6mo ago

Finnish startup says it can speed up any CPU by up to 100x using a tiny piece of hardware with no recoding

www.techradar.com 'One of the wildest claims ever heard': Finnish startup says it can speed up any CPU by 100x using a tiny piece of hardware with no recoding — and it might just be true

Flow Computing's PPU optimizes CPU tasks

46 comments

I highly doubt that unless they invented magic.
Edit: oh... They ommitted the "up to" in the headline.
- Added it
  
  I meant tech radar. Thanks

This change is likened to expanding a CPU from a one-lane road to a multi-lane highway
This analogy just pegged the bullshit meter so hard I almost died of eyeroll.
- You've got to be careful with rolling your eyes, because the parallelism of the two eyes means that the eye roll can be twice as powerful ^1
  (1) If measured against the silly baseline of a single eyeroll
- Apparently the percentage of people actually understanding what they are doing in the management part of the industry is now too low to filter out even such bullshit.

The TL;DR for the article is that the headline isn't exactly true. At this moment in time their PPU can potentially double a CPU's performance - the 100x claim comes with the caveat of "further software optimisation".
Tbh, I'm sceptical of the caveat. It feels like me telling someone I can only draw a stickman right now, but I could paint the Mona Lisa with some training.
Of course that could happen, but it's not very likely to - so I'll believe it when I see it.
Having said that they're not wrong about CPU bottlenecks and the slowed rate of CPU performance improvements - so a doubling of performance would be huge in this current market.
- Putting the claim instead of the reality in the headline is journalistic malpractice. 2x for free is still pretty great tho.
  
  Just finished the article, it's not for free at all. Chips need to be designed to use it. I'm skeptical again. There's no point IMO. Nobody wants to put the R&D into massively parallel CPUs when they can put that effort into GPUs.
- I get that we have to impress shareholders, but why can’t they just be honest and say it doubles CPU performance with the chance of even further improvement with software optimization. Doubling performance of the same hardware is still HUGE.
  
  They... they did?
- I'm just glad there are companies that are trying to optimize current tech rather than just piling over new hardware every damn year with forced planned obsolescence.
  Though the claim is absurd, I think double the performance is NEAT.
  
  This is new hardware piling. What they claim to do requires reworking manufacturing, is not retroactive with current designs, and demands more hardware components. It is basically a hardware thread scheduler. Cool idea, but it won't save us from planned obsolescence, if anything it is more incentive for more waste.
- Ah, good ol’ magic wishful thinking…

Why is this bullshit upvoted?
Already the first sentence, they change from the headline "without recoding" to "with further optimization".
Then the explanation "a companion chip that optimizes processing tasks in real-time"
This is already done at compiler level and internally in any modern CPU for more than a decade.
It might be possible to some degree for some specific forms of code, like maybe Java. But generally for the CPU this is bullshit, and the headline is decidedly dishonest.

Cybercriminals are creaming their jorts at the potential exploits this might open up.
- Please, hackers wear cargo shorts and toe shoes sir
  
  Oof. But yeah. Fair.
  I want to go on record that sometimes I just wear sandals with socks.

🚨 ⚠ 🚨 Hoax alert! 🚨 ⚠ 🚨

The techradar article is terrible, the techcrunch article is better, the Flow website has some detail.
But overall I have to say I don't believe them. You can't just make threads independent if they logically have dependencies. Or just remove cache coherency latency by removing caches.
- Can't have cache latency if there is no cache!
  
  So THIS is what the communists were talking about when they told me about the benefits of transitioning to a cacheless society!

You can download more ram too!

Overclockers:
"Give me some liquid nitrogen and I'll make that 102x."
- Meh, I just spit on it.

Hmm, so sounds like they're moving the kernel scheduler down to a hardware layer? Basically just better smp?
- Processors have an execution pipeline, so a single command like mov has some number of actions the CPU takes to execute it. CPU designers already have some magic that allows them to execute these out of order as well as other stuff like pre calculating what they think the next command will probably be.
  It's been a decade since my cpu class so I am butchering that explanation, but I think that is what they are proposing messing with
  
  That's accurate.
  Its done through multiple algorithms, but the general idea is to schedule calculations as soon as possible, accounting for data hazards to make sure everything is still equivalent to non out of order execution. Individual circuits can execute different things at the same time. Special hardware is needed to make the algorithms work.
  There's also branch prediction which is the same thing kind of except the CPU needs a way to ensure if the prediction was actually correct.

Has anyone been able to find an actual description of what this does? I clicked two layers deep and neither explains the details. It does sound like they’re doing CPU scheduling in the hardware, which is cool and makes some sense, but the descriptions are too vague to explain what the hell this is except “more parallelism goes brrrr” and it’s not clear to me why current GPUs aren’t already that.

Others have already laughed at this idea, but on a similar topic:
I know we’ve basically disabled a lot of features that sped up the CPU but introduced security flaws. Is there a way to turn those features back on for an airgapped computer intentionally?
- The kernel option is mitigations=off, if you want to try adding it to your Grub command line? From the testing I've done, provides no benefits whatsoever - no more frames in games, compilation runs no quicker, battery life on a laptop is no better.
  https://wiki.archlinux.org/title/Improving_performance#Turn_off_CPU_exploit_mitigations
- Haven't used it in years, but it might still work:
  https://www.grc.com/inspectre.htm

Haha okay
Edit: after a skim and a quick Google, this basically looks like a packaging up of existing modern processor features (sorta AVX/SVE with a load of speculative execution thrown on top)