This is fine...
HTML-код
- Опубликовано: 27 авг 2024
- Today's video is sponsored by ME! Get your JayzTwoCents Merch Here! - www.jayztwocen...
○○○○○○ Items featured in this video available at Amazon ○○○○○○
► Amazon US - bit.ly/1meybOF
► Amazon UK - amzn.to/Zx813L
► Amazon Canada - amzn.to/1tl6vc6
••• Follow me on your favorite Social Media! •••
Facebook: / jayztwocents
Twitter: / jayztwocents
Instagram: / jayztwocents
SUBSCRIBE! bit.ly/sub2Jayz...
Derbauer made a few interesting videos with similar experiments however he did delid the CPU and got about 20 degree improvements which is very impressive
that video was awesome!!
He also brutally murdered a 7700x in the process. RIP.
It's pretty straight forward, you will either use direct die cooling with custom parts from der8auer and enjoy lower temps or you will run stock. And with how Amazon returns work it's never been easier to get into de-lidding and direct die cooling. I have personally destroyed and broken 3-5 processors so far. All of which I just put the IHS back onto and shipped back to Amazon and got a full return on my credit card. I then repeated until I figured it out and did it properly and it worked. If you want your AMD Ryzen 7000 series CPU to be 75-80c under load you will have to use the aforementioned methods. Otherwise just accept 95c is the stock accepted temperature and there is nothing you can do about it without de-lidding and direct die cooling.
@@yakivpopavich ah yes, let's just commit fraud to learn delidding. come on, man.
@@pirojfmifhghek566 F
I had a Core2Quad Q6600 back in the day. Stock clock was 2.4ghz and I had it overclocked to 3.55ghz and it ran at 96C constant. If I bumped it to 3.6ghz it would thermal shutdown in 3 seconds. I ran it that way for 1.5 years and then put it back to stock and used it as a file server for 6 years.
i ran an old bulldozer 8350 at 5.0ghz for 7 years on a big AIO. but the max on those was only like 70C. and i ran it right up against that all day every day. it wasnt until after 3rd gen ryzen launched that i started seeing degradation, and it would only run at 4.6. so i retired it to wall art and got a 3950x
same!!!! not the server part the overclocking part, but the Q6600 was a beast for the amount I got for it (free99), and a gtx 1050 was good enough for my entry into pcs and pc gaming
Interesting... I think silicone degrades slower than people act. I think the need for safety margin is more about having margin for when the cooler gets dusty when stuffed under a desk and the AC is broken on a hot summer day, rather than silicon degradation being a real threat.
My Q6600 was running 3.4ghz for years with 70-80C. Additional 15C for +100-150mhz was just not worth for me as I had to bump up the voltage quite high.
lol i have a 9590 at 5.1 ghz with a 5.3 ghz turbo and it will draw mad wattage and spike to 51 . my x5690s will hit 82 easy . cpus just run hot
Other tech youtubers: let's use the most scientifically relevant testing methodology. 🧐
Jay: strap an air conditioner to the radiator with cardboard and tape. 😍
You're the best. 😂
If it's stupid but it works, it's really stupid?
@@black.phoenix. I've never said it is stupid.
@@black.phoenix. no, but it's really stupid if it doesn't
@@dayvie9517 it's not? Testing stuff that should work in theory is not stupid even if it doesn't work.
Who needs PBO2 curve adjustments when you can just introduce freon.
If Gif is Jif, then GN should be Jamers Nexus
A silver lining of this extremely toasty generation, we’re probably going to be getting a lot more custom loop content
Or alot of Copper,IHS modding
Ryzen 5 will come with an LN2 bottle in the box.
And more airflow focused cases!
AIO will cool it exactly the same.
Maybe intel and AMD have stock in EK and alphacool.
Watching you nerd out over a CPU is the reason I love your channel.
I like how he pointed out how snappy the window opening and other task were.... I always felt a difference wasn't sure if it was real or just me imagining
Watching Jay nerd out this much over a stock, out of the box, CPU is actually something that doesnt happen alot. He usually has those moments about other stuff like XOC, water cooling, GPUs. But who can blame him for loving this CPU? You gotta give props to AMD. Not only is this new chip a beast in terms of raw specs, the architecture is revolutionary. The way the chip tunes itself and the way he was telling us in his last video about how the system initial boot is longer because the EXPO ram is self tuning. After years of Intel making marginal improvements on the same damn 14nm process, AMD is really lighting a fire under their asses. I LOVE IT! This is such an exciting time to be a PC enthusiast.
So literally every PC channel
@@AndrewB23 Wouldn't know this is the only tech channel I watch and enjoy.
Derbauer pointed out the gap between the IHS and the die is quite large (was done on purpose to keep z height the same for cooler compatibility) making it harder to cool the CPU from the top of the IHS. Wendell said he used curve optimizer to hit like 5.8 GHz while keeping the voltages lower.
Not the gap is large but the IHS itself is larger than it needs to be
Optimum Tech did pbo2 + power limit and was able to preserve stock performance while keeping it at 60°c
The indium solder (the gap) is not that high, it's the IHS that's thick.
