Coffeelake thread, benchmarks, reviews, input, everything.

[moonbogg]

Yet more evidence that 5GHz will be an easy to achieve overclock. They just hit the the automatic 5GHz setting in the MB and got it.

BUT, you seem to need monstrous cooling. They hit 92C with a liquid cooler.

It won't matter much I don't think. To be clear, there is no such thing as monstrous cooling when there is thermal paste under the hood. The heat can't be transferred to the cooler fast enough for it to matter. I expect a lot of delids and direct die cooling for these chips.

 

[elhefegaming]

It won't matter much I don't think. To be clear, there is no such thing as monstrous cooling when there is thermal paste under the hood. The heat can't be transferred to the cooler fast enough for it to matter.

I totally agree, that would be an amazing benchmark to do, soldered vs thermal paste with the same cooler.

 

[TheF34RChannel]

It won't matter much I don't think. To be clear, there is no such thing as monstrous cooling when there is thermal paste under the hood. The heat can't be transferred to the cooler fast enough for it to matter. I expect a lot of delids and direct die cooling for these chips.

That's a myth that started with SKL-X and has spread onto every other Intel family like a disease. Sorry

 

[tamz_msc]

That's a myth that started with SKL-X and has spread onto every other Intel family like a disease. Sorry

Myth? Really?
https://www.gamersnexus.net/guides/...iquid-metal-vs-thermal-paste-benchmarks-7900x

Cost of the cooling solution is a concern. The near-necessity to purchase high-end 240 & 280mm coolers and run them at max or near-max fan speeds means that there is a hidden cost to these CPUs, and it’s in the cooler. Overclocking starts demanding exotic solutions, custom loops, or 360-420mm radiators with high-end fans. Prices are high on all of these components and noise is high, and yet, not one of these solutions is remotely as efficient at improving thermal performance as a $5 liquid metal application.

What we do know is that, just from the above testing, Intel isn’t doing the best it can, and the company is failing to exploit its biggest advantage over AMD – significant overclocking headroom given controlled thermals.

There are more arguments for Intel to consider than just “we want to overclock higher because we’re enthusiasts,” and those arguments must be made to convince a giant like Intel to listen. Enthusiast overclocking is insignificant. Noise emissions, higher cooler costs, OEM fear of high liquid temperatures that could breach Asetek specification – these are all strong arguments against Intel’s present HEDT TIM practices. Just looking at some of the liquid temperatures hitting the 50-55C range, it’s clear that we’re rapidly approaching the 60C limiter before tripping Asetek’s out-of-spec concerns. Inside of a case, that’s easily done in heavy load scenarios (like AVX workloads). We’d implore OEM giants like Dell and HP to perform internal testing of CLC-enabled HEDT products under various workloads, particularly AVX, to determine if Intel’s TIM is forcing those companies to border on the cooler spec or forcing higher-end cooler purchases, thus potentially losing competitive edge.

 

[PeterScott]

Myth? Really?
https://www.gamersnexus.net/guides/...iquid-metal-vs-thermal-paste-benchmarks-7900x

I think people get a bit carried away when the rant about how bad the heatspreader TIM is. Using liquid metal isn't really a fair comparison.

Sure it would be nice if Intel soldered them, but when they use a TIM, it has to be practical, stay in place, not dry out, etc.

Thermal consideration aren't all that matters.

 

[tamz_msc]

I think people get a bit carried away when the rant about how bad the heatspreader TIM is. Using liquid metal isn't really a fair comparison.

Sure it would be nice if Intel soldered them, but when they use a TIM, it has to be practical, stay in place, not dry out, etc.

Thermal consideration aren't all that matters.

One can't expect a Youtube reviewer to do a metallurgical treatment using solder on these chips and compare them against out-of-the-box processors, hence the use of LM because it has similar thermal conductivity as solder.

