Intel is spreading FUD about supposed Ryzen 4000 performance drops on battery | Hacker News Reader

Intel is spreading FUD about supposed Ryzen 4000 performance drops on battery

extremetech.com

118 pointsexar081532 comments

Same behavior from Intel, using MBA tricks instead of innovating.

[1] Intel "'Forgot' to Mention 28-Core, 5-GHz CPU Demo Was Overclocked to such an extreme that it required a one-horsepower industrial water chiller" https://www.tomshardware.com/news/intel-28-core-cpu-5ghz,372...

[2] Intel decides Cinebench "which has a really niche usefulness" in their Tiger Lake presentation https://www.youtube.com/watch?v=aFHBgb9SY1Y

[3] Intel allegedly paid manufacturers "rebates" to limit their purchases of AMD, ended up paying $1.25B to AMD and $1.45B more to the EU for the antitrust practices. https://en.wikipedia.org/wiki/Advanced_Micro_Devices,_Inc._v....

From the article:

  Intel’s explanation for why AMD CPUs lose so much performance
  on battery is that the systems wait for 7-10 seconds before
  engaging turbo mode

  ... the settings that control the amount of time before turbo
  modes engage and the overall performance delta between AC and
  DC power are settings that the OEM controls, not AMD.

  ... Intel’s five comparison systems for itself came from MSI,
  Lenovo, Intel itself (in the form of a laptop kit), and two
  from HP. Intel is drawing on a much wider range of manufacturers
  for its own systems. I don’t know anything about the NUC laptop
  kit - haven't had the opportunity to test one - but I would have
  preferred the fifth system be a standard commercial comparison,
  and the AMD systems should have been drawn from an equally diverse
  pool of hardware as the Intel ones were. There are four
  manufacturers represented for Intel, and two for AMD.

  ...

  When Intel gave its presentation, it made a point of calling
  out the fact that Cinebench R20 doesn’t show the same behavior
  as the other benchmarks it had chosen to highlight.

  The "oddly" is straight-up FUD. Cinebench R23 also doesn’t
  show the 30-48 percent pattern of decline Intel claims. Neither
  does Corona Render. Neither does Handbrake. Neither does
  JetStream 2. Neither does Blender 2.90. Neither does the
  Blender 1.0Beta2 benchmark (not shown, but I ran it).

There are also some recent suspicions that Intel may be playing games with the Math Kernel Library (again?): https://level1techs.com/video/intel-throttling-mkl-performan... .

Intel's "Chief Performance Strategist" Ryan Shrout is not universally well-esteemed: https://www.youtube.com/watch?v=Uw0ZzA9wTFE .

When you listen to AMD, or at least the new AMD under Dr Lisa Su, it is all about their drives, passion, innovation and trying to succeed while building a brand that enthusiast loves and working hard to crack the Enterprise Market. Not a single mention of Intel.

When you listen to Intel, whether it was tech forum, Investor meeting or marketing presentation. They seems to talk more about AMD than their own offering.

Exec loves these FUD. Business purchase made up to 50% of the PC market. HyperScaler have long lead time and planning of new instances which makes Intel's future Roadmap, promise, marketing and FUD works better. Delaying the purchase of AMD at all cost.

As I have been saying, Intel's marketing machines has been working in full force since last year. From Leaks ( notice the amount of Intel leaks coming out are two to three times the volume over the past 12 months ? ), PR, Investor notes to sales partner's promises, Channel Clearing discount etc. It is the same playbook as they did in Athlon 64 era, and credit where credit is due they are exceptionally good at it.

>ended up paying ... $1.45B more to the EU for the antitrust practices

You wish. Like Microsoft (also never paid the ~$1B fine) they are still trying to weasel their way out of this one. https://www.pcgamer.com/intel-is-still-fighting-a-10-year-ol...

Thanks, I missed that. Good to know.

My 4750u (in T14 thinkpad) doesn't go to 4GHz clock in the first 10s of load when on battery. After 10s it jumps to 4GHz+. Which means that for my bursty workloads like running unit tests in python, linters, etc. I get 60% of performance. To circumvent that I often use external powerbank just to make it 'plugged in'. Running kernel 5.9 btw. Annoying AF.

You can tune power management behavior to suit your needs.

I'd appreciate a resource on how to do so.

Commenting to follow up if op delivers.

What operating system are you using?

First step in Windows is this https://www.instructables.com/How-to-Adjust-CPU-Frequency-in...

In Linux you can control it much more gradually https://wiki.archlinux.org/index.php/CPU_frequency_scaling

>First step in Windows is this

It's even easier than that. Just click the battery icon in the taskbar and there's a slider that goes from "best battery life" to "best performance". Choosing the latter basically forces the highest boost frequency available.