By gap, I meant the distance from the top of the IHS to the top of the die.
Debauer just delided the 7900X and was able to get 20c lower temps. I feel like this thick lid was a marketing choice for cooler compatibiity not an engineering choice,
Optimum Tech showed that you can easily drop the temperature by just limiting the power draw, maintaining performance. You can even have the option to get more power while keeping temperatures and power consumption. (All in PBO settings)
Would be great if you could confirm the validity of the procedure
In a way it makes a lot of sense that it does boost till 95 degrees celsius. It's kinda doing the same that laptop parts have done for god knows how long. Pretty sure linus once said as well that it was because an engineer said that "if its not hitting the thermal limit that means there is performance left on the table". Feel like that is exactly what is going on here.
Edit: so I actually saw some people use the eco mode for the 7000 series chips (eco mode locks it to 105 watts) and the performance was almost the same. They didnt test extensively and I'd love to see more testing but from the looks of it the high power draw is mostly a dick measuring contest while if you reduce the power draw using that eco mode the performance is still almost the same with normal heat out as well.
Exactly, we're seeing competition at work here. Back in the day you'd have to buy a special more expensive version of the chip for either HE (low power), or extreme. Now you get all rolled into one. Its going to give you 95% of your top performance auto-tuned for you based on whatever your silicon lottery / setup is. Or if you clamp the power to 65W-88W you get the efficiency model. I prefer this because like Linus said, it always sucks if you later find out you could have spent a couple days overclocking your CPU to get another 30% performance (at higher power draw of course and then wonder if you're hurting the chip lol).
but its terrible user experience, unless amd comes with some kind of offset to calm down cooler fans. As zen 3 user i can assure you its pretty bad for day 2 day usage as fans are ramping up and down for the smallest of workloads.
@@h1tzzYT get fan management and set your fan speed limits, will fork for all load events other than boot.
The last statement is wrong. Intel has turbo boost 2.0 on laptops that will hit a high ghz for a short burst until the laptop gets to 85c or above and that is the highest clocks the laptops usually get then clock down from there. This statement might have been true on bulldozer desktops from 2012 or maybe ivy bridge. and what going on with AMD is completely different and a first time for a company on mainstream computer desktops to do this.
@@jamestaylor3805 im obviously aware of fan profile management, but when your cpu has up to ~70c temp spikes in web browsing or other simple tasks its hard to dial it correctly.
So basically, now we need Summer/Winter PC builds.
I hate how true this is
Going to have to undervolt during the summer!
Just get a 7950x and 4090 for winter. It'll keep your house plenty hot. Especially if you water cool it.
Eco mode for summer.
I'll put my winter build in the closet with my winter clothes
I am absolutely calling them girafffics cards from here on out!
this is definitely an exciting time again... much like the pc space was between 2000~2009 but the major difference was back tehn people could afford to go out and experiment with different hardware without selling their organs
yeah not many people will be eager to experiment with CPUs that cost 500$ on boards that cost 400$
@@marcogenovesi8570 actually, chips are like 650-700 and boards like 500-600$.. shit is fucking expensive now..
Yes Jay we did learn something from this.
Jay does not keep the AC shield from previous tests and so has to make new ones each time.
Jay loves he modding.
I had to get used to that increase in safe temps when I went from 3900x to 5950x. Despite having a custom loop with dual 360 rads the 5950x just races up towards high 80's but it's happy with it. Just like in your test, the water temps aren't saturated, and as soon as the test finishes the CPU temp plummets back down to mid 30's. I even repasted the heatsink as I was worried the block wasn't drawing the heat off the IHS. It's just down to physics, if you cream that many cores working that fast in such a small space there is a limit to the thermal transfer from die to IHS to heatsink. The 5950x was at the limit, the 7950x seems to redefined the limit.
Same experience I had with my 5950x.
i think these new ryzen cpus behave a lot like GPUs. Until they run out of temps or power, they'll just send it, pedal to the metal.
@@TheHighborn yup. They seem happy with it just the human element that needs to adjust to the increased temps. Will be interesting to see when jay gets a bios that allows voltage changes as on the 5950x I spent a lot of time dialing in the curve optimiser and ppt, EDC & tdc to allow pbo enough room to play but not so much power that it throttled. If I could keep it just under 90 degrees it boosted really well but if I left pbo at what the motherboard gave it, it just threw voltage at it and it performed worse. With AMD have really refined in on 7000 otherwise I wouldn't mind betting there is more performance to be had.
I have a copper air cooler and the temps are very much nearly the same on a r5 3600 at 4.20 ghz. maybe 5 degrees cooler and with a copper water block instead its still 70 over sustained load. Shaving the ihs down is going to be easier with no pins on the bottom . its shouldn't have to be a thing but if you want that little bit extra ...
It will be really interesting to see benchmarks between a decent air cooler and a decent water cooling solution to see how much difference it makes now that its determined to run at 95 no matter what, and see what the clocks end up doing
You can do a manual OC up to 115C. So, ya; the new normal is 95C.
