Intel's Future Chips: News, Rumours & Reviews

Page 17 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
intel at computex, roundup:
Intel intros 'Devil's Canyon,' Pentium Anniversary overclockable CPUs
http://techreport.com/news/26544/intel-intros-devil-canyon-pentium-anniversary-overclockable-cpus
intel's rcp seems closer to the rumored prices. but we still don't get fluxless solder because of "aggressive launch" in june instead of september.

Intel continues its ‘journey’ into mobile at Computex 2014
http://vr-zone.com/articles/intel-continues-journey-mobile-computex-2014/78435.html
"core m" looks like y series broadwell for fanless operation. brw and skylake might finally bring "big cores" to the fanless devices.

ADATA DDR4 OC Module Spotted on a Working Haswell-E HEDT System
http://www.techpowerup.com/201568/adata-ddr4-oc-module-spotted-on-a-working-haswell-e-hedt-system.html
that looks like a haswell-e cpu cooled by a stock intel cooler.

EVGA Prototype X99 Motherboard On Display at Computex
http://www.techpowerup.com/201562/evga-prototype-x99-motherboard-on-display-at-computex.html

MSI Shows off its X99 Motherboard
http://www.techpowerup.com/201558/msi-shows-off-its-x99-motherboard.html

MSI Goes Black and Green with its ECO Line of Motherboards
http://www.techpowerup.com/201560/msi-goes-black-and-green-with-its-eco-line-of-motherboards.html
posted for the color scheme. 😀

Sapphire Also Shows off Atomic Z97 Motherboard
http://www.techpowerup.com/201577/sapphire-also-shows-off-atomic-z97-motherboard.html


Crucial ships DDR4 for servers, desktop modules coming in August
http://techreport.com/news/26546/crucial-ships-ddr4-for-servers-desktop-modules-coming-in-august
Panram Unveils Ninja-V DDR4 Memory Modules
http://www.techpowerup.com/201539/panram-unveils-ninja-v-ddr4-memory-modules.html
HyperX Sets Overclocking World Record Mark at 4500 MHz, Partners with HWBOT
http://www.techpowerup.com/201554/hyperx-sets-overclocking-world-record-mark-at-4500-mhz-partners-with-hwbot.html


ASUS Brings Two USB 3.1-Enabled Motherboards to Computex 2014
http://www.techpowerup.com/201569/asus-brings-two-usb-3-1-enabled-motherboards-to-computex-2014.html
MSI Also Shows off First Motherboard with 10 Gbps USB 3.1 Ports
http://www.techpowerup.com/201559/msi-also-shows-off-first-motherboard-with-10-gbps-usb-3-1-ports.html


GeIL DDR4 Memory Smiles for the Camera
http://www.techpowerup.com/201567/geil-ddr4-memory-smiles-for-the-camera.html

edit:
17 ivy bridge cpus get teh aex
http://www.techpowerup.com/201586/intel-plans-the-retirement-of-17-ivy-bridge-core-processors.html
Core i5-3330, i5-3330S, i5-3340, i5-3340S, i5-3470, i5-3475S, i5-3470S, i5-3470T, i5-3550, i5-3570, i5-3570S, i5-3570T, i5-3570K, i7-3770, i7-3770K, i7-3770S and i7-3770T.
 
Intel Devil’s Canyon hits 5.496 GHz on Air/Liquid.

http://wccftech.com/intel-devils-canyon-processors-officially-announced-core-i74790k-core-i54690k-march-4-ghz-barrier/
 
Bye bye FIVR (Fully Integrated Voltage Regulator)

http://wccftech.com/intel-abandon-internal-voltage-regulator-skylake-microarchitecture/

That could cut down on the CPU temps.
 

nice find. before going into why fivr got ditched, a few minor details stick out:
skylake platform will be called "sky bay".
p.c.h. code name - sunrise point.
diagram indicates ddr4 imc. so what happened to the earlier rumor of having hybrid ddr3/4 imc...

as for the fivr, it wasn't fully "i". the inductors were integrated while the capacitors were on-package, iirc. it was a first attempt and it was for reducing power consumption accross the lineup. IVR was one of the techniques to cope with power supply voltage fluctuations. rwt has a better explanation:
http://www.realworldtech.com/steamroller-clocking/
i suspect intel might be ... ahem.. inspired by beema and kaveri's new power saving features. i expect fivr to make a comback in the future.
 
The IVR may have been overestimated in it's improvements. Motherboard vendors seemed to be switching to better regulators around the same time. They're really paying attention to both the regulators and the cooling now.

I'd expect it to be hybrid DDR3/DDR4 just due to the ramping of DDR4. The dual launch of desktop parts still seems oddly timed but Intel loves throwing out hundreds of SKUs.
 
Taking fanless NUC to a new level.

136b.jpg


http://www.techpowerup.com/201747/streacom-also-shows-off-nc3-hexagonal-fanless-nuc-chassis.html
 


I wonder where they got that 72 number from. GT1 (10) -> GT2 (20) -> GT3 (40). The prior ones all doubled and Broadwell even added 8 more for its GT3 version (48). I was expecting 80 or even 96 for GT4.

