This got a little out of hand... but it worked!
# Extreme Overclocking: A Journey Through Liquid Nitrogen and Precision
## Introduction to Extreme Overclocking
In the world of extreme overclocking, pushing hardware to its limits is both an art and a science. This video delves into the intricate process of overclocking a GPU using liquid nitrogen (LN2), exploring the challenges, techniques, and triumphs encountered along the way. The journey is filled with trial and error, meticulous data tracking, and a deep understanding of thermal dynamics.
---
## Corsair's 4000D AF: A Breath of Fresh Air for Cooling
The video begins by introducing Corsair's 4000D AF case, which offers optimized front panel airflow and maximum cooling capabilities. The case not only keeps the system tidy with its rapid route cable management but also ensures that components stay extremely cool. Equipped with Corsair 120 air guide fans, this case allows for fine-tuned airflow, directing it precisely where it's needed most.
For those interested in exploring the full list of features and configurations for the Corsair 4000D AF, clicking on the link in the video description provides further details. This case proves to be an essential tool in the overclocker’s arsenal, ensuring that even the most demanding components stay within manageable temperature ranges.
---
## The Overclocking Process: Precision and Patience
The narrator shares insights into their overclocking journey, emphasizing that all their success has been stumbled upon through trial and error. Initially hesitant to embrace LN2 cooling, they found themselves drawn into a world of precision and patience. The process involves meticulous tracking of variables such as memory speed, core speed, voltages, load line settings, target temperatures, and the results of each test.
A spreadsheet is used to log all these parameters, allowing for a clear understanding of how changes in one variable affect performance. This data-driven approach is crucial, as it enables the overclocker to isolate issues and refine their techniques systematically.
---
## Mounting Challenges: The Importance of Pressure
One of the key challenges encountered during the process is the mounting pressure on the GPU. The narrator demonstrates a loose mount, highlighting how uneven pressure can lead to hot spots and poor thermal transfer. They explain that their initial c-clamp method provided excellent results, but even this setup has its limitations.
The discussion turns to the importance of consistent mounting pressure and the potential consequences of stripped screws. Despite these challenges, the narrator remains resourceful, using extra screws provided by Vince (their mentor) to address the issue. The focus is on ensuring that the GPU is securely mounted to maximize thermal efficiency and stability.
---
## Memory Management: Balancing Speed and Voltage
The importance of memory settings in overclocking is emphasized, with the narrator experimenting with different speeds to find the optimal balance between core and memory performance. They reveal that achieving 1800 MHz memory speed was possible but acknowledge that further refinement is needed to maximize overall system stability.
The process involves adjusting voltage and load line settings carefully, as even minor changes can lead to significant improvements or crashes. The narrator’s approach is methodical: they focus on maximizing core performance before tweaking memory settings, ensuring that only one variable is changed at a time to isolate its effects.
---
## Thermal Challenges: Maintaining LN2 Pot Temperature
Thermal stability is another critical factor in successful overclocking. The narrator explains the importance of maintaining a consistently cold pot temperature using LN2. They demonstrate how uneven cooling can lead to rapid temperature swings, causing instability and crashes during tests.
The video highlights the patience required to bring the pot to an optimal temperature, which can take up to 15 minutes. The narrator emphasizes that until the pot is sufficiently cold, any test run is likely to fail due to insufficient cooling capacity.
---
## Progress and Setbacks: Learning from Each Crash
Throughout the video, the narrator shares their experiences of crashing tests at various stages, often due to temperature-related issues. They explain how each crash provides valuable insights into system behavior, allowing for adjustments that bring them closer to their goal.
The journey is filled with both frustration and satisfaction as they navigate through countless test runs, tweaking settings, and learning from each failure. The narrator’s determination shines through as they push their system to its limits, driven by the desire to achieve higher clock speeds.
---
## Re-pasting and Adjustments: Fine-Tuning for Success
After several unsuccessful attempts, the narrator decides to re-paste the GPU to improve thermal transfer. They demonstrate the process of removing the backplate, drying it off, and applying new paste with precision. The focus is on ensuring an even application that maximizes contact between the GPU and the cooling pot.