Edit: It boost like a video card so it will always try to hit the set TjMAX. If you cool the crap out of it, it will boost higher but still try to hit 95C. You have an LN2 pot, it should boost to 5400 on all cores with stock BIOS settings.
Second Edit: To be clearer; the throttling at 95C is is set in the BIOS by default and it is not the thermal limit of the CPU. The CPU will start thermal throttling at 115C.
This is really stupid. The fan noise is going to be absurd
@@ksenchy i disagree. I mean gpu fan noise aint unbearable either
@@ksenchy it'd barely hit 60-80°C while gaming (depending on the game). No game can max out even 8 of those monstrous cores.
@@ksenchy This CPU will probably need a 360 rad/AIO to run at stock speeds and be silent.
I was actually coming here to point that out. Lol
I think we're getting to the point of computers where if you care more about fan noise, you're gonna have to take a hit to performance. You're gonna have to disable the auto overclock, allowing you to keep the temps where you want while also having control over fan speeds.
Will there be software tweaks for this? Probably. But this is basic new generation growing pains. Things aren't going to completely perfect out of the box. Same for Intel or Nvidia.
If you're someone who wants a silent system, you're gonna want to wait. If you care more about performance, then the fan noise isn't really gonna matter
DerBauer did a really nice deliding video, showing that the heat transfer through the IHS is the main reason why temps re reported that high.
It's pretty straight forward, you will either use direct die cooling with custom parts from der8auer and enjoy lower temps or you will run stock. And with how Amazon returns work it's never been easier to get into de-lidding and direct die cooling. I have personally destroyed and broken 3-5 processors so far. All of which I just put the IHS back onto and shipped back to Amazon and got a full return on my credit card. I then repeated until I figured it out and did it properly and it worked. If you want your AMD Ryzen 7000 series CPU to be 75-80c under load you will have to use the aforementioned methods. Otherwise just accept 95c is the stock accepted temperature and there is nothing you can do about it without de-lidding and direct die cooling.
@@yakivpopavich Hello Amazon
@@n00buo And goodbye CPU's I've broken lmao, all returned and were processed within 48-72 hours with a full refund. Rinse and repeat until I was successful. Jeff Bezos can kiss my ass.
De-lidding an expensive CPU is not something 99% of people are going to want to even try to do.
@@Ryan_DeWitt Exactly.
Even in the more Enthusiast oriented marked is de-lidding niche.
AS Derbauer measured in his vid - most probably the cause for at least 10-12c higher temps is the THICKNESS of the cpu Lid/shield/top. Maybe - trying to lap the excess material would help without the need to delid.
The word you're looking for is IHS
Gamers Nexus also mentioned possible improvement with lapping
But yet a Hyper 212 is perfectly fine.
From the thumbnail I expected you to provoke an RTX 4000 GPU to light up 🤣 Also, your iFixit ad skit would be perfect for that 😂
Adam Savage describes cardboard as a gateway product to making.
So to use ananalogy, the air in the room is the cooling capacity of the system. A room heater is the processor. If you have that cranked to the max and measure in front of the heater it'll be a high temperature compared to the room.
Edit: in short, the processor core is getting really hot really quick but since it hasn't equalized with the cooling capacity of the heatsink (the metal in an air cooler or the liquid in a water cooler), it can drop really quickly as shown in the video.
Thats a lot of words to say 'ambient temperature' :P
@@MagGray haha
Law of thermodynamics, heat only flows from hot sources to cool unless you add work to the system. Moving water around in a closed loop isn't doing work either, that simply moves the heat to the radiator which also is cooler than the cpu & liquid.
@@Mike__B Doesnt the liquid move the heat faster to the ambient air than heatpipes in an air cooler?
@@L1ft0ff I'm not sure how fast heat pipes move their liquid/vapor around, but in a water cooler you tend to have a LOT more fluid than in heat pipes and that means a much greater heat capacity so you can pull lots of energy to the radiator where it can then transfer that heat to the radiator which then gets blown out of the case via fans. It's a little more complicated than saying one technology is better than the other, but water cooling definitely has the potential to be better at cooling than "air coolers" (with heat pipes), I believe the big limitation with heat pipes is they can't be terribly long because the vapor in them condenses quickly as it gets away from the heat source.
After watching some other videos before this one I really appreciate the camera quality and how good the screen stabilization is. Great work to all the team over at JTC!
It's under amd overclocking and precision boost overdrive, set advanced and curve optimizer
Totally get you with the CPU heat, it's easy to understand but hard af to explain. I wonder if it's a thermal lock on the CPU or the board. Great work Jay!
I see the fact that Ryzen 7000 doubles as a heater as an absolute win, now I don't have to buy one to stay warm in the winter. lol
And a nightmare in the summer lol, my current PC was melting me in my office this summer, I can't imagine what adding all that extra heat would do
What about using in the summer???
