Hi people of ocuk. You may be aware that 8pack is one of the top extreme overclockers in the world and does bench on LN2 and does sometimes try to put forward that there are people out there who buy the hardware for purely overclocking and performance as well as day to day tasks and gaming.
Well i hope this thread gives you a little insight to see into this amazing world of extreme overclocking and that the overclocking community in the Uk is coming back after a quiet period. we have actually had 3 meets this year now and with them growing in size a little i thought i would post here, because being a UK forum you may be interested in coming to join us or just see what goes on in future.
So this event was held in Liverpool from the 12-13th this past weekend and there were 5 people from my hwbot team and 1 from another team. We had 135L litres of LN2 for the weekend.
2 of the people had never had any experience with LN2 before, so we showed them how to insulate there boards and mount the pots to get them started, we also helped them get pretty decent scores and give them help and tips as they increased the speed and score of their systems.
I have no idea how to set this thread out so i'll just go with each person with what they brought and scores. i think i'm going to call this format Multiple 'cool' stories bro
Dragon Soop is a very new member to our team who has just started with mostly air and aio cooling foor the past few months. He came on the Saturday and like most people who have never done LN2 before he did not have a pot.
The fisrt day he benched a Q9550 on a h80 cooler and a core 2 duo e8500 for the hwbot team cup. i did try and help him a little between my system cold bugging and drying out, but he pretty much had maxxed out his socket 775 system on the aio cooler.
E8500 Pifast score
http://hwbot.org/submission/3625459_
Once i had finished with my pot i lent it to him as another member of my team had sent us his as he couldnt make the meet and thought we would need a spare. i showed him how to insulate his board which was an Asus Z170 Impact with Vas. The vas creates a water proof layer between the pcb and components on the motherboard which prevents ice and water shorting the board as it cools down.
We also delided his 7700K with thermal grizzly Kryonaut as he had liquid ultra or pro from aio cooling. Liquid metal does perform better at ambient temps, but will crack and contract at lower temps where as the kryonaut is design to handle very low temps without cracking.
We then applied thermal paste on top of the ihs and mounted the pot with the inner insulation only, due the the board being so small we couldnt get the out layer on. The insulation around the ln2 helps keep the pot cold as well as stopping the board and memory surrounding the pot from getting as cold and built up with ice and water.
We then checked the system was working before pouring ln2 in the pot and checked both the bios and booted into windows.
skylake and kabylake are known for cracking paste as they are very small dies that put out a very concentrated heat load. This means that if the paste is not applied evenly over the die and ihs it can get hot spots which can cause cracking where the paste contracts and the die and ihs become disconnected and your cpu gets fried
To stop this terrible situation from occurring on his first ln2 run ever we let the system run cpuz benchmark on 2 cores until the idle temps were around ~+50c. this makes the paste spread evenly as it is warm. once it gets cold it doesnt spread as well.
Once we had completed that we started to pull the cpu down. at -100c we enabled the ln2 mode jumper on the board, on the impact and most other ROG boards, this ln2 mode puts higher voltages through the cpu at boot up to combat a cold boot bug. This is when the cpu is frozen to such a degree when booting that it wont start. most kabylake chips dont tend to have a cold boot bug or cold bug when in ln2 mode on the asus 9 apex and asrock z170m ocf, but for some reason we found the system seemed to cold bug and cold boot bug around -150c even with the ln2 mode enabled.
I have never benched skylake on ln2 or kabylake before, so i didnt know the voltages to change that may have got this temp lower. we tried a few things in the bios, but in the end we stuck with -150c as it is still far colder than any water loop.
Now that we had figured out the coldbug he started to ramp up the clocks for the cpu and cache and memory as well as the voltages using a combination of the bios and asus turbo v software. once he had found the maximum clock speed of the cpu he started to run a few 2d benchamrks benchmarks
super pi 1m at 6.55ghz
http://hwbot.org/submission/3625687_
cinebench R15 at 6.06ghz
http://hwbot.org/submission/3625701_
Hwprime @5.9ghz
http://hwbot.org/submission/3625708_
After a few 2d benches he also decided to put his power modded 1080ti in and see what it could do in some 3d benchmarks
3dmark firestrike
http://hwbot.org/submission/3625753_
3dmark vantage
http://hwbot.org/submission/3625714_
He also had a couple of runs with a GTX970, but did not manage to get any scores/screenshots as the board had some memory issues.