"We presume that the GT4 has the maximum number of execution units (72). "

 
that's a bit of a pile up in 2015. at least bdw-e is confirmed for hedt.
in this tpu article, in the hsw-e slide, at the bottom, there's a mention of a hsw-e cpus with 8 cores and 3.4ghz clockrate
http://www.techpowerup.com/201936/intel-desktop-cpu-roadmap-updated.html
it's compared against a quad core hsw-k cpu at 3.5ghz clockrate. i wonder if the quad is a lower binned hsw-e or core i7 4770k. clockrate matched 4770k. in degarmo's posted link, the 5960x e.s.'s cpu-z screenshot shows approx. 3.3 ghz, but the vcore reads .978v. seems awfully low vcore for 3.3 ghz across 8c/16t.

wccdefghtech says skylake's successor will be 10nm cannonlake with union point 200-series chipset.
http://wccftech.com/intel-desktop-cpu-roadmap-confirms-broadwelle-q3-2015-broadwell-kseries-q1-2015-skylake-mid2015/
and there might not be a skylake dt with gt3/e igpu, only gt4e for the top 95w sku. the rest of them will have gt2. and (tpu says) intel will hold off k-series skylake until 2016. i guess the pesudo-cadence-stability from sandy bridge to haswell is over. now we'll be getting mixed up launches and mixed up options from mixed gens. so glad i bought ivb. amd users should be happy they bought a thuban or fx83xx cpu.
 
Review: Intel Pentium Anniversary Edition G3258
http://hexus.net/tech/reviews/cpu/70977-intel-pentium-anniversary-edition-g3258/
We've run the engineering-sample processor, provided by Intel, at the default 3.2GHz. Obtaining these numbers provides a frame for reference once we overclock it. Using the stock heatsink and dialling voltage up to 1.20V (from 1.10V) we managed to increase the base multiplier to 44x, or a final, stable speed of 4,400MHz.4.5GHz was semi-stable, falling over very occasionally, and it would likely be stable with a beefier cooler on top. We also increased the graphics frequency from 1.1GHz to 1.4GHz, thus making the most of the unlocked nature of the chip. The ensuing benchmarks have both Pentium Anniversary Edition configurations - stock and overclocked - in the graphs.
stock cooler. 4.4 ghz. hmm...

moar Intel Core i7-4790K Devil’s Canyon Processor Reviews
http://www.legitreviews.com/intel-core-i7-4790k-devils-canyon-processor-review_143880
http://www.guru3d.com/articles_pages/core_i7_4790k_processor_review,1.html
http://www.phoronix.com/scan.php?page=article&item=intel_corei7_4790k&num=1
 


If AMD doesn't release a Steamroller version of the Athlon X4 750K soon that G3258 will likely take over the budget gaming CPU spot.
 
just like that time...

Leaked BIOS Enables Pentium Anniversary Edition OC on Some MSI H97 Boards
http://www.techpowerup.com/202040/leaked-bios-enables-pentium-anniversary-edition-oc-on-some-msi-h97-boards.html

edit:

Some details of Intel Broadwell-DE processors
http://www.cpu-world.com/news_2014/2014061301_Some_details_of_Intel_Broadwell-DE_processors.html
halfway between an atom based soc and low power xeon "big core". 8 cores, HT, 12MB llc with ddr3/4 support. the confusing part is that it supports 2 DIMM per channel. it'll be bga-packaged.
 
Guess I overlooked that before. Silvermont is already getting pushed aside for a trimmed down and I'm guessing lower clocked Broadwell.

So much for their small cores.
 

i recently discovered arm has imposed limitations in their current sbsa spec so that core count won't go higher than 8 per die (or similar).

i think intel has similar limitations in space for their cpus. atom core socs might be limited to protect "bigger" xeons. atoms might not get robust server oriented features like big core xeons. imo, those limitations will be arm's advantage in terms of features and performance. bdw-de has same core count as 8 core atom but has much better features. according to that roadmap, intel still has 14nm atom core in development (prolly show up in next xeon phi, too) but they brought bdw down just the right amount to put more pressure on a57 socs.
 


I read that wrong. Was thinking Broadwell-DE meant DEnverton. 😉

Haven't heard of this 8 core ARM limit. LSI already has a 16 core ARM part and Cavium just announced a 48 core ARM part. ARM offers different grades of cache-coherent interconnects so maybe that's what you saw.
 

according to the s/a article i read, arm has limited core count on sbsa level 1 and 0 because of the interrupt controller supporting up to 8 cores/soc. on level 2.0 it supports gic v 3.0 which massively increases core support. cavium built a custom sbsa lvl 2 compliant (gic v 3.0 based) interconnect in the c.p.i. that connects the 48 cores. i looked into gic and sbsa from older articles and s/a's analysis lines up.
http://semiaccurate.com/2014/06/03/cavium-thunder-x-ups-arm-core-count-48-single-chip/
linux already has gic v 3.0 support.
 
Status
Not open for further replies.