Perfect for those in Cali
It's pretty straight forward, you will either use direct die cooling with custom parts from der8auer and enjoy lower temps or you will run stock. And with how Amazon returns work it's never been easier to get into de-lidding and direct die cooling. I have personally destroyed and broken 3-5 processors so far. All of which I just put the IHS back onto and shipped back to Amazon and got a full return on my credit card. I then repeated until I figured it out and did it properly and it worked. If you want your AMD Ryzen 7000 series CPU to be 75-80c under load you will have to use the aforementioned methods. Otherwise just accept 95c is the stock accepted temperature and there is nothing you can do about it without de-lidding and direct die cooling.
@@yakivpopavich or just not buying it period
I think the thing that bothers me most about how hot this runs is that my fans are going to ramp up to 100% every time I do anything on my pc and it's going to make my system very noisy.
Nobody is talking about this!
We went straight from sound being a frequent topic of discussion and priority for a lot of people, to you now have an F16 in your room on gaming loads.
Adjust your fan curves then
Or just use Eco mode and set it to 105 watt tdp. Gordon over from pc world did that and then ran it through benchmarks and the results were ridiculously close to stock all things considered.
It only gets to 95c under 100% all core workloads or when your consistently at 80%+ utilization from what I have seen. Smaller loads have much more tame temps.
Plug them to molex and enjoy the music like a real man
I do think its worth remembering that in gaming the cpus don't go to the max like they do with rendering.
Your CPU can create bottlenecks long before it hits 100% Usage.
Jay mentioned that in the video.
Run ZEN 4 at 105W ECO mode - problem solved.
7950X even in ECO 65 in production is faster than 12900K so...
Don't waste your time/money on these chips. Wait for the 3D vcache versions; the 5800x3D is on par with the 7950x at almost half the cost. Plus you need a new board+ram+cpu for AM5, so around $1200-1500 (in Canada anyways). 5800x3D is the way; 3D vcache is the way. Hold out for now gaming enthusiasts!
AMD has literally marketed their new chips as "best for gaming" so to comment in ignorance that 7900x and up are "for productivity" is pretty funny
On par where exactly? Gaming?
@@MaksKCS Yes, gamers think they are only one's who buy cpu's lol. It's funny.
Exactly what I’m doing. I will wait for the current 5800X3D to drop in price some. It will be perfectly fine for several years with the games I enjoy.
@@MrBashem Right? It gets so old to only hear about gaming.
Smart move, those versions are going to be awesome.
Another thing to consider is the heat spreader on Zen 4 is quite thick. It stands to reason that it takes a while to conduct all the heat from the CCDs to the cold plate of whatever cooler you're using. Longer than it would have on Zen 3. This is a side effect of AMD's decision to allow backwards compatibility with AM4 coolers and mounting brackets. For every design decision like that, it causes a trade-off of some sort, and in this case it's less efficient heat transfer from the CCDs to the heat spreader to the cooler.
I wonder, since the heat exchange is from above, if they couldn't just make the base ticker. It would transfer less heat to the board, but the heat spreader would be more efficient.
@@sysbofh The reason the base is thinner this generation is because they moved the SMDs from the bottom side of the CPU to the top. That's why the heat spreader has all those cutouts in it and just little legs gluing it to the CPU substrate. That is another design choice they made in order to allow for more data pins in the socket while still having the same general dimensions as socket AM4. Had they left the SMDs on the bottom and stuck with a thinner heat spreader it would have made the CPU socket noticeably larger and forced everyone with an AM4 cooler to at least buy a AM4 > AM5 conversion kit or an entirely new CPU cooler to ensure proper cooler contact. This on a new platform where the adoption cost is already high due to requiring DDR5 and a new motherboard in addition to the CPU itself. So all in all, there is a perfectly rational logic to it. And realistically, while the CPUs do run hot, they are stable at those speeds and at those temps. So as long as you can live with those temps helping to heat your computer room all year round, it's really not that big of a deal.
you should also check the ECO modes
5:30 it is localized heating. If you used HWINFO you can even see the L3 temps, and considering it is on the same die as the cores, with 4 cores sandwiching each block of cache, watch the temps. L3 will be in the 50's while the cores are in the 90's. It is all about the inability to get the heat effectively into the cooler through the IHS. Same as 7nm, just magnified further.
I would love to see a video of testing cooler on those new CPU, starting with a 212 up to custom water cooling to see what kind of performance you get when upgrading your cooler with the new AMD 7000 CPUs.
I second this!
Third! I just upgraded to 5800x with 212 evo and ran curve optimizer. Would be interesting to see it with 5950x.
same, especially the "kings" like the Noctua D-15 or the Deepcool Assassin or whatever else and if they're still considered equal to mid-grade AIOs
Im honestly interested to see what happens when i plug this thing into my custom loop. When i designed it initially, i overkilled the amount i should have needed by quite a bit since it included CPU and GPU+Active backplate. Seems like i might be okay if not making my rig into more of a space heater than it already is.
