HW News - NZXT BLD Responds to GN, Intel Facing Staff & Legal Issues Around 7nm, NVIDIA & ARM

By Published August 05, 2020 at 2:35 am

Last week, Intel dropped the bombshell that its 7nm process would be delayed. As such, much of the news discussion this week centers on that news and its various repercussions. As is evident below, the first half of our news this week will be dedicated to the Intel news show. As it relates to Intel’s 7nm delay, Intel may be dealing with a potential lawsuit, has already reshuffled its leadership and internal organization, and is reportedly preparing to do more business with TSMC.

We also have news regarding NZXT revising its BLD warranty (finally) regarding users enabling XMP with their memory, which was ridiculous to begin with. There’s also AMD’s Q220 earnings report, and more. 

Within GN, we recently took a look at a cringy waifu computer case from AliExpress, and we also compared a $1 thermal pad to a $10 thermal pad. Overall, GN has been experiencing a bit of down time as we’ve been revising some of our methodology and implementing new test tools. We posted a major piece exposing MSI’s shady and (what we believe to be) unethical review practices, too, and that’s worth watching. The improvements from these efforts should show up within the next couple weeks, so stay tuned. 

News article and video embed are below.

01:07 | BLD Charity Auction Re-Listing

Discussed exclusively in the video. We’re running a charity auction for the NZXT BLD we recently reviewed. 100% of proceeds benefit Cat Angels. We have visited this charity organization before and believe in their cause, and have personally determined the charity to be up to our quality standards for how it allocates funding and care for animals. We built a PC for them in 2019, if you’re curious to see more on them.

02:31 | Intel May Be Getting Sued Over 7nm Delay

As if Intel didn’t have enough to deal with already, it may get to add another class action lawsuit to its list. Intel delivered huge news last week, where it announced that its 7nm process was delayed until 2022-2023, with yields approximately 12 months behind internal target goals. The stock market went crazy, Intel’s stock plummeted, and its market cap shed roughly $43 billion. 

Now, Hagens Berman law firm is looking to represent investors who “have suffered significant losses” under a possible class action lawsuit. The law firm is currently investigating the possibility of securities fraud perpetrated by Intel. The investigation -- and subsequent lawsuit, should it materialize -- would hinge on proving that Intel willingly and knowingly concealed manufacturing and performance issues around 7nm from investors. 

In the law firm’s press release, the firm quoted Stacy Rasgon of Sanford C. Bernstein, who said the 7nm delay news was "the worst we have seen in our career covering the company" and “whatever little credibility they had is out of the window."

The law firm is also issuing a call to whistleblowers with non-public information to come forward in an effort to prove that Intel misled investors.  

Source: https://www.newsfilecorp.com/release/60513

04:06 | Intel Reshuffles the Deck After 7nm Delay

It appears as though some internal restructuring is in order following Intel’s news that it won’t be able to deliver 7nm on time. The biggest change will be the exit of Intel’s chief engineering officer (Murthy Renduchintala). Renduchintala was over Intel’s Technology, Systems Architecture, and Client Group (TSCG) and has been since at least 2018.

Now, following Renduchintala’s exit, Intel’s TSCG will be split into five different divisions, each with its own leader who will report directly to CEO Bob Swan. The changes are as follows:

  • Technology Development, led by Dr. Ann Kelleher
  • Manufacturing and Operations, led by Keyvan Esfarjani
  • Design Engineering, led in the interim by Josh Walden (Intel is still looking for a permanent team leader)
  • Architecture, Software and Graphics will continue to be led by Raja Koduri.
  • Supply Chain will continue to be led by Dr. Randhir Thakur

Intel states that its intentions with these changes is to improve focus and create more accountability in process technology. Murthy Renduchintala will resign on August 3, 2020. 

