• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."
  • Community Question: What makes a good motherboard?

Intel Skylake / Kaby Lake

Page 598 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

IndyColtsFan

Lifer
Sep 22, 2007
33,491
553
126
Yeah, if the i5 6C/6T is indeed priced under $200, that's going to be a strong competitor, especially for gaming, with the Ryzen 5 1600/1600X CPUs, especially if it can be clocked above 4Ghz.

I don't hate CoffeeLake, quite the opposite, really. I'm thrilled that Intel is finally upping the core-counts of "mainstream" CPUs. But I do thank AMD for prodding Intel into finally doing so, after all these years, of only quad-core max on mainstream sockets. (Q6600 came out when???)

I might even get an unlocked 6C/6T CoffeeLake to mess with, or use as my "gaming" rig. But I likewise cannot deny the strong value proposition that my Ryzen 5 1600 CPUs have given me. For the same price as a Kaby Lake locked i5 4C/4T CPU, I got an unlocked 6C/12T CPU. Can't really beat that, can you? (Well, not currently.)

Edit: And then, there's this:
https://forums.anandtech.com/threads/incredible-deal-on-ryzen-r7-1800x-mircrocenter-349.2512725/

Ryzen 7 1800X for $349.99 at Microcentr.
While AMD deserves a lot of credit recently, they had very little to do with Intel introducing 6+ cores to mainstream. This move was planned years ago. AMD may have prodded Intel to push their releases up, but the 8700K was planned long before Ryzen came along.
 
  • Like
Reactions: pcp7 and Phynaz

R0H1T

Platinum Member
Jan 12, 2013
2,566
142
106
While AMD deserves a lot of credit recently, they had very little to do with Intel introducing 6+ cores to mainstream. This move was planned years ago. AMD may have prodded Intel to push their releases up, but the 8700K was planned long before Ryzen came along.
You'd have to be naive to think that Intel didn;t know much about Ryzen, unlike leakers or speculators they knew a lot. That's the same for AMD/Nvidia btw since their product cycles & all kinds of investments depend an awful lot on knowing something, or many things about the competitor. Then there's your theory that 6 core would've launched anyway, despite Ryzen btw previously there was the 4790k, well between the 7700k & 8700k what would you choose? The same goes for the new i3 vs previous gen i5 & the latest i5 vs 7700(k) so you're telling us that Intel made a whole lot of their products obsolete, jut because?
 
Last edited:
  • Like
Reactions: USER8000

IndyColtsFan

Lifer
Sep 22, 2007
33,491
553
126
You'd have to be naive to think that Intel didn;t know much about Ryzen, unlike leakers or speculators they knew a lot. That's the same for AMD/Nvidia btw since their product cycles & all kinds of investments depend an awful lot on knowing something, or many things about the competitor. Then there's your theory that 6 core would've launched anyway, despite Ryzen btw previously there was the 4790k, well between the 7700k & 8700k what would you choose? The same goes for the new i3 vs previous gen i5 & the latest i5 vs 7700(k) so you're telling us that Intel made a whole lot of their products obsolete, jut because?
This post is off on so many levels I don't even know where to begin.

1. Let's talk about core count, shall we? AMD had "8 core" processors how many years ago? Why didn't Intel launch more cores then?
2. AMD said that Zen would improve IPC by 40%. That would've placed it around SB. They came in a little higher than that but they're still well below Skylake and KabyLake. If Intel took AMD at their word, why would they be afraid?
3. Intel launching a 6 core now isn't my "theory." As MANY others have also stated, this move was roadmapped years ago for all to see. The thing Intel missed on was Cannonlake, which was supposed to be out about now and likely would've had 8+ cores. So if anything, Intel is behind their original roadmap so AMD should count their blessings.

