TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Did startup Flow Computing just make CPUs 100x faster?

28 点作者 cosmic_quanta11 个月前

13 条评论

wmf11 个月前
This is based on legitimate (although second-tier) academic research that appears to combine aspects of GPU-style SIMD&#x2F;SIMT with Tera-style massive multithreading. (main paper appears to be <a href="https:&#x2F;&#x2F;www.utupub.fi&#x2F;bitstream&#x2F;handle&#x2F;10024&#x2F;164790&#x2F;MPP-TPA-Article.Final.pdf?sequence=1&amp;isAllowed=y" rel="nofollow">https:&#x2F;&#x2F;www.utupub.fi&#x2F;bitstream&#x2F;handle&#x2F;10024&#x2F;164790&#x2F;MPP-TPA-...</a> )<p>Historically, the chance of such research turning into a chip you can buy is zero.
iamyemeth11 个月前
&quot;But the company can’t quite show any of that today&quot;, in case you want to save yourself a click
bell-cot11 个月前
&gt; A Finnish startup called Flow Computing is making one of the wildest claims ever heard in silicon engineering: by adding its proprietary companion chip, any CPU can instantly double its performance, increasing to as much as 100x with software tweaks.<p>Sounds like some of those 1970&#x27;s miracle filters, that would let you fuel your car with microenergized water and hand-waves, instead of gasoline.
评论 #40659463 未加载
评论 #40651319 未加载
评论 #40651543 未加载
bee_rider11 个月前
100X for all workloads is more or less impossible to believe, especially if it is a parallel coprocessor of some sort (I mean Ahmdal’s law would seem to indicate that this is impossible).<p>100X on selected workloads is easy to believe, I mean, it be a 100 wide vector unit and the workload they get 100X on could be adding up a big line of numbers, haha.
评论 #40653329 未加载
HumblyTossed11 个月前
Sigh. Are we doing this again?
fwlr11 个月前
From a brief skim of the whitepaper, it seems like the general idea is something like “GPUs are &lt;big-multiple&gt; faster than CPUs, because GPUs are built ground-up for parallel computation. The bottleneck now is parallel memory access, let’s build a chip from the ground up that is &lt;big-multiple&gt; better at parallel memory access”.<p>Not an entirely unconvincing idea, in fairness, but what strikes me as odd is that they’re trying to patent and sell the idea, rather than the technology.
mafuyu11 个月前
Eh... Not buying it. At least not the way they&#x27;re marketing it. This looks like a group that dreamt up another one of those massively SMP multicore designs and are shopping it around as a coprocessor IP to license. Maybe there&#x27;s interesting stuff in there, maybe not, but it&#x27;s not a new idea. Cooking up some massively parallel benchmark and telling journalists it&#x27;s going to make computers 100x faster is unethical and counterproductive.<p>The usual speed bumps with these SMP designs show up in the white paper. There&#x27;s a section on how they can use recompilation to automatically accelerate existing code. There&#x27;s a rich history of failed attempts at automatic parallelization at compile time. So this section is just an admission that you&#x27;re going to have to write code specifically for this thing to get anything out of it.<p>This coprocessor concept looks like it needs to be tightly coupled with the host CPU to work. The upside needs to be massive to justify a vendor integrating with this IP and also overcome the software adoption costs. Especially since they&#x27;re directly competing with GPGPU...<p>Also some strange red flags, like the comparison to quantum computing in the FAQ. (Did they feel the need to include this because an investor asked?)<p>My guess is they&#x27;ve done some interesting core research around memory bottlenecks&#x2F;latencies in a massively SMP architecture, and picked up investment to attempt to productize it. But the way they&#x27;re marketing themselves right now doesn&#x27;t inspire much confidence.
antisthenes11 个月前
The headline says 100x, but the article says the company only claimed double, 2x.<p>Is the journalism quality these days so low that people don&#x27;t understand multitudes vs percentages?
评论 #40651702 未加载
vsuperpower202011 个月前
To answer the question in the headline: No.
Sohcahtoa8211 个月前
Extraordinary claims require extraordinary evidence.<p>I&#x27;ll believe it when I see 3rd party tests.<p>I slightly skimmed the white paper and FAQ, and it really sounds like they&#x27;re being extremely hand-wavy about making a multi-core CPU.<p>&quot;The blocks inside the CPU die are optimized and meant for different purposes - vector units for vector calculation, matrix units for matrix calculation - Parallel Processing Unit is optimized for parallel processing.&quot;<p>Parallel processing of <i>what</i>? GPUs are essentially simple CPUs but massively parallel. If you&#x27;re claiming to be able to parallel processing of a general purpose CPU, then aren&#x27;t you essentially just making a multi-core CPU? How are you supposedly scaling that up to 100X?<p>This reeks of being a scam to get money from VCs.
g15jv2dp11 个月前
<a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Betteridge%27s_law_of_headlines" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Betteridge%27s_law_of_headline...</a><p>&gt; Any headline that ends in a question mark can be answered by the word no.<p>As confirmed by the article, of course.
评论 #40650918 未加载
评论 #40651129 未加载
评论 #40651139 未加载
sydbarrett7411 个月前
Is this another Mill CPU?
评论 #40656104 未加载
onion2k11 个月前
I think this is possible. In order to double the speed of computing on a CPU you install a special &quot;parallel processing unit&quot; that is ... a second CPU! Ta da! Doubled the speed!<p>Any investors please join the queue.