frozensolid Posted October 7, 2017 Share Posted October 7, 2017 Hi, I'm a newbie overclocker just started learning about it So I have Galax 1050 TI, I did overclock with +100 voltage, 100% power limit(already max, can't increase), +1000 Mem clock & + 90~100 core clock When I test it in firestrike benchmark sometimes the test pass until finish sometimes it crash during the last tick when loading test (mostly graphic test 2 loading) and sometimes during the test, but never crash on physic test and combined test The temp never reach past 60 C during whole test So I test it using unigin heaven extreme preset about 10 minute+ and it never crash. I even manage to up the core to +150 core clock and test it again at unigine heaven and never crash and the temp can reach 70+ What happen with my overclock? Quote Link to comment Share on other sites More sharing options...
Crew Leeghoofd Posted October 7, 2017 Crew Share Posted October 7, 2017 just re-test firestrike with eg +100 core clock and only +950 mem clock. if that passes you know, it is the memory on the gpu holding back the card in that test. Continue testing till you find the max for your card. I usually test first max stable core clocks. Than add the memory into the mix. Remeber that each benchmark stresses the card differently, so one needs to find different operational aka stable values per benchmark. OC is all about logic debugging Quote Link to comment Share on other sites More sharing options...
frozensolid Posted October 7, 2017 Author Share Posted October 7, 2017 (edited) just re-test firestrike with eg +100 core clock and only +950 mem clock. if that passes you know, it is the memory on the gpu holding back the card in that test. Continue testing till you find the max for your card. I usually test first max stable core clocks. Than add the memory into the mix. Remeber that each benchmark stresses the card differently, so one needs to find different operational aka stable values per benchmark. OC is all about logic debugging thanks for the reply i tried +110 core and 0 mem clock, still same. i just update my driver and i think +100 core and +1000 mem clock is ok. havent test it many times +110 core & +1000 mem mostly failed but sometime pass and 1 more thing, i checked OSD and i saw the power limit never went pass 95%, i already set it to 100%, why is that? could it effect the crash? and also the volt mostly stuck at 1.1V sometimes it go 1.112V but most of the time its 1.1v PS : this is not for daily use, im trying to get high graphic score as high as i can for a competition the main question for me right now is why the power limit / consumption never go over 95%, it went up and down 88-94% because i believe the correct one have to be maxed at 99/100% Edited October 8, 2017 by frozensolid Quote Link to comment Share on other sites More sharing options...
Crew Leeghoofd Posted October 8, 2017 Crew Share Posted October 8, 2017 so your testing shows it is not the memclock making it fail, it is the core clock. Try +80 core clock and mem at +1000 and see if it passes all the time... like mentioned each benchmark stresses the gpu differently, so it could be it passes one benchmark at +110/1050 and onother one only at +90/1000. Quote Link to comment Share on other sites More sharing options...
frozensolid Posted October 8, 2017 Author Share Posted October 8, 2017 so your testing shows it is not the memclock making it fail, it is the core clock. Try +80 core clock and mem at +1000 and see if it passes all the time... like mentioned each benchmark stresses the gpu differently, so it could be it passes one benchmark at +110/1050 and onother one only at +90/1000. ya i can run the +80 core and mem +1000 perfectly stable and the power limit is still the same going up and down between 88-94% on graphic test 1 & 2 do you have any information why my GPU cant use power limit over 95%? i already set it 100% Quote Link to comment Share on other sites More sharing options...
Guest Wimpzilla Posted October 8, 2017 Share Posted October 8, 2017 Try to lower your gpu t° as much you can also, changing thermal paste, fan 100%, additional fan on the card/vrm. The best would be to slap a water cooling on it. It should also help with reaching your max clocks. Quote Link to comment Share on other sites More sharing options...
frozensolid Posted October 8, 2017 Author Share Posted October 8, 2017 Try to lower your gpu t° as much you can also, changing thermal paste, fan 100%, additional fan on the card/vrm.The best would be to slap a water cooling on it. It should also help with reaching your max clocks. i did crank up the fan to 100%, and also i did test it with room ambient 27C but for changing the thermal paste, i cant do that for now. because in my country, they put the warranty seal on the screw so i open it, my warranty will void lol but thanks for the reply and suggestion Quote Link to comment Share on other sites More sharing options...
Guest Wimpzilla Posted October 8, 2017 Share Posted October 8, 2017 (edited) You can gently remove the warranty sticker, heating it just just a bit with an hairdryer to help the removal and using a clean cutter blade. For your information, manufacturer will not refuse your card if you removed/destroy the sticker, it's more a deterrence move. The most important thing is to have the card as it was at stock without any evident damage caused by you. Usually there is a law to cite if the stubborn manufacturer still refuse the warranty, that state you own/bought the card and could remove the sticker if you wanted to without breaking the warranty, as you paid for the ownership of the goods. Or do this if you are skilled enough and/or give you an example of what NOT to do: BE AWARE if you SCRATCH the twizzer on the pcb, you could damage the copper traces killing the gpu, in this case for example, it would be NO warranty for you, worst case than breaking the sticker. Hope it help! Edited October 8, 2017 by Wimpzilla Quote Link to comment Share on other sites More sharing options...
