Uh, don't you think that a chip that was intended to have been soldered could have TIM applied using the same IHS? Or would you prefer to scrap 1000s of IHS and reorder without the hole? But for Intel to solder without a hole (or another way to relieve pressure) would potentially be bad.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Skylake-X (Core i9) - Lineup, Specifications and Reviews!
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
The production chips may have or may have not been soldered, but the holes in the IHS clearly shows that they were originally intended to be soldered. Intel, as per usual, probably decided to cheap out...which they have a penchant for doing often
Right, because that's what your entire argument is based on. Evil company and you having no clue.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
His argument for holes on IHS at last makes sense than yours for you don't have one at all.
That aside, the fact that Intel got rid of solder for good makes Zeppelins the better crafted product even if performance is inferior. Fun times we live in.
Like I said earlier either in this thread or another Intel one; Intel would be looking for a performance envelope regarding voltage-temp-frequency and this may end up being around 1.2V to 1.24V acceptable ceiling with regards to a well done die-TIM-IHS package/design.
Just need to see how this looks when tested, and tbh we know that these CPUs all hit the wall pretty quickly even with water for 24/7 when soldered; you do not get much more benefits over the disadvantages going from 1.2V to 1.35V with the soldered 8C+ models.
The sweetspot for Intel seems to be around 1.2V, and that is all cores 4GHz at 148W - 8C Broadwell-E.
So if design goes well the Skylake-X should be 4.5GHz all cores at around 1.2V to 1.24V.
Cheers
See the pic? See the gap in the epoxy and IHS to equalize pressure? This is why the 1366 chips and 2600k chips have no hole.
Intel being an evil company? I never said anything of the sort. They're not inherently evil, but being in front for ten years have made them exceptionally and extraordinarily arrogant and inconsiderate of the needs of certain market segments. The marketing strategy of "let's increase profits by tightening the screws on features and maximizing segmentation" has paid off over the last 10 years because people have had no real alternative. Now with a viable competitor, this strategy is no longer viable and Intel will have to very quickly relearn that you gain customers in a competitive environment by producing a top quality product with more features than your competitor.
Intel being an evil company? I never said anything of the sort. They're not inherently evil, but being in front for ten years have made them exceptionally and extraordinarily arrogant and inconsiderate of the needs of certain market segments. The marketing strategy of "let's increase profits by tightening the screws on features and maximizing segmentation" has paid off over the last 10 years because people have had no real alternative. Now with a viable competitor, this strategy is no longer viable and Intel will have to very quickly relearn that you gain customers in a competitive environment by producing a top quality product with more features than your competitor.
Attachments
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
Here is a 7700K delided. See the hole? Now 7700K was to be soldered too, right? I mean its what your entire argument is founded on.
5775C? Also one.
A 4770K then? Ups, also one.
Sure, IB may not have one right? Oh wait it does.
The hole got nothing to do with soldering. But you had to make up a BS story about it.
5775C? Also one.
A 4770K then? Ups, also one.
Sure, IB may not have one right? Oh wait it does.
The hole got nothing to do with soldering. But you had to make up a BS story about it.
Last edited:
How so?
What I've said is that all CPUs that use solder either have holes in the IHS or some other way to relieve pressure from under the IHS during manufacture. Basically, holes or a gap in the epoxy.
Shintai has shown that TIM can be used in place of solder under an IHS with holes. That's it. What he showed was probably a last minute change from solder to TIM, and a large quantity of holed IHS had probably already been ordered.
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
Ivy bridge, Haswell, Broadwell, Skylake, Kabylake, KNC, KNL, all last minute changes now?
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
His entire argument is that those CPUs that are soldered without vent hole in IHS have epoxy gap. Your strawman does not work here. So please, you can do better than that.
Here, BDW-E for reference.
A gap in the epoxy has been shown to be enough of a relief to equalize the pressure (Socket 1366). Just because a gap exists doesn't mean is was intended to have solder, it just means solder could have been used if desired. If no gap in the epoxy exists and no hole exists, solder cannot be used!!
