One of the best channels, hands down. And people wonder why there is pro overclocker lol there is just no people that knows more about these kinds off stuff. Keep up the good work der8auer!
Thank you for taking the time to work Tom at OC3D! It sounded like you spent considerable time with him. It's great to see people working together to understand the problems; in this case help Tom understand the limits of tools and some uncommon options. It helps us especially to find and understand exactly how things are setup and problem causes. Thanks again for doing dual languages as well!
El Diablo the 72C limit is tcase. If I remember correctly it's for package temperature for CPU which can be read via Hwinfo. Tjmax is separate from that.
Thanks der8auer for all the work you do and time you take to make these videos and keep everyone informed. I like you hold all the vendor's feet to the coals and keep them honest. We, as consumers, should get a quality product and having people like you out there helps us all.
It's refreshing to see people who actually know what they are taking about. It's a shame that tech youtubers who have a large following are often no better informed than a forum troll with entertainment value.
I feel the same about Level 1 Tech's. These are 2 of the very few channels that actually gives me accurate information that I can't find just by reading a spec's sheet.
>be asus >call from intel >"we're releasing a new platform next week so if you'd have boards done by then that'd be greeeeaaaat" >"also make sure the vrms can do 300A without overheating" >hangs up >mfw
It's really mobo vendors fault for prioritizing aesthetics over functionality. X99 WS boards had proper finned and heatpiped heatsinks on the VRMs that could handle 10 core 6950X max overclocks all day long with zero airflow. For X299 where power delivery is actually easier to design due to FIVR, they decided to put a lump of aluminum on them and call it good. It's fine if your loads are lumpy but it's bad if you're under sustained load. MSI at least should have known this since their Z270 M.2 "heatsinks" ended up heating up more than not using them.
+Actually Hardcore Overclocking Yeah, i really would like see someone go deep in why 7900X is slurping so much wattage! A maximum power draw battle video of 6950X vs 7900X could be awesome too.
You are one of many few tech UA-camrs who actually know what they are talking about, love the in depth, precise and in-depth research, explanations and facts. No fanboy-ism or BS. Nice job!!
Thanks for the coverage, funny that my 6 year old Rampage IV Extreme has more effective VRM heatsinks than brand new motherboards. This whole RGB/looks-over-performance nonsense has to stop. I can't honestly find a single X299 motherboard that doesn't look like it's been designed by children for children. Vendors, keep it simple, performance always first!
MrTeeh sadly boards that are designed for children sell better as many gamers are children or just have a very childish taste. Not generalizing all gamers, just saying many. I don't mind a nice looking board but my idea of a nice looking board is for it to be all blacked out because an over the top board should not be the attention grabber in a build IMO.
EVGA's x299 flagship board (I think it was called the x299 Dark) is RGB free and clean looking if i remember correctly. Might be what you are looking for.
I was looking forward to getting an X299 as a new system, and even considering just getting some parts over next few months before the i9's come into play, I'm very pleased to watch your videos not just to break my heart but also to thank you for now before making a mistake. I think waiting is a very wise move. Liked and subscribed for life.
It is good to see there is a follow up of the topic after 1st one. This video is even better by explaining to us on the testing method and the reasons behind it. This show how seriously one can push the cpu to its limit, and not just plug and play. Great work.
Really excellent video that thoroughly dives into his testing methodology and responding to many comments/criticisms. The P95 segment in particular was really illuminating - but not a total surprise for me as the debate on P95 small vs large FFTs is as old as time. But particularly since the addition (and then removal) of FIVR in recent years, people have begun questioning, in earnest, what it means to be stable. Thanks for the hard work Roman! IMO, it is extremely rare to find someone that is detail-oriented in the UA-cam era. I would hope more people treat him with respect and take note of his knowledge and experience.
Have you been having USB issues with any of the boards you have tested yet? because the Aorus Gaming 7 x299 I have been testing has had really bad USB 3.1 Gen 2 issues. When I was stress testing the 7900x @ 4.7ghz Aida64 was telling me that the VRM was at 86c but I am not sure how accurate that number is, as it said the core voltage was 1.8v (really was 1.28v)
Thanks to you, I waited to pull the trigger on an Asrock Taichi X299. (i9-7900X system) They recently released the X299 Taichi XE for the voltage and temps, and has the proper pins on the board. I pulled the trigger today and got it at Newegg, as they are the only ones who have it. Thanks for what you do Roman! You're damned good at it!
Great job explaining it. I experienced this VRM throttling years ago with a 3930k and it was exactly as you described - Massive drops in core clock. Hopefully those that rushed to judge your methods will think twice next time! :)
I commented on Toms video when he posted and told him unless he replicated the exact methodology his testing wasn't going to be able to replicate your findings. Thankfully some of us understood what you were trying to show us and I'm happy to hear Tom now understands too. This will be very interesting to see how the manufacturers respond and I'm sure that they'll have to come out with new boards before the 12-18 core chips arrive. Thanks for sharing your findings, keep up the good work.
I'm not a CPU OCer at all, still learning, der8auer thanks for all the stuff you make esp the Delid tool, really top videos. A hero that we enthusiasts, geeks need. Cheers ! Keep up the good work ! Thanks a lot again man, you won our hearts.
Validate and Verify! Great work! Your Prime95 testing makes me want to go back and fully test out my 6950x using a Hall effect sensor on the EPS side to see what's happening in fuller detail. See how it compares to x299 more.
Thanks again, De8auer. Hopefully the "Flash" will come in 2nd, compared to the function of heat sinks. Takes guts to do what you did... Danke Herr Roman
Thank you Der Power to make things clear. We Do need guys like you to tell the truth. A bit pity not seeing you at ComputeX this year, but we met the one at 2015. Good vid, keep pushing.
