Official Skylake-E Owners Club

[moonbogg]

Now that 2016 is here, the countdown can begin for when we bypass all of this Broadwell and Zen nonsense and get to the only CPU that actually matters: Skylake-E. I'll see you back here with my 8 Skylake cores @ 4.8Ghz. Benchmarks to soon follow.

 

[jpiniero]

You realize it's gonna be 18 months before you see Skylake-E?

 

[ShintaiDK]

2016=Broadwell-E Club.

 

[jhu]

Now that 2016 is here, the countdown can begin for when we bypass all of this Broadwell and Zen nonsense and get to the only CPU that actually matters: Skylake-E. I'll see you back here with my 8 Skylake cores @ 4.8Ghz. Benchmarks to soon follow.

Only 8 cores? What a n00b. I already have 36 cores with my 2x Xeon E5-2699v4 machine. I use it for Facebook mostly.

 

[Burpo]

Imagine that!

Intel Broadwell-EP Xeon lineup detailed, E5-2699 V4 to offer 22 cores and 44 threads

http://vr-zone.com/articles/intel-b...22-cores-44-threads/102758.html#ixzz3wPxgUj5V

 

[moonbogg]

2016=Broadwell-E Club.

2016 is the year to close our eyes and wait for the crappy tech to pass. Skylake-E is the only CPU available and it snot yet available.

 

[guskline]

Interested to see how a Broadwell-E 10 core stacks up to a 5960x.

 

[Aristotelian]

2016 is the year to close our eyes and wait for the crappy tech to pass. Skylake-E is the only CPU available and it snot yet available.

I don't understand. There are users, like me, waiting to build or upgrade a system around a new Broadwell-E CPU. It is not yet available. Skylake-E is also not yet available and will necessarily come after. Yet you're saying that Skylake-E is both available and not yet available...?

Interested to see how a Broadwell-E 10 core stacks up to a 5960x.

Me too - as well as the 8 core to 8 core comparison.

 

[ShintaiDK]

2016 is the year to close our eyes and wait for the crappy tech to pass. Skylake-E is the only CPU available and it snot yet available.

I agree, 2016 will be one of my cheapest years I can remember.

 

[moonbogg]

Interested to see how a Broadwell-E 10 core stacks up to a 5960x.

I expect it to stack up similarly to how Ivy-E stacked up against Sandy-E except it will have two more cores. If you need or want two more cores and are already on X99 then its a good thing.

I don't understand. There are users, like me, waiting to build or upgrade a system around a new Broadwell-E CPU. It is not yet available. Skylake-E is also not yet available and will necessarily come after. Yet you're saying that Skylake-E is both available and not yet available...?

For an enthusiast/gamer not already on X99, Skylake-E is the only chip coming out in the next half decade that means anything. Everything else is blah blah, old news, quad core, boring, herp as well as derp, or ZEN and well...Have fun with that one.

The reason is this. Haswell wasn't enough to ditch Sandy. Broadwell isn't fast enough to ditch haswell, and by extension not fast enough to ditch sandy. Skylake is just fast enough to ditch sandy, but it only has 4 cores, which is old and boring. So the only chip coming out in the next half decade for non X99 enthusiast gamers is an 8 or 10 core Skylake. Its the only chip that is a reasonable upgrade in terms of features, IPC and cores.

I agree, 2016 will be one of my cheapest years I can remember.

With a 6700K, I'm afraid you will have nothing exciting to upgrade to for a very, VERY long time unless you just want more cores. That chip will last for pretty much ever. For damn ever. Like, Xbox twenty/PS 23 era forever. Like humans evolved a third arm era forever. Like earth looks like mars era forever.

 

[ShintaiDK]

With a 6700K, I'm afraid you will have nothing exciting to upgrade to for a very, VERY long time unless you just want more cores. That chip will last for pretty much ever. For damn ever. Like, Xbox twenty/PS 23 era forever. Like humans evolved a third arm era forever. Like earth looks like mars era forever.

My own guess is it will last me till Icelake. Tho with the joker being the Haswell in our second gaming box. But the Skylake isn't going anywhere until Icelake.(2019?)

GPU wise is turning out to be the same. Next 14/16nm GPU will most likely last 5 years+.

Else the only thing to look forward to in the future is 3DXpoint. But Its not something I need anytime soon. Same with NVME.

Mobile is also slowly hitting the wall. Only server keeps running at full speed and accelerating.

 

[ComputerGuy2006]

I agree, 2016 will be one of my cheapest years I can remember.

I think it will depend a lot on price. If the 8 core has a decent price I will pull the trigger. There should also be massive developments in the GPU space due to the die shrink, which will be sucking up money.

