SKL-X moved forward to week 25. CFL coming in Q4.

Shintai

Supreme [H]ardness
Joined
Jul 1, 2016
Messages
5,678
https://benchlife.info/intel-x299-platform-might-launch-at-computex-04092017/

SKL-X is moved forward 6 weeks.

Coffee Lake is still on track for Q4.

skylake-x-1297x450.jpg
 
Last edited:
Great news. I expect Coffee Lake to bring the best out of both worlds, the much antecipated mainstream 6C/12T from Intel on top of top-notch per core performance for games. It might clock even higher than 7700K with this new 14nm++ proccess.
 
The Coffee Lake Q4 relates to mobile?
Raises the question down the line what would be the differentiator between 6C Skylake-X and 6C Coffee Lake on desktops apart from in theory I assume more PCIe on the HEDT but performance wise.
Any idea to when news mentioned desktop 6C Coffee Lake?
Cheers

CFL-S is a known product for quite a while, including die size. But its now Q4 2017 and that's actually not news unlike SKL-X.
Intel-Coffee-Lake-X-Coffee-Lake-S-Coffee-Lake-H-2018-Processors.jpg

And I dont get your question. You could ask that about all HEDT pretty much. But SKL-X does use server type cores with 1MB L2 etc.
 
CFL-S is a known product for quite a while, including die size. But its now Q4 2017 and that's actually not news unlike SKL-X.
Intel-Coffee-Lake-X-Coffee-Lake-S-Coffee-Lake-H-2018-Processors.jpg

And I dont get your question. You could ask that about all HEDT pretty much. But SKL-X does use server type cores with 1MB L2 etc.

My point being for consumers, most will ask why bother with the more expensive 6C HEDT Skylake when in Q4 we can expect an i7 Coffee Lake 6C.
The PCIe is probably more important than the cache (yeah ok debatable depending upon size of cache), but in general performance terms most will then think with Coffee Lake so soon should they pay a premium for HEDT 6C.

Personally and probably like quite a few I am more interested in the 8-10C SKUs so does not influence me, but there are always a few interested in 6C looking to buy HEDT at a more cost effective price, and cost effective will now be compared to the Coffee Lake.
Cheers
 
Its due to better than expected Xeon ramp and yields.

Server first, from now and always. HEDT sales have always been a very tiny niche. In reality it shouldn´t even exist.
 
HEDT sales have always been a very tiny niche. In reality it shouldn´t even exist.

HEDT serves two purposes:
1. Low volume, high profit margins...a little extra cash never hurt anyone, right?
2. Stress-tested real-world field environment to apply to future product r&d...example: unlocked hexacore available since X58 (i7-980X) in early 2010, and an unlocked hexacore will be avail for the mainstream consumer and business segments with the upcoming Coffee Lake/300series chipset at a fraction of the HEDT pricing.

To all the X58/X79/X99 users out there: thank you for beta testing my potential future platform! :ROFLMAO:
 
2. Stress-tested real-world field environment to apply to future product r&d...example: unlocked hexacore available since X58 (i7-980X) in early 2010, and an unlocked hexacore will be avail for the mainstream consumer and business segments with the upcoming Coffee Lake/300series chipset at a fraction of the HEDT pricing.
Sorry, but that was a stupid attempt at making argument, X58/X79/X99 are still nothing akin to Z270/Z370.
 
Its due to better than expected Xeon ramp and yields.

Server first, from now and always. HEDT sales have always been a very tiny niche. In reality it shouldn´t even exist.

look at it this way: HEDT is a way for consumers to get a Xeon while still having nice desktop features like LED's, a black board, audio, and SLI. Without it, folks who wanted a single socket workstation would have to resort to buying server boards, which are pretty expensive, butts ugly, and missing most desktop features.
you pay a huge tax for the unlocked multiplier, but it is still better to have the option than not.
 
I think we talk past one another. The business case for it is just extremely weak at best.

And you are not getting LEDs etc on a Dell/Lenovo/HP/Tyan/Supermicro/etc workstation board ;)
 
I my case, I have enjoyed the shit out of my 980x. Maybe it has a bad business case, but it kicked the bejesus out of everything I had ever seen at the time. Now, its still ok, but slowish. Im now seriously looking into getting a replacement. Coffeelake or maybe kaby x.
 