Source: https://newsroom.intel.com/news-releases/intel-changes-technology-organization/#gs.bbpryz

06:16 | Intel Reportedly Uping Orders At TSMC

When Intel announced its delay at 7nm, it also suggested it would be using third-party fabs wherever it made sense, in order to maintain its roadmap as much as possible. Intel already does a certain amount of business with TSMC. TSMC already makes some of Intel’s chipsets for instance, and certain non-leading edge CPUs. 

According to reports, it seems Intel could already be in talks with TSMC about reserving capacity, presumably for one of TSMC’s 7nm nodes. So far, we don’t have any indication as to what products would use TSMC’s nodes, other than Intel’s Ponte Vecchio GPU, of which Intel already said would use a combination of internal and external process technologies.

Separately, other reports (primarily out of Digitimes) are suggesting that Intel’s and TSMC’s partnership is very temporary, meaning TSMC isn’t treating Intel as a long-term customer and won’t make any additional fab space to accommodate Intel’s orders. TSMC went as far as to say that it viewed itself as more of a “rescuer” than long term partner. Time will certainly tell.   

Source: https://www.tomshardware.com/news/report-intel-increasing-orders-at-amd-partner-tsmc



09:13 | TSMC Stock Price

After Intel’s 7nm delay news broke last week, Intel’s stock took a nasty nosedive while AMD’s got a shot in the arm. While we admittedly aren’t stock experts, we know stocks are a finicky thing, but the news was interesting nonetheless.

We got a tip from a reader pointing-out that AMD’s stock wasn’t the only one that got a boost. On the back of Intel’s 7nm news, it seems TSMC’s stock shot up as well. TSMC’s stock is currently sitting just above $78 per share, and it has a market cap of approaching $400B at time of writing. As the user who sent us the tip pointed out, that puts it above massive companies such as Walmart, Procter & Gamble, as well as AMD and Nvidia. TSMC has also become one of the top 10 most valuable companies, as of this writing, and is one of the top two or three for Taiwan’s export industry.


Source: https://www.tradingview.com/markets/stocks-usa/market-movers-large-cap/

10:15 | NZXT Revises BLD Warranty

We recently took one of NZXT’s pre-builts for a test drive in our review of NZXT’s BLD service. Spoiler alert: we weren't impressed. Not long after our review, we found out via Twitter that a user was unhappy with his machine from NZXT, due to the threat of a voided warranty should he enable the XMP profile on the memory kit that came with his build. NZXT was calling this “RAM overclocking” and claimed it would void the BLD warranty, despite advertising XMP frequency, which we believe to be misleading advertising. NZXT was also refusing to accept a return for the machine.   

After the user went through several days of trouble, we weighed-in on twitter, at which point NZXT relented and agreed to honor the warranty if the user enabled the XMP profile for his memory kit. NZXT also doubled back and updated its warranty on its BLD service. However, NZXT’s blog post has some ambiguous language in it regarding XMP as overclocking that it’s trying to lean on as an excuse here. 

Part of what NZXT is saying is that overvolting equals overclocking, and that’s not true. While it is true that most XMP profiles require a slight bump in voltage to reach XMP speeds, that voltage is well below safe operating thresholds -- it’s literally been defined, tested, and supported to run at those speeds and voltages by the manufacturer, and that manufacturer -- like GSkill -- often warrants that configuration. Gskill has lifetime warranties, for instance. There are some kits that run at 1.35V with XMP, which is quite a bit above the JEDEC spec of 1.2V for DDR4, but not all do. It is feasible to get 3200MHz RAM at 1.2V, too. Also, by the way, NZXT didn’t say that manually typing in the timings on the sticks voids the warranty, just that enabling XMP did. Fortunately, you technically don’t have to “enable XMP” to get the numbers on the stick. You could do it manually.

Either way, if anyone ever actually does end up in a situation where they have to send the system back for some other reason, you could just clear CMOS or reset BIOS and no one would ever know.

