Intel Skylake / Kaby Lake

[witeken]

RIP AMD and GTX 960 and lower... Intel practically destroyed both with the Skylake iGPU... If Gt2 is as strong as a GTX750, Iris Pro could reach easily GTX960 levels....

Hehe. If this was said, say, one year ago, you wouldn't be taken seriously. Although you're exaggerating because GT4 has 1.2TFLOPS, compared with Maxwell's >>2. But sure, Xbox One performance is nothing to sniff at.

Sweepr, your latest link to that pic doesn't work.

 

[myocardia]

You accused different reviewers of being payed by Intel without evidence only because their results don't match your obvious bias.

What does the word "payed" mean? I've never heard of it. I can promise you that I've never in my entire life used that word.

My bias? I'm the one of the two of us who will most likely be buying a Skylake. I've already said so in this thread, and as far as I know, you haven't. I'm always looking for a good excuse to buy new hardware. Hell, I've been salivating since the second that I first saw what the Z170 chipset had to offer. I'd never buy a Skylake to be able to use Z170 if it had a large regression in performance, but as always with Intel, their latest is even better than its predecessor, which was better than its predecessor, ad infinitum. Haha, well back to the P4, at least.

So now all reviewers that show >10% gains per clock for Skylake in CPU limited games are cheating?

No, just the ones that are. For instance, ones that overclock one CPU, and not the CPU to which they are comparing it: http://www.eurogamer.net/articles/digitalfoundry-2015-intel-skylake-core-i5-6600k-review Ones that vastly improve the performance of one of the CPUs by not only using the fastest DDR4 available at any price, but also using a vastly overclocked Bclk, on top of their CPU overclock, but only on the Skylake: https://www.youtube.com/watch?v=lWsEYGA1JbY&feature=youtu.be&t=19m47s

Skewing results in these ways is dishonest. There's really nothing else to say about it, in my opinion. And yes, I do realize that they have been asked to review these CPUs this way, or else Ian Cutress/AnandTech wouldn't be the only one forcing Skylake to compete on its own merits, which it does just fine, btw. It just isn't nearly as dramatic as Intel was wanting it to be, I'm guessing.

These people will rather trust Crucial

Speaking of Crucial, please post the link to someone at Crucial posting what you say they did, in your pretend quote. Link to faked quote: http://forums.anandtech.com/showpost.php?p=37639451&postcount=3949 You faked a quote in your last response to me that said:

DDR3-1600 CL8/CL9 DDR3 (used in most Haswell tests) is superior to DDR4-2133 CL15 DDR4 in true latency (used in AnandTech's Skylake review).

than a random poster trying to put Skylake in the worst light possible:

Yes, I guess I must absolutely hate Skylake. That's why you and I have talked about nothing other than DRAM in the vast majority of our posts, and I've stated in this thread that I most likely will be buying a Skylake. You seriously aren't very good at drawing conclusions, or making arguments, are you?

You need much faster DDR4 to match the real latency of a simple DDR3-1600 CL9 kit

Yes, if 166 Mhz/14.3% is your idea of "much faster". Isn't that the third time you've said more or less that exact thing? I've answered it the same way every time...

Let me help, AnandTech's calculation might be too complex for you which is why you're only making a fool of yourself.
DDR3-1866/CL9 = 207
DDR4-2133/CL15 = 142.2
DDR4-2666/CL15 = 177.73

lmfao, kid, Ian uses simplified estimations like the ones you continue to quote because he wants people who don't have enough intellect to be able to comprehend latency between two types of DRAM will be able to comprehend the point he's making. It has zero to do with him not knowing the proper way to find the true latency for DRAM that has already been speed rated by the manufacturer, I can promise you. Seriously, did you not see where he calls it an approximation? Feel free to contact him about it for verification. BTW, here he is proving that he does in fact know how to compute actual latency:

So here we have the values for True Latency:

DDR3-1600 C11: 13.75 nanoseconds
DDR4-2133 C15: 14.06 nanoseconds

BTW, you'll notice that he used the identical formula that I've been using all along, the same formula that everyone on Earth uses, when they don't have access to a $100,000 or $500,000, or however much the machine costs that can "spec" out a DRAM chip by itself, before it has been matched with other similar speed DRAM chips, and turned into a DIMM, with published specifications. CL 11/800= .01375 X 1,000= 13.75 ns, CL 15/1066.5= .01406 X 1,000= 14.06 ns

http://anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/7

To those of you actually reading this, the way to find a DIMM's true latency is as follows:

(CAS / Frequency (MHz)) × 1000 = X ns

Example:

(7 / 667) × 1000 = 10.49475 ns

https://en.wikipedia.org/wiki/DDR3_SDRAM

Skylake would still be at a slight disadvantage with that DDR4-2666 CL15 kit according to AnandTech

Please link where any of the handful of AnandTech employees said that.