The discussion turns to the importance of mounting pressure once again, as the narrator tightens the screws carefully to avoid stripping them. They explain how even minor adjustments can have a significant impact on thermal performance, highlighting the need for attention to detail in every step of the process.
---
## System Fatigue: Pushing Beyond Limits
The video concludes with a discussion of system fatigue, as the GPU begins to show signs of wear after extensive testing. The narrator acknowledges that while they achieved impressive clock speeds, there are inherent limitations in their setup that prevent further progress.
Despite these challenges, the narrator expresses pride in their achievements and gratitude for the support received from their mentor, Vince. They encourage viewers to sound off in the comments if they would like to see more videos in this series or suggest alternative approaches to extreme overclocking.
---
## Final Results: A New High Score
The video ends with a triumphant moment as the narrator achieves a new high score of 2610 MHz, surpassing their previous best. They reflect on the arbitrary yet meaningful milestone of cracking 2600 MHz and acknowledge that while there are still limitations in their system, they have come closer to achieving their goals.
The final words of encouragement remind viewers that overclocking is as much about the journey as it is about the destination. Whether through trial and error or meticulous planning, every overclocker’s story is unique and值得 celebrating.
---
## Conclusion
This video provides an intimate look into the world of extreme overclocking, showcasing the patience, precision, and problem-solving skills required to push hardware beyond its factory limits. The narrator’s journey serves as inspiration for anyone willing to embrace the challenges of extreme overclocking, proving that success is often just a matter of time, effort, and the willingness to learn from setbacks.
If you’re interested in exploring the world of extreme overclocking or simply curious about the inner workings of high-performance computing, this video offers valuable insights and a glimpse into the passion that drives overclockers to push their systems to the brink.
"WEBVTTKind: captionsLanguage: enall right welcome to what might be the actual last l two video that we do i don't know you know i hate i don't like giving up never give up never surrender corsair's 4000 d af delivers optimized front panel air flow and maximum cooling for your pc and with its built-in rapid route cable management keeps your system looking tidy while also staying extremely cool the equipped corsair 120 air guide fans provide directed airflow allowing you to fine tune the components you want to keep cool to see the full list of features and configurations for the corsair 4000d and 4000d af click the link in the description below all right so there's two different very different approaches now i'm taking to this one all of the success success that you've seen me have so far with ln2 has truly been stumbled upon a prawn i didn't started breathing the nitrogen yet has been stumbled upon on accident quite honestly through trial and error but the thing is i can't make any headway without data so one of the major things we started doing here is we're keeping a spreadsheet and the data on here is basically what mem what's the memory speed what's the core speed what voltages what load line settings what was my target temperature what was i running at during the test uh how far did the test complete and what was the result so i'm keeping track of like these settings gave me this result and then i'm changing one thing at a time which i've preached a million times but i'm keeping track of it which is something i need to do all along because vince bless his heart kingpin he's uh he's been very very he's pushing for me he's like come on man you can do this i know you can you're not the idiot you say you are he doesn't know me like you guys do but um he was like you need to find a true bass line through trial and error which is what we're doing because now that i've been able to get this card as of yesterday up to 250 megahertz at 1.35 volts which is actually insane because check this out look how far i was able to get on stock voltage well technically i just did 1.25 volts but 20 400 and 90 megahertz on 1.25 volts that means this card is a good card if we can reach just under 2500 megahertz on 1.25 volts that says a lot now we were able to get 2550 to pass at 1.35 volts average temp kept at -133 so that's where we're going to start today now one of the things that i realized is that my mounting pressure was really starting to suffer now check this out this is the second card here do you see how whoa it's loose look at how wobbly that is look at that mounting pressure look how poor that is it's like half the core wasn't even getting touched that means it was all crooked and i'm going wait a minute i mounted this using my c-clamp method i've been doing which got me some really good mounts i mean heck even phil saw i was able to run these things super cold without cracking the paste but do you see how uneven that mount is look at the look at the pot that is why i was crashing very very bad mount the areas that are not getting touched under proper pressure are going to be hot spots on the dye and the area like that's basically squished is going