*Official Benchmark Thread* - Post it here or it didn't happen :D

[cylix]

Really want to see sapphire toxic version of 7900xtx, maybe in white😂

 

https://videocardz.com/newz/amd-radeon-rx-7900-gpus-are-designed-to-scale-up-to-3-0-ghz

 

[johnksss]

Okay, so I can now boot 8000 on the Hero Z790, but not fully stable and am stable @ 7800.

 

[jaybee83]

22 hours ago, Rage Set said:

 

No recall incoming, but a vBios nerf is. They are going to lower the power draw, watch and see. 

 

watch me not ever flashing such a performance nerfing vbios! ill trust the community to come up with a non-nerfing solution, as always 😂

 

22 hours ago, Ashtrix said:

 

Class Action Lawsuit is brewing.  Perfect. This is what Jensen and his gang deserves for ruining the PC market pricing since Ampere to now Ada and then forcing the trash 12VHPWR standard.

 

I do not even know much on electronics part until I watched Ahoc and revealed how garbage POS this standard is. The maximum for the new 12VHPWR is under 684W and the standard retail is already pushing it to 600W that means already 87% of limit on the connector. Meanwhile the 8 Pin standard is 50% since it has more headroom for it's max limit as it's standard at 150W, but in reality it can handle 306W. So the safety margins on the 12VHPWR is just 14% and the 8 Pin is 64% considering worst power supply wire gauges. That alone speaks volumes for this TRASH connector pushed by everyone at PCI-SIG and Nvidia and all companies who did a green light on this into Retail channels.

 

AMD Radeon R9 295 X2 with 2x8Pin connectors with 500W power.

 

 

More on the above GPU

 

 

Imagine designing a new GPU in 2022 vs 2014 GPU which broke the compliance even. And losing it out...TRASH.

 

Junk Nvidia over engineered BS, to add insult the pricing scam for 4 damn years. 3080 -> 3090 100% price increase for 15% performance they got away because mining crap. Now they use it to make 4090 price acceptable. Meanwhile 3090Ti is now at $1000 MSRP which is actually 10-15% faster than 3090. Add the Ultimate Trash though which is Frame Interpolation, fake frames not even rendered by GPU inserted for that muh numbers + latency !!

 

The RTX 3090Ti despite using same power connector as 4090 16 Pin 12VHPWR got saved because it doesn't have Sense Pins variables they are shorted and disabled permanently so of 16 Pin only 12 Pins are used. Second the Power circuitry of the RTX 3090Ti has load balancing land splits it into 3 150W type connections while the hot fire hazard Ada 4090 does not. RTX 3090 is not effected because its 350W only and also had a slanted connection plus it's only 2x8 Pin connections.

 

preach brother!

 

17 hours ago, Ashtrix said:

Well I lost respect to this Steve.

 

 

AMD Radeon coverage is not well rounded and cringe. Goes off on how AMD misleading with fake 8K, yes it is FSR obv it's not real it is fake indeed and they got access to slide deck because these Techtubers go on a trip sponsored by all corporations, which is what TechPowerUp and Anandtech post. Anyways so I went and checked his Nvidia coverage, well not a single word on DLSS green bars that Nvidia showed. Here's his Nvidia coverage for Ada RTX 4000 cards.

 

Then how AMD did a jab at Nvidia on power adapter.. and it's not good in his own words lmao. Well AMD is using the free opportunity and the 12VHPWR is a joke no matter what. Saying, how we can also use native gen 5 power adapter (yea just ignore the adapter in the box of a $1600 card and buy others, what about non-native ATX3.0 PSUs ? ah lets forget it right). Boy how well it aged like milk in a day, when we got native 12VHPWR connectors melting out.

 

Also DP2.0 marketing is also not good as to him and somehow misleading by AMD as we do not see any frames... Hmm a $1600 (4090) card with no DP2.0 heck even Intel ARC has it. AMD X670E boards support it too so do their Ryzen processors. Going by that logic, why have PCIe 5.0 X16 lanes on the dGPU PCIe slot on latest mobos ? we dont have GPUs nothing utilizing them. Since when having more is bad..