Different people in this thread provided actual benchmarks results where higher clocked memory doesn't necessarily beat lower clocked memory with lower latency

Right, one guy posted some stuff comparing one type of DRAM to itself, in a completely different type of software than you and I have ever discussed, and in that single software, even when using the largest, most mismatched case, he came up with a maximum of 12.5% difference...in the wrong type of software. That had what to do with gaming performance again?

and none of what you posted proved otherwise

Maybe not to you, but luckily, something you posted makes my point exactly:

So, thanks for doing the legwork on that one. I do appreciate it.<-- That is not sarcasm, btw.

basically busting your bullshit of pro-Skylake reviews.

See the above tables for what it looks like when someone uses the fastest RAM available. That's exactly the percentages of gain someone who isn't wanting to be very honest about their review can add to the performance of the Skylake. And yes, I think if you are too poor to be able to afford to spend $100-200 for some decent DDR3-2400 for the DDR3 CPUs in your review, then you should not be using the ultra-expensive DDR4-3600, either.

BTW, here is the performance difference, when using the identical CPU, and DDR3-1,866 vs DD4-2,133. As can be easily seen, the slowest DDR4 available still outperforms low-latency DDR3, even if not by much.

Since you obviously, for whatever reason, have something against Ian's/Anandtech's opinion of DDR3 vs DDR4, here's more or less the identical opinion, from one of the better-known memory manufacturers:

Ultimately that’s kind of the takeaway here: DDR4 starts at very high speeds with room to scale higher, and at those entry level speeds, it’s faster and more capable than its predecessor in almost every test. Mainstream DDR4 actually winds up with lower overall latency and higher bandwidth than mainstream DDR3.

http://www.corsair.com/en-us/blog/2014/september/ddr3_vs_ddr4_synthetic

Reported for posting plain lies, never said that.

Are you absolutely certain of that? We're talking thousands of posts ago in this thread alone, probably ~300 of them by you. You were awfully excited, if you'll remember. I'm more or less certain it was you, since you have been the only person in this entire thread with whom I've communicated about CPUs. I'll see if I can find the post.

I do think Core i7 6700K is the best gaming CPU at stock. We could also call it the best chip for most games (overall, stock or OCed) due to relatively poor scaling with more cores in most titles today,

Who is we? Anyway, I agree, not that its actually available worldwide yet. It's definitely what I'll be recommending for the near future, once Intel gets their production/supply issues resolved.

but Haswell-E's extra cores might make up for the slightly worse IPC/clocks in future titles, which is why I bought one instead of waiting for Skylake in the first place.

Haha, it took me forever to decide on this 4790k, instead of a 5820k. Like multiple months. I also had considered trying to wait until Skylake was available, but I was really needing an upgrade, and waiting another 9-12 months to see how Skylake turned out just didn't seem like the best idea, at least for me.

Since I decided on the 4790k late last year, I think I made the right decision. Had I known about siliconlottery.com at that time, I would have definitely considered a 5820k even more, although I'm still not sure if I could stand the additional heat of the two additional 22nm cores. Not extra heat during gaming, which I'm sure would be more or less identical, but the extra heat from transcoding video. I had what amounted to close to 2,000 DVDs & Blu-Rays to transcode, not just a couple.

 

[dark zero]

Hehe. If this was said, say, one year ago, you wouldn't be taken seriously. Although you're exaggerating because GT4 has 1.2TFLOPS, compared with Maxwell's >>2. But sure, Xbox One performance is nothing to sniff at.

Sweepr, your latest link to that pic doesn't work.