to be the cold spot so you get these really uneven temperatures so i was really starting to become concerned that these had stripped well see this one i think might actually be stripped now that's fine i think as long as it can pull all the way like tight but it tells me it's not fine based on that mounting pressure it's the screw that's stripped and that's actually good news because he sent me extras it's almost like maybe he knew this right so that's fine i'll deal with this later because one of the things i want to do is when this card becomes fatigued you can only run it so much in fact yesterday we ran 18 times that we logged it so that's way that's way more than you really want to do in a single setting a single sitting i mean you really want to just kind of try and do you want to nail it on the first two tries according to vince and then you're kind of done at that point because one of the things that happened is the board freezes over i had even mentioned on there i don't know if nick made a note i'll have to check and see on here i even made it like i mentioned i'm like i'm getting flickering the screen is flickering and stuff like that seems very indicative of memory and i do recall vince saying the memory can only handle the cold for so long then you start getting weird errors um frozen memory remember i've said it before on the core it gets too cold that's just as bad as too hot the same goes for the memory so we're going to continue today nick is going to be filling my ln2 for me also keeping notes of the things i'm telling him we're going to start with settings that are on this blue line so 1600 on the memory which is low i've actually gotten 1800 memory to run just fine um so i know i'm not going to get my best score unless i get my fast memory going too but i want to try and maximize my core before i play with memory because if i'm changing more than one thing at a time i don't know what caused the crash which is why you do this with your hand these are graphs going up and down but without boring you guys any more than i already have i need to bring the pot temperature down it takes a while for the pot to be able to actually keep up with the card it's kind of funny when the pot is not fully frozen all the way through with ln2 like the inner innards the guts of the metal hey i'm trying to make a video over here like the guts of the metal and down here it's going to be colder than up here and i need the entire pot to be cold all the way through for it to actually have enough cooling capacity the bottom half is copper the top half is aluminum but i've noticed if it's not cold all the way through when i start the test even if i have a full pot of ln2 and i can't do anything else it's full the temperature rises and it will start at like 120 and hit all the way to like 180 and then crash because it's too warm so until i can get the pot temperature to actually hold and not warm up with the gpu the tests aren't aren't ready yet so rather than boring you guys through all that process of bringing the pot temperature down which can take 10 15 minutes i'm just going to do it and then we'll bring you guys along for the first run so when you start getting that that real like it sounds like bacon cooking sound then you know you got the pot nice and cold all right now we're going to go ahead and do our first run this is 2550 1.3 volts i brought the voltage back up a little bit we just have to hope the pod is cold enough though to keep it from sling shotting too warm during the run i believe it is and it and this is the balancing act is i have to pre-load the pod a little bit too right at the start of the run because i don't want to get too warm but then if it crashes at the start then it gets too cold and it pops to grease oh we crashed we crashed it like five percent in or so so one of the things i do with the torch after a crash like that is i use it to keep it from getting too warm i'm sorry too cold yeah i want to add some heat to it because the gpu is not adding a load i honestly believe a lot of this crashing right now is is temperature related i'm going to leave the settings where they are and try again i'm just having a really hard time keeping the temperature as cold as i want so that's why i don't believe it to be a voltage or a mount thing right now that just shows how much incredible heat these cards generate when it can be at minus 140 and then within 15 seconds in the run it's at minus 115 that what 30 c 25c whatever it is like swing could definitely be enough to crash it that one was an instant crash same settings instant crash you see how we do the same thing you get different results that i mean that's the definition of insanity now i guess i go like one four now i'm just see it's funny because it's same thing with a cpu like you can find like oh i can get really far on stock voltage and then as soon as you crash you find you need like a ridiculous amount of extra voltage to get to the next like level and it's frustrating to see like how far something can go on stock voltage and then just suddenly be like nope hard stop if it crashes again instantly at the start i'll do a restart of the system yeah another question okay another instant crash i'm going to be very hard on the system right now i'm just going to shut it down yeah it's it's funny and sometimes when you get crashes like that you have to restart it like now