Practicality isn't the point of these tests. Delidding itself is highly impractical for all but the most negligible number of customers who buy these chips. The point of these tests only to illustrate the claims of @moonbogg that cooling performance doesn't matter once you cross a threshold with these CPUs, and is not some myth aimed at exaggeration as some would believe.

 

[piesquared]

Has there been any actual reviews yet? Are thsese things for sale? Havent been following but wasnt something supposed to happen last monday, or was it a paper launch/ PR show?

 

[RichUK]

Looks like an OC'd 8700k will render Ryzen largely redundant.

 

[elhefegaming]

Looks like an OC'd 8700k will render Ryzen largely redundant.

Well depends for what, and what Ryzen do you mean.

 

[jpiniero]

Well depends for what, and what Ryzen do you mean.

Gaming of course. The 1600 is still pretty useful, esp since you can get cheaper boards that can OC. But that's about it.

 

[StinkyPinky]

I'm actually impressed that it is only 92c with an auto overclock at 5Ghz. If you translate the review, it says 92c was the max it acheived so it looks like it won't throttle even under auto settings.

Interestingly, it looks like this board has a "enhanced multi core performance" option which locks the turbo to 4.7 for all 6 cores. THat would be a nice easy overclock for those who don't want to mess around.

 

[WhoBeDaPlaya]

Looks like an OC'd 8700k will render Ryzen largely redundant.

While I love Zen, this is +2c/4t over Skylake with the same IPC and (potentially) the same max OC.

Roughly speaking (Covfefelake vs. Zen, being conservative on Covfefelake) :
1) IPC : +6% (assuming Zen ~ Haswell)
2) Max OC : +20% (4.8GHz vs 4GHz)
3) Threads : -25% (assuming perfect, linear MT scaling)

Overall MT : 1.06 x 1.2 x 0.75 = ~ 0.96

Guesstimate - about the same / close in MT, fuhgettaboutit in ST

 

[RichUK]

While I love Zen, this is +2c/4t over Skylake with the same IPC and (potentially) the same max OC.

Roughly speaking (Covfefelake vs. Zen, being conservative on Covfefelake) :
1) IPC : +6% (assuming Zen ~ Haswell)
2) Max OC : +20% (4.8GHz vs 4GHz)
3) Threads : -25% (assuming perfect, linear MT scaling)

Overall MT : 1.06 x 1.2 x 0.75 = ~ 0.96

Guesstimate - about the same / close in MT, fuhgettaboutit in ST

I think it'll be more like at least 1:1, cost less and significantly excel in ST.

I also believe 4.8Ghz is quite conservative. I expect 4.9Ghz - 5Ghz to be the norm.

Therefore 8700k > 1800x, IMO.

 

[Aikouka]

The Hero(Wifi) has 2 M.2 slots one is covered by the Asus M.2 heatsink. Also the Hero has 8 usb port on the back
2 USB 2.0
4 USB 3.0
1 USB 3.1 Gen 2 Type A
1 USB 3.1 Gen 2 Type C

The 2 USB 2.0 are also used for the bios flashback and Keybot II

Honestly, I usually never use the BIOS USB port. I've owned a few ASUS motherboards (I'm typing this from a PC with a Maximus IX Hero), and I've always seen weird little issues when plugging devices into it. I also tend to ignore Type C plugs as I'd have to use an adapter for them anyway. The way I look at it is that I'm coming from a board with 10 USB ports where 9 are Type A. I have a ton of USB devices or at least things that demand USB power. For example, my keyboard (Corsair K75 RGB) requires two USB ports. I've used cards in the past, but I had far less issues with the computer after removing them.

And I see how I missed the other M.2 port. It's under that glowing Republic of Gamers shroud. If I look closely, I can see the text on the board say "M.2_1" to the top right of it.

The Hero is still probably the best option, but it's not as good as my current one. Although, it sounds like the board has some USB 2 headers on it, so maybe I can go the simple route and pick up an expansion port USB header. Just like using a card without any of the potential hardware hassle.