You also have the versatility of cranking up fan and pump speed. 1250rpm is usually not too loud for good set of fans - pump varies. But you shoudln't be worried about the cooling capability of a good custom loop, regardless of the components.
At this power draw for a small area, I feel the limiting factor is heat transfer of the thermal paste, and heatspreader. If you do a direct die cooling setup with liquid metal you can drop 20C.
It's so cool watching Jay play with a new CPU like a test fighter pilot getting the go ahead that he may ram the entire plane into the ground as long as they get a full test of what it can do.
I think what a lot of people don't really understand is if a device is generating 250 watts of heat it doesn't really matter if the hotspot temperature is 95 or 80 it's still the same amount of heat being generated and your room won't warm up any faster because 250 watts of heat = 250 watts of heat. Temperature is like voltage, if I have 120 volts at 1 amp it generates 120W and if I have 12v at 10 amps I still have 120 watts. What is missing in the equation is the actual density of the heat because all they are measuring is a tiny spot on the die. That's like lighting a candle in a room, while the temperature is higher right next to the candle the overall room temperature doesn't really change because the density of the heat is so small
What really needs to be done is monitor the water temperature on a new 7950 drawing 220 watts and a 5950 also drawing 220 watts. The fact that it cools down so fast tells me that actual density of the heat isn't any more than a normal 80C system drawing 220 watts. What I suspect will happen in this experiment is the water temperature on both system will be roughly the same even though there is a fairly large difference between the hotspot temperatures after all 220 watts is 220 watts
To cool this CPU properly is really no different than any other CPU, concentrate less on temperature and instead how many watts of heat your cooling system is capable of dissipating. That is what really counts in the end
exactly, more people need to pay attention to the watts instead of the temerature!, im looking forward to testing this with the eco mode aswell
These kinds of temperatures for prolonged periods kind of have me wondering more about the degradation of motherboard components than the chip itself. Are we going to start seeing cooling fans for the back of the motherboard socket?
Also, the second you said "Open the browser" and clicked, my video started buffering. I died laughing 😆
I doubt the temps can transfer well enough through the the chip pcb itself to affect the motherboard. Older ryzen pins would be falling off left and right if that were a thing. Especially since the heat is instantly passed through the ihs. Things generally try to take the path of least resistance. You're not gonna see lightning try to bore a hole through a wall to get to you, it will jump to whatever metallic thing is nearby.
@@Wheres_my_Dragonator I get that, but if you've ever stuck your finger on the back of the PCB of your GPU, you know that gets pretty toasty, even under moderate loads. I'm not specifically suggesting cooling the back of the board is for helping with CPU temps, just overall system temps in general.
Yeah I'm curious to see what the longevity of these systems will be, esp when you undoubtedly have people that aren't cooling them correctly
@@Wheres_my_Dragonator It doesn't have to transfer from the CPU to the mobo, the mobo itself will heat up (unless it is a very expensive one with overboard VRMs) from the 100W+ loads all the time. The VRM will have to handle the increased loads and the CPU socket will have to sink the heat from the CPU and all the amps rushing through it. In 8core + territory we are talking 200W+ now, this is bonkers..... b350 had 1 or 2 boards that could handle 100W, b450 had a few more and b550 can generally run it just fine, but b550 was more expensive, because of that. The next b650 will have to be even more expesive or caps will start blowing up. And probably forget about cheap electrolitic caps, they'll have to use the more expensive solid ones. There is a reason 65W CPUs were/are the norm, with 90-100W ones being for "enthusiasts", more than that is FX9590 territory and where Threadrippers used to be. When does this end ? Next gen 300W ?
Would definitely like to see these tests done with a lapped CPU. My concern is more about dumping too much heat into the room in the summer with the 7600X seeing as AMD has abandoned the 65W specification - so could you show more about limiting the power draw? Thanks!
Yeah, no worries. He's going to make a video just for you
You could always get a 5000 Series CPU Limiting the Power Draw will Limit the CPUs Potential so if the Temps are your Worry go for a CPU that isn't Designed to go to that Temp
I love open loop. Its all I've ever used.
That heat should be made to do something useful too, like warm bagels or dry boots and gloves or even generate some power. You've got a 3-d printer...
Isn't the chip getting to 95c but the IHS is not getting anywhere near that temp because it's a lot thicker for cooler compat, that would explain why Der Baur got 20c off when he delidded the cpu
That's a massive oversight
yup, you're completely right. pretty sure jay was trying to "dumb" it down for the average watcher though
I wonder what the temp of the liquid is compared to the 5xxx chips and what would happen to the temps if no cooler was attached (that might be a tad evil though)
It's pretty straight forward, you will either use direct die cooling with custom parts from der8auer and enjoy lower temps or you will run stock. And with how Amazon returns work it's never been easier to get into de-lidding and direct die cooling. I have personally destroyed and broken 3-5 processors so far. All of which I just put the IHS back onto and shipped back to Amazon and got a full return on my credit card. I then repeated until I figured it out and did it properly and it worked. If you want your AMD Ryzen 7000 series CPU to be 75-80c under load you will have to use the aforementioned methods. Otherwise just accept 95c is the stock accepted temperature and there is nothing you can do about it without de-lidding and direct die cooling.