Not joking, Intel improved dramatically and seems that next year, before Pascal and Ice Islands launch, Kabylake would be ready.... And is supposed to practically being the Presscot to Conroe transition of the GPU... Maybe NVIDIA would start to say bye bye to the low tier if they stick to GDDR5

 

[Sweepr]

No, just the ones that are. For instance, ones that overclock one CPU, and not the CPU to which they are comparing it: http://www.eurogamer.net/articles/digitalfoundry-2015-intel-skylake-core-i5-6600k-review Ones that vastly improve the performance of one of the CPUs by not only using the fastest DDR4 available at any price, but also using a vastly overclocked Bclk, on top of their CPU overclock, but only on the Skylake: https://www.youtube.com/watch?v=lWsEYGA1JbY&feature=youtu.be&t=19m47s

Once again, Eurogamer did a clock per clock comparison using multiple different chips at their Core i7 6700K review. They also pointed out the strenghts of older systems and in which situations it makes sense to upgrade, so they don't look biased at all to me. So did many other websites, it's not like there's not many different articles to read and draw appropriate conclusions about how Skylake and Haswell compare to each other.

Speaking of Crucial, please post the link to someone at Crucial posting what you say they did, in your pretend quote. Link to faked quote: http://forums.anandtech.com/showpost.php?p=37639451&postcount=3949 You faked a quote in your last response to me that said:

You mean the real latency numbers? They were obtained using Crucial's formula.

Yes, if 166 Mhz/14.3% is your idea of "much faster". Isn't that the third time you've said more or less that exact thing? I've answered it the same way every time...

True latency according to memory manufacturer Crucial:

The true definition of latency and the latency equation
At a basic level, latency refers to the time delay between when a command is entered and executed. It's the gap between the two. Because latency is all about this gap, it's important to understand what happens after a command is issued. When the memory controller tells the RAM to access a particular location, the data must go through a number of clock cycles in the Column Address Strobe in order to get to its desired location and &#8220;complete&#8221; the command. With this in mind, there are two variables that determine a module's latency:

The total number of clock cycles the data must go through (measured in CAS Latency, or CL, on data sheets)
The duration of each clock cycle (measured in nanoseconds)

true latency (ns) = clock cycle time (ns) x number of clock cycles (CL)

Once again:

DDR3-1600 CL8/CL9 DDR3 (used in most Haswell tests) is superior to DDR4-2133 CL15 DDR4 in true latency (used in AnandTech's Skylake review).

Haswell DDR3 1600 CL9 = 1.25 ns x 9 CL = 11.25 ns true latency
Skylake
DDR4 2133 CL15 = 0.94 ns x 15 CL = 14.06 ns true latency
DDR4-2400 CL15 = 12.5 ns latency
DDR4-2800 CL15 = 10.7 ns latency
DDR4-3000 CL15 = 10 ns latency

Yes, much faster DDR4 is needed to match the real latency of some bread and butter DDR3 kits (not even talking about the good ones).

lmfao, kid, Ian uses simplified estimations like the ones you continue to quote because he wants people who don't have enough intellect to be able to comprehend latency between two types of DRAM will be able to comprehend the point he's making. It has zero to do with him not knowing the proper way to find the true latency for DRAM that has already been speed rated by the manufacturer, I can promise you. Seriously, did you not see where he calls it an approximation? Feel free to contact him about it for verification. BTW, here he is proving that he does in fact know how to compute actual latency:

Normally in our DRAM reviews I refer to the performance index, which has a similar effect in gauging general performance:

DDR3-1600 C11: 1600/11 = 145.5
DDR4-2133 C15: 2133/15 = 142.2

As you have faster memory, you get a bigger number, and if you reduce the CL, we get a bigger number also. Thus for comparing memory kits, if the difference > 10, then the kit with the biggest performance index tends to win out, though for similar kits the one with the highest frequency is preferred.

It is the approximation AnandTech uses to estimate performance, and according to it:

DDR3-1866/CL9 = 207
DDR4-2133/CL15 = 142.2
DDR4-2666/CL15 = 177.73

Skylake would still be at a slight disadvantage performance-wise using a DDR4-2666 CL15 kit. Nevertheless, it was ~13% faster than Haswell per clock in games @ PCLab's review.

Please link where any of the handful of AnandTech employees said that.