There is a lot of confusion and complacency about tools and methodology for testing. For example, many people now use AIDA64 stress test to validate CPU overclocks and system stability. I found in that tool that default options for its stress test has a memory test option checked in addition to three CPU tests. With that memory test checked, the CPU power/wattage is considerably less (about 10-15 watts based on tool reported wattage) than without it checked. I also found a "stable" overclock with the memory test checked could easily fail when close to the edge when memory test was unchecked. People are looking for run and forget testing and are not very educated in what their testing. It takes more than a download and running of a tool.. Keep up the excellent work and help us learn!
I bet instead there'll just be a new even more expensive "EXTREMEGAMERTURBO" range or something similarly named, that advertises better quality heatsinks.
Hey Roman (Ciro here), I'm waiting for the first X299 boards to come in and I'll try to replicate what you did, in order to confirm your findings! Thanks again for the in-depth analysis!
had similar issue with x79 running xmp i had made vrm heat sinks for the top and under side mosfets then overclocking became stable after having what you explained would of bean vrm throttling. so i made my heat sinks out of copper bar 48mm x 12mm x (length covering the vrm) and got it nickel plated using some thin az ekwb pad. would ya like one made for ya? or my design?
Could you put a thermal pad in the gap between the vrm and heat sinks? Would that make a difference? So the best solution now is to take the heat sink off and have a fan running on it
Thanks for the followup video, much appreciated! There was indeed a large kerfuffle on most forums and websites I frequent, and lots were pegging Jon Gerrow as being in the right. But seeing as you went into careful explanation of settings and including worst case scenarios, I believe this is fairly comprehensive. Thanks a lot! I don't want to hog too much time, but did you notice the same 8-pin cable temps on all of the PSUs, or was that indeed a superflower issue?
Hey @der8auer, I was wondering if you think the Aeorus Gaming 9 might have any difference on VRM temps, looks like it has a heatpipe going to rear IO, possible extra surface area?
Thanks for the update Der8auer! Been wanting a 7820X for a multimedia build (gaming and movies and whatnot, yes I know it's overkill) and was about pull the trigger on a X299 Prime Deluxe. Thus far I've seen MSI have better gaming performance because of Intel's P-States on ASUS supposedly. I custom water cool, so I don't mind the heat, but unfortunately no monoblocks for the VRM. Thank for you allowing me to make a better decision as i didn't just jump onto this platform. I really look forward to the day I can finally buy this!
Dai2 Dai2 if i gamed at 60hz i totally would, but this whole build is meant to be unnecessary and expensive. Will be starting off with 32gb of ddr4-4000 in quad and already have a 512gb 960 pro waiting. I might end up with ddr4 3600 though, because quad channel will give all the bandwidth. As for TR, super excited for the community, but I know it's not going to be as good in games and that's all I do. Please don't take this as I think Intel is a good company or that X299 is a good platform, but I'm stuck with no other choice. I won't "upgrade" to another 4 core, ganes will start using more than 8 threads in a couple years from now and I stay ahead of the curve. So 7820x is that sweetspot. 7800x is useless as coffeelake is coming out and anything over 8 core is certainly not meant for just gaming and thus Is get TR, but no other competition when it comes to the 8 core.
Great video, I have followed TTL for years and years. I knew there was more to it. I am certainly keen to see some videos with better thermal interface. Does the APEX you have use the same IR mosfets? you mentioned these boards all use the best IR components. Does the APEX simply have more phases of the same components or better components that stay cooler / provide greater efficiency?
der8auer, out of curiosity, have you tested AIDA64's stress test? Wonder how hard the FPU test hits the CPU compared to P95's Small FFT's? Or CPU+FPU combined?
Just look to the s755 1366 era - heat pipes, pure copper and fans on vrm! Now we have only designer nonsense made from chinese silumin (low quality fake aluminum) and with vendor logos that strict airflow.
I had a Asus Crosshair nForce 590 SLI AM2 that came with 2 blower fans that could be mounted on its copper VRM heat sinks. Like so: www.frostytech.com/articleview.cfm?articleid=2077&page=2
Oh yeah, my asus smth deluxe for s775 had the same. Damn it was noisy :D But just look at that amount of copper, surface area and heatpipes! P.S btw, modern sabertooth series boards have something similar. Sadly, the noise is also present
Hey nice video i have a 7820x delided from caseking and a msi m7 ack do you think it would have the same vrm problems should i get a gigabite gaming 7 for srs 24/7 oc ?
it amazed me to no end that the few sites and tubers who had day 1 and day 2 reviews for the 7900X didn't even realize it was throttling and that's hwy the performance in half the benchmarks was bad and inconsistent. not one of them even thought of this, even though the few that bothered to publish operating temps all saw the cpu quickly go to 100c under load.
I think that the Phanteks Evolv ATX has one of the most perfect interior layouts for a mid-tower case. But it has very bad interior airflow compared to many mid-tower cases. You can slightly improve the performance of the front intake fans by using longer bolts to push the front plate out about 5mm to 7mm. But the real problem is that the Evolv has extremely restrictive tiny top vent slits combined with the location of the top fan/radiator tray. Many people, including me, have used their Dremel to cut out a big rectangular hole on the Evolv's aluminum top panel and installed a dust filter there. You get far better temperatures once you cut a big hole into the Evolv's top panel.