 

[MrTeal]

Only 8 cores? What a n00b. I already have 36 cores with my 2x Xeon E5-2699v4 machine. I use it for Facebook mostly.

And undoubtedly Facebook flash games still chug along and suck.

 

[Edrick]

GPU wise is turning out to be the same. Next 14/16nm GPU will most likely last 5 years+.

Agreed, the next gen GPUs coming this year will obsolete everything before them. And we will be on that 14/16 node for quite a few years.

Else the only thing to look forward to in the future is 3DXpoint. But Its not something I need anytime soon. Same with NVME.

3DXpoint is very exciting. I do not expect it to be available until after Skylake-E anyways. But NVMe is really nice.

Only server keeps running at full speed and accelerating.

The server world has been exciting for quite a few years already and shows no signs of slowing down. Power8 entering open source was a big win for many. I am pretty sure my Skylake-E build will be on the server chipset rather than the HEDT chipset.

 

[Carfax83]

Now that 2016 is here, the countdown can begin for when we bypass all of this Broadwell and Zen nonsense and get to the only CPU that actually matters: Skylake-E. I'll see you back here with my 8 Skylake cores @ 4.8Ghz. Benchmarks to soon follow.

I was going to wait for Skylake-E myself, but it coming out next year put an end to that; especially when Microcenter was selling the 5930Ks at such a great price.

So far I haven't regretted my decision. My 5930K is noticeably faster than my previous 4930K, even for general desktop performance.. And with AVX2 optimization becoming more and more prevalent, the gap between Haswell and the previous CPUs is getting bigger.

But the biggest change is the platform itself. X99 is just so much better than the X79 platform.

 

[moonbogg]

Agreed Carfax. I can wait it out though. I like big jumps during upgrade time, but X99 is going to be solid for a long LONG time.

 

[Carfax83]

Agreed Carfax. I can wait it out though. I like big jumps during upgrade time, but X99 is going to be solid for a long LONG time.

My plan now is to skip Skykake-E and upgrade when Cannonlake-E becomes available. Hopefully by then that platform will have PCIe 4.0.

If I get impatient or I feel the need to upgrade the CPU, I can always slap an 8 core Broadwell-E cpu after they drop in price..

 

[Aristotelian]

For an enthusiast/gamer not already on X99, Skylake-E is the only chip coming out in the next half decade that means anything. Everything else is blah blah, old news, quad core, boring, herp as well as derp, or ZEN and well...Have fun with that one.

There are a lot of people not already on X99 that are currently making the switch, and Broadwell-E will certainly attract some others - like myself, a gamer (and more, these days) that has not upgraded since sandy bridge.

So there are enthusiasts/gamers who can enjoy Broadwell-E.

And on:

. Benchmarks to soon follow.

Eagerly awaiting those. Think you'll beat me to posting Broadwell-E benchmarks?

 

[Makaveli]

Only 8 cores? What a n00b. I already have 36 cores with my 2x Xeon E5-2699v4 machine. I use it for Facebook mostly.

lol thanks bro that was a good laugh.

 

[RussianSensation]

Now that 2016 is here, the countdown can begin for when we bypass all of this Broadwell and Zen nonsense and get to the only CPU that actually matters: Skylake-E. I'll see you back here with my 8 Skylake cores @ 4.8Ghz. Benchmarks to soon follow.

2016 is the year to close our eyes and wait for the crappy tech to pass. Skylake-E is the only CPU available and it snot yet available.

Wow, that's A LOT of trash talking alright. Let's break it down.

1. We know very little about Zen's real world performance to call it nonsense.

2. Claiming that Broadwell-E is nonsense and crappy tech, but hyping up Skylake-E that may only launch Q2 2017 is odd to me considering how little IPC Skylake has over Broadwell.

Per AT:

Sandy Bridge to Ivy Bridge: Average ~5.8% Up
Ivy Bridge to Haswell: Average ~11.2% Up
Haswell to Broadwell: Average ~3.3% Up
Broadwell to Skylake (DDR3): Average ~2.4% Up
Broadwell to Skylake (DDR4): Average ~2.7% Up

^ That's very unremarkable compared to BW to be calling BW-E crappy tech.

Another source confirms that Skylake's IPC over BW is a very small improvement:

The worst part about Skylake-E is that when it launches Skylake/i7 6700K architecture will be nearly 2 years old and we will be about a year away from Icelake based on current road-maps. That means in the context of 5820K vs. 6700K and 6800K-6950X vs. 6700K, Skylake-E actually looks the worst. We have to wait another full year from BW-E and almost 3 years from 5820K to get the measly IPC increase of SLK in SLK-E form.