Next up is Mickulty who is the team captain for the reddit overclocking team. He only came for a few hours to drop in a GTX580 Matrix, however he did bring a small amd apu based system on a mini itx board which of course needed to be made faster so he blocked up some of the stock heatsink with artist eraser after removing the fan and use the stock cooler as a small pot.
Another person who had never used LN2 before and this was their first meet was No1spank. He has been a member of hwbot for a few years and benches regularly on air cooling and he wanted to take his FX6300 which he had got to 5.2ghz previously, to the next level on LN2. He wanted some advice on how to prepare the board for ln2 and he decided to go with liquid tape and vas rather than vas alone. This is a more permanent solution than just vas alone as it tends to stick to the board better and once dry will not rub off.
As he was just benching 2d with the cpu on LN2 we recommend that he just paint the top half of the board and any areas where the brush couldnt get the LET he put some vas on instead.
he also did not have a pot, but luckily one of the other team members had brought 3 different pots that happens to just fit amd boards, so he borrowed one of those. It was a custom made pot with a copper pipe and a cap with solder making up the base. the pot was held down by a 3d printed bracket and it worked very well.
Now all FX chips are what we call full pot chips, which you can bench with the pot full of ln2 at -196c or a near as the pot and thermal paste allows. This means all you have to do is keep the pot topped up and you dont have to worry about cold bugs and cold boot bugs. This makes Am3 one of the best platforms for getting into extreme oc. Its also cheap to pick up now and the frequencies it is able to achieve are the highest out of any chips ever made.
He pulled the pot down fairly slowly and increased the frequency as he went. He did get a lot of help from Le and some from myself and almost bottled it at 6.7ghz ish, but we encouraged him to keep trying and pushing and before long he was flying past 7ghz and upto towards the top of the ranking for and FX6300
in the end he ran a massive 7.65ghz
http://hwbot.org/submission/3624587_
he also ran a few other benchmarks at 'lower' speeds
super pi [email protected]
http://hwbot.org/submission/3624684_
cinebench r15 at 6.3ghz
http://hwbot.org/submission/3624748_
wprime 32m at 6.35ghz
http://hwbot.org/submission/3625194_
Once he had finished benching on LN2 while he was packing up and waiting for his board and pot to dry out he had a liitle go with the Asus Rampage 9 Apex and 7900X we had set up under water with a few gpus he brought.
http://hwbot.org/submission/3625346_
http://hwbot.org/submission/3625395_
Lee is the man who made the custom pots, he brought 3 with him and he had even started extreme overclocking on phase after watching on of my youtube videos on a phase cooling setup. He had messaged me via emails a few times before moving onto LN2 and joining Hwbot. This was his first meet and he came very well prepared with all his gear and his board had some of the best insulation i have ever seen
he already had and FX8320 at over 7ghz under his belt and had brought an FX4300, phenom ii x2 and sempron 145 to the meet.
Lee started off with his FX4300 and he managed to get 1st place with a massive 7.75ghz oc
http://hwbot.org/submission/3625197_
He then moved onto his sempron 145. the sempron 145 doesnt have an unlocked multiplier like the FX4300, so it requires you to oc with the bclk only. This means you need a very good north bridge and ht link that work at very high frequencies. Normally a 790FX DDr2 based board would be best for this kind of oc but Lee still managed to get over 5ghz with the sempron 145 on his DDR3 board so that is still decent.
http://hwbot.org/submission/3624954_
His final cpu he ran the Phenom ii X2 550 and even though he didnt get 1st he manged to get over 6ghz with it putting him in the top 3 for maximum frequency on this cpu.
http://hwbot.org/submission/3627288_
I ad brought quite a few things to bench over the weekend, i was planning on running Ryzen 7 1800X, Ryzen 5 1600, Core 2 Duo E8600, Pentium E5200 and an i7 950. I had also brought 3 different socket 775 boards, Asus Prime X370 Pro and X58A OC to put these cpus in.
I started off with the Ryzen 7 1800X, my main goals with this chip was to do some 3d benchmarks with my R9 Fury and also run Wprime 1024M, Hwbot X265 4K and Geekbench3. These were all for the challenger division 4 oc esports competition
http://oc-esports.io/#!/round/roadtopro_challenger_season3_division4_round3
I started by mounting my pot and just wrapping some paper towel round the bottom of the pot. Then i got the fury in as well just on the air cooler. I pulled the cpu down to about -40c to start with and increased the voltage to 1.65v and set the cpu to 4.4ghz. I was going to run 3d mark vantage on the Fury as the first benchmark followed by 3dmark firestrike Extreme. Now i didnt know this at the time, but when you go below 0c on ryzen and threadripper, the pcie stops working properly so as you go under 3d load it crashes all the time. I spent quite a while trying to run 3d benches with the fury and apparently the only way to fix it is to set the pcie to 1.0.