Maybe not to you, but luckily, something you posted makes my point exactly:

http://www.hwupgrade.it/articoli/memorie/4431/thief.png
http://www.hwupgrade.it/articoli/memorie/4431/grid.png
http://www.hwupgrade.it/articoli/memorie/4431/luxmark.png

Lol. I have been pointing out since the first reviews came out that Skylake benefits from higher clocked RAM, but keeping latency under control is important too, and the tests above were run at similar CL (unlike DDR3 vs DDR4 comparisons). Actually Skylake needs fast RAM more than Haswell does, look at the scaling from 1600MHz to 2133MHz using the same DDR3 RAM clocks and latency in both Haswell and Skylake systems:

Unfortunately there's no comparison >2133MHz but I'm inclined to believe Skylake's advantage increases using faster RAM. Had AnandTech tested both Haswell and Skylake with a very fast DDR3-3000 CL12-14 kit perhaps their results would be different.

See the above tables for what it looks like when someone uses the fastest RAM available. That's exactly the percentages of gain someone who isn't wanting to be very honest about their review can add to the performance of the Skylake. And yes, I think if you are too poor to be able to afford to spend $100-200 for some decent DDR3-2400 for the DDR3 CPUs in your review, then you should not be using the ultra-expensive DDR4-3600, either.

Equalizing RAM speed/latency is fun for an academical clock per clock comparison but don't forget in real world Skylake does have the advantage to use higher clocked DDR4. Some people do want to know how Skylake compares to Haswell using the fastest RAM at its disposal.

Are you absolutely certain of that?

I am. I understand your points albeit don't agree with many of them but I think there's no reason to drag this conversation any further. I can see some reviews favouring Skylake just as well as others favouring Haswell with the memory and benchmark choice. Summing it up, there are CPU limited games and applications where Skylake does prove itself as a solid tock and the negativity from some people (not talking about you or any specific user) bothers me a bit, perhaps I overreacted at some points. This long conversations are boring for people trying to keep up with Skylake news.

Who is we? Anyway, I agree, not that its actually available worldwide yet. It's definitely what I'll be recommending for the near future, once Intel gets their production/supply issues resolved.
Haha, it took me forever to decide on this 4790k, instead of a 5820k. Like multiple months. I also had considered trying to wait until Skylake was available, but I was really needing an upgrade, and waiting another 9-12 months to see how Skylake turned out just didn't seem like the best idea, at least for me.

Core i7 4790K is still a great chip so nothing to worry about in the near future.

Since I decided on the 4790k late last year, I think I made the right decision. Had I known about siliconlottery.com at that time, I would have definitely considered a 5820k even more, although I'm still not sure if I could stand the additional heat of the two additional 22nm cores. Not extra heat during gaming, which I'm sure would be more or less identical, but the extra heat from transcoding video. I had what amounted to close to 2,000 DVDs & Blu-Rays to transcode, not just a couple.

I've been using E platforms for some years now and I plan to stick with them. I might pull the trigger on a 35W 4C/8T Skylake-S to create a power efficient mini-ITX system though.

 

[Ajay]

Not joking, Intel improved dramatically and seems that next year, before Pascal and Ice Islands launch, Kabylake would be ready.... And is supposed to practically being the Presscot to Conroe transition of the GPU... Maybe NVIDIA would start to say bye bye to the low tier if they stick to GDDR5

I think they'll be out around the same time. The smaller bus widths and lower memory speeds of low end GPUs are their own end. Intel GPUs I experience the reverse of this problem (w/eDRAM and DDR4 helping to bridge the gap).

I think that Intel won't be able to bridge the gap to mid-range GFX cards and higher for a while (with the advent of HBM and Fab companies investing more capital). I originally predicted a cross-over point the end of this decade - not sure if that's going to hold with the process problems Intel has been encountering. Interesting times though.

 

[drteming]

Validation

4.6 GHz = no go. Using Corsair H100i GTX.

 

[cmdrdredd]

Wait so you need 1.7v to get stable at 4.5ghz??????

 

[drteming]

Voltage was set on auto. At stock speed, under full load, HWmonitor reported 1.6v.

 

[gothuevos]

In stock now at newegg as combo purchase only!!

 

[Pandamonia]

Voltage was set on auto. At stock speed, under full load, HWmonitor reported 1.6v.

crazy volts what bios say?

 

[drteming]

With the vcore manually set at 1.3v in bios, it's showing 1.7v in CPU-Z and HW monitor at 4.5 GHz. It's still early in the BIOS development for the Gigabyte board. Maybe it'll get straightened out with later BIOS.