Now if 8-core SLK-E drops to $399-449, that would be another story, but otherwise I don't see how the hype for an 8-core SLK-E is somehow justified. Besides, 8-cores do not even help in games. What happens in 2018 if Intel releases 6-core Icelake i7 on the mainstream platform? All of a sudden the 8-core 2017 SLK-E wouldn't look so great. So I wouldn't be so confident in declaring SLK-E as some amazing value just yet. It may be but we'll have to see by the time we get closer to 2017.

I don't understand. There are users, like me, waiting to build or upgrade a system around a new Broadwell-E CPU. It is not yet available. Skylake-E is also not yet available and will necessarily come after. Yet you're saying that Skylake-E is both available and not yet available...?

Don't worry, you'll have the last laugh in 2016. I remember the hype behind $699 780Ti and 10 months later that card turned into overpriced, under-performing turd. Once NV shifts its optimization resources to Pascal, even a GP104 Pascal will beat 980Ti for less $. Also, if you need a new build, there is no point at all waiting another year for SLK-E because it's not as if BW-E will be slow or something.

 

[Arachnotronic]

RS, good points, but there are some things to consider about SKL-E:

1. It will come w/ 48 PCIe lanes, which should be attractive for those who need to have full x16/x16/x16 for their tri-Pascal/Polaris GPU setups

2. The PCH is much improved too (Kaby Lake PCH), with lots of nice PCIe 3.0 lanes for NVMe SSDs.

3. Should come with enhanced DDR4 memory speed support (DDR4-2666 official)

4. Will use an enhanced version of the Skylake core aimed at servers which Intel architects have hinted will have a lot more to them than the consumer-grade Skylake will have.

I think SKL-E is going to be awesome

 

[jihe]

The only thing exciting about skylake-e is (hopfully) xeon bclk overclocking.

 

[moonbogg]

Wow, that's A LOT of trash talking alright. Let's break it down.

1. We know very little about Zen's real world performance to call it nonsense.

2. Claiming that Broadwell-E is nonsense and crappy tech, but hyping up Skylake-E that may only launch Q2 2017 is odd to me considering how little IPC Skylake has over Broadwell.

Sandy Bridge to Ivy Bridge: Average ~5.8% Up
Ivy Bridge to Haswell: Average ~11.2% Up
Haswell to Broadwell: Average ~3.3% Up
Broadwell to Skylake (DDR3): Average ~2.4% Up
Broadwell to Skylake (DDR4): Average ~2.7% Up

^ That's very unremarkable compared to BW to be calling BW-E crappy tech.

Another source confirms that Skylake's IPC over BW is a very small improvement:

The worst part about Skylake-E is that when it launches Skylake/i7 6700K architecture will be nearly 2 years old and we will be about a year away from Icelake based on current road-maps. That means in the context of 5820K vs. 6700K and 6800K-6950X vs. 6700K, Skylake-E actually looks the worst. We have to wait another full year from BW-E and almost 3 years from 5820K to get the measly IPC increase of SLK in SLK-E form.

Now if 8-core SLK-E drops to $399-449, that would be another story, but otherwise I don't see how the hype for an 8-core SLK-E is somehow justified. Besides, 8-cores do not even help in games. What happens in 2018 if Intel releases 6-core Icelake i7 on the mainstream platform? All of a sudden the 8-core 2017 SLK-E wouldn't look so great. So I wouldn't be so confident in declaring SLK-E as some amazing value just yet. It may be but we'll have to see by the time we get closer to 2017.



Don't worry, you'll have the last laugh in 2016. I remember the hype behind $699 780Ti and 10 months later that card turned into overpriced, under-performing turd. Once NV shifts its optimization resources to Pascal, even a GP104 Pascal will beat 980Ti for less $. Also, if you need a new build, there is no point at all waiting another year for SLK-E because it's not as if BW-E will be slow or something.

That's all very interesting, but for those of us on X79, going X99 makes no sense since the platform is already half way dead unless you just want to upgrade to scratch the itch. Broadwell-E will be good. I'm sure you can detect the sarcasm when I say it will be crappy.
But for people like me on X79? Waiting for Skylake-E makes sense because that's a fresh, new platform and we will have an upgrade path in the form of cannonlake-E? I also saw rumors of the platform supporting PCI-E 4.0, and with all those PCI-E lanes available, you are looking at a potential bandwidth monster for all sorts of high end GPU and SSD combos. Skylake-E will be disgusting and sickening and will be a moment of pride and joy.
Zen? Come on. I'm talking about a real high end rig here. We know the likelihood of Zen competing with this platform is near zero. If Zen turns out amazing, then that's great for everyone, but I'm saying I have my doubts and those doubts are backed by AMD's solid track record of CPU failure.