 

He should have used this slide instead of wasting time on about fake FSR trash 8K marketing fluff nobody cares about 8K unless they are really pushing it native or using it to bench an SLI rig. Anyways here's the slide LTT mentions it as well btw. This is the exact slide which useful. Not that FSR benchmarks garbage that AMD showed. Same with Nvidia they had only 1 slide which was good it showed that 4080 12GB was losing out to 3090Ti.

 

 

Nvidia slide btw which was worth..notice the 4080 16GB perf vs 3090Ti there.

 

 

A small note - This is me who owns an Nvidia GPU and always owned them only. However that is not going to be in the future though.

 

As for Performance estimates here's a LTT video on the charts (ignore that clickbaity thumbnail lol)

 

Skip video and here's the estimates. Note these are Raster ONLY. He shows the RT ones where AMD falls flat totally as expected. Again except Metro I do not see any worthy title out there. Portal RTX, Quake RT are maybe a couple if you really want since both are fully pathtraced like Metro RT only edition.

 

 

 

That 7900XTX will kick the hell out Nvidia flagship part for 60% less price. RTX 3090 TI vs 4080 16GB there won't be massive gains it will be limited as per Nvidia's own slides only reason left is to spend more cash unless they care about RT and DLSS3. This card will KO the 4080 16GB. And fight against the 4090. AMD will release a 7950XT to fight it again once 4090Ti gets released.

 

Finally that awful Frame Interpolation type thing. AMD Is saying they are working to make their FSR3 across other cards than RDNA3 and say it explicitly by that idiot Frank Azor, note AMD is using AI accelerators in RDNA3 and not RDNA2 so if they use it RDNA2 won't but they are hinting it will get.  Also no confirmation if its Frame Interpolation or not.. hoping it is not.

 

 

Imagine that fake FPS booster doesn't use anything AI but also works on Ampere, massive kick in the nut for Nvidia for gating that tech to Ada Lovelace.

 

of course those are 1st party data for now, but its surprising to see AMD placing the 7900xtx as a 4080 competitor and not 4090. the latter will at most only be about 5-15% faster, whereas the 4080 will be absolutely smoked. i guess it really depends on how much emphasis customers will put on RT performance. by AMDs own indirect admission, RT and upscaling algos WILL play a much more important role going forward.

 

8 hours ago, cylix said:

Really want to see sapphire toxic version of 7900xtx, maybe in white😂

 

https://videocardz.com/newz/amd-radeon-rx-7900-gpus-are-designed-to-scale-up-to-3-0-ghz

 

 

oh wow if rdna3 really scales up to 3+ghz then overclocked itll smoke the 4090 easily! AMD touts 2.3 ghz typical gaming clock, so at 3+ghz were talking +30% higher clocks! compare that to the typical 5-7% added performance when overclocking 4090s. iiiiinteresting....

[Papusan]

EVGAs new stuff after they scrapped make graphics cards. All too much nope from me. Weird choices from EVGA. 

 

EVGA's new Octopus

 

 

And here is more from todays new tech..........

 

A new native ATX 3.0 connector melted/burnt (MSI MPG A1000G)

 

And CableMod stop shipping theirs while ModDIY reassured their cable is fine

 

What a disaster this has become. New has to be better, LOOL

[johnksss]

 

[johnksss]

On 11/3/2022 at 11:13 AM, johnksss said:

I just left my MC and they have..er ah, had 4080 Strix in stock.

"For more clarification"

Yes, this had me confused as well considering the card was not even out yet. Considering the system will not even ring up nor process an NDA or not for release item for sale.

 

Manager stated the person in question had it wrong and that he meant the 3080 Strix

 

[tps3443]

17 hours ago, electrosoft said:

 

That WOULD be a refreshing change of pace. Quietly undersell / deflate the hype train and when launched be better than expected.

 

I am sure the lackluster response to AM5 may have played a small part in their approach/pricing.

 

 

Ugh, Azor.....destroyer of Alienware....

 

SFF has its place in hierarchy. I like SFF builds but not for my main rig ever.

 

 

"This card will KO the 4080 16GB. And fight against the 4090."

 

This in spades. Based on presentation and performance estimates (after adjustments), the 7900XTX is going to be an excellent card and price:performance monster.