 

[Pandamonia]

With the vcore manually set at 1.3v in bios, it's showing 1.7v in CPU-Z and HW monitor at 4.5 GHz. It's still early in the BIOS development for the Gigabyte board. Maybe it'll get straightened out with later BIOS.

should have gone asus. Their 170 line is rock solid from launch. good reviews

 

[jpiniero]

Those Skylake-H 45W clocks are pretty "lousy" if you compare it to Skylake-S 35W. The 6700T is 2.8/3.6/35W while the 6920HQ is 2.9/3.8/45 W. The 5950HQ has the same clocks but also includes GT3e at 47 W (the 6920HQ is only GT2) and the 4980HQ is 2.8/4/45W and also has Haswell's Iris Pro.

 

[RaistlinZ]

Wait so you need 1.7v to get stable at 4.5ghz??????

 

[phillyman36]

In Stock TigerDirect as well

http://www.tigerdirect.com/applicat...ffiliateID=rGMTN56tf_w-WDIiCV.X5hAuBoGCny.clg

 

[Deders]

Validation

4.6 GHz = no go. Using Corsair H100i GTX.

Have you tried using load line calibration?

 

[phillyman36]

I chatted with Newegg and the lady told me for now you could only order the i7 6700k in a combo. She said i could order a combo and send the mobo back and get the refund for the motherboard just dont open it.
I ordered. I hope she knows what she is talking about. The cheapest combo is $519.ish

 

[Enigmoid]

Those Skylake-H 45W clocks are pretty "lousy" if you compare it to Skylake-S 35W. The 6700T is 2.8/3.6/35W while the 6920HQ is 2.9/3.8/45 W. The 5950HQ has the same clocks but also includes GT3e at 47 W (the 6920HQ is only GT2) and the 4980HQ is 2.8/4/45W and also has Haswell's Iris Pro.

Likely the mobile chips can run a higher turbo.

 

[jagilbertvt]

I chatted with Newegg and the lady told me for now you could only order the i7 6700k in a combo. She said i could order a combo and send the mobo back and get the refund for the motherboard just dont open it.
I ordered. I hope she knows what she is talking about. The cheapest combo is $519.ish

This pisses me off. I already ordered a board and ram from them. now they want me to order another board (if I return the first one i'll get screwed /w the $25 visa credit I suspect, if I return the second then say goodbye to the $10 discount). *Sigh*

 

[phillyman36]

This pisses me off. I already ordered a board and ram from them. now they want me to order another board (if I return the first one i'll get screwed /w the $25 visa credit I suspect, if I return the second then say goodbye to the $10 discount). *Sigh*

I know I already got the z170 Deluxe on the 5th. I don't want to risk having to wait another 2 weeks or longer so im just going to send the mobo that came with the deal back. tigerdirect is charging $399 then $24 tax on top.

 

[jagilbertvt]

I know I already got the z170 Deluxe on the 5th. I don't want to risk having to wait another 2 weeks or longer so im just going to send the mobo that came with the deal back. tigerdirect is charging $399 then $24 tax on top.

I might do the same, but they you are out return shipping and the $10 combo discount.. (or if I return my first board it'll probably mess up the $25 visa discount for orders >$250).

 

[phillyman36]

I might do the same, but they you are out return shipping and the $10 combo discount.. (or if I return my first board it'll probably mess up the $25 visa discount for orders >$250).

I have their newegg premier membership so I wont get charged a restocking fee and they pay for shipping back. Thats something to think about i guess. If you don't open it will you still get the fee.

 

[jagilbertvt]

I have their newegg premier membership so I wont get charged a restocking fee and they pay for shipping back. Thats something to think about i guess. If you don't open it will you still get the fee.

I don't think they can charge a restocking fee on unopened items (and I don't think I've ever had them charge me a restocking fee), but good point. I do not have premier membership, as I don't do much ordering these days from them.

 

[phillyman36]

I'm still going to chat with them tomorrow. The motherboard i brought on August 5th is part of the combo deal. Doesn't make sense to ship me a motherboard I'm going to send right back. Not sure who at Newegg came up with that bright idea.

 

[jagilbertvt]

I'm still going to chat with them tomorrow. The motherboard i brought on August 5th is part of the combo deal. Doesn't make sense to ship me a motherboard I'm going to send right back. Not sure who at Newegg came up with that bright idea.

LOL, cpu went out of stock while I was chatting w/ them (didn't order). I was contemplating the return of the original board when I noticed it was OOS. Ah well, I'll be waiting a bit longer, I guess. It does seem quite foolish of them.