Unless there is a huge surprise for P3Dv4 the new CPU performance numbers required for me to change from my rock steady (delidded) 2700K and GTX1080 combo have not yet reached the tipping point.

Although I am sure most (and possibly all) new equipment will exceed the performance shown here, it remains doubtful that it will be by sufficient margins to warrant me to make the expense and trouble to do so...yet. 

I look forward to being lured into a new build when the next performance CPUs arrive.

https://www.flickr.com/gp/156377156@N06/S536g0

On the other hand I am really intrigued that Nvidia claims the new 388.00 driver significantly improves SLI performance.

Wonder how 2x1080's would change things since my single card runs wide open?

In any case, am looking forward to exciting new horizons to be reached by those now testing the better and best! 

Kind regards,

1 hour ago, SpiritFlyer said:

Unless there is a huge surprise for P3Dv4 the new CPU performance numbers required for me to change from my rock steady (delidded) 2700K and GTX1080 combo have not yet reached the tipping point.

In any case, am looking forward to exciting new horizons to be reached by those now testing the better and best! 

Kind regards,

I'd love to get 4.9Ghz on my 3930K which is one step better than delidded in terms of heat transfer, but has a few more cores to cope with.  I haven't gone beyond 4.42 because I now have almost 5y of nearly daily use of 3-6h per day at 4.42Ghz at 1.32v which seems decent, and I don't want to take a greater chance of killing my CPU w/ any more overclock.  I'm not too tickled w/ heat management on the newest intel CPUs so am putting it off as long as possible w/ the faint hope that they will return to E class chips w/ soldered IHS some day.   What a shame they didn't solder  I9-7820X/7900X  As a result I have to pick and choose my sliders according to specific scenarios.  I fly the super easy Maj Dash 8 Q400 which I can fly in and out of most anywhere w/ high settings.  W/ PMDG stuff I have to dial back in the big terminals sometimes quite a bit.

We need a new updated FSMark11 test for P3Dv4 so can make a standard comparison of the different hardware available and coming soon. I guess we could still use the old one and change the default plane and test settings. These new CPUs are so close to each other in performance it is going to be difficult to compare the hardware without a standard test.

1 hour ago, Noel said:

What a shame they didn't solder  I9-7820X/7900X

Noel, someone posted a link to an article in another thread about why Intel was not soldering the heat spreaders anymore. From what I remember it was too difficult with the reduced size of the current CPUs. I don't know if this is true or not and have no idea what thread the post and link was in.

Ted 

1 hour ago, Ted Striker said:

Noel, someone posted a link to an article in another thread about why Intel was not soldering the heat spreaders anymore. From what I remember it was too difficult with the reduced size of the current CPUs. I don't know if this is true or not and have no idea what thread the post and link was in.

Ted 

Yes I read that I think but it doesn't explain this whatsoever!  When people de-lid their Kaby/Coffee they don't solder the IHS back on--they use another form of metallic TIM or something like that, and/or apply it in a way that somehow yields 10-20C temperature drop.   What possible explanations do we have for this massive improvement in thermal performance?  There's only one I can think of:  to limit the processors' ultimate overclocking headroom in order to save this untapped potential for when it's time to release a new processor at a time when they begin having problems w/ meeting increased performance targets.  If you have a better idea than this conspiracy theory go for it!

1 hour ago, Noel said:

Yes I read that I think but it doesn't explain this whatsoever!  When people de-lid their Kaby/Coffee they don't solder the IHS back on--they use another form of metallic TIM or something like that, and/or apply it in a way that somehow yields 10-20C temperature drop.   What possible explanations do we have for this massive improvement in thermal performance?  There's only one I can think of:  to limit the processors' ultimate overclocking headroom in order to save this untapped potential for when it's time to release a new processor at a time when they begin having problems w/ meeting increased performance targets.  If you have a better idea than this conspiracy theory go for it!

Good point Noel, and just the kind of stuff that good (and maybe true) conspiracy theories are made of. The 2600K and 2700K (like mine) were soldered and leave little doubt that they had excellent thermal characteristics. Whether any were actually improved by delidding is questionable as many never survived the tabletop experience. In fact my previous 2600K was also cool at high speeds (but at lower overclock speeds), but still far better than today's offerings, so maybe todays inferior heat resistant CPUs are indicative of Intel's future-proofing back up plan, who knows?