 

 

Linus going into more about AMD's new cards and Nvidia:

 

 

 

The 7900XTX is gonna really be awesome. I just wish these manufacturers could get enough of them out there in to everyone’s hands. Nvidia is literally milking supply of RTX4090 to artificially inflate the price. So right now the RTX4090 is roughly $2,000 bucks+ taxes it’s even more to buy one online. I’m gonna say about $2,300 after taxes and shipping.

 

Now with the 7900XTX costing $999, it’s only gonna end up being $1,500+ taxes on eBay maybe more due to no inventory and scalpers.
 

GPU manufacturers lived on a high of 2 years selling GPU’s at any price they wanted. Nvidia only learned how to control the market better. The 7900XTX will only fill the niche of what the RTX4090 actually cost on paper. AMD is not good with GPU inventory lol.

 

In the end we will have the $2,000 dollar 4090. Or the $1,500 dollar 7900XTX lol.

Take your pick!

[Papusan]

1 hour ago, tps3443 said:

In the end we will have the $2,000 dollar 4090. Or the $1,500 dollar 7900XTX lol.

Take your pick!

Thats not bad😎 Cheapest 4090 available here home is 2,736.47 USD (included tax). So don't whine brother😀

 

This is the price we had to pay for 3090 under the mining race (if we could find cards in the stores).

[tps3443]

32 minutes ago, Papusan said:

Thats not bad😎 Cheapest 4090 available here home is 2,736.47 USD (included tax). So don't whine brother😀

 

This is the price we had to pay for 3090 under the mining race (if we could find cards in the stores).

I know it’s a total joke. Nvidia wants scalpers to live. Because they are coming back with a vengeance that’s for sure. AMD will fall in to the $1,500+ price point and also be unavailable. Still cheaper than a 4090 so at least there is that. 

[electrosoft]

1 hour ago, tps3443 said:

 

The 7900XTX is gonna really be awesome. I just wish these manufacturers could get enough of them out there in to everyone’s hands. Nvidia is literally milking supply of RTX4090 to artificially inflate the price. So right now the RTX4090 is roughly $2,000 bucks+ taxes it’s even more to buy one online. I’m gonna say about $2,300 after taxes and shipping.

 

Now with the 7900XTX costing $999, it’s only gonna end up being $1,500+ taxes on eBay maybe more due to no inventory and scalpers.
 

GPU manufacturers lived on a high of 2 years selling GPU’s at any price they wanted. Nvidia only learned how to control the market better. The 7900XTX will only fill the niche of what the RTX4090 actually cost on paper. AMD is not good with GPU inventory lol.

 

In the end we will have the $2,000 dollar 4090. Or the $1,500 dollar 7900XTX lol.

Take your pick!

 

We are less than a month into 4090 launch and unlike Ampere no Cryptoboom or Pandemic with which to contend. While you can't get a, "NOW NOW NOW!" on it, by Jan/Feb they will be available. Even if Nvidia is artificially holding them back, demand is nowhere near of the sustained levels of 2020-2021 where supply never could catch up to demand as you had miners snatching up anything to throw into a rig even at grossly inflated costs.

 

The market will adjust.

 

It just feels like a mild case of PTSD because of before but this is NOTHING like then.

 

 

 

 

[Mr. Fox]

9 hours ago, johnksss said:

 

This really clarified a lot. It's hard to make sense of the media nonsense with all of the talking heads and emo crybabies on the internet. They're mostly a bunch of stupid goofballs that enjoy running their mouths more than using their brains. 

The quest for low power draw is very anti-enthusiast. While the specs might sound high for an uneducated consumer or gamerkid, the 4090 doesn't have a higher TGP or pull more watts than our Kingpin GPUs and shunt-modded cards have for several years. I view the high power draw, high TGP for a GPU and high TDP for a CPU as being very positive, not negative. You just aren't going to get more horsepower without it. You can gain efficiency and do more with less, but you won't accomplish as much as you will be doing more with more.