It is a little tiresome having to continue to argue with diehards on BOTH sides. Let's assume for the sake of argument that you're right and that Intel magically developed Coffee Lake in the span of a year or so out of "fear" of Ryzen. Guess what? Who cares and you know why no one cares? Because when Coffee Lake lands, it will likely not only beat the Ryzen 5 badly, but it could be right up there with the Ryzen 7 in MT and will slaughter it in ST.
 
Last edited:

R0H1T

Platinum Member
Jan 12, 2013
2,566
142
106
This post is off on so many levels I don't even know where to begin.

1. Let's talk about core count, shall we? AMD had "8 core" processors how many years ago? Why didn't Intel launch more cores then?
You know they weren't comparable to the current Zen or KBL, they were even sued for it ~ wrongly so IMO.

Performance, that's why, many of the resident Intel followers on this forum were also pleasantly(?) surprised by what Zen did. Would an Ivy Bridge level performance illicit a six core or 18 core response from Intel? If not then you have your answer.
2. AMD said that Zen would improve IPC by 40%. That would've placed it around SB. They came in a little higher than that but they're still well below Skylake and KabyLake. If Intel took AMD at their word, why would they be afraid?
They're not really, take a look at best case SMT performance. Scroll down to the 7zip, Winrar & AES benchmarks & see if you can spot a trend?
http://www.anandtech.com/bench/product/1932?vs=1934
http://www.anandtech.com/bench/product/1932?vs=1905
3. Intel launching a 6 core now isn't my "theory." As MANY others have also stated, this move was roadmapped years ago for all to see. The thing Intel missed on was Cannonlake, which was supposed to be out about now and likely would've had 8+ cores. So if anything, Intel is behind their original roadmap so AMD should count their blessings.
And Zen was in the works for close to half a decade. Where does this invalidate my point, as you rightly said that the 6 core was probably moved up a few quarters?

Intel probably have their 10nm delays & Zen's pricing to blame for releasing a sub $400 hexa core. You can't say that without the aggressive pricing from AMD a mainstream hexa core would've been anything less than $400 while looking at previous gen HEDT prices.
 
Last edited:
  • Like
Reactions: USER8000

IndyColtsFan

Lifer
Sep 22, 2007
33,491
553
126
You know they weren't comparable to the current Zen or KBL, they were even sued for it ~ wrongly so IMO.

Performance, that's why, many of the resident Intel followers on this forum were also pleasantly(?) surprised by what Zen did. Would an Ivy Bridge level performance illicit a six core or 18 core response from Intel? If not then you have your answer.
Intel was going to release the six core regardless. At most, Ryzen caused Intel to change their schedule a few months.

As I've said on numerous occasions here, clearly Intel did not intend to release 12-18 core HEDT CPUs this soon, but they obviously DID have them in their back pocket as a contingency plan.

BTW, I have 2 Ryzen systems. Once again, AMD is going to be stuck in a position where they have to kill their margins to be competitive with Intel (in terms of sales). This is EXACTLY what Lisa Su said they did not want to do, but they have no choice.
 
  • Like
Reactions: Phynaz

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,886
9,082
136
Its really interesting to see all this speculation. Also, nobody may believe me, but for me its all about TODAY. Where is the performance/dollar and how much juice do they take to run and cool, and how much electric and AC is required to run them. I just bought FOUR E5-2683v3 CPU's and systems in the last 6 months. Then 2 Ryzen and a threadripper. Am I biased to Intel ? No.. AMD ? NO... Just who is better the day I am buying. And I am actually excited to see the competition. I don't care who wins, I just miss the fight, and it benefits us all.
 

