Overclock.net › Forums › Graphics Cards › NVIDIA › NVIDIA GTX 1080 Ti Owner's Club
New Posts  All Forums:Forum Nav:

NVIDIA GTX 1080 Ti Owner's Club - Page 734

post #7331 of 10504
Quote:
Originally Posted by KedarWolf View Post

I disagree with you again, Slim. Palit BIOS has its uses for those it works for and for some it does. If you don't have the shunt mod and/or are on air Palit as I tested it game me decent clocks and the lowest voltages I ever tested on a BIOS and while on benching it not the best performer, for 24/7 use for gaming to avoid power limiting and/or thermal throttling it is a decent BIOS as it hardly power limits at all and can get good clocks are really low voltages.

You know though, @SlimJ87D, you come across and an elitist that everyone HAS to agree with you and you act like you're always right which at times your not. At least when you HAVE been right and I've been wrong, I owned up to it. But, you're ALWAYs right and have ALL the answers, it's really frustrating for me you act like this.

Anyhow, it's not a forum to voice issues really but I'd wish you not respond in absolutes and realise that you are not the end all to beat all when it comes to the forums. I'm sure you are an intelligent guy, no doubt, it just seems that the 'I am right and you are wrong' kinda attitude and statements is not conducive to a learning and growing community that is here to help each other, not to cause division and strife. redface.gif

Anyhow, my rant is done.

And yes, this person is having an issue with the Palit BIOS but I've not had this issue, no power limiting until 117%.

I'm thinking even though my card is a reference FE somehow because it's a first model Gigabyte edition I got on release day there might somehow be some kind of differences where some get varying results then I do while some have similar results and I've seen both sides of that coin in the forums and elsewhere.

Palit BIOS does have it uses for those it's working for though. smile.gif

All I did was pull statistics, notnsure how that's being an elitist? So anyone that believes in math is an elitist?

You have a power supply that measures watts, you had all the time in the world to test and see how much watts the Palit bios delivers, which is 30 watts, but instead you just flashed it, and then "confirmed" it gives you 50 watts without confirming anything. You also went on about it being the best bios out there until everyone started giving it negative feedback. And now I'm an "elitist" because all this happened?

Honestly, I wouldn't even make comments or suggestions to users about bios, but people don't keep up with this forum and they might have read an old post that didn't give them the full scoop. Maybe my post have come off as arrogant, and I'll work on that. But the attitude in my post is due to retaliation of the above paragraph. And I'm sure your entire post in in retaliation of me questioning your methods all the time.

I'm not always right, but we have a forum where smart people can gather together to perform thorough testing. We get more done if people as a group set out to perform research together. And the results speak for themselves.

Just like the Asus bios,several post back you had 5 of us perform test and only measured 15 watts of extra power using nvidia commands, HWINFO64 and PSU. We gathered together to investigate and do the community some good.

All the Palit bios did for me was lower my scores across the board in all benchmarks. Same as other people have reported. So I'm not sure how it's beneficial to flash a bios that just scores worse even with more power draw. That's all I told the previous user.

Truth be told, we both need each other on this forum to conduct research and make discoveries with one another.
Edited by SlimJ87D - 4/27/17 at 8:17am
post #7332 of 10504
I assume memory voltage is seperate?
i seem to get artifacts around +450 occasionally when i was running heaven, my gut is that +500 should easily be achievable, i havnet played too much with the memory, the curve last night took ages, but i seem to have a much stronger overclock due to it, much less power limiting smile.gif
    
CPUMotherboardGraphicsRAM
7600k gigabyte z270 ultra Zotac 1080TI FE 16Gb corsair 3200 LPX 
Hard DriveHard DriveCoolingMonitor
Samsung 950 NVME 256GB Samsung 850 1TB Mayhems Havoc 360, Alphacool crossflow 280 Philips 40" 4K 
PowerCase
Aerocool Xpredator 750W Phanteks Evolv ATX TG 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
7600k gigabyte z270 ultra Zotac 1080TI FE 16Gb corsair 3200 LPX 
Hard DriveHard DriveCoolingMonitor
Samsung 950 NVME 256GB Samsung 850 1TB Mayhems Havoc 360, Alphacool crossflow 280 Philips 40" 4K 
PowerCase
Aerocool Xpredator 750W Phanteks Evolv ATX TG 
  hide details  
Reply
post #7333 of 10504
Quote:
Originally Posted by alanthecelt View Post