frozensolid Posted October 8, 2017 Author Share Posted October 8, 2017 (edited) You can gently remove the warranty sticker, heating it just just a bit with an hairdryer to help the removal and using a clean cutter blade. For your information, manufacturer will not refuse your card if you removed/destroy the sticker, it's more a deterrence move. The most important thing is to have the card as it was at stock without any evident damage caused by you. Usually there is a law to cite if the stubborn manufacturer still refuse the warranty, that state you own/bought the card and could remove the sticker if you wanted to without breaking the warranty, as you paid for the ownership of the goods. when i first had it, i want to replace the paste with mine and tried to open the sticker with flat head screwdriver and it broke a little bit since its like fragile pieces. so i was like "oh bunnyextraction, nevermind then" lol sadly its abit different here in my country, all product is shipped by a distributor and all warranty claim goes to them. they have their own warranty sticker seal. so if i broke the seal, i cant claim warranty to them. well we do can claim warranty directly to the manufacturer but it will cost us shipping cost, and its expensive. so yeah, little fact about my country for you oh btw, the screw seal is just a sticker with a purchase date and stick across the PCB and on top of the screw haha Edited October 8, 2017 by frozensolid Quote Link to comment Share on other sites More sharing options...
Guest Wimpzilla Posted October 8, 2017 Share Posted October 8, 2017 I do not think the distributor warranty worth something, since they will ship back the card to the manufacturer, so again customers laws should protect you. Nevertheless i do not know how good/bad distributors are when handling with these stuff in your country, so yeah if you think it's a "no go" keep the stickers. Just add some fans all over the place to try to lower the gpu/vrm t°. Quote Link to comment Share on other sites More sharing options...
frozensolid Posted October 8, 2017 Author Share Posted October 8, 2017 I do not think the distributor warranty worth something, since they will ship back the card to the manufacturer, so again customers laws should protect you. Nevertheless i do not know how good/bad distributors are when handling with these stuff in your country, so yeah if you think it's a "no go" keep the stickers. Just add some fans all over the place to try to lower the gpu/vrm t°. yeah, like what linus said on their video about removing warranty sticker that i just watch its just like a "contract" and if it broke, they wont take it and ship it to the manufacturer, and i already tear the sticker a little bit so i wont take the risk and going to court is a no no for me lol thanks for your time helping me out Quote Link to comment Share on other sites More sharing options...
frozensolid Posted October 8, 2017 Author Share Posted October 8, 2017 ya i can run the +80 core and mem +1000 perfectly stableand the power limit is still the same going up and down between 88-94% on graphic test 1 & 2 do you have any information why my GPU cant use power limit over 95%? i already set it 100% Does PSU and motherboard have an impact that causing my GPU power limit instability? Does all GPU use 100% it's power limit when fully loaded? Do I get a bad 1050 TI? Quote Link to comment Share on other sites More sharing options...
Crew Leeghoofd Posted October 8, 2017 Crew Share Posted October 8, 2017 Nope consider it is a build in protection for the GPU, Buildzoid and Der8auer have made some nice articles about it. Only way to circumvent that power limit is to do a shunt resistor mod, voiding your warranty again... Quote Link to comment Share on other sites More sharing options...
Guest Wimpzilla Posted October 8, 2017 Share Posted October 8, 2017 Nope consider it is a build in protection for the GPU, Buildzoid and Der8auer have made some nice articles about it. Only way to circumvent that power limit is to do a shunt resistor mod, voiding your warranty again... Instead using liquid thermal paste, what about the old trick of using graphite? Using a soft pencil, drawing on the top of the resistor and apply a thin layer of graphite could help, giving a small current boost? Should not be an issue with the warranty since it is very easy to remove and leave no traces usually. Quote Link to comment Share on other sites More sharing options...
Crew Leeghoofd Posted October 8, 2017 Crew Share Posted October 8, 2017 depends if you need to remove the cooler yes or no right Quote Link to comment Share on other sites More sharing options...
Guest Wimpzilla Posted October 8, 2017 Share Posted October 8, 2017 (edited) depends if you need to remove the cooler yes or no right True, i forgot the cooler. I did not leveraged current sensing/shunt on these cards, dunno if they required/have it due the low power usage. Edited October 8, 2017 by Wimpzilla Quote Link to comment Share on other sites More sharing options...
frozensolid Posted October 9, 2017 Author Share Posted October 9, 2017 Nope consider it is a build in protection for the GPU, Buildzoid and Der8auer have made some nice articles about it. Only way to circumvent that power limit is to do a shunt resistor mod, voiding your warranty again... Instead using liquid thermal paste, what about the old trick of using graphite? Using a soft pencil, drawing on the top of the resistor and apply a thin layer of graphite could help, giving a small current boost? Should not be an issue with the warranty since it is very easy to remove and leave no traces usually. thanks for both of you but i think theres nothing i can do furthermore i will just accept it the fact that i lose the hardware lottery cheers Quote Link to comment Share on other sites More sharing options...
frozensolid Posted October 9, 2017 Author Share Posted October 9, 2017 LOL ok guys, i found something funny when i enabled LOD +3, i can do +150 core +1000 stable most of the time now im really have no idea how that can happen Quote Link to comment Share on other sites More sharing options...
Guest Wimpzilla Posted October 9, 2017 Share Posted October 9, 2017 Well, positive LOD should alleviate the gpu load, in some ways when rendering the displayed image. So not surprising that with lower load = lower stress on the gpu = lower power/T° required = highers oc clocks! Quote Link to comment Share on other sites More sharing options...
frozensolid Posted October 10, 2017 Author Share Posted October 10, 2017 Well, positive LOD should alleviate the gpu load, in some ways when rendering the displayed image. So not surprising that with lower load = lower stress on the gpu = lower power/T° required = highers oc clocks! Lol cool! Quote Link to comment Share on other sites More sharing options...
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.