VirtualLarry

Lifer
Aug 25, 2001
49,531
5,601
126
Let's assume for the sake of argument that you're right and that Intel magically developed Coffee Lake in the span of a year or so out of "fear" of Ryzen. Guess what? Who cares and you know why no one cares? Because when Coffee Lake lands, it will likely not only beat the Ryzen 5 badly, but it could be right up there with the Ryzen 7.
It just strikes me as a little weird, that if this 6-core mainstream CPU wasn't a "rush job", that if it had been "on the roadmaps", then:
1) Why are we getting re-used cores, with no real IPC improvements in the cores themselves, with just layout / floorplan changes? I mean, even the 65nm to 45nm Core2 shrink had nearly a 3-5% IPC improvement. If they had so much time, why didn't they "tweak" the cores?
2) Why isn't it compatible with existing Socket 1151 motherboard infrastructure? If it had been in development / planning for as long as the "it's not due to AMD" crowd says it was, then surely, they could have made initial changes to socket 1151 to accommodate CFL-S.

These two factors indicate to me that it WAS a "rush job", and my best explanation why, is because AMD is finally competitive again.
 

TahoeDust

Senior member
Nov 29, 2011
557
404
136
Looks like it is going to be a hell of a chip for a "rush job". Crushing their competitor in single core performance, while matching it, or better, in multicore performance...with 25% fewer cores. Just imagine what intel could do if they took their time.
 
  • Like
Reactions: Kuosimodo and pcp7
Mar 10, 2006
11,719
2,003
126
It just strikes me as a little weird, that if this 6-core mainstream CPU wasn't a "rush job", that if it had been "on the roadmaps", then:
1) Why are we getting re-used cores, with no real IPC improvements in the cores themselves, with just layout / floorplan changes? I mean, even the 65nm to 45nm Core2 shrink had nearly a 3-5% IPC improvement. If they had so much time, why didn't they "tweak" the cores?
2) Why isn't it compatible with existing Socket 1151 motherboard infrastructure? If it had been in development / planning for as long as the "it's not due to AMD" crowd says it was, then surely, they could have made initial changes to socket 1151 to accommodate CFL-S.

These two factors indicate to me that it WAS a "rush job", and my best explanation why, is because AMD is finally competitive again.
Wrong.

Intel had an 8 core Cannon Lake-S on the roadmap which was cancelled in late 2015 when it was obvious that 10nm just wouldn't be ready in time. Cannon Lake-S died, and in its place was inserted Coffee Lake-S, which had fewer cores and lower IPC than the Cannon Lake part, but in exchange it had the benefit of a tuned physical design and an improved mfg process to get a solid frequency improvement.

The meme that Intel would've just done more quad core parts when Intel had clearly been planning to increase mainstream core counts since before AMD even went public on Zen really needs to die. But people will keep parroting it to suit their "evil greedy Intel" and "Good Guy AMD" narrative until the end of time.

As far as the old cores vs new cores go, backporting CNL or ICL to 14nm++ was obviously a possibility but because Intel's cores are so tightly tied to the processes that they're designed for, you would've seen a sub-optimal result. IPC would've gone up, but frequency would've probably gone down the tubes or the project would've been later to market than what we are seeing now.

Intel's 10nm mess has had serious repercussions on the product offerings, but considering what the product teams have had to work with, it could've been a lot worse -- they could've shipped a total dud like Broadwell instead of CFL (higher IPC, much lower frequency) or they could've shipped a "Skylake Refresh" with no process/physical design improvements like they did with Haswell in 2014.

Credit where it is due.
 

R0H1T

Platinum Member
Jan 12, 2013
2,566
142
106
Looks like it is going to be a hell of a chip for a "rush job". Crushing their competitor in single core performance, while matching it, or better, in multicore performance...with 25% fewer cores. Just imagine what intel could do if they took their time.
You're assuming Intel can just cram in some more IPC in their SKL derivatives, not unlike the argument that Atom failed (in mobiles or tablets) because Intel didn't pay enough heed to them, or that they weren't serious even after throwing billions at them.
 

TahoeDust

Senior member
Nov 29, 2011
557
404
136
You're assuming Intel can just cram in some more IPC in their SKL derivatives...
So, you do not think Coffee Lake will be significantly better in single core and equal or better in multicore compared to their competition?
 