I assume memory voltage is seperate?
i seem to get artifacts around +450 occasionally when i was running heaven, my gut is that +500 should easily be achievable

Unfortunately your guts are wrong. It seems memory OC differs more than maximum core OC. The range seems to be between +250 to +750, with most falling between 350 to 600 Mhz, with the average probably somewhere between 450-500Mhz.

From my 2 FE 1080Tis, one does 650Mhz the other one about 450Mhz.

Also, some memory OC's u see might actually already trigger the error correction in the VRAM modules and therefore cause some performance penalty over a slightly lower overclock without ECC kicking in.
Fun
(15 items)
 
AntiAlias
(11 items)
 
 
CPUMotherboardGraphicsRAM
4770K @ 4.5Ghz @ 1.38v delidded Gigabyte GA-Z87X-OC f9c 2 x Zotac 980Ti Amp Omega @1530, @1530 G.SKILL TridentX 2400 CL10 
Hard DriveHard DriveOptical DriveCooling
Samsung 830 256GB 4 x Samsung EVO 850 1TB RAID0 Samsung Blu Ray Cooler Master Seidon 240M 
OSMonitorKeyboardPower
Windows 7 Professional 64 bit Samsung 27" IPS 2560x1440 Logitech Wireless Seasonic Platinum 860 
CaseMouseAudio
NZXT Phantom 630 white Razer ASUS Xonar U7 external 
CPUMotherboardGraphicsGraphics
7700K@5Ghz ASUS ROG STRIX Z270F Asus 1080Ti FE Gigabyte 1080Ti FE 
RAMHard DriveHard DriveOS
Corsair Vengeance LED 32GB (4x 8GB) DDR4 3400MH... Samsung 960 EVO 500GB 4 x 1TB 850 EVO SSD RAID0 Windows 10 Pro 64 Bit 
MonitorPowerCase
LG 27MU67-B 4K BeQuiet Straight Power 10 800W anidees AI-Crystal black 
  hide details  
Reply
Fun
(15 items)
 
AntiAlias
(11 items)
 
 
CPUMotherboardGraphicsRAM
4770K @ 4.5Ghz @ 1.38v delidded Gigabyte GA-Z87X-OC f9c 2 x Zotac 980Ti Amp Omega @1530, @1530 G.SKILL TridentX 2400 CL10 
Hard DriveHard DriveOptical DriveCooling
Samsung 830 256GB 4 x Samsung EVO 850 1TB RAID0 Samsung Blu Ray Cooler Master Seidon 240M 
OSMonitorKeyboardPower
Windows 7 Professional 64 bit Samsung 27" IPS 2560x1440 Logitech Wireless Seasonic Platinum 860 
CaseMouseAudio
NZXT Phantom 630 white Razer ASUS Xonar U7 external 
CPUMotherboardGraphicsGraphics
7700K@5Ghz ASUS ROG STRIX Z270F Asus 1080Ti FE Gigabyte 1080Ti FE 
RAMHard DriveHard DriveOS
Corsair Vengeance LED 32GB (4x 8GB) DDR4 3400MH... Samsung 960 EVO 500GB 4 x 1TB 850 EVO SSD RAID0 Windows 10 Pro 64 Bit 
MonitorPowerCase
LG 27MU67-B 4K BeQuiet Straight Power 10 800W anidees AI-Crystal black 
  hide details  
Reply
post #7334 of 10504
Quote:
Originally Posted by BrainSplatter View Post

Yes, just that in the way the curve editor works, you have to raise the clock speed of the lower voltage points. Another good starting point is to check out what's the maximum stable core for 1.0v (set 1v to test core clock and flatten curve behind it) because 1v is almost never power limited.

