The Cellar - View Single Post

tw · 05-19-2006, 11:08 PM

According to EE Times, vPro and Viiv are limited functions that don't transport well across the product line. This so different from a productive Intel that once created a roadmap for PC advancement (DIB) that included new video standards, PCI bus (that includes PnP abilities), North/South bridge with memory interface, USB, hardware functions that made sleep and hibernate possible, and even banged Sony Toshiba heads together so violently as to create a single DVD standard.

An industry analyst says, "vPro is not being well-received by the Pc channel partners as it adds undue cost and complexity to the enterprise market". This would be consistent with so many changes announced by Otellini that sound more like rebranding and no technical innovation. These are characteristics also found in AT&T when it started a 20 year self maculating process to cost control itself to death.

Intel's problems first became apparent when AMD introduced HyperTransport - and Intel did not even have anything in planning and eventually came out with 3GIO.

Intel remained in denial about NAND non-volatile memory. It remained with NOR technology and is said to be losing money. Recently it teamed with Micron to play catchup in the NAND market - too late. But Intel is now said to be separating memory production from processor production - step one in selling off its memory business. If true, then all this promise from Intel for OVC memory may have been mythical.

One shocking hint that Intel management does not get it. Intel intends to create a new architecture every two years instead of every four. Reminds me of another American IC manufacturer who decreed a new IC every week would make a market leader. Therefore they introduced numerous ICs that disappeared - nobody wanted them. But on spread sheets analysis, this was a perfect solution. How could the accountants have gotten it wrong? Maybe they had no idea what innovation was?

Intel's problem is not that others 'catch-up' with Intel architectures within four years. The problem is that Intel's architects since and including the P4 had pathetic designs. Since the limits of transistors (gates that are only 3 atoms thick) have created a brick wall, the Intel crown jewel (semiconductor manufacturing advances) have little room to keep advancing. This brick wall and a superior architecture is why AMD with less manufacturing abilities have now created superior high end processors.

Yes, Intel's crown jewel does mean Intel can do more with less power / less heat. But this crown jewel by itself no longer can make Intels faster than AMDs.

IOW a 'new architecture every two years' says Intel's new MBA boss does not get it. He does not understand where the problem lies and instead implements an MBA 'numbers' solution. This myopia also explains why Intel had to cancel some almost completed single core processor designs - because management finally listened to technical people too late. Intel's architecture blindly stuck to a single core design when they long ago should have realized a problem created by architectures that just were not competitive.

And then there is this blind allegiance to Itanium long after they should have had a 64 bit Pentium to fall back on. Readers of Tracey Kidder's "Soul of a New Machine" will appreciate how major this paragraph is.

Finally even long time Intel user, Dell, has finally conceded to AMD processors. Whereas Intel still has a low power Centrino dominance, Intel has lost its title in server application - high end processors - 32 bit processors that don't have the speed, no 64 bit Pentium, and the flawed Itanium that long ago would be canceled if not protected by HP.

BTW, one symptom demonstrates why Intel lost that title - Hyper Transport created by AMD and used by numerous other processor and sysetm companies such as Transmeta, Apple Computer, Cisco, PMC Sierra, Sun Microsystems, Broadcom, and NVIDIA. AMD is now doing what many years ago is what Intel did.

05-19-2006, 11:08 PM	#11
tw Read? I only know how to write. Join Date: Jan 2001 Posts: 11,933	According to EE Times, vPro and Viiv are limited functions that don't transport well across the product line. This so different from a productive Intel that once created a roadmap for PC advancement (DIB) that included new video standards, PCI bus (that includes PnP abilities), North/South bridge with memory interface, USB, hardware functions that made sleep and hibernate possible, and even banged Sony Toshiba heads together so violently as to create a single DVD standard. An industry analyst says, "vPro is not being well-received by the Pc channel partners as it adds undue cost and complexity to the enterprise market". This would be consistent with so many changes announced by Otellini that sound more like rebranding and no technical innovation. These are characteristics also found in AT&T when it started a 20 year self maculating process to cost control itself to death. Intel's problems first became apparent when AMD introduced HyperTransport - and Intel did not even have anything in planning and eventually came out with 3GIO. Intel remained in denial about NAND non-volatile memory. It remained with NOR technology and is said to be losing money. Recently it teamed with Micron to play catchup in the NAND market - too late. But Intel is now said to be separating memory production from processor production - step one in selling off its memory business. If true, then all this promise from Intel for OVC memory may have been mythical. One shocking hint that Intel management does not get it. Intel intends to create a new architecture every two years instead of every four. Reminds me of another American IC manufacturer who decreed a new IC every week would make a market leader. Therefore they introduced numerous ICs that disappeared - nobody wanted them. But on spread sheets analysis, this was a perfect solution. How could the accountants have gotten it wrong? Maybe they had no idea what innovation was? Intel's problem is not that others 'catch-up' with Intel architectures within four years. The problem is that Intel's architects since and including the P4 had pathetic designs. Since the limits of transistors (gates that are only 3 atoms thick) have created a brick wall, the Intel crown jewel (semiconductor manufacturing advances) have little room to keep advancing. This brick wall and a superior architecture is why AMD with less manufacturing abilities have now created superior high end processors. Yes, Intel's crown jewel does mean Intel can do more with less power / less heat. But this crown jewel by itself no longer can make Intels faster than AMDs. IOW a 'new architecture every two years' says Intel's new MBA boss does not get it. He does not understand where the problem lies and instead implements an MBA 'numbers' solution. This myopia also explains why Intel had to cancel some almost completed single core processor designs - because management finally listened to technical people too late. Intel's architecture blindly stuck to a single core design when they long ago should have realized a problem created by architectures that just were not competitive. And then there is this blind allegiance to Itanium long after they should have had a 64 bit Pentium to fall back on. Readers of Tracey Kidder's "Soul of a New Machine" will appreciate how major this paragraph is. Finally even long time Intel user, Dell, has finally conceded to AMD processors. Whereas Intel still has a low power Centrino dominance, Intel has lost its title in server application - high end processors - 32 bit processors that don't have the speed, no 64 bit Pentium, and the flawed Itanium that long ago would be canceled if not protected by HP. BTW, one symptom demonstrates why Intel lost that title - Hyper Transport created by AMD and used by numerous other processor and sysetm companies such as Transmeta, Apple Computer, Cisco, PMC Sierra, Sun Microsystems, Broadcom, and NVIDIA. AMD is now doing what many years ago is what Intel did.