 

Drawing less power  almost always means less performance than you would find by using more power. I always have been, and always will be, an advocate by doing more with more. Doing more with less only applies to gaining efficiency from where you used to be compared to where you are now. I'd much rather go the route of using as much or more power than ever before and taking the performance higher than possible if you're doing more with less. Achieving similar performance using less power is progress, but it should also be a cue that you're not doing nearly as much as you could be by limiting yourself to the same amount of power as before instead of pouring on more power than you ever have in the past and just pushing the performance even  higher along with it. The "efficiency" should be used to facilitate that rather than using less power than before.

[Papusan]

33 minutes ago, tps3443 said:

I know it’s a total joke. Nvidia wants scalpers to live. Because they are coming back with a vengeance that’s for sure. AMD will fall in to the $1,500+ price point and also be unavailable. Still cheaper than a 4090 so at least there is that. 

4090 is the only ADA card launched. There isn't any other options if you want a brand new card. And new cards also means people will fight to get the cards first. Aka prices will go up above MSRP. At release/launch day I coud get a card a hell lot cheaper than what they cost now. But once the card was out for over a week or so the prices went up. Thats the  normal with a launch. The first one to buy have to pay higher prices. This is how the game is.

 

Once AMD is out with their new cards and nvidia launch 4080 there will be easier and easier to get 4090. And at more normal prices. The worst problem for me is what models that will be available. I don't want jump on a model I don't like or want. Then I will rather sit and wait what to come.

 

 

[Mr. Fox]

3 hours ago, Papusan said:

4090 is the only ADA card launched. There isn't any other options if you want a brand new card. And new cards also means people will fight to get the cards first. Aka prices will go up above MSRP. At release/launch day I coud get a card a hell lot cheaper than what they cost now. But once the card was out for over a week or so the prices went up. Thats the  normal with a launch. The first one to buy have to pay higher prices. This is how the game is.

 

Once AMD is out with their new cards and nvidia launch 4080 there will be easier and easier to get 4090. And at more normal prices. The worst problem for me is what models that will be available. I don't want jump on a model I don't like or want. Then I will rather sit and wait what to come.

I know, right. Buying something you don't want just because it is new, almost what you want, and some people love it, is a great way to end up disappointed... and broke. Always better to wait for what you want, or if you're going to buy something you don't want due to the certainty that availability will continue to be poor or the price is expected to remain unreasonable then make it something adequate, but do it as cheap as possible. Your expectations are already in the gutter. If it turns out to be crap, it is exactly what you expected and your surprised look isn't on your face. You have validated it to be crap. But, it was cheap crap... not the end of the world. However, if it turns out better than what you expected then you have something to be thankful for and happy about. 

[Papusan]

8 minutes ago, Mr. Fox said:

I know, right. Buying something you don't want just because it is new, almost what you want, and some people love it, is a great way to end up disappointed... and broke.

Yes. bro Fox. I would put Ryzen Zen 4 in the same basket. I prefer not to by anything from AMD as well. 

 

This doesn't look fun. Even LN2 can't save it. I prefer have fun with what I pay for. And this isn't.

 

https://hwbot.org/submission/5114830_papusan_pifast_core_i9_13900k_(8p)_9sec_50ms?recalculate=true

 

https://hwbot.org/submission/5114826_papusan_cinebench___r23_multi_core_with_benchmate_core_i9_13900k_44878_cb

[Mr. Fox]

36 minutes ago, Papusan said:

Yes. bro Fox. I would put Ryzen Zen 4 in the same basket. I prefer not to by anything from AMD as well. 

 

This doesn't look fun. Even LN2 can't save it. I prefer have fun with what I pay for. And this isn't.

 

https://hwbot.org/submission/5114830_papusan_pifast_core_i9_13900k_(8p)_9sec_50ms?recalculate=true

 

https://hwbot.org/submission/5114826_papusan_cinebench___r23_multi_core_with_benchmate_core_i9_13900k_44878_cb

100%. Looking really amazing, bro.

 

How are you keeping the CPU that cool? Is the radiator outside in freezing weather?

 

I delidded my 13900K this evening and the temperatures before and after essentially did not change at all. They are within 1°C of the same. I regret having wasted my time doing it now. (And, yes... @tps3443 it did bugger up the IHS a bit, as it did with 12th Gen every time. I was able to fix it with sanding the burr off the wing.)

