TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

PCIe 4.0 Card Hosts 21M.2 SSDs: Up To 168TB, 31 GB/s

100 pointsby ohmyblockabout 2 years ago

12 comments

eqvinoxabout 2 years ago
(Pulling up from child comment)<p>The chip this uses is likely a PM4x100 (x ∈ {0, 1, 2}) from Microchip (formerly Microsemi (formerly PMC-Sierra)):<p><a href="https:&#x2F;&#x2F;www.microchip.com&#x2F;en-us&#x2F;product&#x2F;PM40100" rel="nofollow">https:&#x2F;&#x2F;www.microchip.com&#x2F;en-us&#x2F;product&#x2F;PM40100</a><p>^ runs you $800 without bulk discounts [<a href="https:&#x2F;&#x2F;www2.mouser.com&#x2F;ProductDetail&#x2F;Microchip-Technology-Atmel&#x2F;PM40100B1-FEI?qs=TuK3vfAjtkVxKfKx2Q%2F8sw%3D%3D" rel="nofollow">https:&#x2F;&#x2F;www2.mouser.com&#x2F;ProductDetail&#x2F;Microchip-Technology-A...</a>] — if you can get them, that is.<p><a href="https:&#x2F;&#x2F;www.microchip.com&#x2F;en-us&#x2F;product&#x2F;PM41100" rel="nofollow">https:&#x2F;&#x2F;www.microchip.com&#x2F;en-us&#x2F;product&#x2F;PM41100</a><p><a href="https:&#x2F;&#x2F;www.microchip.com&#x2F;en-us&#x2F;product&#x2F;PM42100" rel="nofollow">https:&#x2F;&#x2F;www.microchip.com&#x2F;en-us&#x2F;product&#x2F;PM42100</a><p>^ these latter two I don&#x27;t see publicly listed prices for anywhere.<p>The PCIe 5.0 equivalent is in &quot;Samples available&quot;, i.e. not full production yet, which is likely why the card only does PCIe 4.0:<p><a href="https:&#x2F;&#x2F;www.microchip.com&#x2F;en-us&#x2F;product&#x2F;PM50100" rel="nofollow">https:&#x2F;&#x2F;www.microchip.com&#x2F;en-us&#x2F;product&#x2F;PM50100</a>
h2odragonabout 2 years ago
&gt; the manufacturer confirmed that the X21 offers 100 PCIe lanes, suggesting the presence of a PCIe switch.<p>Almost like it&#x27;s custom designed for a particular application where money&#x27;s no concern... Perhaps someone in Utah needs big rainbow tables?
评论 #35140022 未加载
评论 #35140802 未加载
评论 #35140162 未加载
评论 #35139834 未加载
评论 #35142786 未加载
评论 #35139963 未加载
throitallawayabout 2 years ago
Does anyone else marvel at data throughputs nowadays? People talk about 5GB&#x2F;s NVME cards as being &quot;slow.&quot; Same with Internet speeds. It&#x27;s unreal the progress that we&#x27;ve made (and continue to make.)
评论 #35140271 未加载
评论 #35140216 未加载
评论 #35140853 未加载
评论 #35140381 未加载
评论 #35141150 未加载
评论 #35141199 未加载
评论 #35142562 未加载
评论 #35140421 未加载
jjoonathanabout 2 years ago
What are the current strategies for leveraging NVMe speed &amp; volume in a NAS?<p>When I look at NAS offerings, I see lots of 2.5&quot; and 3.5&quot; bays and 1Gbe (maaaybe 10Gbe at the high end) which is a bit stifling.
评论 #35140366 未加载
评论 #35140341 未加载
评论 #35147747 未加载
Havocabout 2 years ago
Really hope they release a smaller version for home use.<p>Something with say 8 slots would turn all those gen 4 pcie gaming motherboards retiring soon into a great NAS.<p>Asus I think already makes a similar one but it isn&#x27;t fanless
评论 #35140553 未加载
lyindabout 2 years ago
This nice approach has at least these drawbacks:<p>1. Swapping drives is hard<p><pre><code> * may be overcome by declaring failure domain = node </code></pre> 2. No powerloss protection advertised to OS, ie. slow synchronous writes<p><pre><code> * may be overcome by software hacks and whole-system battery supply </code></pre> 3. Potential slowdown on continuous write load (weeks or months, depending on drive)<p><pre><code> * may be overcome by software in _some_ situations </code></pre> At least the last two points are a no-go for enterprise use-cases, if not addressed.
评论 #35140769 未加载
评论 #35142028 未加载
raszabout 2 years ago
&gt;Apex Storage doesn&#x27;t reveal the inner details of the X21<p>&gt;In a single-card configuration, the X21 delivers sequential read and write speeds up to 30.5 GBps and 28.5 GBps, respectively.<p>did they test it or reprint press release?<p>&gt;According to Apex Storage<p>ah<p>&gt;The AIC has an average read and write access latency of 79us and 52us<p>that doesnt make sense unless its additional latency of controller or they ship it populated with drieves.<p>&gt;However, Apex Storage didn&#x27;t expose the type of RAID arrays. The X21 also flaunts &quot;enterprise-grade reliability,&quot; NVMe 2.0 support, advanced EEC, data protection, and error recovery. Apex Storage didn&#x27;t reveal the pricing or availability for the X21.<p>so only revealed performance figures and pictures<p>Their previous product was a fancy looking bracket for holding 16 SATA M.2 drives <a href="https:&#x2F;&#x2F;www.kickstarter.com&#x2F;projects&#x2F;storage-scaler&#x2F;storage-scaler-16-x-m2-sata-iii-ssd-expansion-card" rel="nofollow">https:&#x2F;&#x2F;www.kickstarter.com&#x2F;projects&#x2F;storage-scaler&#x2F;storage-...</a>.<p>&gt;A cross platform drop in card that can massively increase the amount of storage for your computer using cost effective m.2 SSDs.<p>You had to read fine print to realized its just a m.2 stand requiring proper 16 port SATA controller to function. It still hasnt shipped to this day. Im mildly optimistic.
wdbabout 2 years ago
Nice, I am currently in the jungle of SSD enclosures for my Apple M2 Pro device and it’s pretty confusing. But 31 GB&#x2F;s seems wild.
评论 #35139824 未加载
评论 #35140152 未加载
qwertoxabout 2 years ago
I wonder how the SSDs are exposed to the OS.<p>While dealing with the Samsung Pro Firmware issue, I read that SSDs mounted on a hardware RAID controller need to be removed from the RAID in order to have their Firmware update applied, since Samsung&#x27;s tool won&#x27;t see the SSDs if they are placed on the controller.
评论 #35143421 未加载
rektideabout 2 years ago
CXL will also make attaching drives &amp; ram a much easier experience, much more regular.
rkagererabout 2 years ago
Finally! A card worthy of replacing my Areca controllers.<p>Now, is there an NVMe equivalent?
_joelabout 2 years ago
I&#x27;ll take two please