I actually removed the dust filter in the top of the case so the AIO can pull a lot more fresh air in. I always do this with this case otherwise the air intake is really not that great :/ I also wish that the front had a bit more holes or space for air intake
I am surprised that some people even doubted your video in the first place. Faith in humanity lost again. Then again, forums are full of noobs and that's exactly you need to keep doing what you are doing, keep up the good work. Everything I have seen on the forums posted by you are quality stuff right on spot. I have the Strix X299 and the VRM runs burning hot when CPU is pushed to 4.6 at 1.26V. It goes up to 98c and I didn't even check the power consumption and all settings...board is inside a case. For me, X299 is quite the complete failure. Slower in gaming than my 6950X, runs hotter than hell and I cannot get SLI to scale properly. :D I am probably going to change back and possibly check out the 12-18 cores later on + possible bug fixes and Rampage VI Extreme board.
I can understand that it's quite difficult to figure this all out if you don't have the background knowledge. So it's very easy that you think everything is fine while the clocks go downhill in the back and you don't even notice it.
Hi der8auer. I am using the Gigabyte X299 AORUS Gaming Motherboard. My TJ-Max Offset is Auto. And the range of the setting is 0-40. But you said set it to 105'C in the video. I think there must be some differences among different BIOS systems. Can you try it and tell me what should I set for TJ-Max Offset on my board?
#der8auer, is there a chance that the limitation on the cheaper x299 boards could be bad/worse VRMs, or do you think the limiting factor for reaching above 4,5ghz on all of the x299 boards is the heatsink on the Vrms ?
Core temp has quite fast temp reading, but you can still set it to read even faster. I think the setting in Core temp is called polling interval and is set by milliseconds.
Hey Roman. Great video! I have a question: I recently build a 7820x on a msi X299 SLI PLUS motherboard. Unfortunately That board comes with only 1 8pin ATX connector. For PSU i Use a corsair AX650GOLD series. That PSU also has 1 8pin cable. Based on your video I now know ideally you have to use 8+4 or better 8+8 cables but i didn't want to buy a new board and psu so i thought i 'll just give it a try and see where it is capable of. So i overclocked the system to 4.7GHZ on all cores by only changing the vcore to 1.150 volts. Its stable in cinebench and AIDA64 in the FPU test. Package CPU temps are 87C and vrm 79C CPU power consumption hits just over 200 watts. I don't see freq bumping up and down in HWinfo64. AIDA64 and intel extreme overclock utility both say CPU is NOT throttling . Are those programs capable reading out the real values? Based on this information how can really know if the system is throttling or not ? btw: For cooling i use the Noctua D15
Could this testing be done on the older boards where you have the 7700k and the new AMD boards also? Just to see what the results is? I know it takes alot of time and maybe isnt as useful since they doesnt pull that mcuh wattage, but maybe its useful on threadripper/x399 platform?
So much valuable information, so little ego. Bravo der8auer! Maybe, just maybe as a side effect of this problem we may see some real effort put on those monoblocks... I've skipped those things alltogether because they never managed to equal a CPU only waterblock... or maybe we go back to 1366 design days were you could get a propper CPU and propper VRM waterblocks... as you said they may well use this to bring new "thermal solutions" to the table... also in the wake of your video on the APEX heatsink... for extreme overclocks that 50mm fan might be acceptable, but for normal OC'd use is unnacceptable.
out of curiosity what is the max vrm temp mobos were getting on x99 and z270 with oced cpus? Also me thinking about how going from 100% to 140% capability can increase power draw...then it also must have some effect on performance in games etc. reviewers should retest this then..
Great Vid / Channnel, may not of come across your channel if not for this kerfuffal. Is there any board currently available that is worth buying for x299?
@der8auer What do you think of the TUF series from ASUS (in this case, the x299 Mark 1) ? I was hesitating to get one since they are especially made for overclocking (remembering my old but incredible TUF z77 Sabertooth).
Oh, okay then. I thought it would be good for OC since they use military certified components and make the MB undergo punishing burn-in tests. The micro-cooling environment created by the fans in between the MB and the Thermal Armor seemed to me like a huge plus for OC but you're the expert. What MB would you recommend then and do you think I should stick with a z270 + 7700K instead of a x299 + 7800X for OC? That's a lot of questions but I'd really appreciate if you find time to answer. Cheers, Vkno1
Thank you Roman for excellent update. Question though, when's the new Asus or other boards coming out with improved VRM Heatsink designs ? Do you have any info on this ? Also what are the motherboard models that have 8 + 8 CPU pin connectors for Skylake X that are out now or will be released soon that you know of ? Thanking you
End of July / beginning of August :). I know that Apex and Rampage 6 Extreme will be fine. Next week I will test the X299 XPower from MSI. Could also be good
If I were to remove the heatsinks from VRMs and point a 120mm fan at them with reasonable CMF would it be safe/better than using the monolithic heatsinks that are found on new motherboards? I mean I would not risk destroying the VRMa, right?
I would reach out to Gigabyte and see if you can get one of their higher end boards. I currently have a Gaming 7 and it has a significant a heat pipe running from the VRM down along the back plate. Hopefully this added surface area can help dissipate some of the heat.
i have also done some reading with regards to vrm type/ quality before purchasing my x99 motherboard, i went with a irf single package vrm board( asus x99 deluxe ii) it has slightly higher vrm temps vs the gigabyte designer x99( both reviewed by tweaktown with thermal imaging on vrm) but i think the gigabyte board had better heatsink and also maybe more copper in board( gigabyte 20z copper marketing )... maybe motherboard vendors should stack up on vrm phases to help distribute the heat over a wider area wih x299.... great article btw
i dont know much about this so correct me if im wrong but how can they copy that z170 heatsink to the x299 when the space is taken by the additional 4 ram slots and the motherboards dont have much space for a bigger heatsink
great detail. any place we can get config details/bios settings for the asus boards used during your testing? I'm working to reproduce as well with delided 7900X CPU. Direct message/PM me if required. Thanks Daz..