@@yakivpopavich bro do not admit crimes on the internet 💀
test the CPU with diffrent power limits like 65W, 110W, 150W etc to see what kind of performance you can get out of it and the temps on that aswell
Anandtech showed, it outperform stock 12900k and 5950x at 65W and losses about 20% multi-core performance of the original settings.
@@superneenjaa718 tyvm for that reply!
For any cooling method, the temperature gradient across the coolest part of the cooler and the middle of the chip thickness are what determines overall cooling capacity.
Any extra material, distance, whatever increases the physical width of that distance.
Lol I love the cardboard craft! In the fabrication business we call that CAD... Cardboard Aided Design.
With Ryzen 5000, Curve Optimizer was the way to go. Don't trust Ryzen Master's auto tuning it does. It'll zero in on offsets that are way too high. They'll be stable in the short term, sure, but they will cause crashes over time. Although the values it finds could be used as a good starting point for manual tuning. But the curve optimizer lets you undervolt (or overvolt, but that's not what most of the community used it for) each individual core separately. Doing this can really help control the power consumption and heat generated by the CPU. Then after that you can mess with the PBO settings (PPT, EDC and TDC) to use the extra thermal and power headroom you got from the Curve Optimizer to bring up the overall performance. It's definitely a very different style of tuning than just telling it to run higher clocks and increasing the voltage as necessary. I feel it is better for an every day use system though. It just takes a really long time to dial in the values. It took me like 3+ weeks of running stress tests to find the rock solid stable offsets for all my cores, but it was worth it.
Its great until the current agesa still has a bug to where it wont boost over the stock edc limit so you cant even use pbo to oc along with uv
Shulk, can you please edit this to paragraph break? It’s too long for some to read/ understand.
I tried using curve optimizer for my 7950x and it gave such a massive undervolt Windows wouldn't start
I think you swerved into it, but if they're recommending AIO water cooling for normal CPUs, either motherboard makers will have to add fans to VRM cooling, or we'll have 17 pounds of aluminum and heat pipes just to cool power delivery.
Does a top mount radiator or rear mount exhaust fan not provide decent VRM cooling when using an AIO?
@@Yahtzee9820 I think thats a question for tech and testing channels; some say it isn't because there isn't as much air or dedicated cooling for those specific components other than a large lump of metal and whatever happens to be moving through the case. However not all cases are made equal either in terms of directional flow. If you remember back in the day some board makers would have tiny little fans on chipsets specifically.
I wouldn't call the 7950x a 'normal' cpu.
In the UK, GIF like JIF was not common as Jif was a cleaning fluid for kitchens and bathrooms.
To all mon and dad, look at the fans. They are still spinning! LOL!
Because of a small area for cooling, there is a slower thermal transfer, therefore it goes immediately to 95°C while cooling system is still cold...but that's normal ..
It’s impressive that it can maintain max boost at such high temps. Above 70c makes me uncomfortable cause I see my clocks decrease but if 95 is fine then so be it. The more important thing is wattage and power efficiency.
Use Eco Mode. Wait for Eco mode testing.
Apparently tjmax is 115C for this gen.
@@chaon93 It's 95 degrees C tjmax for Zen4.
Just set it and forget it! Lol! 😂😂
WE WANT TO SEE THE UNDERVOLTING when you get it fixed. And the efficiency the Eco mode at 65w and 105w. Can't wait to see ln2 as well.
Hey Jay, I believe it's because under full load, the auto overclocking will just keep rising until it hits that 95 degree safe limit. It only stops increasing its speed once it hits 95 degrees. It's how AMD decided to control it on purpose. It makes it easier for the user, and also safer, as some people bring their systems to the very edge, and run it like that all the time, and with it shorten the life of their processor, often significantly.
Source: Gamers Nexus.
I never heard gamer nexus say AMD will shorten their cpu life running at 95C. Lol another intel fan boy
Saw derbauers video yesterday where he delid and man.. I'm thinking about that.
Maybe even lapping the IHS.
Ordered a 7950X
I'm still running an overclocked 4790k and I'm doing just fine. This CPU was the best buy ever.
These temps bother me. I can appreciate that the entire loop is not at 95C, but even with that just being a spot reading that extra heat has to go somewhere - for me that's the air of the work room I'm sitting in. I don't relish the idea of my workroom being so much warmer than now - I live in Australia, and whilst this has been the coldest and wettest winter in a decade where I am, summers really don't need any extra heat.
Wendell over at Level1Techs found that just by reducing the voltage ⚡ of 7950X it'l boost to a crazy 5.7GHz on all cores of 1 CCD and 5.5GHz on the other.
Supposedly, De-Lidding drops the CPU temps significantly.
