More Clarity on 9th Gen Core Processor Pricing Emerges

[hat]

I don't think liquid metal is better than solder. The only benefit to getting a pasted, "de-liddable" chip is not that you can delid and use liquid metal, as solder should out-perform that anyway, but the ability to run without a heatspreader at all (direct die). That's also quite a risk, though... cracked dies are not good. Also, you would likely have to remove the retention bracket, so the CPU would literally be just sitting in the socket until you get the cooler on, then the cooler would hold it in place (hopefully).

 

[ratirt]

I don't think liquid metal is better than solder. The only benefit to getting a pasted, "de-liddable" chip is not that you can delid and use liquid metal, as solder should out-perform that anyway, but the ability to run without a heatspreader at all (direct die). That's also quite a risk, though... cracked dies are not good. Also, you would likely have to remove the retention bracket, so the CPU would literally be just sitting in the socket until you get the cooler on, then the cooler would hold it in place (hopefully).

I don't know if it is better since I’ve never done that. What I understand is that when using liquid metal, you directly connect the die using liquid metal to the heatsink?

 

[Vayra86]

All this endless chatter about delidding, solder or LM is fine but you guys are seriously only talking about a top 3-5% performance uplift - in a best case scenario compared to the old TIM. There are many chips that just don't get further unless you toss ridiculous voltage into it. When you need to go upwards of 1.4v on Coffee Lake to get your pretty 5 Ghz number, you're not really doing a proper OC, you're way out of bounds in terms of efficiency and longevity of the chip. Intel knows this, and their TIM solution is scaled to that end. They really don't want customers pushing the max recommended 1.5v, simply because such a chip on 24/7 load will never get through 2-3 years of work proper.

The reason Intel used solder on the new parts is not because you guys yelled for it for years, its because Coffee Lake 6 cores run into thermal thresholds and the only way to facilitate 8 cores at those clocks is a more reliable heat transfer.

It has absolutely ZERO to do with getting the last 100 mhz out of your chip. If you haven't got the memo yet, Intel has effectively locked the OC headroom down to what they consider feasible under the hood - you run into voltage and/or thermal thresholds anyway long before you get a noticeable performance boost. And with 9th gen and its higher boost out of the box, that has only become more true than ever. Hell, they even test the voltages for every multiplier for you.

Some perspective to think about. And this is also why I consider solder not to be an 'improvement' of sorts, its just a necessity. Coffee Lake was pushing it on the old TIM, but it is still 'sufficient'. It won't be long before Intel K-CPUs are just like Ryzen X CPUs: they OC out of the box and all you really get is a guarantee of the highest clocks in a product stack.

 

[RejZoR]

Better cooling means they'll turbo higher and for longer. It's literally that simple even if you just run them stock. But pair crappy cooler with toothpaste under IHS and you're getting suboptimal experience literally out of the box.

 

[phill]

I've loved the way this thread has gone Certainly some interesting comments I guess

Still I look forward to seeing the reviews and then thinking I made the right choice with what I paid for my 1700X's I'd bought.. I do find there's a lot of 'hate' between Intel and AMD products with people but I'm unsure why that is.. I'm sure we can all agree there good and bad points with both CPUs and just leave it there

 

[FireFox]

I do find there's a lot of 'hate' between Intel and AMD products with people but I'm unsure why that is.. I'm sure we can all agree there good and bad points with both CPUs and just leave it there

It really looks like you want to intensify this 'war

 

[phill]

It really looks like you want to intensify this 'war

Nah not me I have both makes of CPU, they both have the qualities and their downfalls.. I admire what AMD has managed in a short time with Ryzen. What I would like to see is a Ryzen clocked to 5Ghz, same as Intel and then see what the difference is then.. I'd guess it might be a little less or maybe more depending on what tests are run

 

[FireFox]

I'd guess it might be a little less or maybe more depending on what tests are run

Better we stop it here because i am not feeling well and that's not a good sign

 

[Deleted member 178884]

For all the intel fanboys here.
Solder is only an improvement when your not paying for it - which in this case you are.