Something I'd like to comment on. That post on the thermal paste roundup in 2015 was FRIGGIN amazing. I can't believe i listen to linus about ic diamond. That stuff is tough to apply, also conductive paste, hard to clean and performance wise is crap. I remember i used artic mx4 for some reason and it was like 26 C and when i had to take it off and place ic diamond it was like 28-29 C but thats on different days but still imo IC diamond is still crap. I bought the kryonaut paste and man, soon i'm going to repaste my cpu and the graphics card. I hope to get better temps than when i first bought these and pasted. Thanks for review even if its late.
Hallo Roman, super Video, wie immer :) Ich wollte fragen was du von EK-TIM Indigo XS - Intel 115x oder andere Sockel hältst und ob du ein How tu Use it Video machen könntest. Dachte an einen geköpften i7 6700K 7700K mit Thermal Grizzly Liquid Metal plus EK-TIM Indigo XS - Intel 115x :) Liebe Grüße :)
oh my god, now is an actual 5870x2 Ares holy wurst im collecting old awesome looking gpus and missing alot of dual ones :/ so on subject, what are your thoughts, are companies being lazy when it comes to this? i remember some really Nuts boards like Gigabyte SOC series that had alot of features for OC, but this "high end" boards are becoming more like a marketing stunts, i also saw on AMD side a Video by Buildzoid saying from an MSI X370 board that the VRM Desing was terrible
Great info. Im trying to get my Asus Rampage VI extreme running an i9-7920x up to speed. Really good stuff on throttling. I'm at 4.4 Ghz now and the cpu power fluxuates between 150 and 215 watts... throttle going on somewhere... using a Thermaltake AIO 360 cpu cooler and getting ready to upgrade to the EKW mono block and a custom loop.
How do you think the Asus Rampage VI Apex will perform without the VRM heatsink for pre ln2 testing? Testing such as RAM settings for benching to set up profiles... I will be starting with a I7-7740X thus no AVX512 right
Couldn't you just take off the stock VRM heatsink and throw on a bunch of little copper heatsinks or if you could find a big one that would fit? That would surely increase the surface area and ensure better contact.
I do have a question which i am unsure you can answer. Did any of the board manufacturers contacted you to try to get to the bottom of the issue? Thanks and great work!
+Gil ARAÚJO they for sure know about it even tho you'd never get a public statement. But I'm sure we will see some changes from now. I had some tough reactions on the first video
You are so good, taking your time to respond to those reactionary comments.
My favorite Tech channel now.
One of the best channels, hands down. And people wonder why there is pro overclocker lol there is just no people that knows more about these kinds off stuff. Keep up the good work der8auer!
Thank you for taking the time to work Tom at OC3D! It sounded like you spent considerable time with him. It's great to see people working together to understand the problems; in this case help Tom understand the limits of tools and some uncommon options. It helps us especially to find and understand exactly how things are setup and problem causes. Thanks again for doing dual languages as well!
I really like people on UA-cam like you - those, who know the stuff they're talking about, true professionals.
Keep up the good work. Thumbs up!
Love the Asus Ares on the bench, blast from the past. 😊
:D ARES
El Diablo the 72C limit is tcase. If I remember correctly it's for package temperature for CPU which can be read via Hwinfo.
Tjmax is separate from that.
Thanks der8auer for all the work you do and time you take to make these videos and keep everyone informed. I like you hold all the vendor's feet to the coals and keep them honest. We, as consumers, should get a quality product and having people like you out there helps us all.
I personally like the way you test things in term of closed PC test and taking the worst case scenario which is the best approach ever. Keep it up!
It's refreshing to see people who actually know what they are taking about. It's a shame that tech youtubers who have a large following are often no better informed than a forum troll with entertainment value.
I feel the same about Level 1 Tech's. These are 2 of the very few channels that actually gives me accurate information that I can't find just by reading a spec's sheet.
>be asus
>call from intel
>"we're releasing a new platform next week so if you'd have boards done by then that'd be greeeeaaaat"
>"also make sure the vrms can do 300A without overheating"
>hangs up
>mfw
It's really mobo vendors fault for prioritizing aesthetics over functionality. X99 WS boards had proper finned and heatpiped heatsinks on the VRMs that could handle 10 core 6950X max overclocks all day long with zero airflow. For X299 where power delivery is actually easier to design due to FIVR, they decided to put a lump of aluminum on them and call it good. It's fine if your loads are lumpy but it's bad if you're under sustained load.
MSI at least should have known this since their Z270 M.2 "heatsinks" ended up heating up more than not using them.
6950X maxed out uses less power than 7900X
+Actually Hardcore Overclocking
Yeah, i really would like see someone go deep in why 7900X is slurping so much wattage!
A maximum power draw battle video of 6950X vs 7900X could be awesome too.
hangs up or hangs himself?
You do know X99 used FIVR too, right?
Great work, and glad to see der8auer on this, if it was not for him, this could probably slide, like basic reviewers just testing games and RGB.
So true that was funny hahaha!
You are one of many few tech UA-camrs who actually know what they are talking about, love the in depth, precise and in-depth research, explanations and facts. No fanboy-ism or BS. Nice job!!
Thanks for the coverage, funny that my 6 year old Rampage IV Extreme has more effective VRM heatsinks than brand new motherboards. This whole RGB/looks-over-performance nonsense has to stop. I can't honestly find a single X299 motherboard that doesn't look like it's been designed by children for children. Vendors, keep it simple, performance always first!