we're stuck on a memory code how much of my problems is actually this stupid platform x 299 is i mean it's been such a pain in the ass over over the years so that completed that was only 25 20 though but that's promising 17148 it's still quite a waste away from our best score but we're doing this still with a lot less voltage now we'll go back to the 2550 and just run it again with all the same settings so now we've reached this point where it's crashing at the start of the test no matter what and just by bumping up 25 megahertz so now i gotta do the heal process warm it up to about minus one or minus 50 minus 60 which gets the pace to kind of reflow and get soft again then uh just repeat see now that i'm documenting this like i'm like i should be now i can at least see what happened you know i could paint this picture right at the end okay what i'm gonna do i'm gonna go ahead and warm this card up tear it down we'll take a look at the mount dry it out re-paste it go again so now that i've got the back covering off and i've got it somewhat dried off there's like water seeping out of it everywhere i know it's part of my problem here um i want to loosen these and see if the screws became loose and then we're going to remount it let's see how the mount did i'd say it's pretty centered it sounds a good mount for sure that's really cold in there like if i was to take a thermocouple and see what it's still at right now look it's not 3c well vince did say it needs more pressure on the mounting plate so i fixed it okay if we push hard enough it'll probably stab through yeah just like that it's a handle for carrying all right we're back up and running we've got the pot temp on its way down with how matte wide the mounting point is that when you tighten it down it can actually bow the board outward so it pulls the the dial away from the pot some and it may be like microns of distance but that truly does matter when you're doing the extreme temperatures and stuff like this so by having the clamp squeeze it back we might actually get somewhere here um not something i'd ever actually considered until vince told me which is funny because i use it to mount the pot and we saw improvements in getting a nice level mount with that on there but now having the four screws and the direct tension in the middle uh we might actually be able to see some sort of progress here i do like the fact that i got 1900 megahertz to run before we stopped so if i can get 1600 with a high core clock and then i get the core clock to continue to go up i'm curious if we're going to see any sort of improvement so that's why we're kind of starting off right where we left off prior to even starting today like the blue line i showed you which was our last like successful highest uh numbers so with this run i'm doing 2550 1600 on the memory just to see if i can get it to pass and if it does pass with this then we'll push it we're at minus 115 i got to get this down to like minus 130 before i can truly get it to start otherwise like i've showed before it will get far too warm at the start of the test and it will crash there'll be no point so i've got to give the pot time to to get as cold as it can get and i'm repeating myself but that's the thing is this is truly repetitive oh that crack right when it froze that's a bad sign so i need to do a warm-up now i feel like that crack was the pace because now it's just crashing instantly the fact that it crashed right when the pace cracked and there's something i'm realizing now if i've got a really good bond like the pressure that's being pushed on that die is is better than before that might mean now it i don't need to run it as cold because let's say if we're having bad thermal transfer and i've got to run it at like minus 130 to keep the diet like minus 100 because of whatever the difference is if now we have that real nice tight fit i could probably go a little warmer with it uh and maintain a stability because of the fact that we got good thermal transfer so if the pots a little bit warmer but the gpu is closer to the temp of the pot than the delta we had before that'd be a really good sign and with how fast it warms up right now while i'm doing my warm-up look and this is stock this is roughly stock speed because one of the things we do is we remove the clock offset while we're doing this so we're reflowing we're healing the pace essentially is what this is called which means now i don't i think my new target needs to be like minus 120. so that was a 17 357 it's actually the second run back to back we got to complete current settings on this are 1800 on the memory 475 or 2565 on the core um we were having some artifacts or like some screen flickering though which tells me i believe um the memory was causing that and i i think the memory can go higher than what we're currently doing what we sort of found here too watch all the ice come out like those are that's essentially the same speeds i had gotten that 17501 with so there's still some inefficiency somewhere not entirely sure what it is although i do believe the 17501 was with 1900 megahertz memory so without but but the flickering can also be causing frame drops which will dramatically affect the score so what i'm going to do right now is i'm going to push the memory voltage back to where it was i was trying to reduce it some but we're at 1.45 volts load line five which is uh got us