 

[xorbe]

I need to see 5GHz 8700K vs 5GHz 9900K so we can see if the 12MB vs 16MB cache helps min or avg framerates.

 

[hat]

I don't know if it is better since I’ve never done that. What I understand is that when using liquid metal, you directly connect the die using liquid metal to the heatsink?

Most people don't do direct die. Delidding usually just involves cleaning up Intel toothpaste and using liquid metal instead, while still using the IHS.

All this endless chatter about delidding, solder or LM is fine but you guys are seriously only talking about a top 3-5% performance uplift - in a best case scenario compared to the old TIM. There are many chips that just don't get further unless you toss ridiculous voltage into it. When you need to go upwards of 1.4v on Coffee Lake to get your pretty 5 Ghz number, you're not really doing a proper OC, you're way out of bounds in terms of efficiency and longevity of the chip. Intel knows this, and their TIM solution is scaled to that end. They really don't want customers pushing the max recommended 1.5v, simply because such a chip on 24/7 load will never get through 2-3 years of work proper.

The reason Intel used solder on the new parts is not because you guys yelled for it for years, its because Coffee Lake 6 cores run into thermal thresholds and the only way to facilitate 8 cores at those clocks is a more reliable heat transfer.

It has absolutely ZERO to do with getting the last 100 mhz out of your chip. If you haven't got the memo yet, Intel has effectively locked the OC headroom down to what they consider feasible under the hood - you run into voltage and/or thermal thresholds anyway long before you get a noticeable performance boost. And with 9th gen and its higher boost out of the box, that has only become more true than ever. Hell, they even test the voltages for every multiplier for you.

Some perspective to think about. And this is also why I consider solder not to be an 'improvement' of sorts, its just a necessity. Coffee Lake was pushing it on the old TIM, but it is still 'sufficient'. It won't be long before Intel K-CPUs are just like Ryzen X CPUs: they OC out of the box and all you really get is a guarantee of the highest clocks in a product stack.

Toothpaste is never a good thing, unless it's a really low power chip that doesn't even get hot (Pentium). Why run at 80c when you can run at 60c?

 

[phill]

Better we stop it here because i am not feeling well and that's not a good sign

I'll just talk Intel CPUs around you then

I need to see 5GHz 8700K vs 5GHz 9900K so we can see if the 12MB vs 16MB cache helps min or avg framerates.

I would hazard a guess and say, next to nout.... But, if anything I'd guess maybe something in the difference in IPC between the two CPUs.. But again with Intel and previous gen stuff, 3 to 5% increases are the norm but they seem to be just bumping up the clock speeds and not really gaining masses from the actual IPC count from what I can see or understand.. But again, I'll wait and see as I don't want to say things that anger anyone, even @Knoxx29

 

[FireFox]

I'll just talk Intel CPUs around you then

I don't want to say things that anger anyone, even @Knoxx29

Nah, i was kinda messing with you when i quoted your post.

 

[phill]

Nah, i was kinda messing with you when i quoted your post.

Either way, I'm not after annoying anyone I like the forum, the people.. It's a nice place to make some online friends
On the other hand tho @Knoxx29 , does that mean you'd consider an AMD CPU?? (I'm kidding )

 

[Vayra86]

Most people don't do direct die. Delidding usually just involves cleaning up Intel toothpaste and using liquid metal instead, while still using the IHS.



Toothpaste is never a good thing, unless it's a really low power chip that doesn't even get hot (Pentium). Why run at 80c when you can run at 60c?

Because 80 C provides the exact same performance. Its not hard

Dont get me wrong; I prefer solder too, but its benefits are grossly overrated. As is the benefit of running far below max temp - it doesnt do jack shit.

Better cooling means they'll turbo higher and for longer. It's literally that simple even if you just run them stock. But pair crappy cooler with toothpaste under IHS and you're getting suboptimal experience literally out of the box.