MrTeeh sadly boards that are designed for children sell better as many gamers are children or just have a very childish taste. Not generalizing all gamers, just saying many. I don't mind a nice looking board but my idea of a nice looking board is for it to be all blacked out because an over the top board should not be the attention grabber in a build IMO.
Have you looked at any of the ASRock boards?
EVGA's x299 flagship board (I think it was called the x299 Dark) is RGB free and clean looking if i remember correctly. Might be what you are looking for.
preach bro
but Rampage IV is flag ship , any way R.I.P x299 :)
I was looking forward to getting an X299 as a new system, and even considering just getting some parts over next few months before the i9's come into play, I'm very pleased to watch your videos not just to break my heart but also to thank you for now before making a mistake. I think waiting is a very wise move.
Liked and subscribed for life.
It is good to see there is a follow up of the topic after 1st one. This video is even better by explaining to us on the testing method and the reasons behind it. This show how seriously one can push the cpu to its limit, and not just plug and play.
Great work.
Really excellent video that thoroughly dives into his testing methodology and responding to many comments/criticisms. The P95 segment in particular was really illuminating - but not a total surprise for me as the debate on P95 small vs large FFTs is as old as time. But particularly since the addition (and then removal) of FIVR in recent years, people have begun questioning, in earnest, what it means to be stable.
Thanks for the hard work Roman! IMO, it is extremely rare to find someone that is detail-oriented in the UA-cam era. I would hope more people treat him with respect and take note of his knowledge and experience.
Have you been having USB issues with any of the boards you have tested yet? because the Aorus Gaming 7 x299 I have been testing has had really bad USB 3.1 Gen 2 issues.
When I was stress testing the 7900x @ 4.7ghz Aida64 was telling me that the VRM was at 86c but I am not sure how accurate that number is, as it said the core voltage was 1.8v (really was 1.28v)
Thanks to you, I waited to pull the trigger on an Asrock Taichi X299. (i9-7900X system)
They recently released the X299 Taichi XE for the voltage and temps,
and has the proper pins on the board.
I pulled the trigger today and got it at Newegg, as they are the only ones who have it.
Thanks for what you do Roman!
You're damned good at it!
Great job explaining it. I experienced this VRM throttling years ago with a 3930k and it was exactly as you described - Massive drops in core clock. Hopefully those that rushed to judge your methods will think twice next time! :)
You're my favorite overclocker man, I love your content! cheers mate
I commented on Toms video when he posted and told him unless he replicated the exact methodology his testing wasn't going to be able to replicate your findings. Thankfully some of us understood what you were trying to show us and I'm happy to hear Tom now understands too. This will be very interesting to see how the manufacturers respond and I'm sure that they'll have to come out with new boards before the 12-18 core chips arrive. Thanks for sharing your findings, keep up the good work.
I'm not a CPU OCer at all, still learning,
der8auer thanks for all the stuff you make esp the Delid tool, really top videos. A hero that we enthusiasts, geeks need. Cheers ! Keep up the good work !
Thanks a lot again man, you won our hearts.
I've been following your channel for quite some time now, and I just wanted to say that you do a great job. Subscribed and liked.
thank you :)
Respect for validating man!
Very good... I'm glad you did an update video. Hope to see more videos from you in the future!
Validate and Verify! Great work! Your Prime95 testing makes me want to go back and fully test out my 6950x using a Hall effect sensor on the EPS side to see what's happening in fuller detail. See how it compares to x299 more.
Thanks again, De8auer. Hopefully the "Flash" will come in 2nd, compared to the function of heat sinks. Takes guts to do what you did... Danke Herr Roman
thank you for posting these types of videos!! Patiently (barely) waiting for the x299 delid tool in the USA :)
Thanks man :) Working on the availability. I had to change the tool design a bit to make it as durable as possible
Very smart person, great explanation! thanks!
Hey der8auer, KEEP UP THE AWESOME content man!!
Sending you much love from east europe Estonia :D (nohomo)
Excellent explanation, as always. Really appreciate you doing this in English too.
der8auer is a boss.
Thank you Der Power to make things clear. We Do need guys like you to tell the truth. A bit pity not seeing you at ComputeX this year, but we met the one at 2015. Good vid, keep pushing.
Most informative video ive watched this week. Ty D8 for keeping it real.
There is a lot of confusion and complacency about tools and methodology for testing. For example, many people now use AIDA64 stress test to validate CPU overclocks and system stability. I found in that tool that default options for its stress test has a memory test option checked in addition to three CPU tests. With that memory test checked, the CPU power/wattage is considerably less (about 10-15 watts based on tool reported wattage) than without it checked. I also found a "stable" overclock with the memory test checked could easily fail when close to the edge when memory test was unchecked. People are looking for run and forget testing and are not very educated in what their testing. It takes more than a download and running of a tool.. Keep up the excellent work and help us learn!
I am glad that there are still real hardware testers out there.
Funny that motherboards from 8-10 years ago had much better VRM heatsinks than modern ones.
Well after we start complaining they gonna stop putting cheap metals on motherboards i hope.
I think thats what a lot of us thought, too. The whole X299-Platform feels somewhat rushed.
I bet instead there'll just be a new even more expensive "EXTREMEGAMERTURBO" range or something similarly named, that advertises better quality heatsinks.
Hey Roman (Ciro here), I'm waiting for the first X299 boards to come in and I'll try to replicate what you did, in order to confirm your findings! Thanks again for the in-depth analysis!
Thank you man :) Good to see you on here!
