. Sure there are submissions that don't get the boints they deserve, and vice versa, but to give AMD categories advantages just because there aren't several hundreds of users that bench those chips. We all know superpi (and possibly a few other tests) don't show the real performance difference between the different cores (not just manufacturers), but that itself is not enough to split like you suggests. If you can prove that FM and superpi are deliberately designed to be bad on AMD and good on Intel, then it's different.

Looks better now:) If more categories will reach more than 20 results in each benchmark I feel the database function of HWBot has improved alot.

I have a pc2-5300 kit that can do 600mhz CL4, it all depends on what chips are on the PCB.

Drivers optimized for a specific benchmark are not allowed, so that's taken care of already. The superpi discussion started when c2d was launched, but the application itself is much older. BEsides, the "optimization" is so huge no-one would ever believe that it represents the actual performacne difference:p

It should count, yes. You could always check and see if it does, if not it's a bug;)

If it was Intel optimized, it must've been optimized for some ancient core... If Pentium I optimized means Nehalem-optimized as well, then you're right. Perhaps I'm stupid, but can you link me to a reliable source that tells me that it actually IS pro-Intel? IMO if a benchmark is optimized intentionally towards one manufacturer it should have separate world record rankings, as it's not fair to compare. Proof is needed, though.

I started because I wanted to get more performance for less cash, by OC'ing the then top-of-the-line dual cores AMD had. I bought an X2 3800+ bulk CPU, aiming for 2.7GHz, but since what I got was a CCBIE 0608BPMW it got much higher:D Then I got two Opteron 180's for cheap on ebay, that I wanted to sell... but I felkt like testing those, too see if they were any good - and when those were gone, I got a sick Opteron 175... and the rest is history:p

I can't comment on 3dmark, but Superpi wasn't that much faster on Intel systems during the Athlon XP days - and the program has been unchanged since it was released (when was this? many years ago...). It's right that there are more popular categories in the Intel rankings - but how can people expect to get boints for the AMD CPUs when they don't bench them? If you have more than ~100 results you start to get a pretty decent amount of boints for the top HW spots... 100 is not that many, actually. AMD would've released a $1000 killerchip if they could. Since their parts are sloightly slower than Intels they have to focus on price/performance NOW to be able to sell. The HWBot 3.0 system is popularity optimized, and if AMD users are too lazy to bench, then they will never get a nice reward for 1st place.

There's no reason to split AMD and Intel, as it's the different CORES that make the difference. If AMD vs Intel needs a split, then Nehalem and WOlfdale needs it, too... and San Diego and Deneb and so on. I doubt they will get completely separated, but in some future update there may be additional tanks based on those kind of things, so you will get rewarded for stuff like the best Athlon 64 submission, Nehalem core submission etc as well as the overall and for the specific hardware model.

It made it through post at 3.5ghz and 1.5v, so it is definately stable at stock;) But ofc, 10 euro is nothing. The board can OC HTT, at least 280 was 24/7 stable when the previous owner had it. It was probably not maxed out, as he ran on air and just used the highest multi - so it should be able to kick it up a notch or two from stock.

What about 140 + unlock? I've got that GB board, and one of those semprons... give me a couple of days and I'll test it. Going to Germany tomorrow morning, and that combo will be a christmas present to a friend of mine who burned his A64 3700+ rig:p Edit: nvm, realized I should test the parts before giving them away. First impression of the board was nice, my D9GMH PC5300 CAS5 worked at first boot, never seent hat happen with DFI or ASUS boards Unlocking worked, set ECC firmware to Hybrid instead of Normal, let it update - and set ACC to auto, and it shows up as a 4400e during POST. No OS atm, so cant test stability.

Could be some of the bugs, too... at least some of it.

Looks like a bug, no need to rebench.

Open the window and get -10c in your room, that should help as well:p With properly tweaked and maxed out systems I can't imagine that you're way off receiving boints on air, even without opening the windows. But if you can't tweak or use good cooling, of course that's disadvantage. A cheap water block and a pump isn't that expensive, either, btw... you don't even need a radiator for that ghetto chiller I suggested.

True, as long as the "no performance difference requirement" is followed strictly. There's no reason to join categories just to make them look more popular, either;) I'd say if the difference is less than 1-2% (when overclocked) then it's OK to merge ranks.