[Official] NVIDIA GTX 1080 Ti Owner's Club - Page 1553 - Overclock.net

Forum Jump: 
Reply
 
Thread Tools
post #15521 of 15976 Old 12-17-2017, 05:18 PM
Intel Overclocker
 
gta1989's Avatar
 
Join Date: Dec 2015
Location: iowa
Posts: 45
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 2
Quote:
Originally Posted by EarlZ View Post

What game are you using to test for stability?

I play alot of different games. heros of the storm doesnt support sli so just 1 card runs. destiny 2 does support sli and in pvp will usually show you an unstable core clock pretty fast. I have a 4k tv i play on so idk if that makes any difference the resolution that you are playing at. I also use max settings that i can sustain 60fps min. I hate laggy gaming and tearing of the frames so i usually turn off vsync unless that game just plain doesnt play well with high fps
gta1989 is offline  
Sponsored Links
Advertisement
 
post #15522 of 15976 Old 12-17-2017, 05:20 PM
Intel Overclocker
 
gta1989's Avatar
 
Join Date: Dec 2015
Location: iowa
Posts: 45
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 2
Quote:
Originally Posted by nrpeyton View Post

After you've dialled in your +375 and saved the profile etc try rebooting your machine (soft restart only). The GPU will re-initialize memory timings (the same way your regular RAM trains at start-up). When you get back into Windows you may find you can now go a bit higher with VRAM overclock.


I will give this a try next time I play with some bench marks. do you restart more then once and keep going or is it a one time deal?
gta1989 is offline  
post #15523 of 15976 Old 12-17-2017, 06:31 PM
 
l1CappYl1's Avatar
 
Join Date: Nov 2017
Location: China
Posts: 10
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 0
Quote:
Originally Posted by StenioMoreira View Post

Yeah no, XOC has none, at least not for me. I ran 1.200v on it just recently no shutdowns no power issues nothing. It was drawing 500+ish watt.

Im aware that XOC removes the power limits, but you mentioned lucky strix owners like the cards ship uncapped.
l1CappYl1 is offline  
Sponsored Links
Advertisement
 
post #15524 of 15976 Old 12-17-2017, 09:09 PM
 
StenioMoreira's Avatar
 
Join Date: May 2014
Posts: 339
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 19
Quote:
Originally Posted by gta1989 View Post

I will give this a try next time I play with some bench marks. do you restart more then once and keep going or is it a one time deal?

Yeah, that why if it doesn't go back to way it was before 500watt torture, I will be a bit bummed out. Still the card is top 10 % or better, I own 3 and have tested 8 cards, this was the best of all, core wasn't what stuck out, more of the extremely high Vram clock that scales, 770 was the limit, and beyond it would not give any increase. Now its crashing at 770... let alone anything higher, thats the big evidence for it being hurt by my experiments, it never crashed any Vram overclock before. The performance use to stop at 770, but i could still go further. Anyhow I hope it magically goes back to being 11000 timespy graphics score card.

I've had my hand 3 that couldn't put up 400 on vram
Quote:
Originally Posted by l1CappYl1 View Post

Im aware that XOC removes the power limits, but you mentioned lucky strix owners like the cards ship uncapped.
No lol, i meant maybe their clocks are real with that Bios, considering it's a bios for their card, maybe it scales better. 2152mhz on 1.200v for me was less than 1% better performance than my original bios 2088mhz +770vram
StenioMoreira is offline  
post #15525 of 15976 Old 12-17-2017, 09:57 PM
PC Gamer
 
EarlZ's Avatar
 
Join Date: Mar 2011
Location: Cebu City , Philippines
Posts: 2,265
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 27
My card can do +850Mhz on memory but I dont have a reliable way to determine if I am getting more performance or not because I am hitting a power limit on time spy graphics test 2 and same with firestrike, games dont crash or artifact but maybe thats just the ECC on GDDR5X kicking in and slowing the actual clocks. Any suggestions ?