You can't delid these unless you are insane. You need to use a frickin razor blade to cut the IHS off, as it is soldered on.
der8auer just did that and it does
@@DailyCorvid how does that make you insane? It’s super easy
@@DailyCorvid Negative derbauer posted a video already and is going to make a tool for it
pretty sur Der, De-Lid'd and direct-die cooled the cpu. (and it worked, hello warranty)
If it's gonna self-overclock until it reaches a target temp, there needs to be a bios setting to adjust that target temp. 95C is just ridiculous. I'm sure there's plenty of low-noise enthusiasts that would set the target temp to 75 or even lower in order to get their system fans as quiet as possible, even if it means reduced performance.
What many people seem to fail to understand is that the temperature that a part is running at has nothing to do with how much it heats up the room it's in. Power consumption is the only thing that matters. Whether you try to cool a part with a tiny cooler that lets it run at 100C constantly, or go overboard with your cooling and it runs at like 50C, the heat dumped in the room is the same.
It would be nice to have a easy bios setting to set a limit. Lose a little performance and have your CPU for more than a week.
Enable eco mode and you have 12900k performance at 65w.
What you're finding with the temp being reported is the same on even my 3900x. gets pretty toasty under big load but cooling down to 40's is quick. like within seconds on my 280mm aio
Jay, please do videos with different Aircoolers, not only Noctua, but some cheaper ones on 7600X.
I love how we are reaching this point that booting up the bios is easily the longest process in a computer.
Curious to see if AMD takes this knowledge to heart and adjusts the IHS and contact between the chips and the IHS.
Seems like it's less about it being a hot CPU and more about a manufacturing defect that could be adjusted for by changing how the die contacts the IHS.
Very low likelihood they change fabrication at this point in the release cycle.
In his past video, he said the AMD rep he spoke with was the one that said 95C was intended
Hurray for those who don't have jobs 😎
Hi gamer
Who? Jays2cents? This is his job lol
@@Gamingtechgg With all the sub-par games coming out for a while now, consider me an ex-gamer
@@Deadsmegma No. All the people who have ample time to watch, seconds after the video published
I'm watching this while I do my job...
Honestly, now that we are having the theme of going so high in frequency and power consumption, I would love to see a configuration for ultra or high settings gaming but trying to reach the smallest power consumption possible. Like pushing for efficiency rather than just boost everything and overclock. Given the times that are coming with energy cuts and what-not... that could be a cool video theme.
This. This. THIS.
Beyond the cooling concerns for the CPU itself-...
My man, that heat's just going to go to my room.
I undervolted and overclocked my 3600 to 1.25v and 4.2ghz and didn't notice any performance issues, stock would hit 1.4v at 3.8ghz even with the auto overclock stuff on
4.2ghz is considered a slow clock speed these days lol
@@kyles8524 got my 12900k at 4.5. -7 but hard to go up and try to render in blender, blender takes more than gaming
Radeon Monster Profile' tool for RDNA 2: RX 6800 XT beats RTX 3090ti.
I think we just have to get used to the 95°C as a "norm". It makes a lot of sense, when you think about it. AMD guarantees the CPU can run all day every day at that temperature, so keeping there during a test or while running a load guarantees so much more headroom with the frequency because you are already there with the heat.
The aim is to increase part life span as may components are more limited by number of hot-cold cycles than by 15 degrees peak
Thermodynamics 101 - Skeleton case without side panels = optimal cooling for almost all configurations. Antec, Win X-Frame, and Thermaltake have some great designs.
I trust the chip at that temp (95c) but what concerns me is how the sockets and motherboards in general will hold up. Not to mention the added risk of AIO failure due to the constant high temps. We all know heat is a computers worst enemy so I guess only time will tell. Im more interested in 40 series NVidia atm but these chips do seem to be quite amazing.
Water. Water is the worst enemy, heat is required.
@@DailyCorvid lol yes very true.
the fluid only reached 30C its cooler than intel parts were making the fluid a few years ago stop worrying
It probably won’t affect me regardless as I haven’t used AMD since around 2012. Back when I had no money. I prefer Intel\Nvidia. I’m still using my old i9 9900k and i7 10700kf. Im keen to see the Raptor lake figures now. I’m fairly certain my new build will be raptor lake with 40 series nvidia. But if Raptor lake is a bust then I’ll be going back to AMD.
I love the demo of how fast the system is to close down! - of course if the temps get much higher it might shut down even faster whether you want it to or not!
High tech, low tech, and stupid brilliance. The perfect JayZ vid, haha
The infuriating point to of which goes with the 7.0 series and nine point 0 series is the incasement of disconcerting information on or in the creation of arbitratious a.i.
I love seeing Jay so glad to be living in this era of pc technology and able to get his hands on it. I'm also excited for this stuff but I have no money. Still, it's exciting to see this stuff.
5000 series CPUs are going for dirt cheap. Theres a combo on Amazon for only $295.
This reminds me when i first got an AMD cpu, the Fx8350, and a lot of people used to say AMD wasn'T worth because it got crazy temperatures but it wasn't powerful to justify it... good ol times...
Yep. Those were good times. They were so cheap.
