被英特尔糟糕的命名策略“咬”了一口
Getting bitten by Intel's poor naming schemes

原始链接: https://lorendb.dev/posts/getting-bitten-by-poor-naming-schemes/

用户尝试在其戴尔Precision T3610工作站(运行Proxmox)中将CPU从Intel Xeon E5-1650 v2升级到更强大的Xeon E7-8890 v4,两者均使用FCLGA2011接口。尽管Intel列出这两个CPU与同一接口兼容,但升级失败,原因是物理不兼容——E7-8890 v4有额外的触点和不同的键位。 问题源于FCLGA2011接口家族内的差异。T3610使用R接口(LGA2011-0),而E7-8890 v4需要R2接口(LGA2011-1)。Intel不一致的命名和规格页面导致了这种混淆,经常将所有LGA2011变体归类在单个“FCLGA2011”标签下。 用户现在有一个无法使用的CPU,因为退货运费超过了最初的购买价格。这次经历强调了在升级CPU时,验证*特定*接口修订版的重要性,而不仅仅是通用的接口类型。

Hacker News 新闻 | 过去 | 评论 | 提问 | 展示 | 招聘 | 提交 登录 被英特尔糟糕的命名搞砸 (lorendb.dev) 6 分,由 LorenDB 发表于 24 分钟前 | 隐藏 | 过去 | 收藏 | 2 条评论 XCabbage 发表于 7 分钟前 | 下一个 [–] 标题在 HN 上为什么会出错(schemes vs scenes),以及如何让版主修复它?回复 johng 发表于 18 分钟前 | 上一个 [–] 与 USB 命名的闹剧相比,这不算太糟糕……但确实是科技行业长期存在的问题。回复 指南 | 常见问题 | 列表 | API | 安全 | 法律 | 申请 YC | 联系 搜索:
相关文章

原文

I recently came into possession of an old Dell Precision T3610 workstation and promptly installed Proxmox to add it to my Proxmox cluster. After performing some ludicrously silly RAM and storage upgrades (how about 96 GB of DDR3, plus a 13-disk array of 500 GB SSDs?), I decided I wanted to max out the CPU as well.

The Precision T3610 shipped with an Intel Xeon E5-1650 v2. According to the linked Intel product page, this CPU uses the FCLGA2011 socket. Easy enough, I thought to myself. Just find the best CPU that supports FCLGA2011, make sure you have the latest BIOS installed, and everything should be all hunky dory. So I did some research and landed on the Xeon E7-8890 v4. It’s several years newer than the E5-1650 v2, has a whopping 24 cores (and hyperthreading bumps it to 48 logical cores!), and can support having not one, not two, but eight of itself installed in a single motherboard! Most crucially, the Intel product page says it uses the FCLGA2011 socket. When I stumbled across one of these monsters on eBay for just $15, I snapped it up.

Cue my massive shock and disappointment when, a few days later, I found myself unable to install the E7-8890 v4 in my T3610. The new CPU, despite being the same physical size as the old CPU, had extra contacts on the bottom and had a different physical keying. What? I thought Intel said this was the same socket!

Some amount of research later, I discovered that Intel’s LGA2011 socket has many variations. One of these variations is also called Socket R (or LGA2011-0). The T3610, and by extension the old E5-1650 v2 CPU, uses Socket R. The newer E7-8890 v4, meanwhile, uses a different variation called Socket R2 (or LGA2011-1). As if this wasn’t confusing enough, there’s even a third variation of the LGA2011 socket! I’ll refer you to the Wikipedia page for more info on that.

This is obviously not a great naming scheme. Why not use unique numbers for each version of the socket instead of tacking on a suffix? But the real kicker here is that Intel itself doesn’t seem to be able to keep up with its own naming scheme! It appears that its CPU specifications pages refer to all variants of the LGA2011 socket as FCLGA2011. This leaves folks like myself wondering what went wrong when their new-to-them CPUs don’t fit in their motherboards.

So where does that leave me? Well, I now have a fancy paperweight. I could have returned the CPU, but return shipping costs would have been half of what I paid for the CPU itself, so I’m hanging onto it for now in case I ever come into possession of a server with a Socket R2 motherboard that could use a nicer CPU. At least it wasn’t a super expensive CPU, so all in all, this isn’t the worst learning experience ever.

联系我们 contact @ memedata.com