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 791

post #7901 of 10682
Quote:
Originally Posted by Slackaveli View Post

point is still the same. my initial reaction was simply, "Because we all don't have 8 core cpus". Don't bring a cpu benchmark leaderboard up to argue your point about gpu. very simple. Especially when YOU YOURSELF have an old 4 core. Why not bring a 4 core leaderboard? You just tried to make a weak point with a weak pic, and got called out on it. Why? Because, you have jumped on other posters in this thread and I didn't feel like letting you slide on it.

you win I can't argue with a guy of your calibre
I learnt something today TimeSpy is solely a CPU based benchmark I had always thought it was a combined system benchmark with 3 separate GPU and CPU and combined tests

Warning: Spoiler! (Click to show)
When I compare I only look in the red box it seems I might need to buy another $3000 CPU/Mobo/memory combo to gain 5% FPS or I could spend alittle less and just go SLI with another 1080Ti what do you think?
Warning: Spoiler! (Click to show)

feel free to get your buddies to PM me to correct me no need to fill this thread with...........
kamikaze
(17 items)
 
  
CPUMotherboardGraphicsGraphics
3770k @ 4.8 asus maximus v extreme Asus Strix 1080 Ti  Asus GTX 770 DCUii 
RAMHard DriveCoolingOS
GSkill 2400Mhz Sandisk ultra pro xspc 240 raystorm Windows 8.1 Pro 
MonitorMonitorMonitorKeyboard
LG UC9938-w AOC 24"  AOC 24" G19 
PowerCaseMouseAudio
enermax maxrevo 1500w nzxt 820 Phantom saitik rat 9 Asus Xonar DX 
  hide details  
Reply
kamikaze
(17 items)
 
  
CPUMotherboardGraphicsGraphics
3770k @ 4.8 asus maximus v extreme Asus Strix 1080 Ti  Asus GTX 770 DCUii 
RAMHard DriveCoolingOS
GSkill 2400Mhz Sandisk ultra pro xspc 240 raystorm Windows 8.1 Pro 
MonitorMonitorMonitorKeyboard
LG UC9938-w AOC 24"  AOC 24" G19 
PowerCaseMouseAudio
enermax maxrevo 1500w nzxt 820 Phantom saitik rat 9 Asus Xonar DX 
  hide details  
Reply
post #7902 of 10682
Quote:
Originally Posted by Luckbad View Post

Okay, I believe at this point I've found my best memory overclock.

6210 MHz yields the highest scores in SuperPosition. 6220 does not artifact, but it yields much lower scores. 6237 will artifact or freeze.

Test Environment
Note that I'm not running in a 100% perfect benchmark environment for maximum scores. I was only interested in testing the memory on my GPU.
  • My CPU isn't overclocked to its limit. It's just a 6700k running at lowish voltage at 4.4 GHz (my daily speed, not a real benching speed).
  • My RAM is not overclocked beyond its XMP right now. 16GB of G.Skill TridentZ DDR4 3600 at 15-15-15-35
  • My GPU core is not absolutely maxed out. It's running at +92, which is 2075.5 MHz.
  • I'm allowing for some overvoltage, but it's limited to 1.075 and is not currently Locked via a curve

That is a long preamble for nothing earth shattering at all. Basically, I mentioned in my previous post that as I clock the GPU memory clock up or down, I see pretty wild fluctuations in my SuperPosition scores. +407 offset is definitively better than +417 every single time, well beyond the margin of error. I skipped up until I started seeing artifacts and/or freezing and found a comfortable memory overclock where that would never happen.

Then I started changing numbers and running SuperPosition (4K Optimized). I kept the fans on my Zotac 1080 Ti Amp Extreme at 100% and let the GPU cool down to 35 °C between runs (not its actual idle temp, but it was getting to ~55-58 °C during each run and didn't take long to get to 35).

Offset | Clock | Score
+597 | 6210 | 10270
+615 | 6210 | 10281
+628 | 6220 | 10123
+610 | 6210 | 10279
+605 | 6210 | 10282
+616 | 6220 | 10126
+615 | 6210 | 10285
+599 | 6210 | 10280
+596 | 6210 | 10281
+618 | 6220 | 10123
+596 | 6210 | 10276
+620 | 6220 | 10126

Long story short, everything clocked at 6210 MHz gives scores within the margin of error. There are thresholds at which the memory will gain frequency, and it seems like it doesn't really matter what the number is within that frequency band is, just what the actual clock is.

