Vellinious Posted November 7, 2016 Posted November 7, 2016 (edited) WHY WAS THIS BLOCKED?! WHAT THE..... The clocks on the VGA details CLEARLY state 1310MHz on the core. The base clock in GPUz is reading 1000MHz. Unblock this. This is a legit run..."MATE".... TWO runs with identical scores, both above 1300 on the core and 1860ish on the memory. Result What's the problem Edited November 7, 2016 by Vellinious Quote
Crew Leeghoofd Posted November 8, 2016 Crew Posted November 8, 2016 Just check your subtests and it will become more clear... your GT2 score is way too high for the given clocks thus classified as a bugged run... No need for rocket science... Quote
Vellinious Posted November 8, 2016 Author Posted November 8, 2016 The run was done at 1307 on the core. There was no BUG. The screenshot was taken after I had returned the overclock to normal.....so GPUz is only showing 1000mhz. And you delete the score for that?! Even the 3D Mark link showed the clocks correctly. That's some ticky tack crap....and you know it. Rocket science indeed... Quote
Crew Leeghoofd Posted November 8, 2016 Crew Posted November 8, 2016 let us try rocket science.. plz explain me why your GT2 is so darn high: Quote
Vellinious Posted November 8, 2016 Author Posted November 8, 2016 (edited) Graphics test 2 responds better to memory clocks? Even before I cranked the memory up further, and was running sub 1750, I was getting 71+ FPS on graphics test 2. There was something about the 8GB 290X that allowed for MUCH higher memory clocks than the 4GB version of the same card. Couple that with driver improvements when I was making those runs, and....there's a lot of the difference right there. I dunno who was running those tests in the graph you posted, but their GT2 frames sucked. Here's one I ran with the memory at 1750, similar core clocks and...WELL above 70 fps. AMD Radeon R9 290X video card benchmark result - Intel Core i7-5820K Processor,ASUSTeK COMPUTER INC. X99-A You're WRONG, man. Period. Go ahead and test it yourself. Run once with a normal overclock. Now, drop the memory down by 100 base. The variance is much larger in GT2, than it is to GT1. You're welcome. Edited November 8, 2016 by Vellinious Quote
Crew Leeghoofd Posted November 8, 2016 Crew Posted November 8, 2016 We have never witnessed a +6FPS gain on GT2 with just 100Mhz more on the memory. But the moderators are intrigued nevertheless. We need more data and will compare it with our 290X setup. Secondly we will forward your data to Futuremark for analysis. Either you found the ideal OS/driver/setting combo or our suspicions will be confirmed. Plz provide us with a screenshots and the validation files of: 1200 core / 1600 mem run 1200 core / 1700 mem run 1200 core / 1800 mem run This to witness the GT2 scaling per added 100Mhz, this makes it easier to crosscheck our data. Mail it to albrecht@hwbot.org Quote
Vellinious Posted November 8, 2016 Author Posted November 8, 2016 We have never witnessed a +6FPS gain on GT2 with just 100Mhz more on the memory. But the moderators are intrigued nevertheless. We need more data and will compare it with our 290X setup. Secondly we will forward your data to Futuremark for analysis. Either you found the ideal OS/driver/setting combo or our suspicions will be confirmed. Plz provide us with a screenshots and the validation files of: 1200 core / 1600 mem run 1200 core / 1700 mem run 1200 core / 1800 mem run This to witness the GT2 scaling per added 100Mhz, this makes it easier to crosscheck our data. Mail it to albrecht@hwbot.org I don't even own that GPU any more. I sold it last spring..... This is some heavy handed bull.... I trust you scrutinize EVERY run with this much speculation, and biased opinion? Because, that's EXACTLY what this is..... Do what you wish...delete it, keep it, I guess it doesn't matter that much. Just know, that I know it's total crap. And I've told everyone I know about this thread. G'day to ya Quote
Crew Leeghoofd Posted November 8, 2016 Crew Posted November 8, 2016 Well your score got reported and then we cross check what's okay or not... and your GT2 is tha bomb... For my own peace of mind I'll still run the tests on my 290X lightning to see if there is a significant GT2 boost with Win10 and these drivers over Win7 and 14.12 omega. Ow you also told your team captain about this thread? Would be nice to get his 5 cents on the matter in here too. Quote
Vellinious Posted November 8, 2016 Author Posted November 8, 2016 I'd like to hear it as well. I'm more than just a little peeved at this entire line of garbage. Quote
Crew Leeghoofd Posted November 8, 2016 Crew Posted November 8, 2016 I'd like to hear it as well. I'm more than just a little peeved at this entire line of garbage. pmd him, to accelerate stuff Quote
mirzet1976 Posted November 8, 2016 Posted November 8, 2016 when I compare his FPS and memory clock with my gain is 10 FPS for the 220MHz higher mem clock his - AMD Radeon R9 290X video card benchmark result - Intel Core i7-5820K Processor,ASUSTeK COMPUTER INC. X99-A - 1846mhz = 78.5 FPS mine - mirzet1976`s 3DMark - Fire Strike score: 11729 marks with a Radeon R9 290 - 1620mhz = 68.19FPS Quote
Vellinious Posted November 8, 2016 Author Posted November 8, 2016 Best delete this one too....because I somehow, magically happened upon the same exact "bug", with a different driver version AND slightly different clocks. http://hwbot.org/submission/3362780_ I'm done here Quote
mirzet1976 Posted November 8, 2016 Posted November 8, 2016 (edited) @Vellinious, your run is good is 10FPS gain for 220mhz higher memory and not 6FPS for 100mhz, or my to is not valid when I compare with this delevic`s 3DMark - Fire Strike score: 14522 marks with a Radeon R9 290 I have better graphics score in both tests at lower clocks 1315/1620 vs 1330/1770 Edited November 8, 2016 by mirzet1976 Quote
Vellinious Posted November 8, 2016 Author Posted November 8, 2016 Good? Or not good? Doesn't matter...they've deleted the other submission already. /smh Quote
