Sorry for being about a month late to this news. Apparently, someone got their hands on an Intel Core i7-5960X and they wanted to see its eight cores. Removing the lid, they found that it was soldered directly onto the die with an epoxy, rather than coated with a thermal paste. While Haswell-E will still need to contend with the limitations of 22nm, and how difficult it becomes to exceed various clockspeed ceilings, the better ability to dump heat is always welcome.
Image Credit: OCDrift
While Devil's Canyon (Core i7 4970K) used better thermal paste, the method used with Haswell-E will be event better. I should note that Ivy Bridge-E, released last year, also contained a form of solder under its lid and its overclocking results were still limited. This is not an easy path to ultimate gigahertz. Even so, it is nice that Intel, at least on their enthusiast line, is spending that little bit extra to not introduce artificial barriers.
Could someone explain why the
Could someone explain why the picture seems to show a 12 core cpu ?
Binning. The way CPUs are
Binning. The way CPUs are produced is to manufacture the high end (Xeon) chips and the ones that dont meet the specs for the xeon line can be used for lower level chips by disabling features/cores. This reduces the cost of production by allowing for low yields for the very high end CPUs for an LGA socket whilst not needing to throw away the ones that are not quite up to spec.
Basically the CPU is a 12 core Xeon with some cores disabled because one or more of the disabled cores were damaged or required high voltages to be stable.
POWER 8 blah blah
POWER 8 blah blah
“they found that it was
“they found that it was soldered directly onto the die with an epoxy”
I’m no native english speaker, but how do you solder with epoxy? Lol
The chips are soldered. With an added bit of epoxy around the core to hold it in place when mounted to the PCB. Two different adhesive techniques, for two different purposes.
Yeah.. that part confused me
Yeah.. that part confused me as well.
It confused me too, so I
It confused me too, so I spent a significant amount of time researching it last night. Websites like this one implied that certain epoxies exist which cure in such a way that is considered "soldering", rather than a paste. It seemed to be quite similar to thermal paste but, with multiple websites reporting it in a similar way, I figured it was just my inexperience. I could have been wrong, though.
Is this supposed to be news
Is this supposed to be news that Haswell-E is using the same thermal interface as Ivy Bridge-E and Sandy Bridge-E? It’s an -E part. This is how it’s always been done.
Yes Intel does not design
Yes Intel does not design CPUs for the enthusiast market, Intel takes the Binned server CPUs, and fuses off any non already broken cores along with the malfunctioning ones, to make their E chips. More of a way to recoup some of the investment in their high margin server SKUs. To this end, Intel will continue to forgo any meaningful architectural innovation, or competitive integrated GPU improvements, as evidenced by the slowdown in intergenerational performance improvements on its mainstream consumer SKUs. There has been more intergenerational improvement in the Custom wide order superscalar ARMv8 ISA based designs by Nvidia and Apple, along with better integrated GPUs, Nvidia’s SOC GPU uses a discrete desktop grade GPU microarchitecture, that utilizes the full desktop versions of OpenGL, etc., and Apple Uses a Licensed GPU from Imagination Technologies. It should be noted that Imagination Technologies’ PowerVR wizard GPU is the first GPU, integrated or discrete, to include integrated ray tracing circuitry. Intel will continue to use the process node advantage, but even this process advantage is not enough to halt the competition in mobile market, from the custom ARMv8 ISA based SOCs, that even though they are fabbed on a larger process node, still best Intel in low power usage.
Intel can not take the x86 microarchitecture and use it to capture the mobile market with its process node advantage, and those custom wide order ARMv8 ISA based SOCs are approaching Intel’s core i3 in performance metrics, and besting Intel in graphics. There is a definite limit to what can be done with x86’s CISC microarchitecture, in the mobile market, with the RISC designs better able to be engineered for more processing power(IPCs) while retaining the RISC based design’s better intrinsic power usage advantage. If you want to know just what a RISC design can be beefed up to do, just look no further than the Power8 CPU, and compare it to any Xeon server SKU. It’s just a matter of time before one of the Power8’s licensees takes the design, and along with a compatible motherboard, with plenty of PCI 3.0 lanes(40+), and enters the gaming market with a high performance CPU competitor, the first licensed Power8’s are scheduled to begin arriving in 2015. It will only take someone like a Google to begin utilizing the Power8, to start an economy of ARM type scale around the Power8 ISA/IP, and introduce an ARM style lower pricing around the Power8 server SKUs, and PC SKUs that will spring up like mushrooms, to compete with Intel across the board, for the PC/laptop/server market.
That’s not a desktop
That’s not a desktop Haswell-E CPU. The 5960x only has 8 cores on the actual die (per intel’s unreleased slides).