What does everyone think the pricing will be like on Skylake x? Are we talking an 8 core similar to a ryzen 1800x? Or are we going to have to take it in the rear from Intel again. Do you think Ryzen should have an influence on Intel's pricing this round? We know the 1800x beats 6900k in a lot of tasks yet it's half the price and Intel refused to lower their pricing. What do you think?
 
What does everyone think the pricing will be like on Skylake x? Are we talking an 8 core similar to a ryzen 1800x? Or are we going to have to take it in the rear from Intel again. Do you think Ryzen should have an influence on Intel's pricing this round? We know the 1800x beats 6900k in a lot of tasks yet it's half the price and Intel refused to lower their pricing. What do you think?

I hope Ryzen HEDT hands Intel their ass in the HEDT market and illustrates clearly how Intel takes a good chip (the Xeon) and disables many useful features just because they can. I hope Ryzen HEDT shows clearly to the world the extent to which Intel has been sandbagging with their HEDT lineup, consistently denying the HEDT platform their better chips in an unlocked variant, for the sole purpose of restricting available performance potential and having HEDT users illustrate exactly what their best chips are truly capable of. And lastly, I hope Ryzen HEDT (on it's larger and more robust socket) makes Skylake-X look like a toy...the differences between Ryzen HEDT and Skylake-X should be rather striking (and not in a good way for Intel).
 
What does everyone think the pricing will be like on Skylake x? Are we talking an 8 core similar to a ryzen 1800x? Or are we going to have to take it in the rear from Intel again. Do you think Ryzen should have an influence on Intel's pricing this round? We know the 1800x beats 6900k in a lot of tasks yet it's half the price and Intel refused to lower their pricing. What do you think?

Intel historically never changed prices on their processors in response to AMD. They change prices when they change their lineup.
 
Given that the E3 skylake xeon is only able to run with the c23x chipset, I fully suspect the same to be true for the E5 variants so HEDT and Xeon are not likely to be interchangeable, currently you have x99 for HEDT and C606 for xeon. Also the new socket is a monstrous 3647 pins, so good luck getting that on a itx board.
 
Lol intel playing blatant catch up, what a bunch of overpriced, late-to-the-party looooooooosers
Intel drivers suck!
/reverse AMD shill thread crapping

TBH keen to see what this does especially vs Ryzen2 with the updated IMC configuration. Planned to build a 64-128Gb system for a long time now, so decided to hold off Ryzen.
I don't see them coming close to the budget proposition AMD provides. They almost never do.
 
Ryzen 2 is a late 2019 product at best, and dont expect much more than a shrink. That product will go against Icelake SKUs or Sapphire Lake while it may offer IB IPC instead of SB IPC.

SKL-X pretty much murders Ryzen like SKL/KBL already does, so expect prices to reflect it. AMD isn't selling their chips cheap, they sell them for what they are worth and then some. "Moar cores" of a performance that was 5-6 years ago isn't so impressive.
 
What does everyone think the pricing will be like on Skylake x? Are we talking an 8 core similar to a ryzen 1800x? Or are we going to have to take it in the rear from Intel again. Do you think Ryzen should have an influence on Intel's pricing this round? We know the 1800x beats 6900k in a lot of tasks yet it's half the price and Intel refused to lower their pricing. What do you think?

Do you think AMD sells their Ryzen chips cheap, or trying to get the max price they can? Think the later. 1000$ chips isn't going away.
 
I hope Ryzen HEDT hands Intel their ass in the HEDT market and illustrates clearly how Intel takes a good chip (the Xeon) and disables many useful features just because they can. I hope Ryzen HEDT shows clearly to the world the extent to which Intel has been sandbagging with their HEDT lineup, consistently denying the HEDT platform their better chips in an unlocked variant, for the sole purpose of restricting available performance potential and having HEDT users illustrate exactly what their best chips are truly capable of. And lastly, I hope Ryzen HEDT (on it's larger and more robust socket) makes Skylake-X look like a toy...the differences between Ryzen HEDT and Skylake-X should be rather striking (and not in a good way for Intel).

