SK Hynix Begins HBM Production in Q1 2022 - Could Feature in AMD R9 390X/380X Fiji GPU

SK Hynix has officially listed HBM retentivity for Graphic Cards on its official itemize for Q1 2022. This is huge news, considering this means that HBM retentivity is at present readily available for GPU manufacturers such as Nvidia and AMD to utilise in their GPUs. Looks like the specs I speculated in my analysis a long time ago appear to accept been true later on all; regardless of the corporeality of flames I got for em. And then far the highest SK Hynix will go for GPU HBM is a 4-Hullo HBM Stack (four Retentivity stacks plus 1 logic stack).

A screenshot from the public catalog which can exist found here.

HBM retentivity for next gen GPUs and AMD Pirate Islands gets listing at SK Hynix

Basically this memory is available for use both for AMD and Nvidia, however I would be willing to wager that it will announced outset in Pirate Islands, specifically Fiji. On the off chance information technology doesn't appear in Pirates Islands, it will probably so debut in an Nvidia GPUs. In my stance, the chances of that happening (Dark-green Debut) are very slim considering if you recall the leak most AMD'due south Fiji GPU that occurred a while dorsum, the frequency listed was 1Ghz, which nets you lot 128 GB/s of bandwidth on an HBM (equally opposed to an insanely low corporeality on GDDR5 ) - ergo, everything fits snuggly. The chips are created on the JESD235 standard and were developed in collaboration with JEDEC and AMD.

GDDR5 Retentivity Versus HBM Comparison Table
WCCFTech.com	GDDR5	2-Hi HBM 'Stacked DRAM'*	4-Hi HBM 'Stacked DRAM'*
I/O	32	512	1024
Max Bandwidth Per Pin	seven Gbps	ane Gbps	1 Gbps
Max Bandwidth	28 GBps	64 GBps	128 GBps
Voltage	1.35 - 1.65	~1.2	~i.2
Control Input	Single	Dual	Dual
Layers	1	2 + 1	4 + i

In this type of HBM a unmarried logic circuit is planted directly over or nether a stack of wafers. Basically, the fries are stacked and Silicon punched through TSVs (Through Silicon Vias) till the Control Dice. In this instance, information technology means that the DRAMs stacked like are controlled past a unmarried logic excursion and henceforth referred to as a 'Chip-on-Wafer' design. Since nosotros are using multiple conductive pillars punching through the structure, very low voltage is required to send signals through the stacks. As we overcome the learning curve, we can upwards the frequency to 2Ghz to go an effective 256GB/s bandwidth on just one 4-Hi Stack!

The currently leaked specs of the AMD R9 390X/380X Fiji XT core stand up at 4096 stream processors and a memory clock of 1250 Ghz. This means an constructive bandwidth of 160 GB/due south per bit. Since the specs say that it has a full retentiveness bandwdith of 640 GB/south, nosotros can easily deduce that the card will make apply of four, four-Hi Stacks (for a listen blowing bus width of approximately 4096 $.25 ?!? that's most over 9000). However, that is a huge number and I experience very uneasy even admitting about the possibility of a 4096 scrap passenger vehicle. Anyways, hither you become folks, the showtime tangible proof that HBM is indeed on its way to GPUs. Now give me iv-Hullo HBM Stack on a high end GPU with a Price Signal of not more than $650 we'll talk.

Source: https://wccftech.com/sk-hynix-begins-hbm-production-q1-2015-feature-amd-r9-390x380x-fiji-gpu/

Posted by: dorseyfrialking.blogspot.com

Share this post