Massman

Let's continue the discussion about the AMD tesselation slider in this thread: http://hwbot.org/forum/showthread.php?t=17844. Maybe I should also move the posts that are just allowing/disallowing it at hwbot in that thread?

Interesting to check up with participants and get feedback about the MOA VGA cards. Here's one from an anonymous participant: "Mem: best compared to HWBOT 3D11" hmm, ok, yeah ...

FYI, I just fixed the WR points for a couple of scores. Some might have gained/lost a couple of points because of it. //edit: they were not updated because of this JIRA issue. Moderators couldn't find a list of unverified WR submissions and WR points are only handed out to verified scores.

There are two things: - the tesselation levels - the name that vendors give to it The tesselation levels are not vendor-based, but just defined in dx11 (I think). But of course, any vendor can give different lables to these levels. Evidently, you want the labels to logically match the settings, but it's not a must. So, the levels are brand/driver independant, the labes depedant. @ FM_Jarnis You might want to re-check the security of your systeminfo. It's very easy to make scores approved and change other stuff in the validation file. http://3dmark.com/3dm11/1869402 (approved?) http://3dmark.com/3dm11/1869404 (not approved? Of course, the score is already removed from the competition!

That sounds like a mental setup. Awaiting pictures ... :-)

1688 CPUs 1010 VGAs

Any other hardware database list that would come in handy? Just let me know.

Until we find the time necessary to build a new hardware database overview, here's the list. Links: - Processor database list (last update 22/09/2011): http://hwbot.org/news/6192_hwbot_database_cpu_list - Videocard database list (last update 22/09/2011): http://hwbot.org/news/6197_hwbot_database_videocard_list CPU SELECT "<tr><td>",manufacturer.name,"</td><td>", cpu_family.name,"</td><td>",cpu_subfamily.name,"</td><td>",cpu_core.name,"</td><td><a href='http://hwbot.org/hardware/processor/",cpu_model.safe_name,"/' target='_blank'>", cpu_model.model,"</a></td></tr>" FROM cpu_model JOIN cpu_subfamily ON (cpu_model.subfamily_id = cpu_subfamily.subfamily_id) JOIN cpu_core ON (cpu_core.core_id = cpu_model.core_id) JOIN cpu_family ON (cpu_subfamily.family_id = cpu_family.family_id) JOIN manufacturer ON (cpu_family.manufacturer_id = manufacturer.manufacturer_id) ORDER BY manufacturer.name, cpu_family.name, cpu_subfamily.name, cpu_core.name, cpu_model.model ASC; VGA SELECT "<tr><td>",manufacturer.name,"</td><td>", gpu_family.name,"</td><td>",gpu_core.codename,"</td><td><a href='http://hwbot.org/hardware/videocard/",gpu_model.safe_name,"/' target='_blank'>", gpu_model.model,"</a></td></tr>" FROM gpu_model JOIN gpu_core ON (gpu_core.core_id = gpu_model.core_id) JOIN gpu_family ON (gpu_core.family_id = gpu_family.family_id) JOIN manufacturer ON (gpu_family.manufacturer_id = manufacturer.manufacturer_id) ORDER BY manufacturer.name, gpu_family.name, gpu_model.model ASC;

GRAND FINAL INFO - Qualification: Combination of online qualification and live finals - Final: October 6-9th, NTU Sport Center Taipei, Taiwan - Link: http://www.msi-moa.com/ - Judge: GiorgioPrimo - Participants: --- 12x Live qualifier: Romdominance/Splave (USA), Patch/Dentlord (USA), G H Z/Gautam (USA), Monstru/Matose (RO), Stelaras/Aristidis (GR), Slamms/Smoke (RU), Elmor/ME4ME (SE), BenchBros (DE), Cal930/Gyrock (JP), Oc_windforce/Littleboy (KR), Deanzo/Uncle.Fester (NZ/AU), Zolkorn/Audigy (TH) --- 3x Online qualifier: Hondacity/Chew* (USA), SuicidePhoenix/Kurbel (DE), IRIS/Coldest (ID), Perica_barii/Quake (ME) Other overclockers attending this event as press or audience: Booj, Lucky_n00b, Mean Machine, Massman, NeoForce,Xyala Arrival times Already present: Elmor, Xyala 05/10 - 06h00: Booj 06/10 - 05h00: Hondacity 06/10 - 05h30: Patch, Splave, Gautam 06/10 - 06h25: Kurbel, SuicidePhoenix, ME4ME 06/10 - 10h15: Massman 06/10 - 15h25: Slamms, Smoke, Neoforce 06/10 - 17h10: BenchBros 06/10 - 18h15: Pastelidis, Aristidis, Stelaras 06/10 - 21h00: Matose, Monstru, Perica_barii, Quake Departure times 08/10 - 18h10: BenchBros 09/10 - 18h00: Booj 09/10 - 19h30: Perica_barii, Quake 11/10 - 23h10: Patch, Splave 11/10 - 23h15: ME4ME 12/10 - Stelaras, Aristidis 14/10 - 23h15: Kurbel, SuicidePhoenix 18/10 - Mean_machine 19/10 - 21h00: Massman Never - Elmor Never - Xyala

Note, for your personal score only the - top15 best global scores (most gl points) - top20 best hardware scores (most hw points) will be counted. Even if you have 35 scores with hardware points, only the 20 highest scoring submissions will be counted. For the enthusiast league, that's top20 with air cooling and with system picture.

