Editor’s Opinion: A Culture of Information

As an aside to today's announcement, I had a few thoughts on how Intel releases product information. Seeing as this ventures closer to opinion/editorial than news & analysis, I felt it best not to mix it up with the key facts of the Whiskey/Amber Lake announcements. Nonetheless, I wanted to share my thoughts to give everyone a bit more insight into how information sharing has been changing over the past few years.

For readers that regularly follow us, you will note that with each and every generation, Intel has been less than forthcoming with details about new launches. In some aspects, such as the enterprise team, that trend is slowly reversing, but for this launch, almost all the technical info came in two slides, and for most of the specifications we had to request follow-up questions. The data we used to get in a slide deck in previous years has now been relegated to ‘ask if you care about it’, which is a worrying policy from my point of view.
 
For example, if you are wondering where information on the integrated graphics is, well, we’re waiting on it because it wasn’t provided in the group briefing. Info such as the name of the integrated graphics (UHD xxx), number of execution units, base frequencies, turbo frequencies – all of which used to be standard fare in previous generations. As did the per-core turbo frequencies. We also ask for new information these days as our understanding of products increases, such as PL2 data.
 
Perhaps the best example of how Intel has changed is that Intel didn't even disclose information on the underlying microarchitecture or manufacturing node until it was asked. Information that used to be at the forefront of a presentation has been replaced with marketing, and said information is now left at the end.
 
This isn’t a direct attack on Intel - we are constantly engaging with the people we speak to at the company on the way that they disclose materials like this, encouraging them to be more forthcoming on day one, as the company used to be. The differences between notebook, desktop, and enterprise disclosure are down to the different product teams deciding individually what to disclose, rather than a common disclosure set running through the whole company.
 
Intel’s reaction to this, from the people we speak to, has always been one of co-operation. They have been honest when they are told they can’t disclose information, even if we ask every time because the information is arguably trivial to obtain elsewhere (we would rather Intel was the source, given that it is Intel’s product). The way Intel is going about the marketing message for these new platforms is similar to that of how Intel marks new generations of products: people aren’t interested in names, or specific features. This is why we now have multiple manufacturing nodes and microarchitectures all under the ‘8th Generation’ branding. Products are sold on capabilities and user experiences, not in the fine minutiae of technical specifications – and this I do not doubt.
 
However Intel has historically been a company that has delved deep into details consistently over the years, and that seems to be fading – for a company that takes pride in its engineering, it would be great to offer engineering details to the customers and analysts that track its progress.
 

Intel Launches Whiskey and Amber
POST A COMMENT

73 Comments

View All Comments

  • Death666Angel - Wednesday, August 29, 2018 - link

    Why are you not getting this? You are not contradicting anyone or adding anything to the conversation. His problem (and mine and others) with the variable TDPs is that manufacturers will not disclose them anywhere on the product pages or packing materials (most likely). Hence, it will be difficult to impossible to assess the real performance of product a, b or c based on the specs, because the specs might be incomplete. One more reason to wait for reviews, but reviews of more budget oriented notebooks are few and far between. So we can either buy it blind, test it and hope it performs as needed and if not ship it back. Or rely on sometimes unknowledgeable consumer reviews. Overall, while there might be good reason to implement variable TDP values, not disclosing them and Intel not forcing manufacturers to disclose them is not a good thing for consumers. Reply
  • ballsystemlord - Tuesday, August 28, 2018 - link

    Hey Ian! 25W @ 1.8GHz for the i5-8265U is out of line with the other results. You might have made a mistake there.
    Thanks, and I agree with, and love your "Editor's opinion" section!
    PS: I do buy based on those minute technical details, look at Nvidia gsync+hdr on Pascal vs. Freesync+HDR on AMD GPUs. Those details *do* make a difference!
    Reply
  • ajp_anton - Tuesday, August 28, 2018 - link

    "Here Intel told us that the OPI speed is running at 4 GT/s."

    Does this mean that already a single TB3 controller will be bottlenecked to half its speed by the "OPI"? Or an x4 m.2 drive.
    Reply
  • Byte - Tuesday, August 28, 2018 - link

    The ULV processors really need help. I've been using a Yoga 1 with an Ivy bridge i5 since release, its feeling quite slow nowadays, I was able to get a HP Probook with a Skylake 6200U and it feels barely faster. My desktop skylake i7 feels way faster than my i7 ivy bridge both o/c to 4.4GHz. We need some better cooling for these chips to give them the 25W they need. Reply
  • repoman27 - Tuesday, August 28, 2018 - link

    Not that those things won't get bottlenecked, but OPI is an x8 link running at 4 GT/s, so same effective bandwidth as PCIe 3.0 x4. Reply
  • ajp_anton - Wednesday, August 29, 2018 - link

    I somehow missed the article saying it runs at PCIe 3.0 x4 speeds. But it kind of feels odd, as I thought the chipset here was the same silicon as on desktops, just packaged differently, and the desktop DMI is x4 wide. So this mobile chipset is using an x8 wide OPI? Reply
  • repoman27 - Wednesday, August 29, 2018 - link

    Yes. The U and Y processors bring the PCH on package, ditch the PEG lanes, and use OPI x8 instead of DMI x4. For such short trace lengths, using a wider, lower-clocked interface probably offers several advantages, most notably lower power consumption. Reply
  • repoman27 - Wednesday, August 29, 2018 - link

    The PCH is probably no different than on the desktop, but it’s flip-chipped directly onto the same interposer as the CPU, so it may be exposing an entirely different interface than the DMI used by the two chip solutions. Reply
  • zamroni - Tuesday, August 28, 2018 - link

    Does it have hardware fix of meltdown and spectre? Reply
  • GreenReaper - Tuesday, August 28, 2018 - link

    All signs point to no. From the article: "the new chips are not protected in hardware, for those wondering". Reply

Log in

Don't have an account? Sign up now