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。

Why use ECC? (2015)

159 点作者 vsgherzi大约 1 年前

23 条评论

kimixa大约 1 年前
I wish there was a hard requirement for ECC, as a developer working on GPU drivers, there&#x27;s a <i>huge</i> amount of reported issues that just... don&#x27;t make sense? One offs with slightly different symptoms, memory dumps of nonsense, just nowhere to start rooting out the cause for an issue. Even on &quot;widely reported&quot; issues that make it to reddit and similar.<p>Probably not surprising, there&#x27;s a naturally antagonistic relationship between Performance and Reliability here, and it&#x27;s clear which way many of those &quot;enthusiast&quot; forums lean.<p>I haven&#x27;t got actual numbers, but I feel that <i>most</i> [0] of the issues I start looking at just can never be reproduced, or even make sense from the backtrace or similar. I can&#x27;t say it&#x27;s 100% hardware issues for this, as many games are a little... loose... with reliability if it works &quot;well enough&quot;, and is heavily interacting with code and data we work on so might also be a source for &quot;impossible&quot; issues. But even on straightforward code paths, no weird OS interaction, no allocation, nothing async etc. &quot;Impossible&quot; states happen pretty regularly.<p>I would love there to be enough ECC-using gamers out there to statistically see if it makes a difference.<p>[0] Most in terms of number of different issues, not total reports of the same issue. That&#x27;s dominated by one or two things, normally around the latest game or update doing something dumb :P
评论 #40294710 未加载
评论 #40295923 未加载
评论 #40294723 未加载
评论 #40297466 未加载
评论 #40295992 未加载
评论 #40297969 未加载
评论 #40294709 未加载
jrockway大约 1 年前
Not having ECC is the biggest scam in computing. Ever hear of &quot;bitrot&quot;? That&#x27;s memory errors that have been saved to disk. We have made millions of people lose their data so that servers can be artificially more expensive.<p>Intel was responsible for most of this. It is hard to be sad when seeing how they&#x27;ve lost the market lead.
评论 #40296228 未加载
评论 #40295959 未加载
评论 #40298383 未加载
评论 #40295903 未加载
transpute大约 1 年前
PC Engines $150 APU2 (RIP) shipped with 4GB ECC RAM and AMD Embedded CPU. Since it was a headless device used mostly for 1GbE networking, the RAM was throttled and relatively impervious to Rowhammer.<p>QNAP has a $600 1U short-depth (11&quot;) 4x3.5 2xM.2 2x10GbE 2x2.5GbE 4-32GB DDR4 SODIMM Arm NAS that would benefit from OSS community attention. Based on a Marvell&#x2F;Armada CN9130 SoC which supports ECC, has mainline Linux support, and public-but-non-upstream code for uboot [2]. With local serial console and a bit of effort, the QNAP OS can be replaced by Arm Debian&#x2F;Devuan with ZFS. Rare combo of low power, small size, fast network, ECC memory and upstream-friendly Linux. QNAP also sell a 10GbE router based on the same SoC.<p>Ryzen Pro (OEM) can support ECC [3].<p>[1] <a href="https:&#x2F;&#x2F;www.qnap.com&#x2F;en-us&#x2F;product&#x2F;ts-435xeu" rel="nofollow">https:&#x2F;&#x2F;www.qnap.com&#x2F;en-us&#x2F;product&#x2F;ts-435xeu</a><p>[2] <a href="https:&#x2F;&#x2F;solidrun.atlassian.net&#x2F;wiki&#x2F;spaces&#x2F;developer&#x2F;pages&#x2F;309919745&#x2F;CN913x+Build" rel="nofollow">https:&#x2F;&#x2F;solidrun.atlassian.net&#x2F;wiki&#x2F;spaces&#x2F;developer&#x2F;pages&#x2F;3...</a><p>[3] <a href="https:&#x2F;&#x2F;www.tomshardware.com&#x2F;pc-components&#x2F;cpus&#x2F;amd-confirms-ryzen-8000g-apus-dont-support-ecc-ram-despite-initial-claims" rel="nofollow">https:&#x2F;&#x2F;www.tomshardware.com&#x2F;pc-components&#x2F;cpus&#x2F;amd-confirms...</a>
评论 #40295891 未加载
评论 #40295895 未加载
TheAmazingRace大约 1 年前
So I have to say, ECC memory is definitely something we should not have gotten away from for consumer hardware. My current PC, which is rocking a Core i9 14900k (pray for me) and an ASUS W680M ACE SE motherboard, allowed me to install some 5600MHz speed DDR5 ECC memory, and it works flawlessly.<p>The only downside in my view is the cost. Unbuffered ECC and the cost of using a workstation class chipset really pushes this into luxury territory. Plus, I&#x27;m never too sure what Intel&#x27;s future plans are for successor processors and chipsets, which is why I settled on W680. I don&#x27;t really want to go full blown Xeon.