ReHWolution what did you find
@der8auer
Great Vid and thanks for clearing things up. I am glad that the VCCIO hint helped here a little bit.
had similar issue with x79 running xmp i had made vrm heat sinks for the top and under side mosfets then overclocking became stable after having what you explained would of bean vrm throttling. so i made my heat sinks out of copper bar 48mm x 12mm x (length covering the vrm) and got it nickel plated using some thin az ekwb pad. would ya like one made for ya? or my design?
Could you put a thermal pad in the gap between the vrm and heat sinks? Would that make a difference? So the best solution now is to take the heat sink off and have a fan running on it
Thanks for the followup video, much appreciated! There was indeed a large kerfuffle on most forums and websites I frequent, and lots were pegging Jon Gerrow as being in the right. But seeing as you went into careful explanation of settings and including worst case scenarios, I believe this is fairly comprehensive. Thanks a lot!
I don't want to hog too much time, but did you notice the same 8-pin cable temps on all of the PSUs, or was that indeed a superflower issue?
LooneyJuice He mentioned it in the german video. When using another PSU the temp at the 8pin cable reached only 45 °C.
Ah, nice, thank you. Much appreciated.
Hey @der8auer, I was wondering if you think the Aeorus Gaming 9 might have any difference on VRM temps, looks like it has a heatpipe going to rear IO, possible extra surface area?
My respect to you Roman for making this follow up video.
Thanks for the update Der8auer! Been wanting a 7820X for a multimedia build (gaming and movies and whatnot, yes I know it's overkill) and was about pull the trigger on a X299 Prime Deluxe. Thus far I've seen MSI have better gaming performance because of Intel's P-States on ASUS supposedly. I custom water cool, so I don't mind the heat, but unfortunately no monoblocks for the VRM. Thank for you allowing me to make a better decision as i didn't just jump onto this platform. I really look forward to the day I can finally buy this!
Dylan Shaw just get a 1700 or a threadripper, the problems are much less frequent.
Dai2 Dai2 if i gamed at 60hz i totally would, but this whole build is meant to be unnecessary and expensive. Will be starting off with 32gb of ddr4-4000 in quad and already have a 512gb 960 pro waiting. I might end up with ddr4 3600 though, because quad channel will give all the bandwidth. As for TR, super excited for the community, but I know it's not going to be as good in games and that's all I do. Please don't take this as I think Intel is a good company or that X299 is a good platform, but I'm stuck with no other choice. I won't "upgrade" to another 4 core, ganes will start using more than 8 threads in a couple years from now and I stay ahead of the curve. So 7820x is that sweetspot.
7800x is useless as coffeelake is coming out and anything over 8 core is certainly not meant for just gaming and thus Is get TR, but no other competition when it comes to the 8 core.
Great video, I have followed TTL for years and years. I knew there was more to it. I am certainly keen to see some videos with better thermal interface. Does the APEX you have use the same IR mosfets? you mentioned these boards all use the best IR components. Does the APEX simply have more phases of the same components or better components that stay cooler / provide greater efficiency?
any update on your new delidding tool?
der8auer, out of curiosity, have you tested AIDA64's stress test? Wonder how hard the FPU test hits the CPU compared to P95's Small FFT's? Or CPU+FPU combined?
what gpu is that red dino holding? it has a unique cooler for what looks like an agp slot. also love the old asus mars card in the test bench.
Just look to the s755 1366 era - heat pipes, pure copper and fans on vrm! Now we have only designer nonsense made from chinese silumin (low quality fake aluminum) and with vendor logos that strict airflow.
I had a Asus Crosshair nForce 590 SLI AM2 that came with 2 blower fans that could be mounted on its copper VRM heat sinks. Like so: www.frostytech.com/articleview.cfm?articleid=2077&page=2
Oh yeah, my asus smth deluxe for s775 had the same. Damn it was noisy :D
But just look at that amount of copper, surface area and heatpipes!
P.S btw, modern sabertooth series boards have something similar. Sadly, the noise is also present
+Taran
So many people have no idea how much fake aluminum they use these days on motherboards!
+Dmitry Belogub
P5Q3 Deluxe i have that running!
+Samuel Jones
Wow, thats some nice engineering work from Asus! Sad that they fired the engineer and hired an art student..
Check Rampage Extreme X48 and Rampage II Extreme ;)
Hey nice video i have a 7820x delided from caseking and a msi m7 ack do you think it would have the same vrm problems should i get a gigabite gaming 7 for srs 24/7 oc ?
You did the necessary due diligence in your testing method it seems. I would be surprised had you not.
The backpeddling begins!!
it amazed me to no end that the few sites and tubers who had day 1 and day 2 reviews for the 7900X didn't even realize it was throttling and that's hwy the performance in half the benchmarks was bad and inconsistent. not one of them even thought of this, even though the few that bothered to publish operating temps all saw the cpu quickly go to 100c under load.
PureBlackFire because the 7900x is GARBAGE!!!
PureBlackFire What a terrible way to introduce i-9, Intel should have used the i-9 name for the 980x as it would have done the i-9 name justice!
I think that the Phanteks Evolv ATX has one of the most perfect interior layouts for a mid-tower case. But it has very bad interior airflow compared to many mid-tower cases. You can slightly improve the performance of the front intake fans by using longer bolts to push the front plate out about 5mm to 7mm. But the real problem is that the Evolv has extremely restrictive tiny top vent slits combined with the location of the top fan/radiator tray. Many people, including me, have used their Dremel to cut out a big rectangular hole on the Evolv's aluminum top panel and installed a dust filter there. You get far better temperatures once you cut a big hole into the Evolv's top panel.