It also seems that part of the problem is XMP being an Intel defined specification and pairing with an AMD processor, which is another issue that shouldn't have been. The XMP timing tables on the modules can be loaded regardless of CPU. Furthermore, enabling XMP does’t void the warranty on the memory kit; so there’s no reason it should void the warranty of your entire build from NZXT, whom you’re paying in part to consolidate warranty service. That’s the entire point

There’s a lot of ambiguity in the writeup that reads as if it was written by someone who doesn’t actually understand what’s going on, or was maybe over simplified. NZXT says things like “XMP Profiles allow you to run your high performance RAM (above industry DDR specs),” but they really need to be talking about JEDEC, because “industry DDR specs” can mean a lot of things. For instance, AMD’s Ryzen 3000 parts support 3200MHz memory stock now. JEDEC is what NZXT normally means, but JEDEC isn’t mentioned once in the piece.

NZXT also said: “Until now, we’ve regularly allowed customers to make warranty claims through customer service, even when they overclock, so it made all-the-more sense to adopt this into our official policy.”

From what we’ve seen, NZXT was actually preemptively declining warranty service if XMP got enabled, as shown in explicit support messages, so this is sort of hand-waving past behavior. It’s good that NZXT has fixed it now, but there’s really no reason to act as if they always did it this way.

Some key quotes:

“It may seem harmless, but enabling XMP is still a form of overclocking, just done automatically by enabling the XMP profile in your motherboard BIOS. Whenever the amount of voltage sent through a device is higher than stock voltage, there is a risk of hardware failure and higher temperatures. Not pretty. But if you’re willing to take that risk, we’re here for you.”

“However, we think it’s important to note that--while XMP can create some pretty cool results--it also comes with some risks.. No matter how harmless it may seem, Enabling XMP is still a form of overclocking, just done automatically through enabling the XMP Profile in your motherboard BIOS. There is always a risk to overclocking. Usually when overclocking, you will be raising the voltage on which your specific piece of hardware runs. Whenever the amount of voltage sent through a device is higher than stock voltage, there is a risk with hardware failure and higher temperatures.”

“As you can see here, the default RAM speed is 1066mhz, and after the XMP profile is enabled, the current clock speed is 1597mhz. (Keep in mind RAM is double data rate, if your current clock speed is, 1597mhz, your actual clock speed is 3197mhz).Viola! You've just enabled an XMP profile on your RAM and your RAM is currently running at advertised speeds.”

We make commentary on this in the video section (timestamped above) beyond what is said here, but overall, NZXT is taking partial correct steps with some misinformation mixed in.

Overclocking in the true enthusiast spirit often means pushing the component as far as it will go, which means running it out of manufacturer specification. It’s the reason Intel sells its Performance Tuning Protection Plans for its unlocked CPUs. XMP isn’t that, and NZXT shouldn’t pretend that it is.    

Source: https://blog.nzxt.com/updating-our-warranty-an-xmp-story/

Source: https://support.nzxt.com/hc/en-us/articles/360038630493-RAM-XMP-Guide-FAQ

21:50 | AMD 2Q20 Earnings 

Unlike Intel, AMD has a lot to be happy about. The company’s stock has been on a steady upward trend since the beginning of the year, and it continues to break its own revenue records. AMD’s Zen architecture and partnership with TSMC continue to be powerful catalysts for growth for AMD. 

AMD notched its best quarter in history, as it reported a quarterly revenue of $1.93 billion, a 26% jump YoY. AMD attributes this to a record revenue from its Computing and Graphics

Segment, which is where Ryzen and Radeon live, respectively. Quarterly revenue was up 8% QoQ, primarily driven by an increase in Enterprise, Embedded and Semi-Custom segment revenue.

AMD’s gross margin rose to 44% in the second quarter, up three points YoY. AMD also reported an operating income of $173M, up from $59M a year ago. AMD’s net income came in at $157M, up from $35M a year ago. According to Lisa Su, these figures are led by record notebook and server processor sales, and Ryzen and Epyc revenue has more than doubled from a year ago. 