R0H1T

Platinum Member
Jan 12, 2013
2,566
142
106
So, you do not think Coffee Lake will be significantly better in single core and equal or better in multicore compared to their competition?
Competition as in Ryzen ~ yes but nothing has markedly improved over SKL or KBL, except the core count. Though MT results (for 8 core Zen) will largely depend on how applications make use of AMD's SMT, in my last reply you can see AES gain over 70% with it.

As for your argument ~ the theory that Intel can improve IPC of SKL just willy nilly perhaps without sacrificing core speeds is just wishful thinking. If they could've they would've, they'll need 10nm at its best to improve performance over KBL & CFL.
 

TahoeDust

Senior member
Nov 29, 2011
557
404
136
As for your argument ~ the theory that Intel can improve IPC of SKL just willy nilly perhaps without sacrificing core speeds is just wishful thinking.
I do not believe this. I am not sure what I posted that gave you the impression I do. We will not see any significant gain in IPC until 10nm.

They do not need significant improvement in IPC to beat the competition. I will not say the name of said competition, because if I do I will probably get an infraction.
 
  • Like
Reactions: CHADBOGA

R0H1T

Platinum Member
Jan 12, 2013
2,566
142
106
I do not believe this. I am not sure what I posted that gave you the impression I do. We will not see any significant gain in IPC until 10nm.

They do not need to improve IPC to beat the competition. I will not say the name of said competition, because if I do I will probably get an infraction.
Well this is what you said ~
Looks like it is going to be a hell of a chip for a "rush job". Crushing their competitor in single core performance, while matching it, or better, in multicore performance...with 25% fewer cores. Just imagine what intel could do if they took their time.
http://www.anandtech.com/bench/product/1950?vs=1903
Granted we don't have CFL numbers atm, but if Zen isn't the competitor then SKL-X is the one you're talking about? You're also exaggerating about how good regular SKL is in ST tasks, generally 5~15% depending on the application wrt Zen whilst MT tasks, where Ryzen SMT can be fully exploited, are mostly in favor of the latter.
 

IntelUser2000

Elite Member
Oct 14, 2003
7,249
1,839
136
It just strikes me as a little weird, that if this 6-core mainstream CPU wasn't a "rush job", that if it had been "on the roadmaps", then:
1) Why are we getting re-used cores, with no real IPC improvements in the cores themselves, with just layout / floorplan changes? I mean, even the 65nm to 45nm Core2 shrink had nearly a 3-5% IPC improvement. If they had so much time, why didn't they "tweak" the cores?
The "rush job" for Coffee Lake isn't due to competition, rather 14/10nm problems were known pretty late in the development cycle and had to "rush job" a Cannonlake 8 core replacement.

The direct response to Ryzen is the 10+ core SKL-X parts. No, I don't agree with the assertion Coffee Lake is a Ryzen response either. Some are(like the really high core count SKL-X), but Cannonlake with even more cores were in the works before we even knew what Kaby Lake was.

It just happened we saw Intel screw up on the process along with AMD's success. The saying when it rains it pours is true. AMD got Athlon chips when Intel started messing around with Netburst architecture and let their ego run wild.

Cannon Lake-S died, and in its place was inserted Coffee Lake-S, which had fewer cores and lower IPC than the Cannon Lake part,
Arachronic,

Are we 100% sure that Cannonlake wasn't backported to Coffeelake? Because there's evidence that it has been too. There was a leak where it said Coffeelake, but (Cannonlake) in brackets. Initial performance leak seems to support it.
 
  • Like
Reactions: coercitiv

Bouowmx

Golden Member
Nov 13, 2016
1,059
444
146
In CPU-monkey pre-release benchmarks of Cinebench 11.5, Cinebench R15, and Geekbench 3 (minus PassMark because it is estimated, and Cinebench OpenGL for obvious reasons), Coffee Lake consistently shows ~104% score per Hz over Kaby Lake, except Core i5-8600K in Cinebench 11.5 MT and Geekbench 3 64-bit MC. This improvement is because of change in architecture? Or that the benchmarks can be gamed? Or fake news?