Crew Leeghoofd Posted November 8, 2016 Crew Posted November 8, 2016 Well we are still gathering data and wading through the results. The scores have been removed for further investigation... not permanentely deleted... Nr1 at 290 is also impressive +100fps GT1... Quote
mirzet1976 Posted November 8, 2016 Posted November 8, 2016 Well we are still gathering data and wading through the results. The scores have been removed for further investigation... not permanentely deleted... Nr1 at 290 is also impressive +100fps GT1... Maybe in crossfire GT1 100FPS Quote
Crew Leeghoofd Posted November 8, 2016 Crew Posted November 8, 2016 Should be way higher in 140ish region for Crossfire... Quote
Johnd0e Posted November 9, 2016 Posted November 9, 2016 i dont have a 290, but if i run this bench back to back with only changeing memory clock on my 980ti will that suffice in proving/disproving gains from memory clock in GT2? Quote
Crew Leeghoofd Posted November 9, 2016 Crew Posted November 9, 2016 I'll be conducting tests as promised once I'm home... I can't do 1800+ but can test the scaling in GT2... from stock to 1700ish speeds... Quote
xxbassplayerxx Posted November 9, 2016 Posted November 9, 2016 Hey Leeg, got your PM... sorry for the slow reply... wedding is in less than a month! I'd like to say a few things. First, @Vellinious, I know it feels like you're being personally attacked. If your result is 100% legit, don't worry about it. Look at it this way... someone couldn't get anywhere near your score, thought it looked suspicious, and reported it. The mods checked it out, thought it looked suspicious as well, and removed it to investigate. Part of the investigation, and unfortunately the most helpful, is usually the user being able to replicate it at different clockspeeds to show the true scaling. There can be massive differences between drivers. The second issue is that even if your run was completely and totally legit on your end, something could still be breaking in the driver or in the benchmark. Judging by the scores being so difficult to replicate, I'd say this is the issue. I don't believe that you cheated, but I do believe that something bugged. I'd shrug this one off if it gets permanently removed and review your submissions with greater scrutiny in the future. That can do nothing but help you. If your run looks bugged, you can investigate yourself. If your run looks low, you can try to learn how to improve subscores, etc. @Leeghoofd, I'm looking forward to your review. Quote
Vellinious Posted November 9, 2016 Author Posted November 9, 2016 Hey Leeg, got your PM... sorry for the slow reply... wedding is in less than a month! I'd like to say a few things. First, @Vellinious, I know it feels like you're being personally attacked. If your result is 100% legit, don't worry about it. Look at it this way... someone couldn't get anywhere near your score, thought it looked suspicious, and reported it. The mods checked it out, thought it looked suspicious as well, and removed it to investigate. Part of the investigation, and unfortunately the most helpful, is usually the user being able to replicate it at different clockspeeds to show the true scaling. There can be massive differences between drivers. The second issue is that even if your run was completely and totally legit on your end, something could still be breaking in the driver or in the benchmark. Judging by the scores being so difficult to replicate, I'd say this is the issue. I don't believe that you cheated, but I do believe that something bugged. I'd shrug this one off if it gets permanently removed and review your submissions with greater scrutiny in the future. That can do nothing but help you. If your run looks bugged, you can investigate yourself. If your run looks low, you can try to learn how to improve subscores, etc. @Leeghoofd, I'm looking forward to your review. Nothing bugged during the run. No dropped textures, no black screens and no screen flashing. A few artifacts? Absolutely. I wasn't using any of the bios tricks to get higher frames (see the tools that used magic hex).... I ran nearly all of my benchmark runs between 1802 and 1846, because that's where it ran the best. It was an outstanding 8GB 290X that overclocked like no other Hawaii I had seen. I've been accused of cheating, and in the process, been cheated myself. In conclusion: eat me Quote
Crew Leeghoofd Posted November 9, 2016 Crew Posted November 9, 2016 Nobody called you a cheater in this thread... if we can't replicate this it will be classified as a bugged run... we suspect driver wise something is going bonkers due to too unstable GPU memory...CN is also on this and will crosscheck my upcoming findings with the Futuremark peeps... Quote
Vellinious Posted November 9, 2016 Author Posted November 9, 2016 Nobody called you a cheater in this thread... if we can't replicate this it will be classified as a bugged run... we suspect driver wise something is going bonkers due to too unstable GPU memory...CN is also on this and will crosscheck my upcoming findings with the Futuremark peeps... Oh, you have an 8GB XFX 290X that will clock the memory to 1800+? lol Quote
Crew Leeghoofd Posted November 9, 2016 Crew Posted November 9, 2016 (edited) Trust me we don't need 1800+ mems to measure/predict GT2 scaling or not. If a certain moment it goes nuts with just a few Mhz more, it is bugged... not an optimum setting... If your 1750MHz gives 71.8 and 1830 already does 78, I should be able do the same 6FPS or higher boost comparing between 1450 and 1600MHz right ? Edited November 9, 2016 by Leeghoofd Quote
rtsurfer Posted November 9, 2016 Posted November 9, 2016 This is bugged too...? AMD Radeon R9 290 video card benchmark result - Intel Core i5-4690K,ASUSTeK COMPUTER INC. MAXIMUS VII HERO (P.S not my result) Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.