Yeah looks like its as @nrpeyton alluded to, there are frequency points were it changes multipliers and loosens timings. definitely around +616 it drops off in your case. Be interesting to test with OCCT, but I also find different bioses slant that too, FE was good for me at 6003, inno3d, sc2 and Palit seemed to like 6055, and Asus 6156.
MSI
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 6700k MSI XPower Gaming Titanium Edition MSI 980Ti Lightning Gskill RipJaw V 3200 C16 
Hard DriveCoolingCoolingCooling
Samsung 950 Pro Raid0 Water cooled with 750w chiller EK mono block - MSI XPGTi Bitspower - MSI Lightning 
OSMonitorKeyboardPower
Win 10 Home ACER X34 - 3440x1440  Corsair K65 Seasonic 1050w - Silent Snow 
CaseMouseMouse PadAudio
Silverstone RV05 - White Razer Epic Naga Chroma Artizan - Hayete Otsu - Soft Corsair Void Wireless - Carbon 
  hide details  
Reply
MSI
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 6700k MSI XPower Gaming Titanium Edition MSI 980Ti Lightning Gskill RipJaw V 3200 C16 
Hard DriveCoolingCoolingCooling
Samsung 950 Pro Raid0 Water cooled with 750w chiller EK mono block - MSI XPGTi Bitspower - MSI Lightning 
OSMonitorKeyboardPower
Win 10 Home ACER X34 - 3440x1440  Corsair K65 Seasonic 1050w - Silent Snow 
CaseMouseMouse PadAudio
Silverstone RV05 - White Razer Epic Naga Chroma Artizan - Hayete Otsu - Soft Corsair Void Wireless - Carbon 
  hide details  
Reply
post #7903 of 10682
Thanks to all who answerd my original post with my strix 1080ti's.

Firstly, after talking to someone who was not I presume the work experience kid, the e-tailer have agreed to a speedy return.

Secondly, I have recently made another observation, and have spent painful hours trying to replicate and fix issues. My windows is giving me bsod with the fault in non paged erea and other memory controller issues. I'll wait for replacement card to arrive and see if the issues are resolved. Other forums have been steering me towards a faulty memory controller on my cpu, so to eliminate that as a issue I ordered a 6900k... So much money being dumped for seemingly no reason I just want to go back to my 1080 hybrid where everything worked ok lol. Very frustrating especially when you feel all alone with no one else around my town that I know of with x99 stuff to help me test =p thanks to all who replied again you guys make me feel not so alone =D
post #7904 of 10682
Quote:
Originally Posted by Benny89 View Post

This is curve or just good old sliders?

And thanks for that. You convinced me to return my card. Yours can do 2075 in Witcher 3 at 1.032V.... Mine can't do even 2038 at 1.063V in W3. Already ordered another STRIX, ready to pick up. Gonna return this one day after tomorrow.

I have to say that one thing I hate about PC hardware is sillicon lottery, be it GPU or CPU.... everything should be the same. Well, that will be my 4th 1080Ti rolleyes.gif

Hope next one will finally be decent overclocker.

EDIT: and BTW hour in W3 is not enough. I thought I had stable 2050 in W3, crashed after 2,5h of gaming. Play for like 5-6 hours then I can say this is stable. Not trying to say it won't be, just giving you an advice. One hour in one game is not proof of stable clocks. You need couple of hours at least to say that.

Alright i will try and put it on now and let it run for a couple of hours and see smile.gif

EDIT: Oh and it is curve. No locking.
Edited by viz0id - 5/1/17 at 3:12am
post #7905 of 10682
Hello, I accidentally touched the details on the back of the board and 3 details came off, the card keeps working fine, but tell me, can I leave it so?
The photo is highlighted with red circles.

IMG_20170430_211621933.jpg 3753k .jpg file
post #7906 of 10682
Just did a semi-thorough test on the memory clock

with the core stable at 2062.5 throughout the Superposition benchmark, I've got following results:

Vram SP
+925 10311
+900 10303
+875 10289
+850 10254
+750 10221
+500 10071

It starts to artifact at +1000, and fails 20 loops stress test at +950.

