Intel 9900K Reviews

[Phentos]

3 hours ago, Maticks said:

So glad I bought my 8086k 2 weeks ago. Looks like you still need to delid this solder Tim and it's not easy. Along with the chip thicknesses causing heat issues.

Managed to get 5.4Ghz all 6 cores stable. 

Watched all the reviews deb8er was saying the only way to get 8th gen overclocking is to lap 1.8mm off the CPU so it transfers more heat off the chip.

 

Can't imagine many people will be willing to do that. So it's 95c 5ghz for most out of the box. The worst got 4.8ghz all cores at 90+c.

 

Wow that's one hell of an OC. What cooler you got?

[Maticks]

5 hours ago, Phentos said:

Wow that's one hell of an OC. What cooler you got?

Velocity ekwb. Asus formula X motherboard. Monsta 480mm radiator. 

[Maticks]

6 hours ago, leadeater said:

And TjMax is a safe value, that's the safety point Intel is happy with. Thinking about it I'd love to see like 100 CPUs run 24/7 at TjMax until they fail, if ever. That would be really interesting and boring lol.

Pinned at tjmax will likely just mean you will degrade the CPU requiring more voltage to hold the Clock. The question is does the degrading kick in at 90c, 95c or 100c. 85c is a safe level at least in summer when it's hotter ambient wise you will have your temps go up a little. 

[leadeater]

18 minutes ago, Maticks said:

Pinned at tjmax will likely just mean you will degrade the CPU requiring more voltage to hold the Clock. The question is does the degrading kick in at 90c, 95c or 100c. 85c is a safe level at least in summer when it's hotter ambient wise you will have your temps go up a little. 

From my understanding every 10C increment is a decent amount in regard to longevity. The Intel set TjMax shouldn't be too degrading though, it's there to protect the chip. I'd be really interest to see that tested out though.

 

Wonder if someone would be able to make custom microcode that sets a higher TjMax then have half of them running at like 110C TjMax, race to the death.

[Taf the Ghost]

1 hour ago, leadeater said:

From my understanding every 10C increment is a decent amount in regard to longevity. The Intel set TjMax shouldn't be too degrading though, it's there to protect the chip. I'd be really interest to see that tested out though.

 

Wonder if someone would be able to make custom microcode that sets a higher TjMax then have half of them running at like 110C TjMax, race to the death.

Different parts of the CPU would degrade at different rates. And each region would have a different curve for degradation. What the optimal & maximal ranges should be isn't the easiest thing to work out. There's also the issues that crop up with the hotter the non-CPU parts of the packaging get, so even if a CPU somehow ran optimally at 200 C, that doesn't mean it would be the best place for it to be, given the realities of everything around it.

[leadeater]

20 minutes ago, Taf the Ghost said:

Different parts of the CPU would degrade at different rates. And each region would have a different curve for degradation. What the optimal & maximal ranges should be isn't the easiest thing to work out. There's also the issues that crop up with the hotter the non-CPU parts of the packaging get, so even if a CPU somehow ran optimally at 200 C, that doesn't mean it would be the best place for it to be, given the realities of everything around it.

That's what makes it such an interesting test.... while also being the most boring ever Re: That LTT 1 year case test 

 

Typical 90C-100C Tj shouldn't be much of a problem for CPU package and my old ATI GPUs used to always run at 90C under load, one of them which is a Reference 6970 is still being used today (not by me), so I'd guess at around 5 years before we'd get any CPUs dying off under such a test. Total spit ball.

 

LGA1366 Xeons are extremely cheap 100 of those shouldn't cost much......

[Taf the Ghost]

Just now, leadeater said:

That's what makes it such an interesting test.... while also being the most boring ever Re: That LTT 1 year case test 

 

Typical 90C-100C Tj shouldn't be much of a problem for CPU package and my old ATI GPUs used to always run at 90C under load, one of them which is a Reference 6970 is still being used today (not by me), so I'd guess at around 5 years before we'd get any CPUs dying off under such a test. Total spit ball.

 

LGA1366 Xeons are extremely cheap 100 of those shouldn't cost much......

It'd be cheap right until you start talking about 100 motherboards, RAM, disks and then Linus' reaction to the next power bill.

[Taf the Ghost]

4 minutes ago, leadeater said:

That's what makes it such an interesting test.... while also being the most boring ever Re: That LTT 1 year case test 

 

Typical 90C-100C Tj shouldn't be much of a problem for CPU package and my old ATI GPUs used to always run at 90C under load, one of them which is a Reference 6970 is still being used today (not by me), so I'd guess at around 5 years before we'd get any CPUs dying off under such a test. Total spit ball.