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
Or it got absolutely nothing to do with solder to begin with. All CPUs got it, solder or TIM.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Well, all you need to do is to find soldered CPU without vent hole and with BDW-like epoxy appliance. Go!
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
Like AMD CPUs? All CPUs got a vent, either as a hole or a gap in the epoxy.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
Did you even bother to look at the picture? It got 2 vent holes in the epoxy.
But all soldered CPUs need to have a way to equalize pressure under the IHS during manufacture. Period. Either a hole, or a gap in the epoxy.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Did you read my question?
Now that's an answer that i fail to contest.
Damn, that was entertaining.
Well, apparently all non-soldered need that too.
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
So does all TIM. Your argument is simply flawed that it should be due to solder.
I do not know why, but is there a reason all this debate logic is ignoring what Intel did with Xeon Phi?
It has been clearly shown this is not soldered for 2 generations.
Cheers
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
Yes they do, that's why the complete idiocy about having a vent it must be soldered or intended to be soldered is so silly.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
We have presumed Intel decided to cut pennies on it.
Maybe there's a reliability degree to it but if we cannot establish why Xeon Phi uses 2 vent holes instead of epoxy gaps, we cannot know anything.
Then kindly enlighten us as to the real reason behind the holes in Intel's IHS, from Sandy Bridge-EP to Broadwell-EP. Decoration?
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
You mean the holes in ALL CPUs? Its just a matter of what method being used.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Obviously holes in IHS allow you to save more money than leaving gaps in epoxy, duh.
The discussion was around that Intel always uses solder because it is better for their Xeon range.
But we have a clear example with holes that is not using solder.
All the latest arguing has been around 'proving' solder or lack of and the hole, and somehow this is meant to be relevant to this thread *shrug*.
Xeon Phi sort of concluded this argument a little while ago.
Cheers
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Let's call it 5 stages of grief, because realization that Intel cheaps out on it's most expensive parts is kind of.. disarming.
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
Right, because there cant be any other reason. Good to know you already decided that with your lack of empirical data. Thanks for the big laugh! That was a real comic
I recommend reading this to begin with:
http://overclocking.guide/the-truth-about-cpu-soldering/
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Considering that Intel clearly only cares about money (there's nothing wrong with it, just in case), yes. There may be reliability benefits from switching to TIM, but ... Well, Intel will never reveal those numbers to us so all we can do is speculate. And since Intel are greedy bastards, money degree is the obvious one.
Unless of course, you actually have statistics you can provide.
Except that even writer of that article is confused as to why Intel are doing that. Because large dies do not exactly suffer from cracking issues smaller dies have.
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
Dont they? I am looking forward to your documentation.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
It's your claim they have other reasons, prove it. And in fact, it is his claim that they do not exactly suffer from those issues *as much as small dies*.
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
You say the article is incorrect.
So now its a quantity dispute rather than if it does or doesn't? So you do agree that its a downside with solder that TIM doesn't have?
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Do i?
If anything i cite it.
It was always quantity dispute unless TIM and Intel's shitty gluing literally never fails.
I found this years ago when I was wondering about the hole myself:
http://www.compatdb.org/forums/topic/21394-what-is-the-hole-in-the-p4-cpu/
http://www.compatdb.org/forums/topic/21394-what-is-the-hole-in-the-p4-cpu/
The thing is that Intel goes to extremes doing it. It seems as though they put in as much (or more) marketing time deciding what to snip as they do about what to include and end up cutting all the features they possibly can. In that recent video, Linus had a good point when he said "intel needs to focus on making the best product possible for the best price possible." Instead, they quibble over minutae like not soldering their high-end desktop processor line. For what? A few cents saved during manufacture? AMD has maintained solder...even on their cheapest processors because it's worth it.