After that failed i moved onto wprime 1024M I did manage to get the cpu unto 4.95ghz on 1.8v, but it would not run at 5ghz unfortunately. This is due to my only being able to get to about -130c on my Prime X370 due to it not having some of the voltages in the bios go high enough to run the cpu at full pot. At this speed i did mange to get a decent score
http://hwbot.org/submission/3624734_
it hasput my 1st in that stage of the competition however I did not get to complete any other stages as i was trying to help the others get going and bench at the same time, so the system got too much ice built up and stopped working and i didnt manage to get any other scores with that session.
Once my my pot had dried out and i had spent a while helping some of the others i decided that for the last session of the day i should Run my E8600 as this was top of my list to do on socket 775 as we needed some good pifast scores. I had never pifast before and it was a random cpu that i had pick from a tray as i have a few E8600 and haven't binned or tested any of them.
Now to start i booted at about 5ghz and then pulled to cpu down to in the os while running wprime 1024M to find out where the cpu cold bug was. once the benchmark freezes and i can no longer see the 1% of the benchmark that has completed then it has cold bugged. I also then can then find out what the cold boot bug is by restarting the system. while to cpu is too cold the post code reader would read 00 as if the cpu was dead and the machine will not post, it might flick on and off but it wont post. To find the minimum cold bug i get a blow torch and heat the pot back up until the system starts again. for this E8600 the cold bug was around -145 to -150c and cold boot bug was down at -110c. This means that if the system crashes or i change certain things in the bios while course a cold boot i need to heat the pot up with the blow torch to -110c to get going again. i dont think 30c difference is too bad but some cpus can need to be heated 60-70c higher temp to boot which can be annoying because you end up using loads of ln2 pulling it back down and you have to use the blow torch a lot. You can also get cpus on 775 that cold bug at -80c which is not get and they might not oc very high, because they cant go cold enough to use 1.9v.
So once i had found this basic info i needed to start i ramped up the clock speeds using Cinebench R15 to test the oc increase and also using setfsb to oc the fsb from inside windows. I got to about 5.6ghz before i realised that i had put the cpu in the wrong motherboard XD. i was running it on my Asus P5Q Deluxe which is great for quad core cpu's, but it cant go as high on the fsb as my P5Q-E which makes it better for the dual core chips.
to be continued............................
dun dun durn
Well i hope this thread gives you a little insight to see into this amazing world of extreme overclocking and that the overclocking community in the Uk is coming back after a quiet period. we have actually had 3 meets this year now and with them growing in size a little i thought i would post here, because being a UK forum you may be interested in coming to join us or just see what goes on in future.
So this event was held in Liverpool from the 12-13th this past weekend and there were 5 people from my hwbot team and 1 from another team. We had 135L litres of LN2 for the weekend.
2 of the people had never had any experience with LN2 before, so we showed them how to insulate there boards and mount the pots to get them started, we also helped them get pretty decent scores and give them help and tips as they increased the speed and score of their systems.
I have no idea how to set this thread out so i'll just go with each person with what they brought and scores. i think i'm going to call this format Multiple 'cool' stories bro
Dragon Soop is a very new member to our team who has just started with mostly air and aio cooling foor the past few months. He came on the Saturday and like most people who have never done LN2 before he did not have a pot.
The fisrt day he benched a Q9550 on a h80 cooler and a core 2 duo e8500 for the hwbot team cup. i did try and help him a little between my system cold bugging and drying out, but he pretty much had maxxed out his socket 775 system on the aio cooler.
E8500 Pifast score
http://hwbot.org/submission/3625459_
Once i had finished with my pot i lent it to him as another member of my team had sent us his as he couldnt make the meet and thought we would need a spare. i showed him how to insulate his board which was an Asus Z170 Impact with Vas. The vas creates a water proof layer between the pcb and components on the motherboard which prevents ice and water shorting the board as it cools down.
We also delided his 7700K with thermal grizzly Kryonaut as he had liquid ultra or pro from aio cooling. Liquid metal does perform better at ambient temps, but will crack and contract at lower temps where as the kryonaut is design to handle very low temps without cracking.
We then applied thermal paste on top of the ihs and mounted the pot with the inner insulation only, due the the board being so small we couldnt get the out layer on. The insulation around the ln2 helps keep the pot cold as well as stopping the board and memory surrounding the pot from getting as cold and built up with ice and water.