Does that work reliably with Ryzen processors? I've only had Intel laptops and it's a real pain to get them to not throttle regardless of temperature, power management and being plugged into the AC.

The steps are something like the following:

Install ThrottleStop Disable "BD PROCHOT" Set a mild undervolt via FIVR -> CPU Core Voltage -> Unlock Adjustable Voltage Enable FIVR -> Memory -> Disable and Lock Turn Power Limits Increase Turbo Boost Long Power Max, Turbo Time Limit and PP0 Current Limit

If I replace my existing XPS 13 (1065G7) I'll definitely consider AMD if throttling is less of an issue.

In my limited experience there is much more control over AMD CPUs, but it could have more to do with ASROCK bios options. It took me a while to learn about all the controls that were there and it was a bit of a double edged sword that I need to learn them at all (while building a very small computer). You can control the throttling at multiple temperatures if you want, but I opted for just throttling at a much higher temperature (80C or 85C).

GNU/Linux. As in my top post - kernel 5.9. Arch's wiki page is close to useless for Zen2. Renoir currently reports comically low `scaling_available_frequencies` - in my case: `1700000 1600000 1400000` even when plugged in.

Would you like to know more?

Yes... Yes I would.

Checkout CPU Freq Utils: http://www.thinkwiki.org/wiki/How_to_use_cpufrequtils

You can tune all these things, the trade off is of course performance and battery life.

One of the things that sucked about working for Intel is that you just knew you were working for a company that had the institutional personality of a complete raging arsehole. HR people would explain to you that whilst yes, those floating point units did demonstrably produce the wrong answers, Intel as an organisation doesn't accept that it did anything wrong, customer expectations were simply misplaced.

Geopolitically Intel is probably going to need to get bailed out, but it's a toxic company that could do with disappearing. I won't be mourning their Sales & Marketing teams.

From the outside I have to say one thing. If Intel went away we would all lose.

This isn't like the phone company where somebody has won and consumers take the brunt of it.

This is a situation where there's a struggle and Intel HAS to compete and AMD has to compete and the winners are consumers.

I also think what is probably happening inside intel is that the employees will benefit from the competition.

I think it might quell infighting and put good engineering decisions back above shitty dynasties.

How long's it gonna be until people are scraping those 'intel inside' badges off their computer to increase resale value.

Intel is getting desperate. Sounds like they want to go down in an ugly way.

https://www.extremetech.com/wp-content/uploads/2020/11/Perfo...

oh yes, I totally see this performance drop on battery, Intel simply switched laptops around. Its i7 that drops like a rock on battery.

The new Oracle. Now I am just awaiting the lawsuits

Very happy to see Apple roll out their own processors.

Intel is well know for these tricks.

Quotable from the article.

>is an overreach that recalls Intel’s behavior from the early 2000s in the most unflattering of ways. If I want to know whether the company building the fastest CPU core on a per-clock, per-watt basis thinks AMD’s product stack is valid on the basis of its on-battery performance, I’ll ask Apple.

Walking a fine line here:

I'm not saying it's okay for manufacturers to present misleading data like this—they really shouldn't—but, this has basically been standard practice since the beginning of time, hasn't it? Journalists shouldn't be taking a competitor's claims at face value, and I expect the vast majority are smart enough to know this.

I'd like to see Intel make a better product and sell it based on its own merits rather than attacking the accomplishments of their competitors. Smear pieces like this smack of desperation and the lack of a competitive product, or the strategy to create one.

A great product would sell itself.

>I'd like to see Intel make a better product and sell it based on its own merits rather than attacking the accomplishments of their competitors.

How do you market a "better" product without comparing it to the rest?

Accurately represent the product, including its strengths and weaknesses compared to the competition, which is also accurately represented. Alternatively, just compare against your own product portfolio, and leave the benchmarking against the competition to impartial third parties.

Certain companies do this way more than others, and we should absolutely be calling them out on it.

So what are you saying? Did you mean to say the tone of the article should reflect that it's totally ok that intel obscure the truth with dishonest representations? Or what that meant more as a kudos to extreme tech for doing what apparently is required, dissecting the lies of chip manufacturers? Or something else entirely?

AMD has consistently downplayed their performance gains in their press releases for the Zen architecture.

Stop playing devil's advocate, it doesn't add to the discussion.

This was a very smart cold blooded strategy. They didn't need to do anything at this point because within a few days after the launch there were plenty of reviews to confirm they had the best high-end CPUs. They knew they were winning.

Yes, it was a smart pr move. It wasn't FUD, however, which is the classic Intel/Microsoft strategy.

Reply on news.ycombinator.com