No they dont turbo longer. MSDT intel K chips run full power or go thermal shutdown. There is no throttling in between at lower frequency like there is on laptop CPUs.

They also dont turbo higher - that only happens if you OC them yourself.

Where did you even get this info anyway... its been like this for years., stop spreading BS

Intel Turbo is only managed and adjusted by active core count.

Clock adjustments through speedstep and C states are related to load and idle times. Nowhere does any of it relate to temperature. Intel still uses a pretty archaic form of "boost", really. Its miles behind something like AMDs XFR

 

[hat]

Because 80 C provides the exact same performance. Its not hard

Dont get me wrong; I prefer solder too, but its benefits are grossly overrated. As is the benefit of running far below max temp - it doesnt do jack shit.

I'd consider that only half correct (as I do my own following statements)...

There's a number of benefits to running 60c compared to 80c (in this case, thanks to a hypothetical delid/alternative solder option where all other parameters remain the same, e.g. the chip is still generating the same heat). Many overclockers have reported stability issues that seem to relate to high temperatures. GPU boost also kinda supports this theory, given the number of users that have mentioned Pascal's low thermal "throttling" threshold (not really throttling like an overheating CPU, but there's a low temp ceiling set where it begins to affect GPU boost). Then there's the theory that hardware will last longer the cooler it runs; I remember reading lifespan is doubled for every 10c cooler the component is. Then there's the fact that, though the chip generates the same amount of heat still, crappy thermal transfer thanks to Jergens thermal paste impedes thermal transfer, so there's more heat sitting around in the case than there would be if it were soldered (or delidded).

No they dont turbo longer. MSDT intel K chips run full power or go thermal shutdown. There is no throttling in between at lower frequency like there is on laptop CPUs.

They also dont turbo higher - that only happens if you OC them yourself.

Where did you even get this info anyway... its been like this for years., stop spreading BS

Intel Turbo is only managed and adjusted by active core count.

Clock adjustments through speedstep and C states are related to load and idle times. Nowhere does any of it relate to temperature. Intel still uses a pretty archaic form of "boost", really. Its miles behind something like AMDs XFR

Well, it could at the very least turbo longer if sufficient cooling is in place, before "turbo" conditions send your CPU to thermal throttling land...

 

[R-T-B]

Absolutely the best processor to pair with the 2080 ti to ensure no CPU bottleneck. I've been saying this all the time. It's a match made in heaven. It's so good, I've already pre-preordered it mentally.

Sadly no vendors accept mental preorders. Something about requiring more money, less... this.

But has a toothpaste under IHS which sucks big time.

Except it doesn't.

Compared to 10-15 year old x86 chips too.

Take the heatsink off your x86 of any age and see how well it performs vs ARM.

 

[hat]

Except it doesn't.

Not sure what's going on with the K series. I thought what I read originally was the 9700k and 9900k would have solder, 9600k still has toothpaste... then I read that the 9600k is soldered.

 

[R-T-B]

Not sure what's going on with the K series. I thought what I read originally was the 9700k and 9900k would have solder, 9600k still has toothpaste... then I read that the 9600k is soldered.

the quoted text was referencing a post talking about the top tier 9900k.

 

[hat]

the quoted text was referencing a post talking about the top tier 9900k.

The 9600k was mentioned as well.

 

[R-T-B]

The 9600k was mentioned as well.

You are right. Forgive me, a toad gave me the sicks and my head hurt.

 

[Vayra86]

The 9600k was mentioned as well.

Still doesn't make sense to solder a 6c6t CPU (read: an 8600K, that little one that could - stay comfortably below Tjmax under the old solution). They simply don't need it. And I'm still a believer in the initial reports about solder, and it still being reserved for the 8 core parts.

 

[Robcostyle]

LMAO, 9900K for 520$, 9700K 405$, 9600K 300$, say ty to our taiwan friends.

Spoiler: proof for those who understand ruski

https://3dnews.ru/975350