So see what the highest core clock I can get at 1.000v then even out the line after that? Do I end up raising it later though or just leave it flat? Thanks. I will def try this when I get home. As long as I am at 2000mhz @ 1.000v I will be happy. I easily was hitting 2136mhz on my 1080 though.
post #7335 of 10504
Quote:
Originally Posted by eXistencelies View Post

So the only way to fix that is the shunt mod? I am curious why the power jumps higher than 120 though?

It's the only reliable way to trick the card into drawing more power. We are testing various non-FE BIOSes but results have been mixed.

It is likely jumping above 120 momentarily, which the card detects, and downclock/downvolts itself.
Quote:
Originally Posted by BrainSplatter View Post

Unfortunately your guts are wrong. It seems memory OC differs more than maximum core OC. The range seems to be between +250 to +750, with most falling between 350 to 600 Mhz, with the average probably somewhere between 450-500Mhz.

From my 2 FE 1080Tis, one does 650Mhz the other one about 450Mhz.

Also, some memory OC's u see might actually already trigger the error correction in the VRAM modules and therefore cause some performance penalty over a slightly lower overclock without ECC kicking in.

It's a shame NVIDIA dont let us monitor ECC flags on these cards. Would be a good way to reliably test mem OC. NVSMI has the capability, but I assume it's only for their profession/compute products.

Mine has a sweet spot at +452, no more no less, according to benchmarks, but can go +600 without crashing.
Edited by mshagg - 4/27/17 at 7:00am
Main rig
(9 items)
 
  
CPUMotherboardGraphicsRAM
i7 5930k Asus X99 Deluve Nvidia GTX 1080Ti  Corsair Dominator 3000Mhz 
Hard DriveCoolingOSPower
Intel 600p 1x420 rad, 1x480 rad, D5 Pump Windows 10 Pro x64 EVGA 1000w 
Case
Enthoo Primo 
  hide details  
Reply
Main rig
(9 items)
 
  
CPUMotherboardGraphicsRAM
i7 5930k Asus X99 Deluve Nvidia GTX 1080Ti  Corsair Dominator 3000Mhz 
Hard DriveCoolingOSPower
Intel 600p 1x420 rad, 1x480 rad, D5 Pump Windows 10 Pro x64 EVGA 1000w 
Case
Enthoo Primo 
  hide details  
Reply
post #7336 of 10504
Quote:
Originally Posted by eXistencelies View Post

So see what the highest core clock I can get at 1.000v then even out the line after that? Do I end up raising it later though or just leave it flat?

All voltage points after 1.0v should have a core clock equal/lower than the value for 1.0v (when u press apply in Afterburner it will raise all lower clocks after the 1.0v voltage point to the clock of 1.0v).

That curve u should store in a AB profile. After u found the max clock for 1.0v u replace the profile with it. Then u can then use that curve as a good starting point to find the max OC for 1.012v, 1025v, ... because u only have to raise the core clock of the next voltage point.

For most games u can usually use higher voltages (-> higher core) before running into a power limit compared to many benchmarks.
Fun
(15 items)
 
AntiAlias
(11 items)
 
 
CPUMotherboardGraphicsRAM
4770K @ 4.5Ghz @ 1.38v delidded Gigabyte GA-Z87X-OC f9c 2 x Zotac 980Ti Amp Omega @1530, @1530 G.SKILL TridentX 2400 CL10 
Hard DriveHard DriveOptical DriveCooling
Samsung 830 256GB 4 x Samsung EVO 850 1TB RAID0 Samsung Blu Ray Cooler Master Seidon 240M 
OSMonitorKeyboardPower
Windows 7 Professional 64 bit Samsung 27" IPS 2560x1440 Logitech Wireless Seasonic Platinum 860 
CaseMouseAudio
NZXT Phantom 630 white Razer ASUS Xonar U7 external 
CPUMotherboardGraphicsGraphics
7700K@5Ghz ASUS ROG STRIX Z270F Asus 1080Ti FE Gigabyte 1080Ti FE 
RAMHard DriveHard DriveOS
Corsair Vengeance LED 32GB (4x 8GB) DDR4 3400MH... Samsung 960 EVO 500GB 4 x 1TB 850 EVO SSD RAID0 Windows 10 Pro 64 Bit 
MonitorPowerCase
LG 27MU67-B 4K BeQuiet Straight Power 10 800W anidees AI-Crystal black 
  hide details  
Reply
Fun
(15 items)
 