some stability here we're finally on to something here ah i knocked over a whole thermos of ln2 on my arm and you can see my arm is still attached didn't break off there's no burns this is looking really promising this run what what just because i said it if this will complete the last few percent which it loves to just troll me with this might be my best score based on the frames drawn i just saw is that exactly your score 17 501 is the one i have online and then and then this and then the score i have saved that i never put up is a 507. i landed exactly in between them well i guess 104 had been closer but hey why don't we just go for more right it's working right now ah fingering fast fan whoa it's a new high score i mean i'll take a couple of points it means i'm go getting somewhere you may save i want to get my name on the board with 2600 or above right now it's at 25.95 or 25.65 i can't remember look progress 26 ten one seven five seven four i'm gonna do this over and over and over okay unless something goes then i'll show you and i'll come back and be like oh man when it crashes on that clip so that was 25 or no 26 25 i'm so used to saying 25. so that was 20 we're gonna call it here um how many runs have i done since this replaced 24 runs since that replaced and that's because i keep blowing out the water and the ice and stuff trying to get as much time as i can out of it we'll take a look and see if any ice built up that i couldn't see i think i've done a pretty good job keeping it under control but here's the thing in the middle of that run i don't know if nick or phil caught it but i went oh like at one point i looked at the thing i'm like my face get all excited because there's a particular part of the test i like to look at and see what the fps is and that's when it's pointing down at the ship as it's going into the dock or the bay or whatever it is and at the start of all of this i was getting like 125 120 and then it will drop down to like 123 and earlier today i went man if i could get that number up to like 1 30 and i saw at 1 30. that's when i was like okay we're on to getting some real results here now and we crashed all right so that's what we're going to call it the card's done it's frozen it can't take anymore um we'll go ahead and upload this score look this is where i need you guys to tell me do you want to keep seeing these videos albeit less frequent um i'm not a full-time xoc so or extreme overclocker so it's um really difficult to take the amount of time it does like we lose an entire day we do an entire day on this which is funny because if i did it the right way i would do two or three runs and be done and trust me right now my brain goes you're making progress now's not the time to stop i want to keep going so anyway if you guys want to know if i kept going and got a better score then all you need to do is go to poor the port royal uh whatchamacallits the hall of fame and then see if my score is higher uh 17 574. what makes me happy about this run though is i wanted to crack 2600. that was this arbitrary number in my head that i was like if i could beat that then i'm i'm validated as an overclocker regardless of where i land on the leaderboard but as you can see right there 2610 megahertz i think there's other things in my system that are holding back this overall efficiency here uh or this this number i feel like system related wise i don't think i'm on the proper platform if you wanna know the truth so anyway thanks for watching this video sound off below if you guys want me to continue doing this maybe once or twice a month at the most uh take you guys along for the ride otherwise go to 3dmark hall of fame go to port royal sort by x1 or x2 and you'll see our scores and you'll see if i beat that or not thanks for watching guys as always see in the next oneall right welcome to what might be the actual last l two video that we do i don't know you know i hate i don't like giving up never give up never surrender corsair's 4000 d af delivers optimized front panel air flow and maximum cooling for your pc and with its built-in rapid route cable management keeps your system looking tidy while also staying extremely cool the equipped corsair 120 air guide fans provide directed airflow allowing you to fine tune the components you want to keep cool to see the full list of features and configurations for the corsair 4000d and 4000d af click the link in the description below all right so there's two different very different approaches now i'm taking to this one all of the success success that you've seen me have so far with ln2 has truly been stumbled upon a prawn i didn't started breathing the nitrogen yet has been stumbled upon on accident quite honestly through trial and error but the thing is i can't make any headway without data so one of the major things we started doing here is we're keeping a spreadsheet and the data on here is basically what mem what's the memory speed what's the core speed what voltages what load line settings what was my target temperature what was i running at during the test uh how far did the test complete and what was the result so i'm keeping track of like these settings gave me this result and then i'm changing one thing at a time which i've preached a million times but i'm keeping track of it which is something i need to do all along because vince bless his heart kingpin he's uh he's been