"IPC" ratio of multi-core benchmarks have an additional factor of dividing by 6/4 to remove the factor of extra cores. Single-core benchmarks do not have such factor.
 

eek2121

Senior member
Aug 2, 2005
728
778
136
This post is off on so many levels I don't even know where to begin.

1. Let's talk about core count, shall we? AMD had "8 core" processors how many years ago? Why didn't Intel launch more cores then?
2. AMD said that Zen would improve IPC by 40%. That would've placed it around SB. They came in a little higher than that but they're still well below Skylake and KabyLake. If Intel took AMD at their word, why would they be afraid?
3. Intel launching a 6 core now isn't my "theory." As MANY others have also stated, this move was roadmapped years ago for all to see. The thing Intel missed on was Cannonlake, which was supposed to be out about now and likely would've had 8+ cores. So if anything, Intel is behind their original roadmap so AMD should count their blessings.

It is a little tiresome having to continue to argue with diehards on BOTH sides. Let's assume for the sake of argument that you're right and that Intel magically developed Coffee Lake in the span of a year or so out of "fear" of Ryzen. Guess what? Who cares and you know why no one cares? Because when Coffee Lake lands, it will likely not only beat the Ryzen 5 badly, but it could be right up there with the Ryzen 7 in MT and will slaughter it in ST.
They aren't "well below SL or KL". Intel's top chips trade blows with AMD's chips. Where have you been? Go pull up Anandtech bench if you don't believe me. CL 6 core is a pressured response to AMD stealing marketshare. If Ryzen had not been what it is, coffee lake 6 core would have never made it out the door. Intel has been milking those cheap quad cores for years.
 

Zucker2k

Golden Member
Feb 15, 2006
1,263
631
136
Yeah, if the i5 6C/6T is indeed priced under $200, that's going to be a strong competitor, especially for gaming, with the Ryzen 5 1600/1600X CPUs, especially if it can be clocked above 4Ghz.

I don't hate CoffeeLake, quite the opposite, really. I'm thrilled that Intel is finally upping the core-counts of "mainstream" CPUs. But I do thank AMD for prodding Intel into finally doing so, after all these years, of only quad-core max on mainstream sockets. (Q6600 came out when???)

I might even get an unlocked 6C/6T CoffeeLake to mess with, or use as my "gaming" rig. But I likewise cannot deny the strong value proposition that my Ryzen 5 1600 CPUs have given me. For the same price as a Kaby Lake locked i5 4C/4T CPU, I got an unlocked 6C/12T CPU. Can't really beat that, can you? (Well, not currently.)

Edit: And then, there's this:
https://forums.anandtech.com/threads/incredible-deal-on-ryzen-r7-1800x-mircrocenter-349.2512725/

Ryzen 7 1800X for $349.99 at Microcentr.
Lacking an igp, and being slower than the Intel i7 chips in lightly threaded tasks and gaming, the 1800x should be selling well below $300 to be "viable." The faster 7700k (in gaming and light-threaded tasks) is only $279 at microcenter.
 
  • Like
Reactions: Sweepr

witeken

Diamond Member
Dec 25, 2013
3,876
154
106
In CPU-monkey pre-release benchmarks of Cinebench 11.5, Cinebench R15, and Geekbench 3 (minus PassMark because it is estimated, and Cinebench OpenGL for obvious reasons), Coffee Lake consistently shows ~104% score per Hz over Kaby Lake, except Core i5-8600K in Cinebench 11.5 MT and Geekbench 3 64-bit MC. This improvement is because of change in architecture? Or that the benchmarks can be gamed? Or fake news?
Kaby Lake Refresh has an improved ring bus, although no details given. Maybe Coffee Lake too.
 
  • Like
Reactions: Ajay

ASK THE COMMUNITY