 

You know where I stand with AMD. I am glad that NVIDIA and Intel have more competition than before. Based on the fact that I haven't had a positive experience with anything  from their brand in more than a decade, I am really not open to the idea of taking another chance. I still haven't recovered pschologically or monetarily from my lapse of judgment with 5950X abortion. I am not interested in purchasing a product that runs too hot and doesn't respond favorably to overclocking efforts even when you keep it cool. Based on what I want and expect to receive in exchange for my money, I can see no point in owning such a product.

[Clamibot]

11 minutes ago, Mr. Fox said:

100%. Looking really amazing, bro.

 

How are you keeping the CPU that cool? Is the radiator outside in freezing weather?

 

I delidded my 13900K this evening and the temperatures before and after essentially did not change at all. They are within 1°C of the same. I regret having wasted my time doing it now.

 

You know where I stand with AMD. I am glad that NVIDIA and Intel have more competition than before. Based on the fact that I haven't had a positive experience with anything  from their brand in more than a decade, I am really not open to the idea of taking another chance. I still haven't recovered pschologically or monetarily from my lapse of judgment with 5950X abortion. I am not interested in purchasing a product that runs too hot and doesn't respond favorably to overclocking efforts even when you keep it cool. Based on what I want and expect to receive in exchange for my money, I can see no point in owning such a product.

 

Wait really? Delidding the 13900K didn't do much to drop the temperatures? That's very interesting. Guess Intel did a really good job with the solder on these CPUs.

[Papusan]

24 minutes ago, Mr. Fox said:

100%. Looking really amazing, bro.

 

How are you keeping the CPU that cool? Is the radiator outside in freezing weather?

 

I delidded my 13900K this evening and the temperatures before and after essentially did not change at all. They are within 1°C of the same. I regret having wasted my time doing it now. (And, yes... @tps3443 it did bugger up the IHS a bit, as it did with 12th Gen every time. I was able to fix it with sanding the burr off the wing.)

 

You know where I stand with AMD. I am glad that NVIDIA and Intel have more competition than before. Based on the fact that I haven't had a positive experience with anything  from their brand in more than a decade, I am really not open to the idea of taking another chance. I still haven't recovered pschologically or monetarily from my lapse of judgment with 5950X abortion. I am not interested in purchasing a product that runs too hot and doesn't respond favorably to overclocking efforts even when you keep it cool. Based on what I want and expect to receive in exchange for my money, I can see no point in owning such a product.

I just open the door so fresh air come in. But 13900K run quite ok regarding temps as long you don't max it out. A lot better than Alder lake. And yet, my bin is about average or slightly above. I think I will gain from adding a second pump. Will see if I find one used and cheap. I have a spare pump but the pump tops is ecual expensive as a new pump, so. 

 

Jep. I have seen some with good gain from delidding and some not. I think much is up to how good the soldering was done from Intel.

 

Regarding AMD. I mean if you have an AMD graphics card the Cpu scores will be lower in many benches. Have seen it myself (intel cpu and AMD graphics). Not sure whats up with AMD. Maybe I do something wrong, but I don't think so. So from what I have seen, I avoid this brand. Even if benches was a bit better than Intel. The Red pill just doesn't fit me 🙂

[Mr. Fox]

19 minutes ago, Clamibot said:

 

Wait really? Delidding the 13900K didn't do much to drop the temperatures? That's very interesting. Guess Intel did a really good job with the solder on these CPUs.

 

14 minutes ago, Papusan said:

I just open the door so fresh air come in. But 13900K run quite ok regarding temps as long you don't max it out. A lot better than Alder lake. And yet, my bin is about average or slightly above. 

 

Jep. I have seen some with good gain from delidding and some not. I think much is up to how good the soldering was done from Intel.

 

Regarding AMD. I mean if you have an AMD graphics card the Cpu scores will be lower in many benches. Have seen it myself. Not sure whats up with AMD. But from what I have seen, I avoid this brand. Even if benches was a bit better than from Intel.

I also have probably better than average P-cores and not so great E-core, probably average E-core quality. Yeah, seems like I must have had a good factory solder job. I've not seen this happen before with an Intel CPU. I always have seen a benefit from delidding until now. The temperatures per core saw essentially no change. I was like "Wait, what? The same?" If I did not mark the before and after screenshot you would not be able to tell which temperatures were delid or stock solder except by looking at the clock in the system tray.