very very he's pushing for me he's like come on man you can do this i know you can you're not the idiot you say you are he doesn't know me like you guys do but um he was like you need to find a true bass line through trial and error which is what we're doing because now that i've been able to get this card as of yesterday up to 250 megahertz at 1.35 volts which is actually insane because check this out look how far i was able to get on stock voltage well technically i just did 1.25 volts but 20 400 and 90 megahertz on 1.25 volts that means this card is a good card if we can reach just under 2500 megahertz on 1.25 volts that says a lot now we were able to get 2550 to pass at 1.35 volts average temp kept at -133 so that's where we're going to start today now one of the things that i realized is that my mounting pressure was really starting to suffer now check this out this is the second card here do you see how whoa it's loose look at how wobbly that is look at that mounting pressure look how poor that is it's like half the core wasn't even getting touched that means it was all crooked and i'm going wait a minute i mounted this using my c-clamp method i've been doing which got me some really good mounts i mean heck even phil saw i was able to run these things super cold without cracking the paste but do you see how uneven that mount is look at the look at the pot that is why i was crashing very very bad mount the areas that are not getting touched under proper pressure are going to be hot spots on the dye and the area like that's basically squished is going to be the cold spot so you get these really uneven temperatures so i was really starting to become concerned that these had stripped well see this one i think might actually be stripped now that's fine i think as long as it can pull all the way like tight but it tells me it's not fine based on that mounting pressure it's the screw that's stripped and that's actually good news because he sent me extras it's almost like maybe he knew this right so that's fine i'll deal with this later because one of the things i want to do is when this card becomes fatigued you can only run it so much in fact yesterday we ran 18 times that we logged it so that's way that's way more than you really want to do in a single setting a single sitting i mean you really want to just kind of try and do you want to nail it on the first two tries according to vince and then you're kind of done at that point because one of the things that happened is the board freezes over i had even mentioned on there i don't know if nick made a note i'll have to check and see on here i even made it like i mentioned i'm like i'm getting flickering the screen is flickering and stuff like that seems very indicative of memory and i do recall vince saying the memory can only handle the cold for so long then you start getting weird errors um frozen memory remember i've said it before on the core it gets too cold that's just as bad as too hot the same goes for the memory so we're going to continue today nick is going to be filling my ln2 for me also keeping notes of the things i'm telling him we're going to start with settings that are on this blue line so 1600 on the memory which is low i've actually gotten 1800 memory to run just fine um so i know i'm not going to get my best score unless i get my fast memory going too but i want to try and maximize my core before i play with memory because if i'm changing more than one thing at a time i don't know what caused the crash which is why you do this with your hand these are graphs going up and down but without boring you guys any more than i already have i need to bring the pot temperature down it takes a while for the pot to be able to actually keep up with the card it's kind of funny when the pot is not fully frozen all the way through with ln2 like the inner innards the guts of the metal hey i'm trying to make a video over here like the guts of the metal and down here it's going to be colder than up here and i need the entire pot to be cold all the way through for it to actually have enough cooling capacity the bottom half is copper the top half is aluminum but i've noticed if it's not cold all the way through when i start the test even if i have a full pot of ln2 and i can't do anything else it's full the temperature rises and it will start at like 120 and hit all the way to like 180 and then crash because it's too warm so until i can get the pot temperature to actually hold and not warm up with the gpu the tests aren't aren't ready yet so rather than boring you guys through all that process of bringing the pot temperature down which can take 10 15 minutes i'm just going to do it and then we'll bring you guys along for the first run so when you start getting that that real like it sounds like bacon cooking sound then you know you got the pot nice and cold all right now we're going to go ahead and do our first run this is 2550 1.3 volts i brought the voltage back up a little bit we just have to hope the pod is cold enough though to keep it from sling shotting too warm during the run i believe it is and it and this is the balancing act is i have to pre-load the pod a little bit too right at the start of the run because i don't want to get too warm but then if it crashes at the start then it gets too cold and it pops to grease oh we crashed we crashed it like five percent in or so so one of