EarlZ is offline  
post #15526 of 15976 Old 12-17-2017, 11:39 PM
 
StenioMoreira's Avatar
 
Join Date: May 2014
Posts: 339
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 19
Quote:
Originally Posted by EarlZ View Post

My card can do +850Mhz on memory but I dont have a reliable way to determine if I am getting more performance or not because I am hitting a power limit on time spy graphics test 2 and same with firestrike, games dont crash or artifact but maybe thats just the ECC on GDDR5X kicking in and slowing the actual clocks. Any suggestions ?

of course there is, run 6 + runs of each memory state around same temps, has to be hot, to simulate real prolonged game and not a premature quick session, i've gathered results from valleybenchmark and timespy, and no it doesnt have to be 4k + 8k ect,.... lol there is always a performance difference, also sometimes the clocks dont really take effect unless u do software restart or even pc... Not absolutely sure but the first 2 runs were always unreliable. And yeah, for me personally, i see consistent difference with clocks doing the above ^^. All benchmarks have shown it, even at 1980p
StenioMoreira is offline  
post #15527 of 15976 Old 12-18-2017, 02:38 AM
PC Gamer
 
EarlZ's Avatar
 
Join Date: Mar 2011
Location: Cebu City , Philippines
Posts: 2,265
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 27
Quote:
Originally Posted by StenioMoreira View Post

Quote:
Originally Posted by EarlZ View Post

My card can do +850Mhz on memory but I dont have a reliable way to determine if I am getting more performance or not because I am hitting a power limit on time spy graphics test 2 and same with firestrike, games dont crash or artifact but maybe thats just the ECC on GDDR5X kicking in and slowing the actual clocks. Any suggestions ?

of course there is, run 6 + runs of each memory state around same temps, has to be hot, to simulate real prolonged game and not a premature quick session, i've gathered results from valleybenchmark and timespy, and no it doesnt have to be 4k + 8k ect,.... lol there is always a performance difference, also sometimes the clocks dont really take effect unless u do software restart or even pc... Not absolutely sure but the first 2 runs were always unreliable. And yeah, for me personally, i see consistent difference with clocks doing the above ^^. All benchmarks have shown it, even at 1980p

What Im trying to say is that I need a reliable way to establish if I am gaining rt losing performance from overclocking my mems from a +500 to +850, I intended to use Firestrike and Timespy but the problem with that is hitting the power limit even on 1080P, once I am able to tell until which mem oc gives me performance thats when I will test for stability.

EDIT:
Ive been testing superposition from 500-600-700-800Mhz memory overclock and Ive been getting a score increase little by little for every 100Mhz all the way up to 800Mhz so Im thinking my memory is not throttling down what so ever I just need to find a stable clock for it.

EarlZ is offline  
post #15528 of 15976 Old 12-18-2017, 07:16 AM
 
StenioMoreira's Avatar
 
Join Date: May 2014
Posts: 339
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 19
Quote:
Originally Posted by EarlZ View Post

Ive been testing superposition from 500-600-700-800Mhz memory overclock and Ive been getting a score increase little by little for every 100Mhz all the way up to 800Mhz so Im thinking my memory is not throttling down what so ever I just need to find a stable clock for it.

Yeah, hmm check out the gtx1080ti bios flash thread on this site, find the batfile that comes with the zip folder for "best Bioses", use that bat file to give you 50 watt head room, to help ur situation, and yea 3dmark is almost like a GPU power virus lol.

next, I'd test runs at different clocks like you have been, and recording not just score but minimums if possible. Discard any clock that does not show performance increase, keep lowering it until there's a drop off in performance that's more of an outlier compared to the rest, like if it stands out because I know pinpointing memory gains/losses is tough due to how it jumps around. 770 for me is great on first boot, top score always, but doesn't work if core is realistically hot, so I backed off to 750, where there is least variable in scores and fps low's also, 750 scores for me were closest within each run than the higher ones.

