If they grow desperate I have GBs of DDR2 AND DDR3 in a drawer.
Meta reuses old RAM in new servers with custom bridge chip
31–40 of 240 posts
Re: Meta reuses old RAM in new servers with custom bridge chip
#32Earlier quoted context omitted.
Chip fabrication processes are not fungible: GPUs and CPUs might be made on roughly the same process, but DRAM is not (flash is a different process again, as is power electronics, analog electronics, MEMS, etc. And even within those broader categories there are different variations). While there are some overlaps in machines and techniques, a fab set up for one cannot generally switch to the other, and the economics…
You're not the first person to say so (and I don't mean to dispute it), but I have never been able to find a clear answer for /why/ those processes are incompatible. Is it built in different silicon, is it physical steps that's incompatible (ie its actually incompatible), is it different physical preparations that needs to be made (making it economically infeasible to combine) I cannot help but wonder, even if the an…
[flagged]
Re: Meta reuses old RAM in new servers with custom bridge chip
#33With regards to RAM price I never understood the following: A 16GB RAM stick has 16*8=128 billion bits, with 1 transistor per bit, thats still 128B, yet its supposed to cost like $60 before the price hikes? In contrast, a 5090 GPU was $2000 (true it has RAM, but you're paying for the GPU ASIC really, I guess the rest of the GPU was less than $500), it had 93B transistors. GPU transistors are smaller due to the more a…
Why would you expect smaller transistors to be cheaper?
That is what Moore's Law said.
Re: Meta reuses old RAM in new servers with custom bridge chip
#34From the paper:
"Our CXL solution achieves substantial gains for diverse workloads, including up to a 25% reduction in server count for disaggregated ML inference"
How does using worse RAM result in 25% reduction of server count for given workloads?
Re: Meta reuses old RAM in new servers with custom bridge chip
#35From the paper: "Our CXL solution achieves substantial gains for diverse workloads, including up to a 25% reduction in server count for disaggregated ML inference" How does using worse RAM result in 25% reduction of server count for given workloads?
Because it's used in addition to, not in place of, the better RAM.
Re: Meta reuses old RAM in new servers with custom bridge chip
#36Re: Meta reuses old RAM in new servers with custom bridge chip
#37Earlier quoted context omitted.
Source paper linked is https://aisystemcodesign.github.io/papers/isca26/vistara_cam... From a quick skim, you could think of this as roughly equivalent to shoving a large amount of DDR4 on a PCIe card and using it as a swap space. It's more sophisticated (see CXL protocol), but that gives you an idea of the tradeoffs. It seems there is some OS-level support for moving hot/cold pages between the main fast DRAM and the…
> and contains significant embedded carbon emissions Hi - thanks for the insightful comment - could you please expand on the above? Genuinely curious :)
From the paper:
"Second, memory dominates the carbon footprint of the fleet [8], accounting for 69% of CO2 emissions and posing a significant sustainability challenge [4]. DRAM dominates datacenter embodied CO2 largely because it is ubiquitous and deployed in large quantities across essentially all servers. Based on our internal fleet data, and aligned with studies from other hyperscalers such as Microsoft [33], memory is one of the largest single embodied-emissions contributors"
[8] U. Gupta, M. Elgamal, G. Hills, G.-Y. Wei, H.-H. S. Lee, D. Brooks, and C.-J. Wu, “ACT: Designing Sustainable Computer Systems with an Architectural Carbon Modeling Tool,” in Proceedings of the 49th Annual International Symposium on Computer Architecture (ISCA’22), 2022.
[4] D. Azevedo, M. Patterson, J. Pouchet, and R. Tipley, “Carbon usage effectiveness (cue): A green grid data center sustainability metric,” White paper, vol. 32, 2010.
[33] J. Wang, D. S. Berger, F. Kazhamiaka, C. Irvene, C. Zhang, E. Choukse, K. Frost, R. Fonseca, B. Warrier, C. Bansal, J. Stern, R. Bianchini, and A. Sriraman, “Designing Cloud Servers for Lower Carbon,” in Proceedings of the 51st Annual International Symposium on Computer Architecture, ser. ISCA ’24, 2025, p. 452–470.
Not a reference, but I found https://www.interface-eu.org/publications/semiconductor-emis... which goes into great detail on the subject. I hadn't realized there were significant emissions of fluorinated gases directly from the fabs, which is mildly alarming. Although it looks like there has been a crackdown on this either politically or through ESG policies.
Re: Meta reuses old RAM in new servers with custom bridge chip
#38ServeTheHome already reported on CLX memory expansion controllers back in December:
https://www.servethehome.com/hyper-scalers-are-using-cxl-to-...
Re: Meta reuses old RAM in new servers with custom bridge chip
#39if not the prices, no one would have implemented this in large scale solution..
Re: Meta reuses old RAM in new servers with custom bridge chip
#40The interesting part of this "RAM crisis" is similar to other fields where a problem results multiple parties looking for alternative solutions. This yields for exciting ideas or workarounds that might result a post-crisis memory boom (hopefully) also for local machines. 1. Lowest, Apple is evaluating new Chinese manufacturer which means change of supply demand if indeed it has reasonable QA. ( https://www.ft.com/con…
I would love it if we started designing software with hardware constraints in mind again.
We do already, if it ooms at 32g I have to prompt again /s