We really could use a consistent standard platform to platform set of P3D performance tests to help us with more objective comparisons. I nominate you, Ted Striker and Westman to head up the task force.

Kind regards,

Of interest: http://www.pcgamer.com/meet-the-overclocker-who-made-delidding-your-cpu-word not allowed-proof/

15 hours ago, Noel said:

There's only one I can think of:  to limit the processors' ultimate overclocking headroom in order to save this untapped potential for when it's time to release a new processor at a time when they begin having problems w/ meeting increased performance targets.

I don't think its so much to just limit the processor performance as not incurring the cost of soldering the heat spreader if they don't have to. Intel has the fastest CPUs now so they are enjoying all the profit margin they can get. I agree with you that if AMD starts catching up they will do all they can to maintain the speed lead and their market share. What I don't understand is since they seem to like to offer so many models that its confusing, why don't they offer soldered versions and just charge more for them to cover the cost and more profit. Actually after a little thinking about it, its probably because most would not buy it but delid a cheaper model instead.

Ted

14 hours ago, SpiritFlyer said:

I nominate you, Ted Striker and Westman to head up the task force.

I haven't even had time to install P3Dv4 yet so you don't want to be waiting on me. . I agree with you though that a new FSMark version should be developed with input from Westman and probably Rob Ainscough also.

Thanks for the link to the de-lidding article. Unfortunately it looks like Intel is trying to make it harder to delid their newer CPUs. Maybe so they charge more for soldered versions in the future! 

Ted

 

1 hour ago, Ted Striker said:

I don't think its so much to just limit the processor performance as not incurring the cost of soldering the heat spreader if they don't have to. Intel has the fastest CPUs now so they are enjoying all the profit margin they can get. I agree with you that if AMD starts catching up they will do all they can to maintain the speed lead and their market share. What I don't understand is since they seem to like to offer so many models that its confusing, why don't they offer soldered versions and just charge more for them to cover the cost and more profit. Actually after a little thinking about it, its probably because most would not buy it but delid a cheaper model instead.

Ted

Unfortunately cost to solder doesn't explain why it is people who delid don't solder, use a DIFFERENT TYPE OF LIQUID TIM, yet seem to see 10-20C drop in temps, so w/o solder.  What possible explanations are there for this beyond intentionally lowering heat transfer efficiency by using what amounts to a relative insulator (compared to whatever the de-lidders use) between die and IHS? 

Absolutely on your second point.  Most of the folks who want the highest performing CPU aren't going to risk delidding--I can't see how they can fail by taking something like 8700K at $399 using their rubberized (!) TIM, then charging $599 for a soldered 8700X, or something along those lines.  Or offer a Coffee Lake 'E' that uses no transistors for graphics and uses solder for the IHS.  I would happily pay the other $200 for the soldered version and am guessing a few others would as well.  I"m assuming solder is better than liquid metal TIM or whatever the delidders use.

On 24/10/2017 at 4:17 PM, Noel said:

 

Fair, schmair, I think it's really pathetic to send these out w/o solder.  At the very least it kills longevity to be running at 80C+.  

 

80 degrees is only during an abnormally intense workload though Noel. I recall Tom from OC3D uses his own custom blender test. Very intense. In normal use it wouldn't be that hot.

 

Quote

The really weird part is--when people de-lid these things they aren't even soldering the IHS back on, instead that are using a some kind of superior liquid metal heat conductor to get that 10-20C temp drop?   This implies Intel is literally using a relative insulator between the IHS and the die!!!   Awful!

 

The TIM Intel use between die and IHS isn't exactly awful, it's actually reasonable TIM. Not the best by a long way but not the worst. Intel have to consider how the TIM will age, what will it be like in years to come, will it still do the job. The paste they chose for that, and I would imagine other technical reasons too, is Dow Corning. We can't say the same for the longevity of our own TIM that we apply when we delid, but then it doesn't matter as it's easy to reapply once the lid is popped.

When delidding, Liquid Pro/Ultra or Thermal Grizzly. I used Thermal Grizzly. Rather than being a paste like conventional thermal interface material, it's actually a liquid metal. It has better thermal conductivity than paste, due to it's nature, and also it's in a "very" thin layer. 