 

[SPBHM]

Well depends for what, and what Ryzen do you mean.

yes...
I think Ryzen is still a very compelling option, 1700+B350 @3.9GHz is likely a good bit cheaper and close enough for most MT tasks,
also Z370 + cooler + higher CPU price (maybe also delid and high end thermal paste?), it can become very expensive compared to how low you can go with Ryzen,

obviously the 1800x is even less interesting now, but it was never a good option when the 1700 exists.

8600K and the locked ones might be a bigger problem for Ryzen!?
(because I think it attracts gamers trying to save some $)

 

[raghu78]

Looks like an OC'd 8700k will render Ryzen largely redundant.

Oh you mean like how the 1700 made the 6900k irrelevant. We are going to see Intel and AMD keep trying to outdo each other with their CPU generation launches over the next few years. We are going to see a golden era of competition just like the original Athlon K7 / K8 days.

 

[maddie]

Oh you mean like how the 1700 made the 6900k irrelevant. We are going to see Intel and AMD keep trying to outdo each other with their CPU generation launches over the next few years. We are going to see a golden era of competition just like the original Athlon K7 / K8 days.

It obvious that a lot of posters weren't around then. They're going to be shocked. I wonder how much the recent experience of expecting multi-year CPU longevity is going to still exist. Not that present CPUs will become obsolete or unusable, but that the loudest, most flamboyant voices here, tend to be those wanting the cutting edge. The next 2-3 years should have at least double the advancement rate. IPC, MHz and cores combined.

 

[Excessi0n]

That's a myth that started with SKL-X and has spread onto every other Intel family like a disease. Sorry

I upgraded from a Hyper 212 Evo to a custom loop with a top-tier block and a 360mm radiator and got barely any improvement. Like, right around five degrees.

With a quad core.

The TIM works fine at stock, does a mediocre but acceptable job if you aren't pushing things too far, and fails miserably if you are pushing it. You can leap headfirst into the realm of diminishing returns if you want to and you will see a modest improvement, but that improvement will be completely out of line with the cost of the cooling setup.

 

[elhefegaming]

I upgraded from a Hyper 212 Evo to a custom loop with a top-tier block and a 360mm radiator and got barely any improvement. Like, right around five degrees.

With a quad core.

The TIM works fine at stock, does a mediocre but acceptable job if you aren't pushing things too far, and fails miserably if you are pushing it.

I noticed almost no improvement when I installed my h60, comparing againt my 212+, so much so that I ended returning the h60, i'm not a fan of having liquids inside my rig

I'm planning to use my 212+ on the 8700k and see how much oc i can get, then i'll probably buy an h100iv2 and if can't get that much further in oc, then i'm going back to the 212

 

[StinkyPinky]

I noticed this from that Romanian review

But I'm telling you a secret. In terms of gaming, the star is not 8700K, but 8600K overclock. 8600K is a real beast. It's cool and can run steady at 5.3 GHz using an AIO cooler. You will see on Oct. 5 when I know that I can publish that processor to what I mean.

 

[raghu78]

It obvious that a lot of posters weren't around then. They're going to be shocked. I wonder how much the recent experience of expecting multi-year CPU longevity is going to still exist. Not that present CPUs will become obsolete or unusable, but that the loudest, most flamboyant voices here, tend to be those wanting the cutting edge. The next 2-3 years should have at least double the advancement rate. IPC, MHz and cores combined.

I agree. We were stuck at 4 cores on mainstream desktop for what seems an eternity now. But we have moved to 8 cores on mainstream desktop with Ryzen in 2017 and will have moved to 12 cores on mainstream desktop by 2019. Great for the consumer.

 

[epsilon84]

While I love Zen, this is +2c/4t over Skylake with the same IPC and (potentially) the same max OC.