If Nvidia neglects Maxwell like they did with Kepler, I am switching over to AMD, monitor and all. That's all there is too it. The idea of an entire architecture being neglected so badly makes me sick. I can't buy a product like that. After what happened with Kepler, I am very sensitive to this and will be watching very closely. Its not just about the 980ti's. Its about choosing what kind of GPU company I want to support. It won't be NVidia if they do this again.

RS, good points, but there are some things to consider about SKL-E:

1. It will come w/ 48 PCIe lanes, which should be attractive for those who need to have full x16/x16/x16 for their tri-Pascal/Polaris GPU setups

2. The PCH is much improved too (Kaby Lake PCH), with lots of nice PCIe 3.0 lanes for NVMe SSDs.

3. Should come with enhanced DDR4 memory speed support (DDR4-2666 official)

4. Will use an enhanced version of the Skylake core aimed at servers which Intel architects have hinted will have a lot more to them than the consumer-grade Skylake will have.

I think SKL-E is going to be awesome

Fully agree. That RS guy has no idea what he's talking about. Who is that guy anyway? Also, don't forget about the PCI-E 4.0 possibility.

 

[PPB]

Wow, that's A LOT of trash talking alright. Let's break it down.

1. We know very little about Zen's real world performance to call it nonsense.

2. Claiming that Broadwell-E is nonsense and crappy tech, but hyping up Skylake-E that may only launch Q2 2017 is odd to me considering how little IPC Skylake has over Broadwell.

Per AT:

Sandy Bridge to Ivy Bridge: Average ~5.8% Up
Ivy Bridge to Haswell: Average ~11.2% Up
Haswell to Broadwell: Average ~3.3% Up
Broadwell to Skylake (DDR3): Average ~2.4% Up
Broadwell to Skylake (DDR4): Average ~2.7% Up

^ That's very unremarkable compared to BW to be calling BW-E crappy tech.

Another source confirms that Skylake's IPC over BW is a very small improvement:

The worst part about Skylake-E is that when it launches Skylake/i7 6700K architecture will be nearly 2 years old and we will be about a year away from Icelake based on current road-maps. That means in the context of 5820K vs. 6700K and 6800K-6950X vs. 6700K, Skylake-E actually looks the worst. We have to wait another full year from BW-E and almost 3 years from 5820K to get the measly IPC increase of SLK in SLK-E form.

Now if 8-core SLK-E drops to $399-449, that would be another story, but otherwise I don't see how the hype for an 8-core SLK-E is somehow justified. Besides, 8-cores do not even help in games. What happens in 2018 if Intel releases 6-core Icelake i7 on the mainstream platform? All of a sudden the 8-core 2017 SLK-E wouldn't look so great. So I wouldn't be so confident in declaring SLK-E as some amazing value just yet. It may be but we'll have to see by the time we get closer to 2017.



Don't worry, you'll have the last laugh in 2016. I remember the hype behind $699 780Ti and 10 months later that card turned into overpriced, under-performing turd. Once NV shifts its optimization resources to Pascal, even a GP104 Pascal will beat 980Ti for less $. Also, if you need a new build, there is no point at all waiting another year for SLK-E because it's not as if BW-E will be slow or something.

Your math is wrong, Broadwell to Skylake gains wont apply speculate to Broadwell-E to Skylake-E gains.

Current Broadwell offerings have eDRAM acting as LLC and skew the gaming results a lot, upping unnecesarily the averaged results. In reality Haswell to Broadwell was just a 2.5-3% IPC gain.

What will happen is that Broadwell-E will have a painfully low IPC gain against Haswell-E, only to make Skylake-E look better. In the most paralel workloads, Skylake is about 9-10% IPC gain from Haswell at same LLC cache sizes (L3 in this case).

 

[Carfax83]

If Nvidia neglects Maxwell like they did with Kepler, I am switching over to AMD, monitor and all. That's all there is too it. The idea of an entire architecture being neglected so badly makes me sick. I can't buy a product like that. After what happened with Kepler, I am very sensitive to this and will be watching very closely. Its not just about the 980ti's. Its about choosing what kind of GPU company I want to support. It won't be NVidia if they do this again.

Off topic, but I'm still not convinced that there was any neglect on NVidia's end regarding Kepler. I think that it just wasn't a forward enough looking architecture, especially in regards to it's compute performance.

Kepler was introduced at a time when compute shaders were just starting to become popular. After Maxwell launched, there was a massive uptick in the amount of games that were using compute shaders, and even more they were doing so to a far greater extent.

That became Kepler's Achilles heel so to speak.. NVidia saw the industry trend and made sure that Maxwell had significantly beefed up compute capabilities, on par with or better than GCN..

So I don't think there was any conscious neglect by NVidia. I just think they misread the tea leaves.