So it's not just the quality of Intel's TIM that's the issue, it's also the "amount" of paste and silicone adhesive. Both factors act to reduce cooling efficiency. Intel don't care of course, the chips still function within their normal temp range and still handle a moderate overclock. It's we enthusiasts though, proud members of the PC master race, that curse and growl because our CPU's get close to or beyond TJ Max when we push them to 5 GHz and beyond. 

8 minutes ago, martin-w said:

Intel don't care of course, the chips still function within their normal temp range and still handle a moderate overclock. It's we enthusiasts though, proud members of the PC master race, that curse and growl because our CPU's get close to or beyond TJ Max when we push them to 5 GHz and beyond. 

Just release a bloody CPU that uses solder like AMD is still doing, and charge accordingly!  I'm sure, worldwide, they could justify the added expense and smaller market, and it's not like they don't know how to solder! 

OK, what are typical temps then for water cooled 8700K in P3D?  Are we below 70 yet?  I can't imagine w/ the temps reported that I will get 5y out of a 7820 or 8700 based on what I read w/ temps.

9 minutes ago, Noel said:

Just release a bloody CPU that uses solder like AMD is still doing, and charge accordingly!  I'm sure, worldwide, they could justify the added expense and smaller market, and it's not like they don't know how to solder! 

OK, what are typical temps then for water cooled 8700K in P3D?  Are we below 70 yet?  I can't imagine w/ the temps reported that I will get 5y out of a 7820 or 8700 based on what I read w/ temps.

 

I just edited my post with conductivity data.

Conductivity...

Solder 81.8 W/(M*K)

Thermal Grizzly 12.5 W/(M*K)

Conventional TIM 5-10 W/(M*K)

Don't know temps in P3D Noel. But not overclocked would be below 70 I would have thought. Upper 70's I used to regard as acceptable when running the sim with an overclocked CPU. Anything under 80 is not a significant issue for CPU longevity. Yes,  extra heat is a negative but it's primarily voltage that kills.

Most enthusiasts swap their CPU's long before mid 70 temps cause an issue.  

Quote

I can't imagine w/ the temps reported that I will get 5y out of a 7820 or 8700 based on what I read w/ temps.

 

Overclocked or stock? How overclocked?

Overclocked at lets say 75 degrees is still 30 degrees below TJ Max. [Not sure of exact TJ Max for 8700K] So within Intel spec. So warrantied to last the three year warranty period. I would bet it would last you your 5 years easily, and more. I've ran at all kinds of CPU temps and never had a CPU fail. 

Stock: Piece of cake.

2 hours ago, Ted Striker said:

I don't think its so much to just limit the processor performance as not incurring the cost of soldering the heat spreader if they don't have to.

Ted

 

Yep, there are technical reasons why it's harder to solder, covered in the article I previously liked to. But yes, it's also due to the fact that it means more profit per CPU. Not sure how many thousands of CPU's it would amount to, but over many years and many CPU architectures it saves big money for Intel. 

Why would they bother to solder a CPU when they don't have to, when they still sell? They sell their so called "enthusiast" CPU's at big bucks, and we still buy them, soldered or not.

Thinking about it, Noel's conspiracy theory might have some validity I suppose. At the moment, AMD CPU's run quite warm, despite being soldered and having lower frequency. If that changes and AMD become more of a threat, Intel could solder, drop temps by 30 degrees, increase frequency and or cores and once again take the lead.

  

Haven't run a stock CPU for 15 years or so, so of course, overclocked to the max that water cooling or a Noctua along w/ my incredibly effective a/c unit!  I'm voltage shy so w/ my SB-E never more than 1.32v.

2 minutes ago, Noel said:

Haven't run a stock CPU for 15 years or so, so of course, overclocked to the max that water cooling or a Noctua along w/ my incredibly effective a/c unit!  I'm voltage shy so w/ my SB-E never more than 1.32v.

 

I can only go by the reviews I've seen. I'll post the Tom Logan video below. As I said, he was seeing 80 degrees with a Corsair H110. That's at 5 GHz and a very intense workload. Not sure if it was running AVX. I think that's a very good temp at 5 GHz. I bet if you ran at something like 4.9 GHz, with an equivalent 280 rad AIO, that your aim to not exceed 70 degrees in the sim would be achievable. ( Silicone lottery permitting) If not, dropping to 4.8 would still net excellent performance and drop temps even more.   

As you and I have said in the past, 100 MHz OC amounts to bugger all in terms of frame rate.