mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware

Reply
 
Thread Tools
Old 2017-10-06, 12:05   #1
VictordeHolland
 
VictordeHolland's Avatar
 
"Victor de Hollander"
Aug 2011
the Netherlands

117610 Posts
Default Intel Toffee Cake launched - mainstream now 6 core

Intel Coffee Lake (8th gen Core) launched

i7 = 6C12T
i5 = 6C6T
i3 = 4C4T

Finally more cores for the same price from Intel! I wonder if this is a response to AMD Zen (Ryzen) or if Intel was planning this all along.....

Initial review:
https://www.anandtech.com/show/11859...nitial-numbers
VictordeHolland is offline   Reply With Quote
Old 2017-10-06, 13:26   #2
M344587487
 
M344587487's Avatar
 
"Composite as Heck"
Oct 2017

3·52·11 Posts
Default

Quote:
Originally Posted by VictordeHolland View Post
Intel Coffee Lake (8th gen Core) launched

i7 = 6C12T
i5 = 6C6T
i3 = 4C4T

Finally more cores for the same price from Intel! I wonder if this is a response to AMD Zen (Ryzen) or if Intel was planning this all along.....

Initial review:
https://www.anandtech.com/show/11859...nitial-numbers
The release date at least seems to be a response to Ryzen, it's not a paper launch AFAICT but the supply does not seem to be there to meet demand (for the longest time intel has been very good at keeping supply stocked, this is out of character). Maybe it's a short term issue, it's early days so we'll see how the situation pans out in a few weeks.

6 core mainstream is great, it's been a long time coming even if the majority aren't hampered by 4 core. I'm interested in how this lineup (and its pricing) affects the Ryzen lineup. I think Ryzen 3 will drop more than the average to stay competitive with price/perf at the low end. I wouldn't be surprised if the Zen+ lineup is simplified to Ryzen3+ 4C8T, Ryzen5+ 6C12T, Ryzen7+ 8C16T, allowing Ryzen3+ to be a little more on par with i3. Maybe they will use 4C4T Zen+ parts as a Pentium equivalent, or in APUs.
M344587487 is online now   Reply With Quote
Old 2017-10-06, 14:11   #3
VictordeHolland
 
VictordeHolland's Avatar
 
"Victor de Hollander"
Aug 2011
the Netherlands

23·3·72 Posts
Default

Quote:
Originally Posted by M344587487 View Post
I wouldn't be surprised if the Zen+ lineup is simplified to Ryzen3+ 4C8T, Ryzen5+ 6C12T, Ryzen7+ 8C16T, allowing Ryzen3+ to be a little more on par with i3.
That would be a likely response. but according to rumours the yields of Ryzen are very good. I wouldn't be surprised if they go something like:
R7+ 8C16T
R5+ 8C16T (lower clocked than R7)
R3+ 6C12T
VictordeHolland is offline   Reply With Quote
Old 2017-10-06, 14:20   #4
mackerel
 
mackerel's Avatar
 
Feb 2016
UK

3·5·29 Posts
Default

It has been speculated Intel bought forward the launch, which was due to be early next year. This alone would help keep some from switching to Ryzen. Problem is, they simply don't have the stock. When I looked yesterday, the 6 core parts were few and far between, and the few places that had the K versions were milking them, priced way above expected levels. The worst offender was offering 8700k binned for 5.0 at £500 and 5.2 for £800. NOPE! I just want one at expected £360 and I'll take my chances on OC.

Instead, I went to a different retailer and I have in my possession 8350k + mobo. It's only quad core but I want to see if there is much more OC headroom on them, and if I can detect any IPC difference at all (over Skylake).
mackerel is online now   Reply With Quote
Old 2017-10-11, 03:58   #5
thyw
 
Feb 2016
! North_America

22·3·7 Posts
Default

The mobos seem like such a milking, new platform, incompatible with the SL and KL. (CL only works in the new mobo, SL/KL only works in the older ones.)
But uses the same socket (1151) just to add confusion, and only the highest tier(Z370) is available lower ones (H and B) maybe come in 2018 (Q1/H1).
They cannabalized their latest platform (there are plenty of Z270 stocked everywhere, nobody will buy them for the "obsolete" "4core max" platform) in favor of competing with Ryzen. They couldn't stand missing out on Christmas, Ryzen would get even more marketshare.

Last fiddled with by thyw on 2017-10-11 at 04:14
thyw is offline   Reply With Quote
Old 2017-10-12, 14:59   #6
Lemonlurker
 
Lemonlurker's Avatar
 
Aug 2017

78 Posts
Default

Quote:
Originally Posted by mackerel View Post
When I looked yesterday, the 6 core parts were few and far between, and the few places that had the K versions were milking them, priced way above expected levels. The worst offender was offering 8700k binned for 5.0 at £500 and 5.2 for £800. NOPE! I just want one at expected £360 and I'll take my chances on OC.
I don't like this train of thought. Is it so unfair that some company takes their own resources to buy a bunch of cpus, overclock and test them all, and then sell someone a product at a guaranteed overclock rather than luck of the draw if you buy a random one from intel?