We then checked the system was working before pouring ln2 in the pot and checked both the bios and booted into windows.
skylake and kabylake are known for cracking paste as they are very small dies that put out a very concentrated heat load. This means that if the paste is not applied evenly over the die and ihs it can get hot spots which can cause cracking where the paste contracts and the die and ihs become disconnected and your cpu gets fried
To stop this terrible situation from occurring on his first ln2 run ever we let the system run cpuz benchmark on 2 cores until the idle temps were around ~+50c. this makes the paste spread evenly as it is warm. once it gets cold it doesnt spread as well.Once we had completed that we started to pull the cpu down. at -100c we enabled the ln2 mode jumper on the board, on the impact and most other ROG boards, this ln2 mode puts higher voltages through the cpu at boot up to combat a cold boot bug. This is when the cpu is frozen to such a degree when booting that it wont start. most kabylake chips dont tend to have a cold boot bug or cold bug when in ln2 mode on the asus 9 apex and asrock z170m ocf, but for some reason we found the system seemed to cold bug and cold boot bug around -150c even with the ln2 mode enabled.
I have never benched skylake on ln2 or kabylake before, so i didnt know the voltages to change that may have got this temp lower. we tried a few things in the bios, but in the end we stuck with -150c as it is still far colder than any water loop.
Now that we had figured out the coldbug he started to ramp up the clocks for the cpu and cache and memory as well as the voltages using a combination of the bios and asus turbo v software. once he had found the maximum clock speed of the cpu he started to run a few 2d benchamrks benchmarks
super pi 1m at 6.55ghz
http://hwbot.org/submission/3625687_
cinebench R15 at 6.06ghz
http://hwbot.org/submission/3625701_
Hwprime @5.9ghz
http://hwbot.org/submission/3625708_
After a few 2d benches he also decided to put his power modded 1080ti in and see what it could do in some 3d benchmarks
http://hwbot.org/submission/3625753_
3dmark vantage
http://hwbot.org/submission/3625714_
He also had a couple of runs with a GTX970, but did not manage to get any scores/screenshots as the board had some memory issues.
Next up is Mickulty who is the team captain for the reddit overclocking team. He only came for a few hours to drop in a GTX580 Matrix, however he did bring a small amd apu based system on a mini itx board which of course needed to be made faster so he blocked up some of the stock heatsink with artist eraser after removing the fan and use the stock cooler as a small pot.
Another person who had never used LN2 before and this was their first meet was No1spank. He has been a member of hwbot for a few years and benches regularly on air cooling and he wanted to take his FX6300 which he had got to 5.2ghz previously, to the next level on LN2. He wanted some advice on how to prepare the board for ln2 and he decided to go with liquid tape and vas rather than vas alone. This is a more permanent solution than just vas alone as it tends to stick to the board better and once dry will not rub off.
As he was just benching 2d with the cpu on LN2 we recommend that he just paint the top half of the board and any areas where the brush couldnt get the LET he put some vas on instead.
Now all FX chips are what we call full pot chips, which you can bench with the pot full of ln2 at -196c or a near as the pot and thermal paste allows. This means all you have to do is keep the pot topped up and you dont have to worry about cold bugs and cold boot bugs. This makes Am3 one of the best platforms for getting into extreme oc. Its also cheap to pick up now and the frequencies it is able to achieve are the highest out of any chips ever made.
He pulled the pot down fairly slowly and increased the frequency as he went. He did get a lot of help from Le and some from myself and almost bottled it at 6.7ghz ish, but we encouraged him to keep trying and pushing and before long he was flying past 7ghz and upto towards the top of the ranking for and FX6300
in the end he ran a massive 7.65ghz
http://hwbot.org/submission/3624587_
he also ran a few other benchmarks at 'lower' speeds
super pi [email protected]
http://hwbot.org/submission/3624684_
cinebench r15 at 6.3ghz
http://hwbot.org/submission/3624748_
wprime 32m at 6.35ghz
http://hwbot.org/submission/3625194_
Once he had finished benching on LN2 while he was packing up and waiting for his board and pot to dry out he had a liitle go with the Asus Rampage 9 Apex and 7900X we had set up under water with a few gpus he brought.
http://hwbot.org/submission/3625346_
http://hwbot.org/submission/3625395_
Lee is the man who made the custom pots, he brought 3 with him and he had even started extreme overclocking on phase after watching on of my youtube videos on a phase cooling setup. He had messaged me via emails a few times before moving onto LN2 and joining Hwbot. This was his first meet and he came very well prepared with all his gear and his board had some of the best insulation i have ever seen
he already had and FX8320 at over 7ghz under his belt and had brought an FX4300, phenom ii x2 and sempron 145 to the meet.