评论 #40295496 未加载
评论 #40294500 未加载
评论 #40301317 未加载
summerlight大约 1 年前
<a href="https:&#x2F;&#x2F;discourse.codinghorror.com&#x2F;t&#x2F;to-ecc-or-not-to-ecc&#x2F;3771&#x2F;62" rel="nofollow">https:&#x2F;&#x2F;discourse.codinghorror.com&#x2F;t&#x2F;to-ecc-or-not-to-ecc&#x2F;37...</a><p>Interestingly, Jeff Atwood has changed his mind on ECC memory.
评论 #40294477 未加载
评论 #40305193 未加载
magicalhippo大约 1 年前
Memory corruptions can impact very differently. A sample of decoded music getting corrupted leads to a small glitch, maybe even inaudible. An instruction in executable code getting corrupted can leads to all sorts of havoc.<p>Since ECC is seemingly not getting mandatory, I&#x27;ve been wishing CPUs would support &quot;soft-ECC&quot;. That is, the OS could mark certain pages as needing &quot;soft-ECC&quot;, and the CPU would then store (at least) three copies of that page in RAM. When reading such pages back from RAM the CPU would read all physical copies and compare. If the majority agrees it can use that, otherwise raise an error.<p>This could then be used for executable pages and important configuration data which occupies relatively few pages, and where integrity matters a lot more than speed.<p>There&#x27;s probably some good reasons why this is non-trivial to implement, I&#x27;ve forgotten most of what I learned about the virtual memory implementation in CPUs. But a man can dream...
评论 #40295380 未加载
评论 #40296502 未加载
Animats大约 1 年前
ECC memory should have a price premium of only 1 - 9&#x2F;8, or 12.5%. It costs more than that, because it&#x27;s &quot;enterprise&quot;.
评论 #40296936 未加载
评论 #40300613 未加载
评论 #40294494 未加载
snvzz大约 1 年前
ECC should be a requirement.<p>The FCC could just not allow computers to ship it without.<p>CPU makers like Intel and AMD could simply have their CPUs not work with non-ECC RAM.<p>Microsoft could e.g. require ECC RAM for Windows 12.<p>It is insanity that most computers shipping today do not use ECC and are thus unreliable.<p>With luck they&#x27;ll crash, but most likely they will fail silently, while corrupting data.
评论 #40296888 未加载
评论 #40296628 未加载
dang大约 1 年前
Related:<p><i>Why Use ECC? (2015)</i> - <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25167288">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25167288</a> - Nov 2020 (98 comments)<p><i>Why Use ECC Memory?</i> - <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=23361577">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=23361577</a> - May 2020 (2 comments)<p><i>Should I buy ECC memory? (2015)</i> - <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=14206635">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=14206635</a> - April 2017 (224 comments)<p><i>Why use ECC?</i> - <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=10638324">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=10638324</a> - Nov 2015 (95 comments)
hi-v-rocknroll大约 1 年前
Other references:<p><a href="https:&#x2F;&#x2F;cr.yp.to&#x2F;hardware&#x2F;ecc.html" rel="nofollow">https:&#x2F;&#x2F;cr.yp.to&#x2F;hardware&#x2F;ecc.html</a> (2001)<p><i>DEF CON 19 - Artem Dinaburg - Bit-squatting DNS Hijacking Without Exploitation (2011)</i><p><a href="https:&#x2F;&#x2F;youtu.be&#x2F;aT7mnSstKGs" rel="nofollow">https:&#x2F;&#x2F;youtu.be&#x2F;aT7mnSstKGs</a><p><a href="https:&#x2F;&#x2F;media.defcon.org&#x2F;DEF%20CON%2019&#x2F;DEF%20CON%2019%20video%20and%20slides&#x2F;DEF%20CON%2019%20-%20Artem%20Dinaburg%20-%20Bit-squatting%20DNS%20Hijacking%20Without%20Exploitation%20-%20Video%20and%20Slides.mp4" rel="nofollow">https:&#x2F;&#x2F;media.defcon.org&#x2F;DEF%20CON%2019&#x2F;DEF%20CON%2019%20vid...</a>
评论 #40299093 未加载
Nerada大约 1 年前
DDR5 comes with on-die ECC. My understanding is this only checks errors occuring within the RAM itself, not errors that occur during transmission to and from RAM.<p>My question is, how common are transmission errors over errors happening within RAM?