I'd only call that milking them if they had been selling random 8700k's they didn't do anything with for those prices.
Lemonlurker is offline   Reply With Quote
Old 2017-10-12, 15:46   #7
fivemack
(loop (#_fork))
 
fivemack's Avatar
 
Feb 2006
Cambridge, England

642310 Posts
Default

Quote:
Originally Posted by Lemonlurker View Post
I don't like this train of thought. Is it so unfair that some company takes their own resources to buy a bunch of cpus, overclock and test them all, and then sell someone a product at a guaranteed overclock rather than luck of the draw if you buy a random one from intel?

I'd only call that milking them if they had been selling random 8700k's they didn't do anything with for those prices.
I agree with lemonlurker; I'm assuming those prices are a function of the fraction of CPUs that don't overclock that far. It seems completely reasonable for a company (overclockers.uk!) to take their whole small allocation, do the testing and sell at a relevant price.
fivemack is offline   Reply With Quote
Old 2017-10-12, 23:42   #8
mackerel
 
mackerel's Avatar
 
Feb 2016
UK

3·5·29 Posts
Default

I wouldn't have a problem with that if there was general availability of unbinned product at normal price. There isn't. You can argue they are generating value by offering some beyond stock performance level. In constraint when there is no choice, it does feel like a money grab. Doesn't matter if it is a notorious store or some ebay flipper doing it. I also note another German seller has gone the same route. The UK seller does not indicate any criteria on their sales pages on how they test for stability. The German one does use an ancient version of Prime95, I'm guessing one that pre-dates FMA3 without checking version, so basically not much of a stress at all. Also, another Uk retailer was selling the 8600k but for almost as much as the 8700k price should be. Note it is not that I don't understand why these stores are doing it, but with Intel's practically paper launch it is frustrating not to be able to get one of these CPUs at a reasonable-for-Intel price. The other annoyance is that the 7800X isn't badly priced either if you want 6 Intel cores right now. The only problem I have with that is the new cache configuration which simply doesn't play well with existing consumer workloads as the more traditional approach. I don't want those FP weak Ryzen cores either. #firstworldproblems

I only have a sample of 1 to go on, but the 8350k I got has been stable at 5.0 GHz for everything I've tried except Prime95 FMA3, which required dropping to 4.6 GHz at current test voltage. I'm a bit uncertain as to exactly what that voltage is, but best guess is about 1.40v based on mobo manufacturer's own utility. Other tools report unbelievable values. Based on what we saw with Kaby Lake before it, 5 GHz doesn't seem much of a stretch. With supply constraint, it will take some time before we get some real user stats on OC distribution of these. I haven't had any usable stability at 5.1 so far.

For small FFT running one each core, I've confirmed IPC is no different from Skylake-S. I'm getting lower than expected performance for either of large FFTs or multi-thread, that I've so far not been able to explain. I do have more things to test like a clean OS install (I was initially recycling an old one to save time) and maybe swap ram too. I don't think ram differences account for it, but I need to do this to rule it out for sure. Note the 8350k has 8MB of L3 cache like 6700k I'm comparing against, both fixed at 4.0 GHz. The 6700k does run a higher cache clock, but increasing same on 8350k only made a small difference and didn't significantly close the gap. The clean OS I've already done one run on but I've yet to compare results. I usually take at least two runs and take the best of each to mitigate against Windows background processes interfering. It happens rather too often with Windows 10, less so with Windows 7.
mackerel is online now   Reply With Quote
Old 2017-10-15, 22:22   #9
mackerel
 
mackerel's Avatar
 
Feb 2016
UK

1B316 Posts
Default

Minor update on the 8350k vs 6700k large FFT results. I redid some testing, found some more mistakes and background Windows things, and I'm now happy enough that with both CPUs at 4 GHz, as are their caches, and the same ram fitted, they are giving close enough results to each other at 4096k FFT 1 worker. I got up to 225 iter/s on one, 220 on the other. Maybe there are minor differences in the ram timings picked by each bios. My confusion stemmed from previously getting 240 iter/s on what I thought was the 6700k system used this time. Maybe I got it mixed up with my other 6700k system which has much faster ram.
mackerel is online now   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Intel Xeon-D-2100 series launched VictordeHolland Hardware 0 2018-02-11 22:50
Intel i7-980X (among other chips) launched ixfd64 Hardware 38 2010-08-25 12:26
Intel i7 ("Nehalem") chips launched ixfd64 Hardware 34 2008-11-25 18:22
Optimising work for Intel Core 2 Duo or Quad Core S485122 Software 0 2007-05-13 09:15
Intel Core Duo processor drew Hardware 5 2006-05-29 07:00

All times are UTC. The time now is 08:12.


Mon Aug 2 08:12:48 UTC 2021 up 10 days, 2:41, 0 users, load averages: 1.14, 1.27, 1.38

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.