Ryzen HEDT is only there because the Zen server dreams completely collapsed. Then you can think about why.
 
Ryzen HEDT is only there because the Zen server dreams completely collapsed. Then you can think about why.

Indeed, Summit Ridge was the HEDT platform, but now they are re-purposing the original server SP4 products as a new HEDT platform (ThreadRipper), whereas we heard nothing about the original SP4. The last time I did heard some spec about SP4 was two years ago. It followed the same path than K12, which is no longer mentioned or appears in any roadmap.
 
SKL-X pretty much murders Ryzen like SKL/KBL already does, so expect prices to reflect it. AMD isn't selling their chips cheap, they sell them for what they are worth and then some. "Moar cores" of a performance that was 5-6 years ago isn't so impressive.

Wow, you are incredibly clairvoyant to know exactly how Skylake-X performs. As more and more optimizations (that have already occured on Intel platforms) occur on the very new Ryzen platforms, the modest single threaded performance gap will shrink considerably. Skylake-X=more of the same low core count refuse we've seen over the last four generations of the HEDT platform.

Ryzen HEDT is only there because the Zen server dreams completely collapsed. Then you can think about why.

Drawing conclusions about a product that hasn't even been released yet...wow. You obviously have no idea why Ryzen HEDT is here...you're so far off the mark it's beyond laughable.
 
Wow, you are incredibly clairvoyant to know exactly how Skylake-X performs. As more and more optimizations (that have already occured on Intel platforms) occur on the very new Ryzen platforms, the modest single threaded performance gap will shrink considerably. Skylake-X=more of the same low core count refuse we've seen over the last four generations of the HEDT platform.



Drawing conclusions about a product that hasn't even been released yet...wow. You obviously have no idea why Ryzen HEDT is here...you're so far off the mark it's beyond laughable.

When you got SKL-EP servers, you know how SKL-X works and performs. And talking about servers, ye, I know exactly why Ryzen HEDT is here. Because Zen is a complete joke in the server segment.

So ye, keep dreaming. But reality is what it is.

And people still wait on the FX optimizations, after 6 years. ;)
 
Last edited:
When you got SKL-EP servers, you know how SKL-X works and performs. And talking about servers, ye, I know exactly why Ryzen HEDT is here. Because Zen is a complete joke in the server segment.

So ye, keep dreaming. But reality is what it is.

And people still wait on the FX optimizations, after 6 years. ;)

Again, already passing judgement on an unreleased product. Let's revisit your statement in a year and see if you're correct.

You have access to Skylake-EP servers...whoopie. It definitely shows how close you or the company you work for is to Intel. I question that someone who has early access to unreleased Intel hardware can truly be impartial, as writing unflattering things about said hardware may lead to the abrupt revocation of your "special" status (or worse). We'll see truly what the performance results are AFTER the launch of the two platforms.

And again, you have no idea why Ryzen HEDT is here...I know the background behind it...you don't...;)
 
Again, already passing judgement on an unreleased product. Let's revisit your statement in a year and see if you're correct.

You have access to Skylake-EP servers...whoopie. It definitely shows how close you or the company you work for is to Intel. I question that someone who has early access to unreleased Intel hardware can truly be impartial, as writing unflattering things about said hardware may lead to the abrupt revocation of your "special" status (or worse). We'll see truly what the performance results are AFTER the launch of the two platforms.

And again, you have no idea why Ryzen HEDT is here...I know the background behind it...you don't...;)

I have access to Naples too. What does that make me then? AMD fanboy? You didn't think that one through did you? I was about to write in an earlier post. But I left it out exactly to see the response from people like you.

So try and drop the lame excuses and come up with something better instead.

And feel free to express your view on why Ryzen HEDT is here then.
 
I suspected that you were full of it when you said you had access to Skylake-EP and your claim that you have access to Naples as well cinches it. Prove that you have access to these systems. I don't think you do. I think you're just a bored troll. And I'll be happy to educate you on why Ryzen HEDT is here...after it launches...;)
 
I suspected that you were full of it when you said you had access to Skylake-EP and your claim that you have access to Naples as well cinches it. Prove that you have access to these systems. I don't think you do. I think you're just a bored troll. And I'll be happy to educate you on why Ryzen HEDT is here...after it launches...;)

Your post is a contradiction.
 