the things i do with the torch after a crash like that is i use it to keep it from getting too warm i'm sorry too cold yeah i want to add some heat to it because the gpu is not adding a load i honestly believe a lot of this crashing right now is is temperature related i'm going to leave the settings where they are and try again i'm just having a really hard time keeping the temperature as cold as i want so that's why i don't believe it to be a voltage or a mount thing right now that just shows how much incredible heat these cards generate when it can be at minus 140 and then within 15 seconds in the run it's at minus 115 that what 30 c 25c whatever it is like swing could definitely be enough to crash it that one was an instant crash same settings instant crash you see how we do the same thing you get different results that i mean that's the definition of insanity now i guess i go like one four now i'm just see it's funny because it's same thing with a cpu like you can find like oh i can get really far on stock voltage and then as soon as you crash you find you need like a ridiculous amount of extra voltage to get to the next like level and it's frustrating to see like how far something can go on stock voltage and then just suddenly be like nope hard stop if it crashes again instantly at the start i'll do a restart of the system yeah another question okay another instant crash i'm going to be very hard on the system right now i'm just going to shut it down yeah it's it's funny and sometimes when you get crashes like that you have to restart it like now we're stuck on a memory code how much of my problems is actually this stupid platform x 299 is i mean it's been such a pain in the ass over over the years so that completed that was only 25 20 though but that's promising 17148 it's still quite a waste away from our best score but we're doing this still with a lot less voltage now we'll go back to the 2550 and just run it again with all the same settings so now we've reached this point where it's crashing at the start of the test no matter what and just by bumping up 25 megahertz so now i gotta do the heal process warm it up to about minus one or minus 50 minus 60 which gets the pace to kind of reflow and get soft again then uh just repeat see now that i'm documenting this like i'm like i should be now i can at least see what happened you know i could paint this picture right at the end okay what i'm gonna do i'm gonna go ahead and warm this card up tear it down we'll take a look at the mount dry it out re-paste it go again so now that i've got the back covering off and i've got it somewhat dried off there's like water seeping out of it everywhere i know it's part of my problem here um i want to loosen these and see if the screws became loose and then we're going to remount it let's see how the mount did i'd say it's pretty centered it sounds a good mount for sure that's really cold in there like if i was to take a thermocouple and see what it's still at right now look it's not 3c well vince did say it needs more pressure on the mounting plate so i fixed it okay if we push hard enough it'll probably stab through yeah just like that it's a handle for carrying all right we're back up and running we've got the pot temp on its way down with how matte wide the mounting point is that when you tighten it down it can actually bow the board outward so it pulls the the dial away from the pot some and it may be like microns of distance but that truly does matter when you're doing the extreme temperatures and stuff like this so by having the clamp squeeze it back we might actually get somewhere here um not something i'd ever actually considered until vince told me which is funny because i use it to mount the pot and we saw improvements in getting a nice level mount with that on there but now having the four screws and the direct tension in the middle uh we might actually be able to see some sort of progress here i do like the fact that i got 1900 megahertz to run before we stopped so if i can get 1600 with a high core clock and then i get the core clock to continue to go up i'm curious if we're going to see any sort of improvement so that's why we're kind of starting off right where we left off prior to even starting today like the blue line i showed you which was our last like successful highest uh numbers so with this run i'm doing 2550 1600 on the memory just to see if i can get it to pass and if it does pass with this then we'll push it we're at minus 115 i got to get this down to like minus 130 before i can truly get it to start otherwise like i've showed before it will get far too warm at the start of the test and it will crash there'll be no point so i've got to give the pot time to to get as cold as it can get and i'm repeating myself but that's the thing is this is truly repetitive oh that crack right when it froze that's a bad sign so i need to do a warm-up now i feel like that crack was the pace because now it's just crashing instantly the fact that it crashed right when the pace cracked and there's something i'm realizing now if i've got a really good bond like the pressure that's being pushed on that die is is better than before that might mean now it i don't need to run it as cold because let's say if we're having bad thermal transfer and i've got to run it at like minus 130 to keep the diet like minus 100 because of whatever the difference