 

LGA1366 Xeons are extremely cheap 100 of those shouldn't cost much......

Correction: film of the reaction of Linus to the power bill would be worth it... from the viewer's perspective.

[IceCold008]

22 hours ago, Valermos said:

8700K seems like the better buy, 9700k if it sold for MSRP.

From a gaming perspective.

"Seems" is the wrong word for it, "Is" is the word your looking for. 

[mynameisjuan]

14 hours ago, leadeater said:

And TjMax is a safe value, that's the safety point Intel is happy with. Thinking about it I'd love to see like 100 CPUs run 24/7 at TjMax until they fail, if ever. That would be really interesting and boring lol.

It will never happen and people will keep spewing info that anything over 80c will kill your CPU. I do know a couple production PC that run at 100c like 16 hours a day and have been since 2011. They were just OCed to the border point of Tjmax and are still running like champs. One other one was at a company in the meeting room and when I started I found the Optiplex in a god damn closet. No ventilation and pack with dust. Thing runs at like 90c idle but its still there to this day. Now add on all the laptops (ultrabooks to be exact) that run 100c constantly for years. 

 

Now I can count all the CPUs that died instead of the mobo on no hands. I have no examples even when I dug for information where people were getting this hypothetical optimal temperature.

 

We know silicon does degrade with heat but for all we know it might decrease life from 25 years to 20. But other hardware will die long before the CPU does and thats even if you still have the CPU by then. 

[yian88]

19 hours ago, D13H4RD2L1V3 said:

Not really. 

 

If your chip is constantly thermal throttling, then I'd worry. The 80s is a little toasty but is otherwise nothing to worry about too much. Laptop CPUs, especially those in gaming machines have seen the 80s as typical load temps and I haven't seen one suffer an early death. 

 

I'd start worrying if it's really close to TJ Max.

Bullshit, if you knew what you were talking about.... i had an intel ivy bridge i7 4c/8t laptop 2 years ago and it constantly hit 80-90c and not even going full load, the motherboard or cpu crapped after some time and it broke something because of the heat the cpu would lock down to lowest power saving frequency as soon as it would hit high temps, turning off cache and making the pc unusable, only restarting would fix it.

So yeah for desktop PC's even if they have better cooling, dumping 80-90c of heat into the socket overheating the motherboard and hence the VRM's even further is a dangerous situation.

[mynameisjuan]

35 minutes ago, yian88 said:

Bullshit, if you knew what you were talking about.... i had an intel ivy bridge i7 4c/8t laptop 2 years ago and it constantly hit 80-90c and not even going full load, the motherboard or cpu crapped after some time and it broke something because of the heat the cpu would lock down to lowest power saving frequency as soon as it would hit high temps, turning off cache and making the pc unusable, only restarting would fix it.

So yeah for desktop PC's even if they have better cooling, dumping 80-90c of heat into the socket overheating the motherboard and hence the VRM's even further is a dangerous situation.

Two different conversations here. What you are mentioning is just shitty cooling and the CPU throttling until failure. That not the same as the CPU failing hardware wise due to heat. Former which is not fixed by reboot. 

[79wjd]

Just now, mynameisjuan said:

Two different conversations here. What you are mentioning is just shitty cooling and the CPU throttling until failure. That not the same as the CPU failing hardware wise due to heat. 

Not to mention that he's also referring to an inappropriate power delivery system and no or significantly inferior VRM cooling.

[TahoeDust]

37 minutes ago, yian88 said:

Bullshit, if you knew what you were talking about.... i had an intel ivy bridge i7 4c/8t laptop 2 years ago and it constantly hit 80-90c and not even going full load, the motherboard or cpu crapped after some time and it broke something because of the heat the cpu would lock down to lowest power saving frequency as soon as it would hit high temps, turning off cache and making the pc unusable, only restarting would fix it.

So yeah for desktop PC's even if they have better cooling, dumping 80-90c of heat into the socket overheating the motherboard and hence the VRM's even further is a dangerous situation.

You can't compare your experience with a laptop mainboard overheating until failure to a high end desktop motherboard running a CPU hot.  I could run my CPU all day at TjMax and my motherboard/VRMs would be just fine. 