AntiAlias
(11 items)
 
 
CPUMotherboardGraphicsRAM
4770K @ 4.5Ghz @ 1.38v delidded Gigabyte GA-Z87X-OC f9c 2 x Zotac 980Ti Amp Omega @1530, @1530 G.SKILL TridentX 2400 CL10 
Hard DriveHard DriveOptical DriveCooling
Samsung 830 256GB 4 x Samsung EVO 850 1TB RAID0 Samsung Blu Ray Cooler Master Seidon 240M 
OSMonitorKeyboardPower
Windows 7 Professional 64 bit Samsung 27" IPS 2560x1440 Logitech Wireless Seasonic Platinum 860 
CaseMouseAudio
NZXT Phantom 630 white Razer ASUS Xonar U7 external 
CPUMotherboardGraphicsGraphics
7700K@5Ghz ASUS ROG STRIX Z270F Asus 1080Ti FE Gigabyte 1080Ti FE 
RAMHard DriveHard DriveOS
Corsair Vengeance LED 32GB (4x 8GB) DDR4 3400MH... Samsung 960 EVO 500GB 4 x 1TB 850 EVO SSD RAID0 Windows 10 Pro 64 Bit 
MonitorPowerCase
LG 27MU67-B 4K BeQuiet Straight Power 10 800W anidees AI-Crystal black 
  hide details  
Reply
post #7337 of 10504
Quote:
Originally Posted by BrainSplatter View Post

All voltage points after 1.0v should have a core clock equal/lower than the value for 1.0v (when u press apply in Afterburner it will raise all lower clocks after the 1.0v voltage point to the clock of 1.0v).

That curve u should store in a AB profile. After u found the max clock for 1.0v u replace the profile with it. Then u can then use that curve as a good starting point to find the max OC for 1.012v, 1025v, ... because u only have to raise the core clock of the next voltage point.

For most games u can usually use higher voltages (-> higher core) before running into a power limit compared to many benchmarks.

OK great! Will def check that out when I get home from work later. So find the highest core clock at 1.000v. Hit the check mark and the rest of the points behind level out. Save that profile. Then test the next point at the next voltage level and see what the highest I can get on that? Then repeat, yes? Also when I am running a benchmark and see the crosshair drop some core where I have the point set I should lower that point to there? Another thing is I should only have 120 power and voltage level on 0, right?
post #7338 of 10504
Quote:
Originally Posted by alanthecelt View Post

I assume memory voltage is seperate?
i seem to get artifacts around +450 occasionally when i was running heaven, my gut is that +500 should easily be achievable, i havnet played too much with the memory, the curve last night took ages, but i seem to have a much stronger overclock due to it, much less power limiting smile.gif

Its only a function of voltage if your IMC is the bottleneck. Even then the IMC on Pascal has its own voltage plane + VRM. The thing that has the biggest impact on Memory OC is the CAS timings of the VRAM which cannot be modified by us, it is up to the manufacturer of your GPu to release a BIOS that loosen's the timings to allow the extra OC. Then after that its a function of silicon lottery, after all, the odds of landing 11 GGDR5X chips with not a single one being the bottleneck to go 500+ is actually on the uncommon side.
post #7339 of 10504
So Zotac were good enough to RMA my card for the cosmetic defect.

New card seems to hold the same core clocks, around 2037 stable. However this one crashes sooner at 2050 than the previous one. But it doesn't jump around as much with clocks stabilising around 2025-2037.

Memory also goes up to +450 on this one without artifacting (previous card could only do +300), this is inclusive of the stock overclock of +100 Zotac has on their Amp Extreme cards.

What I've noticed though is that power usage never goes anywhere near the 120% limit? Max was 115% but that lasted for less than a second before it downclocked and lowered back to around the 110% range.

