原文《Thermal paste scandal with graphics cards? The hotspot problems on cards from various manufacturers have a common cause: greed》
地址：https://www.igorslab.de/en/cause ... rior-thermal-paste/
原文作者注意到有读者提到他们的显卡在使用几个月之后散热性能会下降，甚至导致热点温度达到100℃。随后作者购买了一块全新万丽4080盖拉多并收集了原装硅脂到测试设备上进行测试。中间的测试方法和数据我不是学这个的就不献丑了，直接跳到第四页的结论：
And now let’s look at another image of a homogeneous surface in the middle of the smear and analyze what floats on top as a layer of fat, so to speak, like on a good chicken broth. Particles between 5 and 8 µm are actually just ok, you could live with that, but…
现在让我们看另一张涂抹中间均匀表面的图像，并分析漂浮在上面的脂肪层，可以说，就像在一块好的鸡汤上一样。5 到 8 μm 之间的颗粒实际上还可以，你可以忍受它，但是......
… that’s not the whole truth, because it’s ON the paste and not IN it. That’s why I simply vaporized the top layer as a plasma with the laser at minimum power to take a look inside. And well, it’s best to see for yourself:
…这不是全部真相，因为它是在糊状物上而不是在糊状物中。这就是为什么我只是用激光以最小功率将顶层蒸发为等离子体，以观察内部。好吧，最好亲眼看看：
Particles up to 16 µm are beyond reasonable! For cost reasons, very large particles of aluminum oxide have simply been added to the fluffy paste. The trick is to achieve higher thermal conductivity very easily and cost-effectively, because the many layers with the additional heat transfer resistance between the particles and the silicone are eliminated. The only stupid thing is that these large particles also act as a kind of larger spacer between which the silicone oil can then simply run through and out together with the fine zinc particles. This is a kind of self-dissolution with an announcement. The very cheap silicone oil used would certainly result in a material price of well under 2.50 USD. So per kilo, not per gram!
颗粒大小高达 16 µm 是超出合理范围的！出于成本原因，只需将非常大的氧化铝颗粒添加到蓬松的糊状物中即可。诀窍在于非常轻松且经济高效地实现更高的导热性，因为消除了颗粒和硅胶之间具有额外传热阻力的多层。唯一愚蠢的是，这些大颗粒还充当了一种更大的隔离物，硅油可以简单地穿过它们并与细小的锌颗粒一起流出。这是一种带有公告的自溶解。使用非常便宜的硅油肯定会使材料价格远低于 2.50 美元。所以是每公斤，而不是每克！
With the quantities of paste that are used by a board partner every month, it is certainly much cheaper, especially as the margins here are likely to be astronomically high. Of course, the graphics card manufacturers also benefit from this, which is why my sympathy is very limited. Nobody can claim that they didn’t know, because you could have guessed it from the price alone – in fact, you should have! And what exactly is inside? In addition to the vast quantities of mostly coarsely ground aluminum oxide (Al2O3) and a lot of silicone oil (silicon, hydrogen, oxygen), we also have some finer zinc oxide (ZnO) as a cheap filler material. I can’t find any carbon that might have suggested a more sophisticated silicone (side chains). So it really is just the simplest oil.
以电路板合作伙伴每月使用的焊膏数量来看，这当然要便宜得多，尤其是这里的利润可能高得惊人。当然，显卡制造商也能从中受益，这也是我非常同情他们的原因。没人能说他们不知道，因为单从价格就能猜到——事实上，你应该猜到的！里面到底是什么？除了大量粗磨的氧化铝 (Al 2 O 3 ) 和大量硅油 (硅、氢、氧) 之外，我们还有一些更细的氧化锌 (ZnO) 作为廉价的填充材料。我找不到任何可能暗示更复杂的硅酮 (侧链) 的碳。所以它真的只是最简单的油。
With such minimalist grinding grades, the paste simply MUST degrade and dissolve in a short time. This is a deliberately and cheaply cobbled together concoction with a built-in decay time of only a few months. I can’t understand why a graphics card manufacturer would do this to itself (and several have bought this crap). The rejection of RMA requests due to excessively high hotspot temperatures (up to 115 °C would be “normal”) is also downright cheeky towards customers. I simply have to put it so harshly, because the prices charged for such cards actually suggest quality. Except that this is not even remotely delivered here!
在如此简单的研磨等级下，糊剂必须在短时间内降解和溶解。这是一种故意拼凑起来的廉价混合物，内置衰减时间只有几个月。我不明白为什么显卡制造商会这样做（而且有好几个人买了这种垃圾）。由于热点温度过高（高达 115°C 是“正常”温度）而拒绝 RMA 请求，这对客户来说也太无礼了。我不得不说得这么严厉，因为这种卡的价格实际上表明了质量。只是这里根本没送货！
结合全文来看igors lab的意思应该是有些显卡（厂商）像万丽盖拉多这种使用的硅脂并不是多么好的硅脂，硅脂会在几个月内衰减并严重影响显卡的导热性能，体现在温度上就是核心温度升高，热点与核心温差也明显增加。
文章还提到一款华硕TUF 4080和万丽4080盖拉多均能通过更换硅脂后大幅改善了核心和热点温度：
https://www.igorslab.de/en/asus- ... sis-and-the-rescue/
https://www.igorslab.de/en/even- ... continues-unabated/ |
310112100042806
发表于 2024-7-22 14:56
我首发的TUF4090当时拆开硅脂感觉也是很稀，换了7950相变片之后温度正常很多