I don't know why you people ever argue with Shintai...he's practically part of the Intel marketing team and will argue ad nauseam about how great Intel is and how bad AMD sucks. Better to just ignore him and move on with your life.
 
Wow, you are incredibly clairvoyant to know exactly how Skylake-X performs. As more and more optimizations (that have already occured on Intel platforms) occur on the very new Ryzen platforms, the modest single threaded performance gap will shrink considerably. Skylake-X=more of the same low core count refuse we've seen over the last four generations of the HEDT platform.



Drawing conclusions about a product that hasn't even been released yet...wow. You obviously have no idea why Ryzen HEDT is here...you're so far off the mark it's beyond laughable.


We arlready know how Skylake performs, can't you guess from that? How hard it it see how Intel scales based on architectures? Not too hard right? Look at Haswell from 4 to 6 to 8 cores. Then look at Skylake. Simple, even if Skylake X doesn't hit max, its still going to crush Ryzen.

And by modest you mean the 25% IPC gap Skylake has over Ryzen? That isn't modest, that is much more (double the performance gap of Ryzen's SMT implementation)

And if you haven't been paying attention about Ryzen and whats going on with it, you might want to read up some more and why in many instances it won't be ideal for servers. These are problems with the CPU architecture not with the platform.
 
I think he refers to this. Something SKL-X/SP/EP just improves on. And that chart is made by an educated AMD fan. ;)

lK7gSAo.png
 
Except chart you are citing is outdated.
qKhwq3T.png

It cannot be "outdated" when the percents are the same than in the older graph: 146.91% vs 181.67%. The gap is 24%, which razor1 rounded to 25%.

The only difference is that the new graph adds scores that excluded 256bit code. Ignoring those workloads the gap is reduced to 12% above Zen, which agrees with the 13% that I said in another thread.

I understand the reason for eliminating 256bit workloads when discussing desktop chips. But why would we eliminate 256bit when discussing server/HPC chips? Note as well that SKL-X introduces 512bit, which will increase the gap with Zen in the server/HPC space.
 
Not surprised about the new one either. Cant let down the company that sponsors you and they didn't approve the first one. ;)
 
But why would we eliminate 256bit when discussing server/HPC chips?
Because Himeno used wrong code path for Zen (probably K10's like Prime95), so the chart is a little skewed with 256bit workloads.
llmrCzp.png

That said, yes, in SIMD loads SKL-X>>Zen.
 
Because Himeno used wrong code path for Zen (probably K10's like Prime95), so the chart is a little skewed with 256bit workloads.
llmrCzp.png

That said, yes, in SIMD loads SKL-X>>Zen.

Then eliminate Himeno from the average, but not any 256bit.

And if we eliminate Himeno, then we would start to eliminate other benches as well, like his CB 15 ST scores where his Haswell chip is performing worse than expected, because Haswelll is not below Zen and the IPC gap between Haswell and Kabylake is not the 13% that he reports, but 8%

5.png


We can check that his 3.5GHz score for Skylake/Kabylake is correct

131 * 35/30 = 152.8

and The Stilt got 153. OK

But his score for 3.5GHz Haswell is not correct

121 * 35/30 = 141.2

and The Stilt got 135. FAIL

There are other weird results in his review, like when he obtains that SMT yields for Kabylake are superior to Haswell, but then in the relative performance charts he obtains a larger "relative performance" gap for 4C/4T (13.5%) than for 4C/8T (9.9%) between both chips. Luckily for him all the weird results seem to compensate ones with others to provide acceptable averages.
 
Intel historically never changed prices on their processors in response to AMD. They change prices when they change their lineup.

Intel doesn't regard AMD as a competitor. They haven't for at least a decade. And it's questionable whether they ever took AMD seriously even during the K7 and K8 eras. To Intel, AMD is a pimple on their ass, and a relatively small one at that.
 
Back
Top