Lee started off with his FX4300 and he managed to get 1st place with a massive 7.75ghz oc
http://hwbot.org/submission/3625197_
He then moved onto his sempron 145. the sempron 145 doesnt have an unlocked multiplier like the FX4300, so it requires you to oc with the bclk only. This means you need a very good north bridge and ht link that work at very high frequencies. Normally a 790FX DDr2 based board would be best for this kind of oc but Lee still managed to get over 5ghz with the sempron 145 on his DDR3 board so that is still decent.
http://hwbot.org/submission/3624954_
His final cpu he ran the Phenom ii X2 550 and even though he didnt get 1st he manged to get over 6ghz with it putting him in the top 3 for maximum frequency on this cpu.
http://hwbot.org/submission/3627288_
I ad brought quite a few things to bench over the weekend, i was planning on running Ryzen 7 1800X, Ryzen 5 1600, Core 2 Duo E8600, Pentium E5200 and an i7 950. I had also brought 3 different socket 775 boards, Asus Prime X370 Pro and X58A OC to put these cpus in.
I started off with the Ryzen 7 1800X, my main goals with this chip was to do some 3d benchmarks with my R9 Fury and also run Wprime 1024M, Hwbot X265 4K and Geekbench3. These were all for the challenger division 4 oc esports competition
http://oc-esports.io/#!/round/roadtopro_challenger_season3_division4_round3
I started by mounting my pot and just wrapping some paper towel round the bottom of the pot. Then i got the fury in as well just on the air cooler. I pulled the cpu down to about -40c to start with and increased the voltage to 1.65v and set the cpu to 4.4ghz. I was going to run 3d mark vantage on the Fury as the first benchmark followed by 3dmark firestrike Extreme. Now i didnt know this at the time, but when you go below 0c on ryzen and threadripper, the pcie stops working properly so as you go under 3d load it crashes all the time. I spent quite a while trying to run 3d benches with the fury and apparently the only way to fix it is to set the pcie to 1.0.
After that failed i moved onto wprime 1024M I did manage to get the cpu unto 4.95ghz on 1.8v, but it would not run at 5ghz unfortunately. This is due to my only being able to get to about -130c on my Prime X370 due to it not having some of the voltages in the bios go high enough to run the cpu at full pot. At this speed i did mange to get a decent score
http://hwbot.org/submission/3624734_
it hasput my 1st in that stage of the competition however I did not get to complete any other stages as i was trying to help the others get going and bench at the same time, so the system got too much ice built up and stopped working and i didnt manage to get any other scores with that session.
Once my my pot had dried out and i had spent a while helping some of the others i decided that for the last session of the day i should Run my E8600 as this was top of my list to do on socket 775 as we needed some good pifast scores. I had never pifast before and it was a random cpu that i had pick from a tray as i have a few E8600 and haven't binned or tested any of them.
Now to start i booted at about 5ghz and then pulled to cpu down to in the os while running wprime 1024M to find out where the cpu cold bug was. once the benchmark freezes and i can no longer see the 1% of the benchmark that has completed then it has cold bugged. I also then can then find out what the cold boot bug is by restarting the system. while to cpu is too cold the post code reader would read 00 as if the cpu was dead and the machine will not post, it might flick on and off but it wont post. To find the minimum cold bug i get a blow torch and heat the pot back up until the system starts again. for this E8600 the cold bug was around -145 to -150c and cold boot bug was down at -110c. This means that if the system crashes or i change certain things in the bios while course a cold boot i need to heat the pot up with the blow torch to -110c to get going again. i dont think 30c difference is too bad but some cpus can need to be heated 60-70c higher temp to boot which can be annoying because you end up using loads of ln2 pulling it back down and you have to use the blow torch a lot. You can also get cpus on 775 that cold bug at -80c which is not get and they might not oc very high, because they cant go cold enough to use 1.9v.
So once i had found this basic info i needed to start i ramped up the clock speeds using Cinebench R15 to test the oc increase and also using setfsb to oc the fsb from inside windows. I got to about 5.6ghz before i realised that i had put the cpu in the wrong motherboard XD. i was running it on my Asus P5Q Deluxe which is great for quad core cpu's, but it cant go as high on the fsb as my P5Q-E which makes it better for the dual core chips.
to be continued............................
dun dun durn
Last edited:
