I think Intel could be more aggressive in pushing desktop performance on the HEDT platforms with more cores/higher TDPs.
The IamA did not contain any solid info on the question I asked; I.e. if there was any concrete evidence that Skylake Y will be able to sustain higher clocks without throttling (compared to Broadwell Y).
But there was some other interesting comments though, like this one:
So it looks like Intel's focus on low power will continue. Good for mobile, but not so great for enthusiasts and alike that mainly would like to see better desktop CPU performance.
Yeah, BlackBerry phones are so in vogue
I think Intel could be more aggressive in pushing desktop performance on the HEDT platforms with more cores/higher TDPs.
Most notably, running my mem at 3000 15-15-15-35 1.35v (XMP) vs. default 2133 settings causes my CPU to get hotter. About 10 C hotter under load. Does this sound right? I'm not used to CPU being affected so much by mem settings, reminds me of AMD64 memory controller days.
I would like that too.
However I'm not sure higher TDP would give much improved performance. The TDP goes up quite drastically when increasing frequency further from the current levels. To get significantly higher performance when increasing TDP I think it would take a process tech and uArch designed from scratch with performance as number one priority. And as we all know, Intel's focus is not on that but on mobile.
Also, are suggesting a lower price too? Because even the 8 core CPUs are already priced at $1000, so not that many buy it. I don't believe many enthusiasts would by a 12/16 core CPU if priced at e.g. $2000 either.
Unfortunately I don't see a lower price happening under current conditions, because it would cannibalize on Intel's server CPUs where Intel have very high margins.
Perhaps if Zen can give them some competition in 2016 though, pushing the price downwards...
I noticed the same issue when running Prime95. My CPU uses 12 more watts at full load going form 2133 to 3000 RAM. Crazy right? Just goes to show how much of a bottleneck the slower DDR4 RAM is.
In the Skylake review we stated that 4.6 GHz still represents a good target for overclockers to aim for, with 4.8 GHz being indicative of a better sample. Both ASUS and MSI have also stated similar prospects in their press guides that accompany our samples, although as with any launch there is some prospect that goes along with the evolution of understanding the platform over time.
In this mini-test (performed initially haste pre-IDF, then extra testing after analysing the IGP data), I called on a pair of motherboards to provide a four point scale in our benchmarks. Starting with the 4.2 GHz frequency of the i7-6700K processor, we tested this alongside every 200 MHz jump up to 4.8 GHz in both our shortened CPU testing suite as well as IGP and GTX 980 gaming. Enough of the babble – time for fewer words and more results!
Corsair DDR4-2133 C15 2x8 GB 1.2V or
G.Skill Ripjaws 4 DDR4-2133 C15 2x8 GB 1.2V
Unless quad channel, those octocores and Zen (If we believe that pipedream) would be so bottlenecked you couldnt get any meaningful performance out of it. Just look at Skylake what it takes before the CPU gets to run its full potential. Skylake-E/EP will move to 6 channel memory for the same reason.
Unfortunately they didn't learn the first time:
@ Sweepr Yeah pretty weak you'd think someone clocking up that high would not use crap DDR4 settings.
Skylake will stretch its legs with time and people will see the perf. gains.
DDR4 already has what, 3600 kits being offered, think we saw 4000 and 4200 demo'd also?
Haven't clocked memory in a while but with BLCK settings I'm looking forward to it this weekend
Unless quad channel, those octocores and Zen (If we believe that pipedream) would be so bottlenecked you couldnt get any meaningful performance out of it. Just look at Skylake what it takes before the CPU gets to run its full potential. Skylake-E/EP will move to 6 channel memory for the same reason.
And that would mean boards at twice the cost and locking out the 5x5 and ITX factor.
It does improve performance, yes, but it's not like it's a totally decisive factor.
And anyway, we're just seeing the early generations of DDR4. Later generations will provide even greater bandwidth per channel. I wouldn't be surprised if we see DDR4 that is twice the speed of current 2133 MHz not too far from now.
And 4 channels x 2133 Mhz is effectively the same as 2 channels x 2x2133 MHz. So then dual channel will do just fine with 8 cores.
EDIT: I see there already is 2x2133=4266 MHz DDR4 RAM available. Too costly for now though, but price will come down.
And 4 channels x 2133 Mhz is effectively the same as 2 channels x 2x2133 MHz. So then dual channel will do just fine with 8 cores.
If you need 3000+ today with 4, then 4266 with 8 isnt enough.
Neither did you from what is posted in this thread by users.
You think that it would had clocked that high with the memory controler dealing with fast RAM..?.
Well, Skylake-E will be 6 channel. And it will be rated at 2400mhz from what we are hearing now. So you will need pretty fast RAM to offset that bandwidth with dual channel.
Neither did you from what is posted in this thread by users.
You think that it would had clocked that high with the memory controler dealing with fast RAM..?.
Neither did you from what is posted in this thread by users.
You think that it would had clocked that high with the memory controler dealing with fast RAM..?.
Nice ad hominem.
Some impact on max OC from the added thermals. But I think there's nothing fundamentally hindering SKY from reaching top clocks with fast mem.
The ASUS guru, Raja, had posted 4.85 Ghz stress testing w/ mem @ 3600-something CL17 irc.... not bad.
I'll let you all know once I get my 6700K setup put together. Using 2x8GB of DDR4-3000
Not bad at all but this is the guy that said that Prime 95 is no good for stress tests, so his result is to take with some salted grains..
Not bad at all but this is the guy that said that Prime 95 is no good for stress tests, so his result is to take with some salted grains..