Hm. Checking.

All scores have HW points as far as I can see? Maybe no global points because they're not your best scores?

I'm just looking for the easiest way to moderate results. Like I said:

What a fine example we have here to show your the difference between "tragic accident" and "idiot did something stupid".

Rumours, yes, but as far as I know there's no tool that allows disabling Tesselation on Nvidia. I'm sure it can be done, though ... FYI - if you disable the Tesselation in Heaven on Nvidia cards, the gain is approximately 600p (1850 -> 2450), which is pretty much the same gain as AMD-based cards (30%). I'd have no problems considering adjusting the Tesselation level a 'tweak' for competitive benchmarking purposes. It makes the benchmark result less meaningful, yes, but then again so does using liquid nitrogen. For uniformity sake, we should have the same guidelines for any DX11 benchmark. That means: if it's not allowed for 11, it's not allowed for Heaven. That requires us to find a solution for Heaven, which we currently don't have. There are all sorts of other solutions that work around this problem, eg disallowing specific drivers or disallowing AMD-based cards to be used in DX11-benchmarks, but they are very impractical. Keeping it simple and understandible is the key here. As far as we can see, you can extract the default tesselation level from the driver profile xml, but in 5 minutes we also found a way to have the profile say it's enabled and the benchmark running disabled. Afaik, the dll is the only way to grab the info. Of course, all our guidelines are useless when in a few month's time we can disable it on Nvidia too and we don't have any detection mechanism for it.

- Vgpu - Vmem - Vgpu OCP - CB/CBB Vgpu was at 1.6V measured from coil. Vmem was 1.9V also measure from coil. I adjusted basic vgpu bias with the VR and then used inspector to increase voltage in the OS (~ 0.15V wiggle room).

Awesome. So far, my result has not been deleted by the engine yet!

Hazzan and Giorgio will be fighting it out around the 1400MHz mark . Memchips are Samsung. VRM is similar to the one of a reference card. Definitely not much better ... I wouldn't say this is an overclocker's card. It's sometimes nice to work with limitations, though Computerbase's pictures of the card:

How is that possible when the competition organiser will tell Gainward they have contacted 200k overclockers and there are 5000 participants in the competition I was gonna, but when I was checking the card, the Vgpu mod fell off . I'll make schematics in a minute.

If anyone needs the modifications, just ask and I'll post them here too

So, Gainward send me this card to compete in their VGA OC competition. It's always nice to compete with a free card and I had to make up for the failure that was the Palit competition (card died after 1,5 runs). So, this time, I went up slower and made sure I did everything right to max out the card before eventually sending it to silicon heaven. Important for this card: keep the heatspreader on the card at all times. If you don't use the heatspreader, it will be quite likely that you will blow a hole in the PWM area. I tried my first runs without heatspreader and stopped after 5 minutes as the PWM area was untouchably hot. So, I drove to Leeghoofd to dremel the heatsink a little bit (on the top there was metal in the way of my pot). After that, I could safely return to the benching arena. With Vmem at stock, the card was artifacting around -125°C due to the memory being too cold. Also, the memory was limited to 1150MHz ... not off to a great start. After the Vmem modification, the memory could go up to 1300MHz stable and even 1340MHz with artifacts. The coldbug moved down slightly and as long as I would start the benchmark at -130°C (GT1 loading), I'd be able to bench to -150°C. Still, there were artifacts, but at least the bench got through. After much crashed runs, I finally reached 1375/1325 MHz giving me a final score of 10999. 1 point lower than the goal I set in the beginning of the competition. Overall, I'm quite happy with this result; a lot of hard work to get to this speed, but satisfying nonetheless. Much thanks to Eileen for sending the card. Also thanks to Hazzan and Giorgioprimo for the nice fight in the contest ... I really liked their style of competition with backups (sometimes submitting 10 backups after each other ), but never really secretive about scores. So, in despite of the low amount of participants in this competition, I had fun anyways. I think Leeghoofd has a couple of pictures of the setup when we were benching at his place last weekend. Coldslow BIOS and Vmod guide in this thread: http://hwbot.org/forum/showthread.php?t=26163 Final score: [hwbot=2208383]submission[/hwbot] At least I was #1 at one point during the competition:

That's a good idea, actually. We could make a reference screenshot ourself and check the the image quality against the screenshot taken by the application. Certainly not the nicest solution, but 'doable' unless we get information on how to access the atimgpud.dll to read the tesselation level.

The official driver doesn't allow this, but there are unofficial tweak applications that do allow turning it off completely. It pretty much makes any AMD D3D11 scores suspicious now.

FYI, tesselation can be disabled completely and it'll make the benchmark rediculous. So allowing it is no option either. See attachment.

9-9-9-24 faster? Thanks for sharing the performance optimisation tweak, Captain Gamer!!