Intel has found themselves in a position that's entirely their own doing. They decided to completely segment the Xeon from the HEDT lineup, including the new Socket 3647. AMD did the opposite, creating a nearly identical version of the Naples socket for their HEDT Threadripper, which allows AMD to port tech from Naples to Threadripper nearly at will. This move will hamstring Intel massively with HEDT, as Socket 2066 has very definitive limits, especially with HCC CPUs that AMD can introduce at will and with ease on their large and robust socket. The situation will not get better anytime soon for Intel.
Of course, Intel could turn things around quickly if they had the will to do so and set aside petty and misguided marketing beliefs in pursuit of such a turnaround...
First Intel was an evil company and AMD a non-profit organization. Now both are evil but you claim Intel goes extreme.
You are criticizing Intel by cutting down lanes on the cheaper chips, but you don't do the same when AMD cuts down DP capabilities on the cheaper cards. You criticize Intel by segmenting sockets, but you don't do the same when AMD takes the SP3 socket (Naples/EPYC) transform it into the incompatible SP3r2 socket and relabel it as the TR4 socket for ThreadRipper usage.
There are more examples of this double standard. If Intel applies a price reduction then "they were milking users". If AMD applies a price reduction then "it is a natural reaction to the market needs".
Not-soldering could be a mistake or not. My guess is that most X-series users will not be affected by this. On the other hand all those owners of soldered RyZen chip that cannot break the ~4GHz barrier...
Last edited:
Gideon
2[H]4U
- Joined
- Apr 13, 2006
- Messages
- 3,548
They should be soldered but Intel likes to save money and the thermal cycling argument is a funny joke as that takes extreme temperatures of hot and cold. This will also likely hurt their chance of higher clock speeds on their new HEDT platform as well and make threadripper a even bigger threat.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,340
No dog in this fight, but if I were spitballing, I'd say Intel not soldering on a lot of chips was because they had no real competition. Save a little $. If it means slightly lower clockspeeds, who cares? The competition is miles behind anyway, what difference does it make? AMD, on the other hand, needed to do it, to get maximum potential out of their design to claw back to competitiveness. Now that AMD is back in the game, even if still somewhat behind in some metrics, Intel ought to consider soldering again. But I imagine the beancounters probably got used to the savings and extra profits, so it's not so easy to convince them.
I'm not saying any company is evil, but Intel has been firmly in front for 10 years (largely due to missteps by AMD). This has created an Intel that is exceptionally arrogant and much less responsive in dealing with the needs of the market, as they haven't had to for a quite a while. The market was expected to adapt itself to what Intel offered, especially the HEDT market. As AMD is still getting back in the game, they recognize that they have a lot of rebuilding to do in regards to reputation and market share.
Am I criticizing Intel for segmenting sockets? Yes, I am and I'll explain better why. The Threadripper socket and the Epyc socket are largely identical in terms of their size and capacity. I don't know why there are two sockets, but I imagine we'll know for sure after the launch of Epyc and Threadripper.
The fact is that the Threadripper socket won't serve to impede the function of AMDs HCC desktop chips (it will supply ample power and cooling area) and it's fairly clear that the socket was intended for 10-16 core CPUs (and beyond) due to its Epyc roots.
In comparison, Socket 2066 clearly was NOT designed for high clock HCC chips. We know that Intel decided on the two sockets for the sole purpose of segmenting the server Xeon. Intel could have used a version of Socket 3647 for the HEDT line but didn't, which was understandable as they didn't intend to exceed 10 cores on the HEDT platform and would've been a waste of money to create a desktop version of Socket 3647 for a maximum of 10-cores. Now with the arrival of Threadripper, they have very little wiggle room as they try to make Socket 2066 perform like a desktop Socket 3647 in an attempt to have it support higher clocked HCC chips to compete with Threadripper.
As for AMD reducing DP abilities of certain video cards, I don't support that or defend that practice. As for prices, I want the best processor possible and I'm willing to pay more if the features and performance are there. I've never criticized any company for lowering prices nor have I criticized any company for charging a certain price. Nice things cost money. My main concern is that if a company chooses to charge heavily for a product that all the features are there to support that price.