[D13H4RD]

6 hours ago, yian88 said:

Bullshit, if you knew what you were talking about.... i had an intel ivy bridge i7 4c/8t laptop 2 years ago and it constantly hit 80-90c and not even going full load, the motherboard or cpu crapped after some time and it broke something because of the heat the cpu would lock down to lowest power saving frequency as soon as it would hit high temps, turning off cache and making the pc unusable, only restarting would fix it.

Yeah....no.

 

My ASUS laptop averages the mid 80s during gaming and I’ve owned it for a year at this point. It has yet to have issues. My 2009 MacBook Pro runs hot enough to fry a freaking egg and it’s still running after all this while.

 

No reasonable quality board should crap out if the CPU is even at 80 degrees. You have bigger issues to worry about if it does.

[Phentos]

6 hours ago, yian88 said:

So yeah for desktop PC's even if they have better cooling, dumping 80-90c of heat into the socket overheating the motherboard and hence the VRM's even further is a dangerous situation.

That's why one should never cheap out on the motherboard if they can help it.

 

Minimum I'd ever spend on a motherboard for any of my builds is a firm $200. And I only trust ASUS boards at that.

[leadeater]

5 minutes ago, Phentos said:

And I only trust ASUS boards at that.

I use to be the same with that though EVGA has been catching my eye lately.

[Phentos]

Just now, leadeater said:

I use to be the same with that though EVGA has been catching my eye lately.

I looked into the Dark Z390 board, but the golden yellow accents would make it very difficult to match my desired color themes.

 

ASRock has been catching my eye more and more as of late, especially as I slowly begin to realize that I might have to make some hard choices with my upcoming build lol. 9700K + Maximus XI Hero or 9900K + ASRock Phantom Gaming n board...

 

FFS it sucks being poor and having expensive tastes.

[leadeater]

7 hours ago, Phentos said:

FFS it sucks being poor and having expensive tastes.

True that lol. I'm still rocking a 4930K and every generation Intel manages to make HEDT worse every time, please stop sucking Intel you used to be better.

[Phentos]

7 hours ago, leadeater said:

True that lol. I'm still rocking a 4930K and every generation Intel manages to make HEDT worse every time, please stop sucking Intel you used to be better.

Meh if I ever go HEDT I'll be more likely to go Threadripper tbh. I might still go that route next year and keep the Intel system for gaming only, which is why I'm giving the 9700K some serious consideration.

 

My friend was trying to sell his 7820X system for emergency rent money earlier this week, and I almost bought it but chose not to. He was able to sell it to another buyer and I still lent him a couple hundred bucks to stave off the eviction process. 

 

I've just felt that Intel's HEDT range is just meh compared to TR. Sure it has better single-thread performance and AVX-512, but if you prioritize single-thread performance then just go with a 9900K or 9700K. AVX-512 is not used in most content creation workloads iirc, unless it's used for H.265 transcoding? Threadripper's price-performance is just unbeatable imo in HEDT so far.

[WereCat]

On 10/19/2018 at 5:23 PM, VegetableStu said:

come to think of it, if the 2990WX could survive running at 5.2 on all cores, what would the power draw be like O_O

 

On 10/19/2018 at 5:27 PM, Deus Voltage said:

I'm definitely no expert on this, but if I had to guesstimate, it would be exponential. Something along the lines of 350 to 500 max (which would probably render the chip useless, unless you genuinely want a good heater for your home during winter)

Around 1200W+

Depending on how much voltage you would need. 

You're also looking at 800A to 1000A of current and there is no X399 motherboard capable of handling so much current so you would be destroying motherboards VRM long before you would start hurting the CPU by pushing it too hard. 

 

But even if you could push it that hard, it would still be a lot more easy to cool it than any comparable Intel CPU since TR 2990WX has 4 CPU dies on a much larger surface area. 

[Andreas Lilja]

Would it be worth it to upgrade for 1440p gaming, Ryzen or Intel? I guess not? 

[jagdtigger]

On 10/19/2018 at 5:20 PM, TahoeDust said:

1920x is such shit in single core performance

It has almost the same score as my 4670k running at 4,5 GHz, its not slow/shit by any stretch of the meaning...

[TahoeDust]

4 minutes ago, jagdtigger said:

It has almost the same score as my 4670k running at 4,5 GHz, its not slow/shit by any stretch of the meaning...

The 4670k is 5 years old.  If you are happy buying a processor in 2018 that has 2013 single core performance, have at it boss.

[Deli]

1 hour ago, TahoeDust said:

The 4670k is 5 years old.  If you are happy buying a processor in 2018 that has 2013 single core performance, have at it boss.

It takes Intel 5 years to improve single core performance by 20%. Great job, Intel.