评论 #40294563 未加载
评论 #40294638 未加载
评论 #40296282 未加载
评论 #40294713 未加载
Sweepi大约 1 年前
I would love to put ECC in my Desktop computers, however its more expensive (ok), is not officially supported on most Desktop Motherboards (and in reality does not work in &quot;ECC-Mode&quot; on the majority of them) and finally: ECC Ram available to purchase is painfully slow, in both bandwidth (:&#x2F;) and latency (:&#x2F;&#x2F;)
sph大约 1 年前
Please, I&#x27;d love someone to tell me how to find and buy computers that support ECC. I am looking to buy a NUC&#x2F;mini-server, and they basically all sell with non-ECC RAM. Last time I asked on this forum, I was told that on Intel, only Xeon processors support ECC, while all modern (?) AMD CPU support them. Elsewhere I read that was matter is the mobo needs to support it. I have no idea how to go about it.<p>So, let me ask again. I was to buy a NUC new or off Ebay, how can I be 100% sure it works with ECC RAM without having to spend half a hour researching CPU, mobo and BIOS specs for each single product I come across?<p>If I had a budget in the thousands, I would go with a Xeon server that comes with ECC pre-installed. I don&#x27;t and have modest needs. I only want to splurge on ECC RAM to replace the original sticks.<p>(No <i>&quot;you don&#x27;t need ECC for a NUC&quot;</i> reply please. That is not the point of my question, yet it is a far too common response)
评论 #40297379 未加载
评论 #40300506 未加载
评论 #40297252 未加载
评论 #40307514 未加载
eadmund大约 1 年前
What’s the best price&#x2F;performance for a home lab server running Linux with ECC these days? Bonus points if it is rackable.<p>Sadly, my go-to Linux hardware manufacturers either don’t offer ECC RAM, or only offer it as an option on their absolute top-end machines. Yes, yes, the extra two thousand dollars for a machine with a six-year lifespan probably is worth it on a monthly basis, but man it still hurts.
评论 #40298270 未加载
评论 #40299497 未加载
评论 #40298235 未加载
BlueTemplar大约 1 年前
&gt; From talking to folks at a lot of large tech companies, it seems that most of them have had a climate control issue resulting in clouds or fog in their datacenters. You might call this a clever plan by Google to reproduce Seattle weather so they can poach MS employees. Alternately, it might be a plan to create literal cloud computing. Or maybe not.
oskarkk大约 1 年前
&gt; For example, at 20nm, a DRAM capacitor might hold something like 50 electrons, and that number will get smaller for next generation DRAM and things continue to shrink.<p>Nice. That got me curious, how many electrons are in today&#x27;s DRAM capacitor? I tried searching but haven&#x27;t found any recent info.
dvt大约 1 年前
I tried building an old rig (maybe ~7 years ago or so) using ECC RAM (since I was running two Xeons). It was <i>such</i> a pain in the butt to get it to boot and find sticks that were compatible with each other, don&#x27;t really want to go down that path again.
评论 #40295070 未加载
评论 #40298276 未加载
评论 #40296469 未加载
评论 #40295230 未加载
nextaccountic大约 1 年前
Why not ECC CPUs and GPUs? They can be hit by cosmic rays too.
评论 #40294764 未加载
评论 #40305205 未加载
nottorp大约 1 年前
I believe the first part could make for the start of a great &#x27;if Google does it it doesn&#x27;t mean it&#x27;s good for you&#x27; article...
forty大约 1 年前
I read somewhere that DDR5 has some kind of internal ECC mechanism even for non ECC stick, is that right? Does it make ECC less relevant?
luckystarr大约 1 年前
That&#x27;s the reason I chose AMD for my own laptop. AFAIR Intel doesn&#x27;t support ECC.
genpfault大约 1 年前
(2015)
评论 #40294384 未加载
ceving大约 1 年前
The only problem with ECC is that it is only an improvement, not a solution. Every error correction has a limit to the number of errors it can detect or correct. There is no such thing as absolute security.
评论 #40296346 未加载
评论 #40296957 未加载