AMD is reporting that it also managed to grab double digit market share in the server CPU segment, something that it’s been pursuing since it debuted its Epyc brand. “Importantly, we met our double-digit server processor market share goal as data center products accounted for more  than 20 percent of our second quarter revenue,” said Lisa Su in a conference call. 

Regarding Epyc, AMD highlighted wins with Google for its virtual machines, AWS adding more Epyc powered instances, as well as Dell, HPE, IBM, and SuperMicro also expanding their Epyc-based offerings. 

For the third quarter, AMD is forecasting a revenue of $2.55 billion, plus or minus $100M, which would be a 42% increase YoY and a 32% increase sequentially. AMD notes that the increase will be driven by strong demand for Ryzen and Epyc, as well as next-generation semi-custom solutions -- that is, the Xbox Series X and PS5. AMD is also reiterating its intent to launch both Zen 3 and RDNA 2 some time this year. 

Source: https://ir.amd.com/financial-information/quarterly-results


24:03 | Update: NVIDIA Trying to Buy ARM Soon

In an update to a story we ran a week or two ago, Bloomberg reports that NVIDIA and SoftBank, the holding group over Arm, are getting deeper into discussions. There is no formal news or announcement pertaining to this and it’s all through back alley discussions, but the Bloomberg report says, quote, “Nvidia is the only suitor in concrete discussions with SoftBank.”

As we mentioned last week, an NVIDIA acquisition of ARM, with both being major silicon players in a small space, would invite regulators in the US and EU to pick apart the deal. This could make the acquisition less likely, but it would be a powerhouse move for NVIDIA to acquire ARM, which is now working with Apple on its departure from Intel and AMD silicon. We’ll follow this story as it develops, but the only advancement is that, according to Bloomberg, NVIDIA is aggressively pursuing a deal and attempting to close within a few weeks.

Source: https://www.bloomberg.com/news/articles/2020-07-31/nvidia-said-in-advanced-talks-to-buy-softbank-s-chip-company-arm

24:57 | Fox Sports’ Virtual Crowds

If you ever wanted to know what a baseball stadium would look like if it were filled with, say, characters from The Sims rather than real people, Fox Sports can scratch that itch. While sports isn’t our usual beat, obviously, this particular story treads into the world of technology, especially as it uses animation and Unreal Engine -- a couple things we know a bit about. Plus, it’s equal parts weird and hilarious, so we felt the need to mention it.  

As we’re currently living under the threat of the human malware pandemic, crowded sports arenas simply aren’t a thing right now, and likely won’t be for the foreseeable future. So, in an attempt to normalize the televised viewing experience, Fox Sports settled on the idea of a virtual crowd. It’s definitely one of those ideas that sounds better than it looks. 

The virtual crowd was designed by Silver Spoon, an animation company Fox Sports chose to work with. According to Engadget, the virtual crowd is the result of augmented reality, camera tracking, and real-time graphics, and was all done in Unreal Engine. The virtual fans can be built with different clothes, can cheer, and can even do the wave. 

Fox Sports debuted this technology recently at the Milwaukee Brewers and Chicago Cubs game. And while the virtual crowd may not have seemed out of place if you didn’t look too close, it wasn’t without some hitches. Some of the “fans” were very obviously sitting and standing on top of their chairs, rather than sitting in them or standing in front of them. Some of the virtual fans also seemingly glitched in and out of the crowd. 

So, it seems this virtual crowd would be a natural fit for Skyrim. In fact, are we sure Bethesda didn’t design them? We’ll look into it.

Source: https://www.engadget.com/fox-sports-virtual-181529869.html

Editorial: Eric Hamilton
Host, Additional Reporting: Steve Burke
Video: Keegan Gallick, Andrew Coleman

Last modified on August 05, 2020 at 2:35 am

We moderate comments on a ~24~48 hour cycle. There will be some delay after submitting a comment.

  VigLink badge