 

 

It was also more difficult to delid than 12900K/KS. The IHS was trying to rotate counter clockwise instead of pushing straight sideways, and the solder was less brittle than normal. It seemed more like lead and it was a bit stretchy.

 

 

[Papusan]

18 minutes ago, Mr. Fox said:

 

I also have probably better than average P-cores and not so great E-core, probably average E-core quality. Yeah, seems like I must have had a good factory solder job. I've not seen this happen before with an Intel CPU. I always have seen a benefit from delidding until now. The temperatures per core saw essentially no change. I was like "Wait, what? The same?" If I did not mark the before and after screenshot you would not be able to tell which temperatures were delid or stock solder except by looking at the clock in the system tray.

 

 

It was also more difficult to delid than 12900K/KS. The IHS was trying to rotate counter clockwise instead of pushing straight sideways, and the solder was less brittle than normal. It seemed more like lead and it was a bit stretchy.

 

 

I saw Jufes results from the delidd. But me think I only saw a gain of 5C. The rest was from the other he did. Yep, I think a lot is into the solder job. Even the IHS has some variance. Intel can't make all chips 100% equal.

[Mr. Fox]

11 minutes ago, Papusan said:

I saw Jufes results from the delidd. But me think I only saw a gain of 5C. The rest was from the other he did. Yep, I think a lot is into the solder job. Even the IHS has some variance. Intel can't make all chips 100% equal.

Had I known it, then of course I would not have wasted my time on a worthless exercise that yielded no benefit. Based on the temps he is seeing, I would also suggest that @tps3443not waste any time on it or risk damaging his CPU. I do not think he would see any benefit, just as I did not.

 

The 13900K does not have the super tiny SMD components on the PCB at opposite corners of the die like 12900K/KS does. There is nothing under the IHS except for the die. Like 12th Gen, it is very difficult to get the stock rubber gasket off of the PCB. More than any other prior generation CPUs, the black rubber sealant sticks extremely well and it is very time consuming and tedious, especially near the resistors and capacitors mounted near the IHS. I used plastic razor blades and even using them it was hard to get the PCB  cleaned off.

Here is the damage to the left wing on the IHS. You can see that it was deformed by pressing against it. I had to sand it down to take the burr off of the edge.

 

 

I also sanded the underside of the IHS to remove the black rubber seal and to make sure it was perfectly flat after the damage to the left wing on the IHS.

[Papusan]

13 minutes ago, Mr. Fox said:

Had I know it, then of course I would not have wasted my time on a worthless exercise. Based on the temps he is seeing, I would suggest that @tps3443not waste any time on it or risk damaging his CPU. I do not think he would see any benefit.

 

The 13900K does not have the super tiny SMD components on the PCB at opposite corners of the die like 12900K/KS does. There is nothing under the IHS except for the die. Like 12th Gen, it is very difficult to get the stock rubber gasket off of the PCB. More than any other prior generation CPUs, the black rubber sealant sticks extremely well and it is very time consuming and tedious, especially near the resistors and capacitors mounted near the IHS. I used plastic razor blades and even using them it was hard to get the PCB  cleaned off.

Here is the damage to the left wing on the IHS. You can see that it was deformed by pressing against it. I had to sand it down to take the burr off of the edge.

 

 

I also sanded the underside of the IHS to remove the black rubber seal and to make sure it was perfectly flat after the damage to the left wing on the IHS.

Really stupid move from Intel put resistors and capacitors mounted so close to the IHS. They know very well that a lot people delidde their Cpus. Only hardcore benchers on sub cooling don't delidde because liquid metal is bad on sub zero temps. Sub zero means only Intels tim is usable.

 

Edit. Good job bro Fox on the job 🙂

 

And for @tps3443He should forget sell his chips if he delidde it. Can't be used for LN2 then.

[Clamibot]

21 minutes ago, Papusan said:

Really stupid move from Intel put resistors and capacitors mounted so close to the IHS. They know very well that a lot people delidde their Cpus. Only hardcore benchers on sub cooling don't delidde because liquid metal is bad on sub zero temps. Sub zero means only Intels tim is usable.