Roughly speaking (Covfefelake vs. Zen, being conservative on Covfefelake) :
1) IPC : +6% (assuming Zen ~ Haswell)
2) Max OC : +20% (4.8GHz vs 4GHz)
3) Threads : -25% (assuming perfect, linear MT scaling)

Overall MT : 1.06 x 1.2 x 0.75 = ~ 0.96

Guesstimate - about the same / close in MT, fuhgettaboutit in ST

That's what I've been saying all along, but got shot down in the other thread when I tried to say Intel has a massive lead in ST performance.

Agreed that you are being conservative on all 3 points though.

IPC difference seems to be greater than 6%, easiest way to see this is to look at single threaded benchmarks or completely CPU limited gaming scenarios. I would say it is 10% - 20% depending on the application (please, AMD fans, don't start a flame war about this)
https://www.cpubenchmark.net/singleThread.html (compare 4770K to Ryzen, Haswell has superior single thread performance)
https://tpucdn.com/reviews/AMD/Ryzen_3_1300X/images/perfrel_1280_720.png (CPU limited gaming chart, compare the 1300X to i5 7500, both 4C/4T chips @ 3.4GHz)

Max O/C: If motherboards like the AORUS have a 5GHz 'auto o/c' function built in, and all reviews/leaks so far hit 5GHz+, we can probably say 4.8GHz is a conservative figure.

With regards to threads/SMT, AMD actually gains more than Intel from SMT here. If you look at the ST to MT scaling in Cinebench or CPU-Z benchmark this is obvious to see.

So whilst your maths is rather conservative on all 3 fronts, I do agree with you as a whole, Coffee Lake will provide comparable MT performance whilst having a big lead in ST.

 

[Excessi0n]

I noticed almost no improvement when I installed my h60, comparing againt my 212+, so much so that I ended returning the h60, i'm not a fan of having liquids inside my rig

I'm planning to use my 212+ on the 8700k and see how much oc i can get, then i'll probably buy an h100iv2 and if can't get that much further in oc, then i'm going back to the 212

I decided to double down and delid the thing. Now it happily runs Prime95 in the low 70s at 4.9GHz and 1.5 volts. And way lower in any other workload. It's great.

Quiet, too.

Anyhow, I think you are being conservative on all 3 points though.

Well, yeah. He said right there in his post that he was being conservative.

 

[tamz_msc]

That's what I've been saying all along, but got shot down in the other thread when I tried to say Intel has a massive lead in ST performance.

Agreed that you are being conservative on all 3 points though.

IPC difference seems to be greater than 6%, easiest way to see this is to look at single threaded benchmarks or completely CPU limited gaming scenarios. I would say it is 10% - 20% depending on the application (please, AMD fans, don't start a flame war about this)
https://www.cpubenchmark.net/singleThread.html (compare 4770K to Ryzen, Haswell has superior single thread performance)
https://tpucdn.com/reviews/AMD/Ryzen_3_1300X/images/perfrel_1280_720.png (CPU limited gaming chart, compare the 1300X to i5 7500, both 4C/4T chips @ 3.4GHz)

Max O/C: If motherboards like the AORUS have a 5GHz 'auto o/c' function built in, and all reviews/leaks so far hit 5GHz+, we can probably say 4.8GHz is a conservative figure.

With regards to threads/SMT, AMD actually gains more than Intel from SMT here. If you look at the ST to MT scaling in Cinebench or CPU-Z benchmark this is obvious to see.

So whilst your maths is rather conservative on all 3 fronts, I do agree with you as a whole, Coffee Lake will provide comparable MT performance whilst having a big lead in ST.

It has been shown again and again that the IPC difference is like 8-10 percent, not 10-20 percent. Gaming performance isn't a measure of IPC. Performance scaling with frequency isn't linear for all but canned benchmarks like Cinebench. So 30 percent improvement is for situations with ideal ST scaling.

 

[maddie]

Why is IPC so misunderstood? It's pretty much self-defined.