Core i7-8700K Review: Coffee Lake Brews A Great Gaming CPU

[InvalidError]

Also I suspect InvalidError is speaking hypothetically about die costs. It's not really clear though. If he literally means that it costs $60 to churn out a 140mm^2 die, I'd like to see a source for that. Otherwise he's right, 100mm^2 -> 150mm^2 is a 50% increase in die space and cost if it scales linearly.

The $60 comes from an article about the cost of making a Ryzen CPU but I don't remember enough details about it to dredge it up. That's only the chip manufacturing, not R&D, makerting and other costs that also need to be recovered somehow. You can get a similar figure by deconstructing retail price: assuming AMD breaks even selling the R3-1200 for $120, subtract the typical 60% middlemen+retailer margins, a token amount of AMD's own non-manufacturing costs to keep the offices lit and you're down to CPUs indeed having to cost $60-70 to manufacture. The real cost recovery and money is made on higher-end, higher-margin SKUs.

Also, I seriously doubt Intel is making losses on those ~$70 ~100sqmm Pentiums (same die as the i3 at least until Coffee Lake), which definitely pegs manufacturing cost below $50/100sqmm.

 

[btmedic04]

posting overclocking with a chilled cooler is at best skewed and at worst, disingenuous. use a high end air cooler or a high end AIO that a vast majority of folks buying these will use, then you'll have meaningful overclocking results

 

[mapesdhs]

you dont have USB3 , you dont have Sata3 and you dont have bootable NVME , you dont even have Trim support for your SSD , and your ports are old DMI which means very very low bandwidth for modern SSD ...

PCIe NVMe is still possible via addin cards, TRIM will work fine with the right OS, real-world performance with a typical decent SSD isn't that much less on such a system, and USB3 can be added via option cards. People keep saying older systems are terrible because of this, that and the other, but who has the data to back it up? I've run tests on older systems, they're nowhere near as bad as people claim.

... and I don't tend to move much data so the limitations of first gen DMI and SATA 2 don't impact me significantly. ...

And even if it was, just use an M2 with an addin card, works very well.

Delidding is not a thing for normal gamers, users, etc..

You're right that there seems to have been a sharp increase in the degree to which people on forums and tech sites have talked about delidding as if it's normal. Indeed, it really is not and shouldn't be encouraged as if it were. Doing so also lifts the pressure off Intel to stop using poor TIM.

Ian.

 

[MaCk0y]

Can you benchmark a couple of games that are GPU bottleneck at lower graphic settings? Especially BF1 MP 64 players matches, even though in the MP there will be variance, it would still be very useful to compare between 4 core and six core processors in that game.

Agreed. Currently on a 6600k that seems to bottleneck on 64 player BF1 matches. Would love to see how the 8700k stacks up.

Too bad there isn't some way to consistently benchmark online MP

If they always benchmark on the same server and benchmark a couple of predetermined maps and making sure that it's mostly full, it should be fairly consistent. Would like to see 8700K, 8600K, 7700K and 7600K all at the same frequency. At 1080p medium/high settings. Would be nice to see if 6 cores 6 threads is enough and if it will be better than 4 cores 8 threads.

 

[mapesdhs]

The key phrase in the earlier post was, "seems to bottleneck"; that's not exactly a great starting point for determining any real differences. What would be the measurable parameters? What determines if system A is better than system B? Such a test risks being too subjective. Also, if cores/threads ends up being what matters the most, it could be AMD's options are better anyway, or at least sufficiently good. I get the desire for testing such scenarios, but the guts of how that would work need to be far better defined in order for the expended testing effort to be worthwhile.

 

[Broadbanned]

Up the cost of the PSU as well if you're mid-high level gaming: you can't get away with a 500W 80+ on these Intel chips, the CPU+Motherboard will suck 200W+ easy.

Did you notice they left out the Ryzen R5 1600X from the torture loop power tests? This certainly is an Intel fanboy review. I can't speak for all 1600X chips, but mine hits 63watts (1.28V vcore avg. all 6-cores boost 3.7Ghz.) running Prime95 all day at 56C (25.5C ambient) with a $35 air cooler (Scythe FUMA).

I'm no blind AMD fan boy, I just prefer to purchase from companies who don't have a history of being that in-game cheater who comes in and wipes all your years of true hard earned progress in a split second with memory address hacks.

Play LoL? Imagine all your characters always die repeatedly for every game you join: because of a hacker named Intel, until there's no point in playing.

Play WoW? Imagine all your gear, gold, and levels just get deleted every time the hacker Intel targets you.

Play any FPS? Imagine you are hit with instant headshots every time you spawn as the hacker's points go up, GO INTEL!!! WOOOO...

Why do you like Intel again? Is it because they cheated to get ahead??? Because they'd surely still be trying to sell Core2Duos right now without the revenue converted to R&D from their illegal marketing tactics

 

[AgentLozen]

That post got crazier and crazier the longer I read it.
Upvoted.

 

[YoAndy]

Someone is a little crazy lol

 

[bloodroses]

Yeah, that dude is definitely off his meds or something... lol

 

[prebenmikal]

So would something like a hyper 212x be able to cool this at stock speeds under load?

 

[AgentLozen]

The 8700k doesn't get super hot or anything. It could probably be cooled by an Intel stock cooler and only have problems when you're doing long bouts of video encoding or running Prime 95 or something.