Now I'm running at +900 on the memory. it seems to be a sweet spot for my card.
post #7907 of 10682
Quote:
Originally Posted by SlimJ87D View Post

If you're not shunt modded, then you should do some thorough test and comparisons, that's all I'm recommending. All I'm saying is it can be a mistake assuming more on the core is going to get you better results, that's evident with my test and kraxkills.

And if you are not shunt modded, you're not getting 400 watts. I can't confirm this, another user did but not even 350 watts according to his measurements. If you have other test, it would be beneficial to share them for knowledge purposes. 350 watts would be pretty sweet, but the only measurements was from a video another user made and he settled at around 325 watts.

Well thats the thing, I matched the curve to give me what the manual oc was giving me so more or less the same core not higher.

Yeah I was using HWM for the raw power numbers so I'm not sure how accurate that is. But I did observe that the normalized TDP was hitting a max of 124% which does make sense since 320w + 24% is roughly 400w.

Second observation is that total system power draw was at ~600w on my AX860i graphing. This is at 8K Supo which isn't that CPU intensive so unless my CPU was drawing more than 200w at 1.245v 4.5ghz in Supo, there can't be anything else that adds up to 600w..

Of course it is not holding at the max though. It somehow hit that max briefly before it power limited itself and throttled down possibly.
post #7908 of 10682
Quote:
Originally Posted by nrpeyton View Post

Possible important info on Memory Overclocking.

I trust some of you have now been using OCCT to find max stable memory O/C, (which we found is great at finding a more precise, stable, upper memory O/C).

But one more thing to note of importance:

I believe there 'could be' different memory timings straps programmed into BIOS which are automatically activated depending on your memory O/C. (for instance a good comparison to this would be how DDR3/4 XMP profiles load optimised (looser) timings for higher speeds you select in motherboard BIOS).

Anyway these timings can affect your score when you're trying to find best memory O/C.

During the 1080 days (at the owners club over there) there was even a roomer circulating that odd overclocks scored higher than even overclock numbers or memory overclocks ending in 0's didn't score as well as those ending in 5. (I.E. +495/+605 always scored higher than +500/+600).

What is more likely to be the case is that: pre-optimised 'memory timings straps' are selected by BIOS depending on the O/C you enter.
For example; +250...lost FPS. But +240 and +255-+260 = gain.

And remember OCCT is still the best we have for testing memory O/C "stability". smile.gif

If anyone has anything to add, or any corrections (or different opinion), fire away :-) Would be good to see some 'memory overclocking' spirited discussion.

Explains why my testing showed anything but linear results. I settled at +350 but if according to this even a +5 or +10 change makes a difference I might have to do more extensive testing.
post #7909 of 10682
Quote:
Originally Posted by Atotarho View Post

Hello, I accidentally touched the details on the back of the board and 3 details came off, the card keeps working fine, but tell me, can I leave it so?
The photo is highlighted with red circles.

IMG_20170430_211621933.jpg 3753k .jpg file

Two them appear to be on the back of memory modules. Im not entirely sure but i think many of those capacitors exist to improve the quality of electrical signals on the board.

I'd be looking to do some stress testing of the vram. However, in my experience it has always been difficult to find one that reliably stresses the huge amounts of vram we have now.

EDIT: Try this, it only allocated 3.5GB for each instance, but I opened three of them and ran them all and GPU memory usage was reported as 11195 in afterburner.

http://www.programming4beginners.com/gpumemtest
Edited by mshagg - 5/1/17 at 7:06am
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 #7910 of 10682
Quote:
Originally Posted by mshagg View Post

Two them appear to be on the back of memory modules. Im not entirely sure but i think many of those capacitors exist to improve the quality of electrical signals on the board.

I'd be looking to do some stress testing of the vram. However, in my experience it has always been difficult to find one that reliably stresses the huge amounts of vram we have now.

EDIT: Try this, it only allocated 3.5GB for each instance, but I opened three of them and ran them all and GPU memory usage was reported as 11195 in afterburner.

http://www.programming4beginners.com/gpumemtest

Thanks, I started 3 tests at once, all ok. But the memory worked at 10,400 MHz, so it should be? I have a bios ichill 4, the memory I usually work at 11400 MHz. So these 2 capacitors can not be soldered? A transistor what role does it perform?
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