Memes aside, have seen Dr Ian Cutress's info packed live stream and all I'll say is wait for the Eco mode testing.
So from a previous video you made you mentioned about having the water cooler cables down at the bottom cause eventually the water will evaporate or the air bubbles may make their way to the CPU cooler or whatever. Well the computer I got a long while back from NCIX from their California branch a few years before they shut down they put in the radiator with the cables near the top. After seeing your video and finally having some time and blowing out all the dust out of my computer finally got around to moving the cables down to the bottom while making sure to try and get all the air up into the top of the cooler. So far seems to be working just fine.
Delid, enjoy ~20C lower temps w/direct die cooling. Easy peasy. It's never been more simple to delid and no reason not to.
Do you think now with the way this cpu temp is set up, it will effect the temp on the gpu in a single custom loop? Thanks for the video
Unlikely, it isn't running at 95C because it can't get rid of heat fast enough. More that it doesn't care about the temp. So the waste heat wattage going into the loop won't be much different than any other equivalent TDP drawing cpu.
I would try windows 10, I have a feeling that windows 11 is the issue.
Has it been confirmed that Zen 4 works on Windows 10? I can't find any info on that anywhere.
I'm sure it's not just me who appreciates your honesty Phil. I love your reviews, you are my go-to for no BS info so thankyou.
I feel bad for my darkrock air cooler trying to cool that thing down.
Maybe they are aggro going for 95 so our PCs burn out. No more keeping pc for over 10 years.
AMD really been making the last 5 years super exciting, by bringing the competition and just even things like chiplet technology.
Can't wait to see the chiplet Gpu's
What? Intel did chiplet design back in P4D and early version of C2D and C2Q, AMD shamed Intel for it.
Only thing Evil Su bring to AMD is Intel-Nvidia's copied-pasted strategy, but she made it much worst.
A good analogy about this temperature behavior would be the cooling system of a car engine. It's designed to keep the coolant in the 80-95°C and oil at 90-120°C no matter the load. Even idling your car eventually will hit these temps w/o any load. Doing top speed run on the highway also would keep them temps if everything is working as intended.
No space heater needed, just air cooler without fans.
Cooling on the VRM is essential because the 1920-vintage design is only about 60% efficient.
jay saying "eiste laido arride" makes me laugh
5:18 I think you can call it as generated heat - as equivalent of the work that the cpu was actually doing by the time that the system reports its temperature reading. And what the AIO is actually cooling is the residual heat that was dissipated from the IHS to the cold plate of the AIO. And that the AIO's job is to continue keeping its cold plate cool so that heat dissipation process keeps on going, keeping the cpu as cool as possible. If cpu's IHS and AIO's or any other cooler's cold plate becomes equal in temperature, heat dissipation stops. And cpu cooling stops as well.
Was fun watching. Note to anyone, physics is a very harsh mistress and there is no free lunch, every watt of electricity pulled from the PSU is exactly one watt of thermal energy radiated from those transistors that used that electricity. Ohmic resistance inside transistors is literally converting electrical energy into thermal energy via (Power = i^2 * R). That means this new generation of CPU's are incredibly inefficient out of the box by design, they are discarding efficiency in the attempt to maximize clock rate and therefor thermal output.
Add in a graphic cards heat in a complete water cooling loop and you will not need to bother with heating the bedroom with hours of gaming. Surly in a case we will see higher than 95c especially on air cooled.
Is there any info on the ECO mode? it seems that we can (forcibly) limit it's power usage to old Opteron style to 105 and 65 watt with some performance loss, but still make more sense and not having to warm your room or even make it acceptable in warmer climate part of the world.
With how large GPUs have become I think Girafphics cards is a viable term 🤣
To bend double-wall corrugated cardboard, cut through the outer layer of cardboard on the bend line, it will bend there easily.
Hey Jay! Make a batch file to restart directly into your bios without mashing the keyboard! Run as admin!
shutdown /r /t 0 /fw
Thumbs up for Jay!
Two out of the five stores are already out of stock. Looks like everyone has been waiting for this lol tbh I didn't even think they were coming out yet.
Two things to note that are important
1) these things auto clock as high as they can go with the 95 degrees as an actual target for Max performance.
2) the feature is entirely optional - meaning you can deactivate it and run more comfortable temperatures.
I see options at the hands of the consumer and one or the other route should end up satisfying both opinionated sides. I don't see a problem here.
You don’t improve your cooling just to low down the temp, rather improving your cooling will increase the cpu clock which results to performance. It is designed to bump up the cpu clock to reach 95C then from there clocks are controlled to maintain 95C. It’s like “if temp > 95C then decrease X clock else keep increasing X clocks”. So if you want performance, beef up your cooling. Air coolers are fine but some cores will be at lower clocks. Either cooling the cpu will always run hot at full load.
What amd did was the reverse of the conventional way we squeeze performance out of the cpu. I think this is good because consumers are assured that your computer won’t crash. This is unconventional way of overclocking your cpu without ever crashing. This time, overclocking is done through the cooling not the bios.