is if now we have that real nice tight fit i could probably go a little warmer with it uh and maintain a stability because of the fact that we got good thermal transfer so if the pots a little bit warmer but the gpu is closer to the temp of the pot than the delta we had before that'd be a really good sign and with how fast it warms up right now while i'm doing my warm-up look and this is stock this is roughly stock speed because one of the things we do is we remove the clock offset while we're doing this so we're reflowing we're healing the pace essentially is what this is called which means now i don't i think my new target needs to be like minus 120. so that was a 17 357 it's actually the second run back to back we got to complete current settings on this are 1800 on the memory 475 or 2565 on the core um we were having some artifacts or like some screen flickering though which tells me i believe um the memory was causing that and i i think the memory can go higher than what we're currently doing what we sort of found here too watch all the ice come out like those are that's essentially the same speeds i had gotten that 17501 with so there's still some inefficiency somewhere not entirely sure what it is although i do believe the 17501 was with 1900 megahertz memory so without but but the flickering can also be causing frame drops which will dramatically affect the score so what i'm going to do right now is i'm going to push the memory voltage back to where it was i was trying to reduce it some but we're at 1.45 volts load line five which is uh got us some stability here we're finally on to something here ah i knocked over a whole thermos of ln2 on my arm and you can see my arm is still attached didn't break off there's no burns this is looking really promising this run what what just because i said it if this will complete the last few percent which it loves to just troll me with this might be my best score based on the frames drawn i just saw is that exactly your score 17 501 is the one i have online and then and then this and then the score i have saved that i never put up is a 507. i landed exactly in between them well i guess 104 had been closer but hey why don't we just go for more right it's working right now ah fingering fast fan whoa it's a new high score i mean i'll take a couple of points it means i'm go getting somewhere you may save i want to get my name on the board with 2600 or above right now it's at 25.95 or 25.65 i can't remember look progress 26 ten one seven five seven four i'm gonna do this over and over and over okay unless something goes then i'll show you and i'll come back and be like oh man when it crashes on that clip so that was 25 or no 26 25 i'm so used to saying 25. so that was 20 we're gonna call it here um how many runs have i done since this replaced 24 runs since that replaced and that's because i keep blowing out the water and the ice and stuff trying to get as much time as i can out of it we'll take a look and see if any ice built up that i couldn't see i think i've done a pretty good job keeping it under control but here's the thing in the middle of that run i don't know if nick or phil caught it but i went oh like at one point i looked at the thing i'm like my face get all excited because there's a particular part of the test i like to look at and see what the fps is and that's when it's pointing down at the ship as it's going into the dock or the bay or whatever it is and at the start of all of this i was getting like 125 120 and then it will drop down to like 123 and earlier today i went man if i could get that number up to like 1 30 and i saw at 1 30. that's when i was like okay we're on to getting some real results here now and we crashed all right so that's what we're going to call it the card's done it's frozen it can't take anymore um we'll go ahead and upload this score look this is where i need you guys to tell me do you want to keep seeing these videos albeit less frequent um i'm not a full-time xoc so or extreme overclocker so it's um really difficult to take the amount of time it does like we lose an entire day we do an entire day on this which is funny because if i did it the right way i would do two or three runs and be done and trust me right now my brain goes you're making progress now's not the time to stop i want to keep going so anyway if you guys want to know if i kept going and got a better score then all you need to do is go to poor the port royal uh whatchamacallits the hall of fame and then see if my score is higher uh 17 574. what makes me happy about this run though is i wanted to crack 2600. that was this arbitrary number in my head that i was like if i could beat that then i'm i'm validated as an overclocker regardless of where i land on the leaderboard but as you can see right there 2610 megahertz i think there's other things in my system that are holding back this overall efficiency here uh or this this number i feel like system related wise i don't think i'm on the proper platform if you wanna know the truth so anyway thanks for watching this video sound off below if you guys want me to continue doing this maybe once or twice a month at the most uh take you guys along for the ride otherwise go to 3dmark hall of fame go to port royal sort by x1 or x2 and you'll see our scores and you'll see if i beat that or not thanks for watching guys as always see in the next one\n"