I discovered some good news guys regarding that XOC bios
StenioMoreira is offline  
post #15529 of 15976 Old 12-18-2017, 07:19 AM
Intel Overclocker
 
gta1989's Avatar
 
Join Date: Dec 2015
Location: iowa
Posts: 45
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 2
Quote:
Originally Posted by EarlZ View Post

What Im trying to say is that I need a reliable way to establish if I am gaining rt losing performance from overclocking my mems from a +500 to +850, I intended to use Firestrike and Timespy but the problem with that is hitting the power limit even on 1080P, once I am able to tell until which mem oc gives me performance thats when I will test for stability.

EDIT:
Ive been testing superposition from 500-600-700-800Mhz memory overclock and Ive been getting a score increase little by little for every 100Mhz all the way up to 800Mhz so Im thinking my memory is not throttling down what so ever I just need to find a stable clock for it.


i have found that synthetic benchmarks do show more improvements with memory above 400mhz oc then games or heaven. i found firestrike i could get a boost if i went to +600. but in heavon or games i would actually loose fps and at 650+ in games it was a drastic decrease in fps. maybe that point depends on the card but thats just what i have noticed with my 2 cards. sli doesnt seem to make a difference that +400 mark is it then im over the hill
gta1989 is offline  
post #15530 of 15976 Old 12-18-2017, 03:45 PM
PC Gamer
 
EarlZ's Avatar
 
Join Date: Mar 2011
Location: Cebu City , Philippines
Posts: 2,265
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 27
Quote:
Originally Posted by StenioMoreira View Post

Quote:
Originally Posted by EarlZ View Post

Ive been testing superposition from 500-600-700-800Mhz memory overclock and Ive been getting a score increase little by little for every 100Mhz all the way up to 800Mhz so Im thinking my memory is not throttling down what so ever I just need to find a stable clock for it.

Yeah, hmm check out the gtx1080ti bios flash thread on this site, find the batfile that comes with the zip folder for "best Bioses", use that bat file to give you 50 watt head room, to help ur situation, and yea 3dmark is almost like a GPU power virus lol.

next, I'd test runs at different clocks like you have been, and recording not just score but minimums if possible. Discard any clock that does not show performance increase, keep lowering it until there's a drop off in performance that's more of an outlier compared to the rest, like if it stands out because I know pinpointing memory gains/losses is tough due to how it jumps around. 770 for me is great on first boot, top score always, but doesn't work if core is realistically hot, so I backed off to 750, where there is least variable in scores and fps low's also, 750 scores for me were closest within each run than the higher ones.

I discovered some good news guys regarding that XOC bios

It does not have the MSI Gaming X as part of its list. I am able to run 800Mhz on COD WW2 about 2hrs that I've played all max settings render scale is at 400% which uses about 10.6GB VRAM. No artifacts and stable during that 2hr time frame

Quote:
Originally Posted by gta1989 View Post

Quote:
Originally Posted by EarlZ View Post

What Im trying to say is that I need a reliable way to establish if I am gaining rt losing performance from overclocking my mems from a +500 to +850, I intended to use Firestrike and Timespy but the problem with that is hitting the power limit even on 1080P, once I am able to tell until which mem oc gives me performance thats when I will test for stability.

EDIT:
Ive been testing superposition from 500-600-700-800Mhz memory overclock and Ive been getting a score increase little by little for every 100Mhz all the way up to 800Mhz so Im thinking my memory is not throttling down what so ever I just need to find a stable clock for it.


i have found that synthetic benchmarks do show more improvements with memory above 400mhz oc then games or heaven. i found firestrike i could get a boost if i went to +600. but in heavon or games i would actually loose fps and at 650+ in games it was a drastic decrease in fps. maybe that point depends on the card but thats just what i have noticed with my 2 cards. sli doesnt seem to make a difference that +400 mark is it then im over the hill

Thats what I am trying to do know, is to know if my +800Mhz is getting me any real performance increase in games. Im currently playing COD WW2 and I'll try to look for a scene with a steady FPS and try a 100Mhz increment see if there is a performance drop/gain

EarlZ is offline  
Reply

Quick Reply

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off