I actually removed the dust filter in the top of the case so the AIO can pull a lot more fresh air in. I always do this with this case otherwise the air intake is really not that great :/ I also wish that the front had a bit more holes or space for air intake
I am surprised that some people even doubted your video in the first place. Faith in humanity lost again. Then again, forums are full of noobs and that's exactly you need to keep doing what you are doing, keep up the good work. Everything I have seen on the forums posted by you are quality stuff right on spot. I have the Strix X299 and the VRM runs burning hot when CPU is pushed to 4.6 at 1.26V. It goes up to 98c and I didn't even check the power consumption and all settings...board is inside a case. For me, X299 is quite the complete failure. Slower in gaming than my 6950X, runs hotter than hell and I cannot get SLI to scale properly. :D I am probably going to change back and possibly check out the 12-18 cores later on + possible bug fixes and Rampage VI Extreme board.
I can understand that it's quite difficult to figure this all out if you don't have the background knowledge. So it's very easy that you think everything is fine while the clocks go downhill in the back and you don't even notice it.
Hi der8auer. I am using the Gigabyte X299 AORUS Gaming Motherboard. My TJ-Max Offset is Auto. And the range of the setting is 0-40. But you said set it to 105'C in the video. I think there must be some differences among different BIOS systems. Can you try it and tell me what should I set for TJ-Max Offset on my board?
#der8auer, is there a chance that the limitation on the cheaper x299 boards could be bad/worse VRMs, or do you think the limiting factor for reaching above 4,5ghz on all of the x299 boards is the heatsink on the Vrms ?
thanks for the update, looking forward to Tom's video
Thanks for the fluid explanation.. keep up the good work! Love it and kudos!
Where can I buy your delid tool in America?
Sorry for the noob question, which is a program that allow me to know the power consumption of cpu and gpu? Thanks!
Core temp has quite fast temp reading, but you can still set it to read even faster. I think the setting in Core temp is called polling interval and is set by milliseconds.
So, what do you think the 7820x would be about to do one there boards? Specifically the carbon.
Hey Roman. Great video!
I have a question:
I recently build a 7820x on a msi X299 SLI PLUS motherboard. Unfortunately That board comes with only 1 8pin ATX connector. For PSU i Use a corsair AX650GOLD series. That PSU also has 1 8pin cable. Based on your video I now know ideally you have to use 8+4 or better 8+8 cables but i didn't want to buy a new board and psu so i thought i 'll just give it a try and see where it is capable of.
So i overclocked the system to 4.7GHZ on all cores by only changing the vcore to 1.150 volts. Its stable in cinebench and AIDA64 in the FPU test. Package CPU temps are 87C and vrm 79C
CPU power consumption hits just over 200 watts. I don't see freq bumping up and down in HWinfo64. AIDA64 and intel extreme overclock utility both say CPU is NOT throttling . Are those programs capable reading out the real values?
Based on this information how can really know if the system is throttling or not ?
btw: For cooling i use the Noctua D15
Top work. Earned a sub and can't wait to see more of your stuff.
Could this testing be done on the older boards where you have the 7700k and the new AMD boards also?
Just to see what the results is?
I know it takes alot of time and maybe isnt as useful since they doesnt pull that mcuh wattage, but maybe its useful on threadripper/x399 platform?
So much valuable information, so little ego. Bravo der8auer! Maybe, just maybe as a side effect of this problem we may see some real effort put on those monoblocks... I've skipped those things alltogether because they never managed to equal a CPU only waterblock... or maybe we go back to 1366 design days were you could get a propper CPU and propper VRM waterblocks... as you said they may well use this to bring new "thermal solutions" to the table... also in the wake of your video on the APEX heatsink... for extreme overclocks that 50mm fan might be acceptable, but for normal OC'd use is unnacceptable.
so even for a moderate oc of 7820x to about 4,5 - 4,6 GHz you don't recommend any of these mobos? It's better to wait for the new ones?
out of curiosity what is the max vrm temp mobos were getting on x99 and z270 with oced cpus?
Also me thinking about how going from 100% to 140% capability can increase power draw...then it also must have some effect on performance in games etc. reviewers should retest this then..
The cpu at full speed is not performing right bicause the vrm is overheating and loosing efficienty, inducing micro throttling?
hi Derbauer do you know if they have fixed the VRM cooling on 299 or is X299 VRM still bad
Does threadripper have this kind of problem? I want to buy a 1950x, and overclock it at something arround 3.8ghz, what mobo should i get?
Great Vid / Channnel, may not of come across your channel if not for this kerfuffal. Is there any board currently available that is worth buying for x299?
I love the enthoo evolv ATX but man the ventilation is really bad in that case, your shit will get hot in there too.
Thank you! Keep up the good work!
Great Video, I just wish it made it easier for me to take the plunge into X299. For now, I am going to wait until this clears up.
der8auer, do you think Asus x299 TUF Mark 1 is a good motherboard? Would really appreciate your response.
Personally I don't think it's a great board :/ VRM cooler doesn't look good to me. VRMs should be the same as on STRIX or X299-A
thanks for the reply man, appreciate it. I guess I'll wait for another month for other boards to be released.
@der8auer What do you think of the TUF series from ASUS (in this case, the x299 Mark 1) ? I was hesitating to get one since they are especially made for overclocking (remembering my old but incredible TUF z77 Sabertooth).
It would not be on the list of boards I'd recommend for OC :D It's really not that great
Oh, okay then. I thought it would be good for OC since they use military certified components and make the MB undergo punishing burn-in tests. The micro-cooling environment created by the fans in between the MB and the Thermal Armor seemed to me like a huge plus for OC but you're the expert.
What MB would you recommend then and do you think I should stick with a z270 + 7700K instead of a x299 + 7800X for OC? That's a lot of questions but I'd really appreciate if you find time to answer.