So am I just being limited by voltage on this chip?

Edit : Also do all 1080tis get stuck on lower bins after a instability or crash occurs until restart? Its a bit annoying to have to keep restarting when trying to find the max stable clocks.
Edited by TheBoom - 4/27/17 at 7:16am
post #7340 of 10504
Quote:
Originally Posted by eXistencelies View Post

Hit the check mark and the rest of the points behind level out. Save that profile. Then test the next point at the next voltage level and see what the highest I can get on that? Then repeat, yes?

Yes. You just have to lower the core clock value of all voltages points after the 1.0v point once before hitting apply. Then AB will straighten out that line. That's a bit of fiddling due to the number of points. But after that u have a nice starting point for all further tests.

Quote:
Originally Posted by TheBoom View Post

But it doesn't jump around as much with clocks stabilising around 2025-2037.

Seems to hava a lower VID than your first one. That's also why u don't see much power throttling. So there is probably some more power headroom for a higher voltage and clock before it will again start to fluctuate a lot due to hitting the power level ceiling constantly.
Edited by BrainSplatter - 4/27/17 at 7:22am
Fun
(15 items)
 
AntiAlias
(11 items)
 
 
CPUMotherboardGraphicsRAM
4770K @ 4.5Ghz @ 1.38v delidded Gigabyte GA-Z87X-OC f9c 2 x Zotac 980Ti Amp Omega @1530, @1530 G.SKILL TridentX 2400 CL10 
Hard DriveHard DriveOptical DriveCooling
Samsung 830 256GB 4 x Samsung EVO 850 1TB RAID0 Samsung Blu Ray Cooler Master Seidon 240M 
OSMonitorKeyboardPower
Windows 7 Professional 64 bit Samsung 27" IPS 2560x1440 Logitech Wireless Seasonic Platinum 860 
CaseMouseAudio
NZXT Phantom 630 white Razer ASUS Xonar U7 external 
CPUMotherboardGraphicsGraphics
7700K@5Ghz ASUS ROG STRIX Z270F Asus 1080Ti FE Gigabyte 1080Ti FE 
RAMHard DriveHard DriveOS
Corsair Vengeance LED 32GB (4x 8GB) DDR4 3400MH... Samsung 960 EVO 500GB 4 x 1TB 850 EVO SSD RAID0 Windows 10 Pro 64 Bit 
MonitorPowerCase
LG 27MU67-B 4K BeQuiet Straight Power 10 800W anidees AI-Crystal black 
  hide details  
Reply
Fun
(15 items)
 
AntiAlias
(11 items)
 
 
CPUMotherboardGraphicsRAM
4770K @ 4.5Ghz @ 1.38v delidded Gigabyte GA-Z87X-OC f9c 2 x Zotac 980Ti Amp Omega @1530, @1530 G.SKILL TridentX 2400 CL10 
Hard DriveHard DriveOptical DriveCooling
Samsung 830 256GB 4 x Samsung EVO 850 1TB RAID0 Samsung Blu Ray Cooler Master Seidon 240M 
OSMonitorKeyboardPower
Windows 7 Professional 64 bit Samsung 27" IPS 2560x1440 Logitech Wireless Seasonic Platinum 860 
CaseMouseAudio
NZXT Phantom 630 white Razer ASUS Xonar U7 external 
CPUMotherboardGraphicsGraphics
7700K@5Ghz ASUS ROG STRIX Z270F Asus 1080Ti FE Gigabyte 1080Ti FE 
RAMHard DriveHard DriveOS
Corsair Vengeance LED 32GB (4x 8GB) DDR4 3400MH... Samsung 960 EVO 500GB 4 x 1TB 850 EVO SSD RAID0 Windows 10 Pro 64 Bit 
MonitorPowerCase
LG 27MU67-B 4K BeQuiet Straight Power 10 800W anidees AI-Crystal black 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA
Overclock.net › Forums › Graphics Cards › NVIDIA › NVIDIA GTX 1080 Ti Owner's Club