 

Edit. Good job bro Fox on the job 🙂

 

And for @tps3443He should forget sell his chips if he delidde it. Can't be used for LN2 then.

 

So does that mean liquid metal isn't ideal for cooling using Peltier elements either? I'm using liquid metal with my MasterLiquid ML360 Sub-Zero, both between the CPU die and IHS, and between the IHS and cooler coldplate.

[Papusan]

24 minutes ago, Clamibot said:

 

So does that mean liquid metal isn't ideal for cooling using Peltier elements either? I'm using liquid metal with my MasterLiquid ML360 Sub-Zero, both between the CPU die and IHS, and between the IHS and cooler coldplate.

Liquid metal lose conductivity with lower temp. Not very bad around zero but then the liquid metal starts being brittle while conductivity going east as well. Not the best if your cold plate is of pure copper. The chemical reaction start speed up from zero and downwards. Bad etc for laptop cooling with the weak and flimsy pressure mounting. Lm will start raising the cold plate from the die.

 

But I wouldn’t worry too much. Replacing liquid metal is a fast job.   And Nickel plated metal make it a lot easier to clean it off. 

[tps3443]

9 hours ago, Papusan said:

Really stupid move from Intel put resistors and capacitors mounted so close to the IHS. They know very well that a lot people delidde their Cpus. Only hardcore benchers on sub cooling don't delidde because liquid metal is bad on sub zero temps. Sub zero means only Intels tim is usable.

 

Edit. Good job bro Fox on the job 🙂

 

And for @tps3443He should forget sell his chips if he delidde it. Can't be used for LN2 then.

Yeah I’m probably not going to delid it. I’m fine with how it performs right now. It’s a great chip with good silicon and the factory tim is doing okay. I have delidded all of my CPU’s I’ve ever owned lol. I guess I could skip just one. 😂
 

@Mr. Fox did you use thermal paste or liquid metal? I’ve found that thermalpaste under the IHS will perform the same as a non-delidded CPU that is soldered sometimes even worse. I would also try LM between the IHS and water block cold plate. Definitely using the liquid metal sandwhich is the only way to go.
 

When I tested direct die I could consistently get 10-11c lower temps by just switching from thermal paste to liquid metal. 
 

Honestly it seems like it would still provide a decent temp reduction. That’s crazy that it didn’t. Maybe give EKWB a shot when their direct die LGA1700 block eventually comes out, then you can remove the IHS all together. It’s supposed to be good.

[Mr. Fox]

58 minutes ago, tps3443 said:

Yeah I’m probably not going to delid it. I’m fine with how it performs right now. It’s a great chip with good silicon and the factory tim is doing okay. I have delidded all of my CPU’s I’ve ever owned lol. I guess I could skip just one. 😂
 

@Mr. Fox did you use thermal paste or liquid metal? I’ve found that thermalpaste under the IHS will perform the same as a non-delidded CPU that is soldered sometimes even worse. I would also try LM between the IHS and water block cold plate. Definitely using the liquid metal sandwhich is the only way to go.
 

When I tested direct die I could consistently get 10-11c lower temps by just switching from thermal paste to liquid metal. 
 

Honestly it seems like it would still provide a decent temp reduction. That’s crazy that it didn’t. Maybe give EKWB a shot when their direct die LGA1700 block eventually comes out, then you can remove the IHS all together. It’s supposed to be good.

I always use liquid metal under the IHS where the solder used to be. The only exception was with the Ryzen 5950X nightmare, which ran hotter after a delid because the chiplets were inconsistent height and there was no longer any solder plugging all the gaps and liquid metal doesn't work for that. Using thermal paste made the temps the same as stock solder (before delid).

 

Using liquid metal on top will absolutely yield better results than thermal paste, but I don't want to destroy the markings on the stock IHS. Maybe I will buy a broken 13th Gen CPU on eBay (something very low end like i5) and use the IHS. I could lap it and use liquid metal and keep the stock IHS looking new. When I had to lap one of the 12900K IHS due to the damage caused by the delid process that made a huge difference. But, the IHS was already destroyed so I had nothing to lose.