Cheers,
Vkno1
what about the higher end Gigabyte boards such as the Aorus 9 with the extra heat pipe and spreader for the vrms?
Nice troubleshooting mate, subbed
In 95% aller Fälle heißt Update=Selbstzensur
Ich finde gut was du machst!
first i just wanted to say thanks for the english vids. second love the content. keep up the awesome work.
Will you do the same testing for X399 based motherboards when they are released?
Thank you Roman for excellent update. Question though, when's the new Asus or other boards coming out with improved VRM Heatsink designs ? Do you have any info on this ? Also what are the motherboard models that have 8 + 8 CPU pin connectors for Skylake X that are out now or will be released soon that you know of ? Thanking you
End of July / beginning of August :). I know that Apex and Rampage 6 Extreme will be fine. Next week I will test the X299 XPower from MSI. Could also be good
Thanks very much for your quick reply. Appreciated :)
If I were to remove the heatsinks from VRMs and point a 120mm fan at them with reasonable CMF would it be safe/better than using the monolithic heatsinks that are found on new motherboards? I mean I would not risk destroying the VRMa, right?
Temps would be a lot better if you do direct air to the VRMs
Thank you for the answer.
I would reach out to Gigabyte and see if you can get one of their higher end boards. I currently have a Gaming 7 and it has a significant a heat pipe running from the VRM down along the back plate. Hopefully this added surface area can help dissipate some of the heat.
i have also done some reading with regards to vrm type/ quality before purchasing my x99 motherboard, i went with a irf single package vrm board( asus x99 deluxe ii) it has slightly higher vrm temps vs the gigabyte designer x99( both reviewed by tweaktown with thermal imaging on vrm) but i think the gigabyte board had better heatsink and also maybe more copper in board( gigabyte 20z copper marketing )... maybe motherboard vendors should stack up on vrm phases to help distribute the heat over a wider area wih x299.... great article btw
i dont know much about this so correct me if im wrong but how can they copy that z170 heatsink to the x299 when the space is taken by the additional 4 ram slots and the motherboards dont have much space for a bigger heatsink
It's the design of the heatsink too. The surface area is basically, non existent . These CPU's pull to much for the poor design.
great detail. any place we can get config details/bios settings for the asus boards used during your testing? I'm working to reproduce as well with delided 7900X CPU. Direct message/PM me if required. Thanks Daz..
Something I'd like to comment on. That post on the thermal paste roundup in 2015 was FRIGGIN amazing. I can't believe i listen to linus about ic diamond. That stuff is tough to apply, also conductive paste, hard to clean and performance wise is crap. I remember i used artic mx4 for some reason and it was like 26 C and when i had to take it off and place ic diamond it was like 28-29 C but thats on different days but still imo IC diamond is still crap. I bought the kryonaut paste and man, soon i'm going to repaste my cpu and the graphics card. I hope to get better temps than when i first bought these and pasted. Thanks for review even if its late.
haha thanks man :D happy that it helped you. That was probably the most time consuming review I ever did
Hallo Roman, super Video, wie immer :) Ich wollte fragen was du von EK-TIM Indigo XS - Intel 115x oder andere Sockel hältst und ob du ein How tu Use it Video machen könntest.
Dachte an einen geköpften i7 6700K 7700K mit Thermal Grizzly Liquid Metal plus EK-TIM Indigo XS - Intel 115x :)
Liebe Grüße :)
oh my god, now is an actual 5870x2 Ares holy wurst im collecting old awesome looking gpus and missing alot of dual ones :/
so on subject, what are your thoughts, are companies being lazy when it comes to this? i remember some really Nuts boards like Gigabyte SOC series that had alot of features for OC, but this "high end" boards are becoming more like a marketing stunts, i also saw on AMD side a Video by Buildzoid saying from an MSI X370 board that the VRM Desing was terrible
Great info. Im trying to get my Asus Rampage VI extreme running an i9-7920x up to speed. Really good stuff on throttling. I'm at 4.4 Ghz now and the cpu power fluxuates between 150 and 215 watts... throttle going on somewhere... using a Thermaltake AIO 360 cpu cooler and getting ready to upgrade to the EKW mono block and a custom loop.
How do you think the Asus Rampage VI Apex will perform without the VRM heatsink for pre ln2 testing? Testing such as RAM settings for benching to set up profiles... I will be starting with a I7-7740X thus no AVX512 right
Just put a fan over the VRMs and you'll be fine :)
der8auer cool thought as much. thsnks for replying and your great videos :)
Asus Ares Vegax2 on test bench? :D
Good video and thx for finding this out for us simple minds :D. I feel bad tho cause i just ordered a 7800x with msi X299 Gaming M7 ACK :(
Alex Ander DUMMY!
Couldn't you just take off the stock VRM heatsink and throw on a bunch of little copper heatsinks or if you could find a big one that would fit? That would surely increase the surface area and ensure better contact.
Cooling issues aside, do you believe the VRM on the ASUS Prime X299-A is good enough to handle an i7-7820X in the 4.8-5.0 GHz range?
No it's by far not good enough. You have to wait for high-end boards to push this high.
Do you know the VRM specs for the different boards? I need to stick to ATX, and all of ASUS' boards are somewhat lacking in describing their VRMs.
I do have a question which i am unsure you can answer. Did any of the board manufacturers contacted you to try to get to the bottom of the issue? Thanks and great work!
+Gil ARAÚJO they for sure know about it even tho you'd never get a public statement. But I'm sure we will see some changes from now. I had some tough reactions on the first video
I'd like a comment on that too...
Very clear.